- Home
- Golf
- Biomechanics
- Psychology of Sport and Exercise
- Physiology of Sport and Exercise
- Motor Behaviour
- Kinesiology/Exercise and Sport Science
- Motor Control in Everyday Actions
Motor Control in Everyday Actions
248 Pages
Motor Control in Everyday Actions presents 47 true stories that illustrate the phenomena of motor control, learning, perception, and attention in sport, physical activity, home, and work environments. At times humorous and sometimes sobering, this unique text provides an accessible application-to-research approach to spark critical thinking, class discussion, and new ideas for research. The stories in Motor Control in Everyday Actions illustrate the diversity and complexity of research in perception and action and motor skill acquisition. More than interesting anecdotes, these stories offer concrete examples of how motor behavior, motor control, and perception and action errors affect the lives of both well-known and ordinary individuals in various situations and environments.
Readers will be entertained with real-life stories that illustrate how research in motor control is applicable to real life:
•Choking Under Pressure examines information processing and how it changes under pressure.
•The Gimme Putt shows how Schmidt’s law can be used to predict the accuracy of golf putts.
•Turn Right at the Next Gorilla examines inattention blindness and its role in traffic accidents.
•The Farmers’ Market describes reasons why a man drives his car through a crowded open-air market, killing and injuring dozens of shoppers in the process.
•Craps and Weighted Bats describes the curious role of myths and superstition in how we play games.
•And 42 other examples of motor control in everyday actions will both entertain and inform.
Each story is followed by a set of self-directed activities that are progressively more complex. These activities, plus the additional notes and suggested readings and websites at the conclusion of each story, provide a starting point for critical thinking about the reasons why human actions sometimes go awry. A reader-friendly writing style and easy-to-follow analysis and conclusions assist students in gaining mastery of the issues presented, conceptualizing new research projects, and applying the content to current research.
The stories are grouped into three parts, beginning with situations involving errors and mistakes in perception, action, or decision making. Next, stories investigating varied techniques for studying perception and action are presented. The remaining scenarios provide readers with a look at research focusing on the motor learning process as well as some of the unexpected discoveries resulting from those investigations.
Motor Control in Everyday Actions will engage its readers—not only through the central topic of the story but also in the fundamental concepts involving perception, action, and learning. Used as a springboard for new research or as a catalyst for engaging discussion, Motor Control in Everyday Actions offers perspectives that will enhance understanding of how human beings interact with their world.
Part I. Stories About Perception and Action Gone Wrong
Chapter 1. Perceptual Errors
The Magnetic Hill
How do visual illusions distort perception and influence action?
The Farmers' Market
What are the roles of motor error and hypervigilance in unintended acceleration accidents?
The Grocery Store
How do population stereotypes shape our interactions with manufactured environments?
Push or Pull?
How do product designs influence people to perform specific actions?
Chapter 2. Decision Errors
Friendly Fire
What role did decision errors play in the death of Patrick Tillman?
Method to His Bratness
Did John McEnroe's verbal abuse of line judges influence their decisions?
Choking Under Pressure
What changes in information processing cause athletes such as Jean Van de Velde to fail under pressure?
Turn Right at the Next Gorilla
What is inattention blindness, and what role does it play in common traffic accidents?
Chapter 3. Action Errors
The Calculator
How can product designs accommodate Fitts' law?
The Gimme Putt
Can Schmidt's law be used to predict the accuracy of golf putts?
Pouring Coffee
How does the information-processing rate create a speed–accuracy trade-off?
Is the Bean Dizzy?
What do Spoonerisms reveal about motor control?
Part II. Adventures in Perception and Action
Chapter 4. Fun With Numbers
Public Opinion Polls
What do central tendency, variability, and statistical significance mean in the context of motor research?
Cutting Wood and Missing Putts
How are constant error, variable error, and absolute error useful for understanding motor control?
The Hot Hand
Do statistics support the existence of hot streaks in sports?
Chapter 5. Perception in Action
Red Light, Green Light
What factors influence reaction time and its measurement?
Jumping the Gun
How can a reaction be distinguished from an anticipation?
Antilock Brakes
How does the complexity of a motor program influence reaction time?
Preventing Penalties and Batting Baseballs
How do athletes use temporal and spatial anticipation?
Craps and Weighted Bats
What role do perceptual illusions play in sport performance?
Chapter 6. Attention
The Toad and the Centipede
Is an internal or external attentional focus better for improving performance?
The Preshot Routine
Why does a consistent mental preparation ritual benefit performance?
Gumbo
What are the limits of attentional capacity?
Fakes
What role does the psychological refractory period play in sport?
Chapter 7. Motor Control
Websites and Silly Walks
How do redundancies help us solve motor problems?
The Curling Draw
What sport skills use open- and closed-loop systems of motor control?
Cool Papa Bell
When vision is interrupted, how does iconic memory guide motor tasks?
Moving Sidewalks and Beer Glasses
How does the end-state comfort effect influence movement planning?
The Tickle
How do motor commands influence sensory feedback during motor control?
The Point of No Return
Is there a point in time after which an initiated motor program cannot be stopped?
Forensic Motor Control
What are generalized motor programs and what do keystroke dynamics reveal about them?
Party Tricks
How does our nervous system use functional linkages to coordinate movements?
Disappearing Act
What makes some coordination patterns more automatic than others?
Part III. Stories About Learning Motor Skills
Chapter 8. Measuring Motor Learning
How You Get to Carnegie Hall
What is the best way to measure progress in motor learning?
The Babe
Can a general motor ability be defined and measured?
Learning to Win From Losing
Why is the learning-performance distinction important?
Zero-Sum Training
What is the practical impact of ineffective training methods?
Chapter 9. Organizing Practice
But I Was Great on the Practice Range!
How does practice repetition influence performance and learning?
The Coach as a Dictionary
What roles does augmented feedback play in motor learning?
The Golfer’s Little Helper
What elements of motor learning are neglected when we use mechanical training aids?
Chapter 10. Skill Development
Bend It Like Becker
What types of models are best to observe when learning a skill?
Sport Snake Oils
Can visual training programs improve sport performance?
The Keypad
How do explicit and implicit memories influence skilled performance?
Wayne Gretzky
What role does skilled perception play in sport performance?
Chapter 11. Skill Retention
Shooting Two From the Line
How does the warm-up decrement affect repeated performances?
Like Riding a Bicycle
How are motor skills stored in memory?
H.M.
What does the amnesia suffered by Henry Gustav Molaison reveal about memory and motor skills?
Timothy D. Lee, PhD, is a professor in the department of kinesiology at McMaster University in Hamilton, Ontario. Lee, whose research on practice and motor learning has been frequently cited, is the author of more than 80 research papers in peer-reviewed publications in the area of motor control and learning. He is also the coauthor, along with Richard Schmidt, of the seminal text Motor Control and Learning: A Behavioral Emphasis, now in its fifth edition.
Lee is the former president of the Canadian Society for Psychomotor Learning and Sport Psychology and also a former editor of both Research Quarterly for Exercise and Sport and Journal of Motor Behavior. As an amateur golfer, Lee was ranked 22nd among senior golfers in Ontario in 2010. He also enjoys playing right wing for the Dundas Oldtimer ice hockey team and is a blues music enthusiast. Tim and his wife, Laurie Wishart, reside in Ancaster, Ontario.
The Farmers' Market
What are the roles of motor error and hypervigilance in unintended acceleration accidents?
The prosecution, at Weller's trial several years later, claimed it was no accident—that Weller deliberately drove his car through the crowded market. The reason, they said, was that Weller had been involved in a minor fender bender just moments before he entered the farmers' market. His response to the fender bender was to flee the scene of the accident. Witnesses for the prosecution painted Weller as a cold-blooded killer, commenting on the determined look on his face during the ordeal and his relatively calm demeanor afterward. It probably also didn't help his case when he emerged from the car immediately afterward and wondered aloud why the people he had hit had not jumped out of his way. Adding his age into the mix, Weller's actions were painted as rather pathetic.
Richard Schmidt, a renowned motor control scholar and human factors expert, testified on behalf of the defense team and argued that the facts of the case shared many similarities to accidents caused by errors of pedal misapplication, or unintended acceleration. Accidents of this type, which, thankfully, are quite rare, occur when the driver intends to apply pressure to the brake pedal, but misses the brake and pushes down on the accelerator instead.
Unintended acceleration accidents had been investigated for many years prior to the Weller case. These accidents were more common when the driver first got into the car, started the engine, and engaged the automatic transmission from the Park position into either Drive or Reverse. Indeed, such frequent episodes of “runaway cars” were the primary reason auto manufacturers added the brake-transmission shift interlock system in the 1980s so that an automatic transmission lever could not be moved from the Park position until the car sensed that a certain amount of pressure had been applied to the brake. However, the brake lock system only prevented this particular type of pedal misapplication; moving the foot to the accelerator instead of the brake would not be prevented once the transmission was successfully engaged out of Park and the car was in motion.
It is important to note that reaching for the brake requires that we steer the foot from a comfortable seated position to a target (the brake) in the absence of any visual guidance. We do this all the time without making any errors. We know where the brake is located from experience, and we also know that the brake feels different underfoot than does the accelerator. So then why would we suddenly miss the brake, push down on the accelerator instead, and then keep the foot there?
Schmidt (1989) presented evidence that unintended acceleration cases frequently involved accidents in which the driver had less experience than usual with that particular vehicle. Therefore, in some of these cases, the exact location and feel of the brake might not have been as familiar to the driver as normally could have been expected. These accidents also occurred more frequently on start-up, compared to later in the driving cycle, perhaps due to temporary factors associated with preparing an action (see “Shooting Two From the Line” in chapter 11). Driver inattention has also been linked with these cases, so it may not come as a surprise that drivers would not immediately notice the difference between the brake and gas pedal if engaged in a distracting activity at the same time (see “Gumbo” in chapter 6).
But for Weller, none of the common profiles for these accidents fit the case: he had already been driving before the supposed pedal misapplication error, so it was not a matter of missing the brake on initial start-up. Weller was quite familiar with his own vehicle, an 11-year-old Buick LeSabre. And he was not talking on a cell phone. Instead, Schmidt argued that Weller's pedal misapplication error was likely triggered by a catastrophic case of panic, termed hypervigilance, which could have been initiated when Weller had been involved in the fender bender just prior to the episode.
But one last issue seemed particularly problematic, according to the prosecution. Once the pedal misapplication error had occurred and the car started to accelerate wildly out of control, why didn't Weller simply remove his foot from the pedal or turn off the engine—actions that would have brought the car quickly to a stop? Again, failure to carry out corrective actions is typical of unintended acceleration cases, and some reasonable accounts have been offered to explain why drivers do not perform them. First, the driver probably does not realize that the foot is on the accelerator rather than the brake. The intention was to press the brake, and the fact that the pedal has gone all the way to the floor could reasonably be interpreted as brake failure rather than human error. And second, once the driver enters into this catastrophic state of panic, all normal modes of thinking cease. Reasoning and problem solving, the kinds of activities that are easy to do when unflustered, become unlikely, if not impossible, to carry out when in this state of hypervigilance.
Thomas Shelton, a member of the California Highway Patrol, testified at Weller's trial that he once investigated an unintended acceleration case in which an elderly woman ended up driving her car onto the top of another vehicle. The woman was in such a panicked state that when Shelton arrived at the accident scene, he had to climb up into the car to shut off the racing engine, at which time he noticed that the woman was still seated, very much alive, staring straight ahead with a death grip on the steering wheel, and with her foot still pushing the accelerator all the way to the floor.
Unfortunately, all of these arguments can only be used to speculate about what may have occurred in George Russell Weller's Buick LeSabre on that fateful day. On October 20, 2006, the jury convicted him of vehicular manslaughter in the 10 deaths resulting from the Santa Monica farmers' market crash. Nobody will ever know whether the verdict was the right one.
Self-Directed Learning Activities
1. In your own words describe the phenomenon known as unintended acceleration.
2. Describe a situation in which you made an action error that you were able to correct. How did you know that you had made the error, and what did you do to correct it?
3. Using our feet to manipulate car pedals involves aiming without visual feedback. What factors influence our ability to make these aiming movements accurately?
4. Propose a research methodology that examines one's ability to (a) move to a target without visual feedback and (b) estimate the accuracy of those aimed movements (again, without vision).
Notes
- Evidence from George Weller's trial during September and October of 2006 was summarized in the Santa Monica Daily Press, which can be accessed through its archives:
- Not all cases of unintended acceleration are generally agreed to be the result of a pedal misapplication. A segment of the television show 60 Minutes, hosted by Ed Bradley and which aired November 22, 1986, claimed that accidents of similar etiology involving the Audi 5000 were the result of a faulty idle stabilizer, which caused the car to accelerate wildly out of control when put into gear. An investigation by the U.S. NHTSA (National Highway Traffic Safety Administration) failed to support 60 Minutes' claim.
- More recently, runaway Toyotas have been a topic of concern. Once again, a government investigation failed to support a claim that these cases of unintended acceleration were due to an electronic fault in the engine. Again, this leaves open the very real possibility that driver error is to blame, as suggested by Richard Schmidt in the New York Times:
Suggested Readings
Castelli, J., Nash, C., Ditlow, C., & Pecht, M. (2003). Sudden acceleration: The myth of driver error. University of Maryland: CALCE EPSC Press.
Pollard, J., & Sussman, E.D. (1989). An examination of sudden acceleration. National Highway Traffic Safety Administration Final Report # DOT-TSC-NHTSA-89-1.
Schmidt, R.A. (1989). Unintended acceleration: A review of human factors contributions. Human Factors, 31, 345-364.
Schmidt, R.A. (1993). Unintended acceleration: Human performance considerations. In B. Peacock & W. Karwowski (Eds.), Automotive ergonomics (pp. 431-451). London: Taylor & Francis.
Schmidt, R.A., & Lee, T.D. (2011). Central contributions to motor control. In Motor control and learning: A behavioral emphasis (5th ed., pp. 177-222) Champaign, IL: Human Kinetics.
Push or Pull?
How do product designs influence people to perform specific actions?
It seems hard to believe, but brand-new buildings are still being constructed with very silly design features. For example, a new building on our campus opened last year and featured many modern technologies, including very nice-looking glass doors through which you enter and exit the building. The problem is that the doors have identical handles on either side. The handles themselves are very attractive, but they don't tell me what to do when I want to enter the building—should I push or pull the door? The handles on these particular doors suggest to me that they are made for grasping and pulling. However, the door goes in only one direction, and the very same handles appear on both sides of the door. Therefore, to enter the building, I have to use the handle to pull the door, and to exit, I have to use the identical type of handle to push the door. How would someone possibly know which way the door swings when approaching it? After a year of using these doors on a daily basis, I still just guess about whether to push or pull. Why would someone design a door like that?
Michael Darnell has created a wonderful website that illustrates many instances of products that have been designed without the user in mind. One of his web pages describes a door handle problem similar to the one I just described, in which a building has a short walkway with one set of doors on either end. Each door has identical handles on either side of the door (the same type of handle is used for pushing and pulling the door). However, according to Darnell, these particular doors were installed such that the problem was magnified even more—the doors at either end of the walkway both swung out from the walkway. He tells the story of a friend who entered the walkway by pulling on the handle of one door; then went to the second door and found that the door would not open when she pulled on the handle. So, concluding that the door was locked, she returned to the first door to exit and found herself trapped in the walkway when the door wouldn't move when she pulled on its handle. After trying to get the attention of others on the outside to tell them that she was trapped in the walkway, she finally realized, with much embarrassment, that the handles had to be pushed when inside the walkway. Darnell reminds us that this funny story could have had tragic consequences if, for example, a panicked person were trying to flee the building in case of a fire.
There are many ways to solve this problem of having identical handles used for opposite actions. One solution would be to put signs on the door that say “push” or “pull.” Although these signs would probably work, a well-designed door (i.e., designed for the user rather than for its looks) should not need a sign to tell people how to use it. Pushing and pulling a door are not rocket science, after all. A better solution would be to install a flat plate on one side of the door, which would make the correct choice obvious: Because there is no handle to grab to pull toward you, there is only one other option—push! A sign would be unnecessary in this case because there would be no question about which action to take.
Signs and symbols that tell us how to use things should be the last option in technology design. Signs are sometimes small and hard to read, they fall off, they wear out, other things cover them up, and they are often presented in only one language, making them less than desirable in a multicultural society. Symbols avoid being unilingual but introduce other problems, such as ambiguity. For example, how would one create simple, unambiguous, and instantly recognizable symbols for “push” and “pull”? As in the door handle example, a better design would be one that made the correct action as obvious as possible, one that would negate the need for a sign or symbol.
A classic example of the failure of technology to exploit this simple idea is a standard topic in ergonomics textbooks and on websites: the arrangement of burners on the stove top. A typical stove top layout is shown in figure 1.5.
The four knobs used to turn on the elements are located at the top of the figure. Quick, say out loud which one is used to turn on the right front burner. My guess is that it is one of the two rightmost knobs. But which one? It is impossible to know for certain without turning one of the knobs and waiting to see which element begins to heat up. So, designers add little labels to tell us which knobs are mapped to which burners. Do they work?
My personal experience is that these labels, if they have not fallen off, worn out, been covered by last night's pasta sauce, or been printed too small to read, are still hard to interpret. More times than I wish to remember, I have returned to a stove expecting to find a pot of boiling water for my pasta but instead found a glowing burner with nothing on it. The relative failure rate of these labels leads me to believe that they are only slightly more helpful than if I had just turned a knob at random.
Have a look at the two alternative stove top designs in figure 1.6. The layout of the knobs in figure 1.6a is the same as before; only the burner layout has been altered. In figure 1.6b, the burner layout is the same as it was before, but now the arrangement of the knobs has been altered.
In both cases, the mapping of each knob to its associated burner has been made obvious by a simple rearrangement of either the knobs or the burner layout. No symbols or labels are needed, and the chances of turning on the wrong burner are minimal. Moreover, the amount of space occupied by the burners and knobs in these more compatible layouts is identical to the space occupied in the incompatible layout. It is interesting to note that these alternative layouts have been available for many years. But have a look at any appliance store flyer or website, and you will see that the classic design still continues to dominate the market.
Sometimes it is simply not possible to avoid the use of labels or signs. In such cases the designer's goal should be to make them as informative as possible. The typical ceiling fan is another good example of a failure to consider the needs of the user when designing the product. The ceiling fan that we have in our home has a wall switch that turns the fan off and on. But the ceiling fan has three additional controls, located on the base of the fan itself (where the fan connects to the ceiling), that are operated by pulling on long strings that hang down from the base. These three strings can be pulled to adjust the speed of the fan, to reverse its direction, or to turn off or on a light located on the bottom of the fan. The problem is that the strings are identical in appearance; I have no way of knowing which control I am activating when I pull on any of the strings.
Actually, that is not completely true. If I stand on a stepladder and get up really close to the ceiling, I can see little labels beside the holes where the strings exit the electrical base of the fan that say “speed,” “light,” and “direction.” I could also stand on the floor and use binoculars to read these labels, I suppose. But, you get the point. Because these labels are next to useless to me, I really just have a one-in-three chance of getting it right each time I pull one of the strings.
How could each control be designed to communicate its function easily to the user? I have used two strategies to modify my own fan. Because reversing the fan's direction is the control that I use least often, I have shortened the string associated with this option to be the most difficult of the three to reach. However, I use the light and speed options about equally often. So, to make these as distinct as possible I borrowed two charms from an old charm bracelet and attached them to the ends of each string. A charm of a rabbit is attached to the string that controls the speed of the fan, and a charm of a book (for which I turn on the light to read) is attached to the string that controls the light. The fan may not look the way the manufacturer had intended it to look, but I have not made the mistake of pulling the wrong string since I attached these charms.
Motor skills such as pushing doors, turning knobs, and pulling strings are simple to learn, and we have all mastered these simple skills over our lifetimes. So why do we have such trouble using them? One reason is that the manufacturer is often paying more attention to aesthetics than to the needs of the user.
Self-Directed Learning Activities
1. Describe, in general, the principle of how product design influences human performance.
2. Search the literature for the term stimulus-response compatibility; then define it in your own words with specific reference to an example presented in this story.
3. Over the next 24 hours, keep a diary of all the objects or things you encounter that you think represent stimulus-response incompatibility. Propose ways each of these things could be made more compatible.
4. Find three research articles in which stimulus-response compatibility has been investigated. What are the similarities and differences in the stimuli and responses used in these studies?
Notes
- Michael Darnell has a wonderful website that features many examples of poorly designed products, with suggestions about simple ways to greatly improve usability: www.baddesigns.com
Suggested Readings
Proctor, R.W., & Van Zandt, T. (2008). Human factors in simple and complex systems (2nd ed.). Boca Raton, FL: CRC Press.
Proctor, R.W., & Vu, K.P.L. (2006). Stimulus-response compatibility principles: Data, theory, and application. Boca Raton, FL: CRC Press.
Schmidt, R.A., & Lee, T.D. (2011). Human information processing. In Motor control and learning: A behavioral emphasis (5th ed., pp. 57-96) Champaign, IL: Human Kinetics.
Learning to win from losing
Why is the learning–performance distinction important?
One day, Michelle Wie might be remembered as one of the greatest golfers of all time. At the age of 14 she was pounding out 300-yard (274 m) drives and not only competing in Ladies Professional (LPGA) Tour events, but also performing quite well in them. Her talent is undeniable. But controversy surrounded this exceptional athlete because of her desire to compete in PGA Tour events, which traditionally have included only male competitors. Wie has performed admirably in some of these events. For example, in the 2004 Sony Open, she shot rounds of 72 and 68 and missed the 36-hole cut by just a single stroke. Her two-round total of 140 placed her in a tie for 80th, which was better than 53 male professionals, including PGA stars Zach Johnson, Hunter Mahan, and Adam Scott. However, Wie has not performed well in most of the other PGA events she has entered, including a 139th-place finish (out of 141 players) at the 2007 Sony Open.
Many argued that Wie should stop competing in PGA Tour events and concentrate her efforts on the LPGA Tour. The typical argument was that her skill development as a golfer was being stalled by repeated failures in PGA Tour events, and that if she concentrated on LPGA events instead, the greater chance for success would escalate her skill development.
I argue that there is a fundamental flaw in this theory, because it falls into a trap known as the learning-performance distinction. The trap underlies one of the most misunderstood concepts in motor learning research, so let's start by defining the two terms. Motor skills researchers use the term performance to refer to a single observation. It could be a score or outcome that reflects the value of a single attempt at a motor skill, or perhaps an average score that statistically summarizes a number of attempts. An 18- or 36-hole score, or a final placement ranking in a tournament, might indicate a representative performance score in the case of a golfer. The term learning is used quite specifically to refer to a stable improvement in skill over time—an improvement that has specifically occurred as the result of practice. The problem illustrated by the criticisms leveled at Michelle Wie is that the critics fail to consider the difference between performance and learning; they mistakenly confuse her failures to improve her performance in men's PGA events as a failure to learn from these experiences. Let's use another example to illustrate this distinction between performance and learning.
Suppose you like to bowl, play regularly in a league, and also practice occasionally. Two years ago your average was 110, last year it was 120, and this year it is 130. Those averages probably indicate that you have become a better bowler as a result of learning: the scores indicate improvements that appear to be rather stable and that have resulted from practice.
So, does this mean that you will score a 130 the next time you bowl? Not necessarily, because there are many reasons for the fluctuations in scores that occur from game to game. Next time you may score well below your average because you are not feeling well, your shoes are too tight, or the crowd is very noisy that night. But this does not mean that you have suddenly lost some of your learned skill. You may score well above your average because you try extra hard to impress someone, or because everything just seems to be well focused (your mojo is working). As before, this does not mean that you have suddenly had a change in learned skill, because there is no indication that the sudden improvement in performance reflects a stable and permanent change in your capability to perform.
At any one time we have a theoretical capability for attaining a certain level of performance. When that theoretical capability changes to a higher level as a result of practice, we can say with confidence that we have learned. The confusion lies in the fact that individual performances may sometimes exceed or not live up to these theoretical capabilities. These fluctuations are expected and in no way diminish or detract from the performer's theoretical capability to perform at a certain skill level.
What I have described is the typical distinction between learning and performance. But, there is another side to this issue, the one that can be applied to Michelle Wie. The issue concerns the situation in which performances do not appear to change, or appear to be getting worse. Does that mean that learning is not occurring? Let's go back to the bowling example.
Suppose that this year your average was 120, which is the same as it was last year. You may take the absence of a stable improvement, despite all of your practice, as evidence that your learning has stalled. But, as it turns out, last year your league bowled Tuesday nights, which is your day off. This year you bowl on Wednesday nights, one hour after completing your 10-hour work shift. You are typically tired and hungry on league nights, and your bowling scores suffered this year as a result. What does all this suggest? What is your theoretical capability to bowl, and has this changed from last year, despite the change in your bowling night? Could it be that you actually have improved (i.e., learned), but the feeling of fatigue and other factors directly related to your work schedule prevented you from actually performing up to these expectations?
The point here is that making inferences about learning requires evidence that goes beyond a single performance or sometimes even a set of performances. Under what circumstances were these performances observed? Are there mitigating factors that might explain why the theoretical true score differs from the observed score? The absence of any observable change in performance does not mean that the unobservable (that theoretical capability to perform at a certain level of skill) has not improved.
Michelle Wie took a very different route in the development of her golfing skill. Her decision to compete in PGA events placed enormous pressures on her to perform, and only an exceptional result would have convinced the public that she was benefiting from these extreme challenges. All we could observe were her scores in these events. The unobservable, however, what she has learned by playing in these events, can never be truly understood. I suspect that if she achieves the rank of one of the greatest players of all time, it will be due in no small measure to these experiences competing against the very best golfers in the world.
Self-Directed Learning Activities
1. In your own words, explain the distinction between performance and learning.
2. What is the difference between a theoretical true score and an observed score?
3. Suggest a reason a person's observed score might exceed the true score, and a reason the observed score might fail to achieve the level of the true score.
4. Identify another sport, athlete, or situation in which the failure to account for the learning-performance distinction has resulted in an inappropriate conclusion.
Notes
- Babe Zaharias (see “The Babe”) was the first woman to compete in a men's PGA event.
- Here is a sampling of some of the criticism of Michele Wie's decision to compete in PGA Tour events, from two sports commentary sites:
- Michelle Wie won her second LPGA Tour title in Winnipeg, Manitoba, at the 2010 CN Canadian Women's Open.
Suggested Readings
Schmidt, R.A. (1972). The case against learning and forgetting scores. Journal of Motor Behavior, 4, 79-88.
Schmidt, R.A., & Lee, T.D. (2011). Motor learning concepts and research methods. In Motor control and learning: A behavioral emphasis (5th ed., pp. 327-346). Champaign, IL: Human Kinetics.
Cutting wood and missing putts
How are constant error, variable error, and absolute error useful for understanding motor control?
Two things I like to do—cutting wood and golfing—remind me often of the importance of measures of central tendency and variability in motor control performance. For example, I use my chainsaw to cut up fallen trees to burn in the woodstove. Cutting logs into burnable lengths is not an exact science. In the middle of a Canadian winter, when the temperature is cold and plenty of snow blankets the ground, it is generally considered a good idea to keep moving. I do not take time to measure the lengths of the logs I saw; rather, I make a quick visual estimate and then start to make my cut. But, in doing so, I must remember two important things related to the mean and the standard deviation of the logs I cut. First, because our woodstove is only large enough to hold pieces of wood that are 50 centimeters (20 in.) or shorter, any pieces that are longer than that are useless. And second, wood is easier to stack when the pieces are all roughly of the same length. Essentially, these features remind me that the mean and standard deviation are both important; I must have a mean that is less than 50 centimeters and a standard deviation that is as small as possible.
When considering data in terms of a specific goal or standard, it is often better to express the mean and standard deviation in terms of error measures—in my case, relative to a goal wood length of 50 centimeters. For example, wood lengths of 45, 42, 48, 47, and 52 centimeters would be expressed as lengths that are -5, -8, -2, -3, and +2 centimeters, respectively, relative to my goal wood length of 50 centimeters. If we calculated the mean of these lengths of wood, then we could express the stack of wood as having either an average length of 46.8 centimeters, or the error scores as having a mean constant error of -3.2 centimeters. Having a negative constant error is good in this case, because the average piece of wood that I have cut will easily fit in the woodstove (i.e., the mean is less than 50 centimeters).
The careful reader will have noted, however, that even though my mean constant error is negative, there is still one block of wood that will not fit (the 52 cm log). Thus, by itself, the mean constant error score (-3.2 cm) is misleading, because it does not provide any indication of how many logs might not actually fit in the woodstove. Our measure of variability, the standard deviation, gives us some indication of this information. The measure of variability of these error scores (called the variable error) represents the deviation of the error score for each log relative to the mean constant error and is calculated in the same way as is a standard deviation. In this case, the variable error is 3.3 centimeters. In general, when I am cutting wood, I want to achieve a negative mean constant error with a variable error that is as small as possible.
Although constant error provides a useful heuristic measure of average performance in cutting wood with a chainsaw, it is a quite misleading measure of central tendency in another activity I like to do. Let's say that I have struck 10 golf putts—5 of these putts go past the hole by 5, 10, 15, 20, and 25 centimeters, respectively, and the other five putts come up short of the hole by equal amounts (-5, -10, -15, -20 and -25 cm). On average, by how much have I missed the hole with these putts? If you were to calculate the average as in the previous wood-cutting example (the sum of the 10 individual error scores, divided by 10) the answer would be a mean constant error of 0. In other words, my “average” putt ended up in the hole. However, we know that this answer surely must be wrong because none of the individual putts actually went in the hole. So, we must use a different way to express these error scores to avoid an answer that makes no sense.
The problem that we sometimes run into with constant error is that it provides a measure of average bias, the tendency to err in a specific way (e.g., by too much or too little; too far left versus too far right). In some cases a specific bias is desirable, such as the tendency to undercut a wood length of 50 centimeters so that all of my wood pieces fit in the woodstove. In the case of my putts that have no consistent bias, we are much better off using a measure of central tendency that removes the bias from each score prior to calculating the mean. Such a procedure is called using the absolute (unsigned) scores. Hence, this measure of central tendency is called the mean absolute error. Here, our measure of central tendency is the mean of the unsigned scores (5, 5, 10, 10, 15, 15, 20, 20, 25, 25), which is 15 centimeters. That is, the golf putts tended to miss the hole by an average of 15 centimeters. In this particular case, the mean absolute error score (15 cm from the hole) represents a more accurate picture of the entire set of individual scores than what is represented in the mean constant error (in the hole).
Averages are convenient ways to express how we tend to perform. But, numbers that represent central tendencies can be misleading if the tendency itself is not a strong one. Measures of variability are one way to identify the strength of a tendency. Depending on the context, different methods of calculating central tendencies may characterize the data in more representative ways.
Self-Directed Learning Activities
1. Define constant error, absolute error, and variable error in your own words.
2. What do the terms algebraic error and total error (or Henry's E) refer to? How are they calculated, and how are they similar to the terms constant error and absolute error, respectively?
3. Look up and briefly describe a research investigation that uses at least two error measures. Why do you think the researchers used these measures in particular?
4. Conduct a brief study on yourself. Close your eyes and, using a pencil, draw five lines as close to 4 inches (10 cm) as possible, one below the previous one, without opening your eyes until you have drawn the fifth line. Measure and record the length of each line you drew; then calculate all of the error measures that have been discussed.
Notes
- Here is a web-based standard deviation calculator:
www.tinyurl.com/standarddeviationcalculator
Suggested Readings
Chapanis, A. (1951). Theory and methods for analyzing errors in man-machine systems. Annals of the New York Academy of Sciences, 51, 1179-1203.
Schmidt, R.A., & Lee, T.D. (2011). Methodology for studying motor performance. In Motor control and learning: A behavioral emphasis (5th ed., pp. 21-56). Champaign, IL: Human Kinetics.
Forensic motor control
What are generalized motor programs, and what do keystroke dynamics reveal about them?
Most of what I know about forensic identification comes from crime shows that I've seen on TV. The suspect is caught because a fingerprint was left on the murder weapon. The evidence is used to convict the suspect because no two fingerprints have ever been found to be identical. Crime shows tell us that other forensic methods can be used as well, such as DNA and eye retina data, but the fingerprint is the oldest biometric tool used for identification. However, another method that reveals a great deal about your identity is how you write or even type your name. For example, the password that you type to log on to your e-mail account may be just as identifiable as your fingerprint.
Try the following task as an example: go into any word processing program and type your name 10 times, once on each line, as I have done here:
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
I am not a skilled typist. I use my right middle finger to hold down the shift key and press the letter T with my left middle finger, then use my right middle finger to press i and my right index finger to press m, and then my right thumb to press the space bar. It is not very efficient, but I do it the same way every time I type my name. Regardless of how skilled a typist you are, you most likely repeat much the same process each of the 10 times you type your name.
Now, let's try to unravel the temporal “fingerprint” that you left behind when you typed your name. Suppose we conducted a simple analysis of the keystrokes and the time between each of the keystrokes that you made. The total time taken to type your name once is simply the time from the first keystroke to the last keystroke. At a more fine-grained level of analysis, the total time constitutes the time each key is held down (the dwell time) plus the time between the release of one key and the depression of the next (the transition time). All of the individual dwell times plus all of the transition times will sum to the total time.
In figure 7.5, I have plotted a hypothetical example of the 10 trials to type my name. The light gray bands denote the dwell times, and the darker gray bands indicate the transition times. I have ordered the lines of bands that represent the 10 trials from the fastest (least total time) at the top to the slowest trial at the bottom (most total time).
If you were to analyze each of the 10 repetitions of your name, you would probably find that all of the total times would be similar, but probably never exactly the same each time, just as in figure 7.5. The “noise” in our central nervous systems, plus other factors, causes the results to vary a little bit each time, resulting in some repetitions to be slower, and others faster, than the average total time.
But, take a closer look at each individual band in figure 7.5 as it changes over the 10 trials. What you will notice is that as the total time increases, each band gets proportionally longer too. We could express these numbers another way by dividing the time for each individual band by the total time for that trial to obtain a relative proportion of time represented by each band. Given the hypothetical evidence in figure 7.5, what we would discover is that the relative time for each band, expressed as a percentage, would stay roughly the same across all of the repetitions. Applying temporal and other methods to deconstruct how we type (e.g., key press forces) is called the study of keystroke dynamics.
According to generalized motor program theory, relative time is one of the key features of movement that is controlled by the central nervous system, especially for brief, rapid actions. Relative time is an expression of our motor control fingerprint. For well-learned tasks such as typing and handwriting, our central nervous system regulates the relative timing of impulses that are sent to the muscles that carry out these tasks. Various factors influence the real time on any particular instance. For example, using a keyboard that requires more force to depress the keys might result in overall slower times than would result using a keyboard that has a light touch, but the relative timing would likely remain the same.
Research in keystroke dynamics may result in the ability to identify people who carry out Internet fraud. In many ways it is similar to the use of handwriting dynamics to analyze the timing of the cursive expressions of a signature. It is fairly easy to forge the spatial representation of someone's signature, but very difficult to forge the temporal dynamics that result in that signature. These expressions of timing behavior represent our motor control fingerprints.
Self-Directed Learning Activities
1. Define the term generalized motor program in your own words.
2. How does the concept of a generalized motor program differ from the concept of a motor program as it was used in stories such as “Antilock Brakes” (in chapter 5) and “Point of No Return” (earlier in this
chapter)?
3. Suggest a method by which the handwriting dynamics of signatures could be used to detect fraud.
4. Some people contend that there is one generalized motor program for the full swing in golf, regardless of which club is used. How could a temporal dynamics analysis be used to assess this contention?
Notes
- An excellent review of fingerprint analysis by David Ashbaugh of the Royal Canadian Mounted Police is available here:
www.onin.com/fp/ridgeology.pdf
- A lot of controversy remains about invariances in motor performance; how they are measured; and what invariance, or lack of invariance, means in terms of motor programs. The following articles provide good arguments for the debate:
Gentner, D.R. (1987). Timing of skilled motor performance: Tests of the proportional duration model. Psychological Review, 94, 255-276.
Heuer, H. (1988). Testing the invariance of relative timing: Comment on Gentner (1987). Psychological Review, 95, 552-557.
Suggested Readings
Schmidt, R.A. (1975). A schema theory of discrete motor skill learning. Psychological Review, 82, 225-260.
Schmidt, R.A. (1985). The search for invariance in skilled movement behavior. Research Quarterly for Exercise and Sport, 56, 188-200.
Schmidt, R.A., & Lee, T.D. (2011). Central contributions to motor control. In Motor control and learning: A behavioral emphasis (5th ed., pp. 177-222) Champaign, IL: Human Kinetics.
The Farmers' Market
What are the roles of motor error and hypervigilance in unintended acceleration accidents?
The prosecution, at Weller's trial several years later, claimed it was no accident—that Weller deliberately drove his car through the crowded market. The reason, they said, was that Weller had been involved in a minor fender bender just moments before he entered the farmers' market. His response to the fender bender was to flee the scene of the accident. Witnesses for the prosecution painted Weller as a cold-blooded killer, commenting on the determined look on his face during the ordeal and his relatively calm demeanor afterward. It probably also didn't help his case when he emerged from the car immediately afterward and wondered aloud why the people he had hit had not jumped out of his way. Adding his age into the mix, Weller's actions were painted as rather pathetic.
Richard Schmidt, a renowned motor control scholar and human factors expert, testified on behalf of the defense team and argued that the facts of the case shared many similarities to accidents caused by errors of pedal misapplication, or unintended acceleration. Accidents of this type, which, thankfully, are quite rare, occur when the driver intends to apply pressure to the brake pedal, but misses the brake and pushes down on the accelerator instead.
Unintended acceleration accidents had been investigated for many years prior to the Weller case. These accidents were more common when the driver first got into the car, started the engine, and engaged the automatic transmission from the Park position into either Drive or Reverse. Indeed, such frequent episodes of “runaway cars” were the primary reason auto manufacturers added the brake-transmission shift interlock system in the 1980s so that an automatic transmission lever could not be moved from the Park position until the car sensed that a certain amount of pressure had been applied to the brake. However, the brake lock system only prevented this particular type of pedal misapplication; moving the foot to the accelerator instead of the brake would not be prevented once the transmission was successfully engaged out of Park and the car was in motion.
It is important to note that reaching for the brake requires that we steer the foot from a comfortable seated position to a target (the brake) in the absence of any visual guidance. We do this all the time without making any errors. We know where the brake is located from experience, and we also know that the brake feels different underfoot than does the accelerator. So then why would we suddenly miss the brake, push down on the accelerator instead, and then keep the foot there?
Schmidt (1989) presented evidence that unintended acceleration cases frequently involved accidents in which the driver had less experience than usual with that particular vehicle. Therefore, in some of these cases, the exact location and feel of the brake might not have been as familiar to the driver as normally could have been expected. These accidents also occurred more frequently on start-up, compared to later in the driving cycle, perhaps due to temporary factors associated with preparing an action (see “Shooting Two From the Line” in chapter 11). Driver inattention has also been linked with these cases, so it may not come as a surprise that drivers would not immediately notice the difference between the brake and gas pedal if engaged in a distracting activity at the same time (see “Gumbo” in chapter 6).
But for Weller, none of the common profiles for these accidents fit the case: he had already been driving before the supposed pedal misapplication error, so it was not a matter of missing the brake on initial start-up. Weller was quite familiar with his own vehicle, an 11-year-old Buick LeSabre. And he was not talking on a cell phone. Instead, Schmidt argued that Weller's pedal misapplication error was likely triggered by a catastrophic case of panic, termed hypervigilance, which could have been initiated when Weller had been involved in the fender bender just prior to the episode.
But one last issue seemed particularly problematic, according to the prosecution. Once the pedal misapplication error had occurred and the car started to accelerate wildly out of control, why didn't Weller simply remove his foot from the pedal or turn off the engine—actions that would have brought the car quickly to a stop? Again, failure to carry out corrective actions is typical of unintended acceleration cases, and some reasonable accounts have been offered to explain why drivers do not perform them. First, the driver probably does not realize that the foot is on the accelerator rather than the brake. The intention was to press the brake, and the fact that the pedal has gone all the way to the floor could reasonably be interpreted as brake failure rather than human error. And second, once the driver enters into this catastrophic state of panic, all normal modes of thinking cease. Reasoning and problem solving, the kinds of activities that are easy to do when unflustered, become unlikely, if not impossible, to carry out when in this state of hypervigilance.
Thomas Shelton, a member of the California Highway Patrol, testified at Weller's trial that he once investigated an unintended acceleration case in which an elderly woman ended up driving her car onto the top of another vehicle. The woman was in such a panicked state that when Shelton arrived at the accident scene, he had to climb up into the car to shut off the racing engine, at which time he noticed that the woman was still seated, very much alive, staring straight ahead with a death grip on the steering wheel, and with her foot still pushing the accelerator all the way to the floor.
Unfortunately, all of these arguments can only be used to speculate about what may have occurred in George Russell Weller's Buick LeSabre on that fateful day. On October 20, 2006, the jury convicted him of vehicular manslaughter in the 10 deaths resulting from the Santa Monica farmers' market crash. Nobody will ever know whether the verdict was the right one.
Self-Directed Learning Activities
1. In your own words describe the phenomenon known as unintended acceleration.
2. Describe a situation in which you made an action error that you were able to correct. How did you know that you had made the error, and what did you do to correct it?
3. Using our feet to manipulate car pedals involves aiming without visual feedback. What factors influence our ability to make these aiming movements accurately?
4. Propose a research methodology that examines one's ability to (a) move to a target without visual feedback and (b) estimate the accuracy of those aimed movements (again, without vision).
Notes
- Evidence from George Weller's trial during September and October of 2006 was summarized in the Santa Monica Daily Press, which can be accessed through its archives:
- Not all cases of unintended acceleration are generally agreed to be the result of a pedal misapplication. A segment of the television show 60 Minutes, hosted by Ed Bradley and which aired November 22, 1986, claimed that accidents of similar etiology involving the Audi 5000 were the result of a faulty idle stabilizer, which caused the car to accelerate wildly out of control when put into gear. An investigation by the U.S. NHTSA (National Highway Traffic Safety Administration) failed to support 60 Minutes' claim.
- More recently, runaway Toyotas have been a topic of concern. Once again, a government investigation failed to support a claim that these cases of unintended acceleration were due to an electronic fault in the engine. Again, this leaves open the very real possibility that driver error is to blame, as suggested by Richard Schmidt in the New York Times:
Suggested Readings
Castelli, J., Nash, C., Ditlow, C., & Pecht, M. (2003). Sudden acceleration: The myth of driver error. University of Maryland: CALCE EPSC Press.
Pollard, J., & Sussman, E.D. (1989). An examination of sudden acceleration. National Highway Traffic Safety Administration Final Report # DOT-TSC-NHTSA-89-1.
Schmidt, R.A. (1989). Unintended acceleration: A review of human factors contributions. Human Factors, 31, 345-364.
Schmidt, R.A. (1993). Unintended acceleration: Human performance considerations. In B. Peacock & W. Karwowski (Eds.), Automotive ergonomics (pp. 431-451). London: Taylor & Francis.
Schmidt, R.A., & Lee, T.D. (2011). Central contributions to motor control. In Motor control and learning: A behavioral emphasis (5th ed., pp. 177-222) Champaign, IL: Human Kinetics.
Push or Pull?
How do product designs influence people to perform specific actions?
It seems hard to believe, but brand-new buildings are still being constructed with very silly design features. For example, a new building on our campus opened last year and featured many modern technologies, including very nice-looking glass doors through which you enter and exit the building. The problem is that the doors have identical handles on either side. The handles themselves are very attractive, but they don't tell me what to do when I want to enter the building—should I push or pull the door? The handles on these particular doors suggest to me that they are made for grasping and pulling. However, the door goes in only one direction, and the very same handles appear on both sides of the door. Therefore, to enter the building, I have to use the handle to pull the door, and to exit, I have to use the identical type of handle to push the door. How would someone possibly know which way the door swings when approaching it? After a year of using these doors on a daily basis, I still just guess about whether to push or pull. Why would someone design a door like that?
Michael Darnell has created a wonderful website that illustrates many instances of products that have been designed without the user in mind. One of his web pages describes a door handle problem similar to the one I just described, in which a building has a short walkway with one set of doors on either end. Each door has identical handles on either side of the door (the same type of handle is used for pushing and pulling the door). However, according to Darnell, these particular doors were installed such that the problem was magnified even more—the doors at either end of the walkway both swung out from the walkway. He tells the story of a friend who entered the walkway by pulling on the handle of one door; then went to the second door and found that the door would not open when she pulled on the handle. So, concluding that the door was locked, she returned to the first door to exit and found herself trapped in the walkway when the door wouldn't move when she pulled on its handle. After trying to get the attention of others on the outside to tell them that she was trapped in the walkway, she finally realized, with much embarrassment, that the handles had to be pushed when inside the walkway. Darnell reminds us that this funny story could have had tragic consequences if, for example, a panicked person were trying to flee the building in case of a fire.
There are many ways to solve this problem of having identical handles used for opposite actions. One solution would be to put signs on the door that say “push” or “pull.” Although these signs would probably work, a well-designed door (i.e., designed for the user rather than for its looks) should not need a sign to tell people how to use it. Pushing and pulling a door are not rocket science, after all. A better solution would be to install a flat plate on one side of the door, which would make the correct choice obvious: Because there is no handle to grab to pull toward you, there is only one other option—push! A sign would be unnecessary in this case because there would be no question about which action to take.
Signs and symbols that tell us how to use things should be the last option in technology design. Signs are sometimes small and hard to read, they fall off, they wear out, other things cover them up, and they are often presented in only one language, making them less than desirable in a multicultural society. Symbols avoid being unilingual but introduce other problems, such as ambiguity. For example, how would one create simple, unambiguous, and instantly recognizable symbols for “push” and “pull”? As in the door handle example, a better design would be one that made the correct action as obvious as possible, one that would negate the need for a sign or symbol.
A classic example of the failure of technology to exploit this simple idea is a standard topic in ergonomics textbooks and on websites: the arrangement of burners on the stove top. A typical stove top layout is shown in figure 1.5.
The four knobs used to turn on the elements are located at the top of the figure. Quick, say out loud which one is used to turn on the right front burner. My guess is that it is one of the two rightmost knobs. But which one? It is impossible to know for certain without turning one of the knobs and waiting to see which element begins to heat up. So, designers add little labels to tell us which knobs are mapped to which burners. Do they work?
My personal experience is that these labels, if they have not fallen off, worn out, been covered by last night's pasta sauce, or been printed too small to read, are still hard to interpret. More times than I wish to remember, I have returned to a stove expecting to find a pot of boiling water for my pasta but instead found a glowing burner with nothing on it. The relative failure rate of these labels leads me to believe that they are only slightly more helpful than if I had just turned a knob at random.
Have a look at the two alternative stove top designs in figure 1.6. The layout of the knobs in figure 1.6a is the same as before; only the burner layout has been altered. In figure 1.6b, the burner layout is the same as it was before, but now the arrangement of the knobs has been altered.
In both cases, the mapping of each knob to its associated burner has been made obvious by a simple rearrangement of either the knobs or the burner layout. No symbols or labels are needed, and the chances of turning on the wrong burner are minimal. Moreover, the amount of space occupied by the burners and knobs in these more compatible layouts is identical to the space occupied in the incompatible layout. It is interesting to note that these alternative layouts have been available for many years. But have a look at any appliance store flyer or website, and you will see that the classic design still continues to dominate the market.
Sometimes it is simply not possible to avoid the use of labels or signs. In such cases the designer's goal should be to make them as informative as possible. The typical ceiling fan is another good example of a failure to consider the needs of the user when designing the product. The ceiling fan that we have in our home has a wall switch that turns the fan off and on. But the ceiling fan has three additional controls, located on the base of the fan itself (where the fan connects to the ceiling), that are operated by pulling on long strings that hang down from the base. These three strings can be pulled to adjust the speed of the fan, to reverse its direction, or to turn off or on a light located on the bottom of the fan. The problem is that the strings are identical in appearance; I have no way of knowing which control I am activating when I pull on any of the strings.
Actually, that is not completely true. If I stand on a stepladder and get up really close to the ceiling, I can see little labels beside the holes where the strings exit the electrical base of the fan that say “speed,” “light,” and “direction.” I could also stand on the floor and use binoculars to read these labels, I suppose. But, you get the point. Because these labels are next to useless to me, I really just have a one-in-three chance of getting it right each time I pull one of the strings.
How could each control be designed to communicate its function easily to the user? I have used two strategies to modify my own fan. Because reversing the fan's direction is the control that I use least often, I have shortened the string associated with this option to be the most difficult of the three to reach. However, I use the light and speed options about equally often. So, to make these as distinct as possible I borrowed two charms from an old charm bracelet and attached them to the ends of each string. A charm of a rabbit is attached to the string that controls the speed of the fan, and a charm of a book (for which I turn on the light to read) is attached to the string that controls the light. The fan may not look the way the manufacturer had intended it to look, but I have not made the mistake of pulling the wrong string since I attached these charms.
Motor skills such as pushing doors, turning knobs, and pulling strings are simple to learn, and we have all mastered these simple skills over our lifetimes. So why do we have such trouble using them? One reason is that the manufacturer is often paying more attention to aesthetics than to the needs of the user.
Self-Directed Learning Activities
1. Describe, in general, the principle of how product design influences human performance.
2. Search the literature for the term stimulus-response compatibility; then define it in your own words with specific reference to an example presented in this story.
3. Over the next 24 hours, keep a diary of all the objects or things you encounter that you think represent stimulus-response incompatibility. Propose ways each of these things could be made more compatible.
4. Find three research articles in which stimulus-response compatibility has been investigated. What are the similarities and differences in the stimuli and responses used in these studies?
Notes
- Michael Darnell has a wonderful website that features many examples of poorly designed products, with suggestions about simple ways to greatly improve usability: www.baddesigns.com
Suggested Readings
Proctor, R.W., & Van Zandt, T. (2008). Human factors in simple and complex systems (2nd ed.). Boca Raton, FL: CRC Press.
Proctor, R.W., & Vu, K.P.L. (2006). Stimulus-response compatibility principles: Data, theory, and application. Boca Raton, FL: CRC Press.
Schmidt, R.A., & Lee, T.D. (2011). Human information processing. In Motor control and learning: A behavioral emphasis (5th ed., pp. 57-96) Champaign, IL: Human Kinetics.
Learning to win from losing
Why is the learning–performance distinction important?
One day, Michelle Wie might be remembered as one of the greatest golfers of all time. At the age of 14 she was pounding out 300-yard (274 m) drives and not only competing in Ladies Professional (LPGA) Tour events, but also performing quite well in them. Her talent is undeniable. But controversy surrounded this exceptional athlete because of her desire to compete in PGA Tour events, which traditionally have included only male competitors. Wie has performed admirably in some of these events. For example, in the 2004 Sony Open, she shot rounds of 72 and 68 and missed the 36-hole cut by just a single stroke. Her two-round total of 140 placed her in a tie for 80th, which was better than 53 male professionals, including PGA stars Zach Johnson, Hunter Mahan, and Adam Scott. However, Wie has not performed well in most of the other PGA events she has entered, including a 139th-place finish (out of 141 players) at the 2007 Sony Open.
Many argued that Wie should stop competing in PGA Tour events and concentrate her efforts on the LPGA Tour. The typical argument was that her skill development as a golfer was being stalled by repeated failures in PGA Tour events, and that if she concentrated on LPGA events instead, the greater chance for success would escalate her skill development.
I argue that there is a fundamental flaw in this theory, because it falls into a trap known as the learning-performance distinction. The trap underlies one of the most misunderstood concepts in motor learning research, so let's start by defining the two terms. Motor skills researchers use the term performance to refer to a single observation. It could be a score or outcome that reflects the value of a single attempt at a motor skill, or perhaps an average score that statistically summarizes a number of attempts. An 18- or 36-hole score, or a final placement ranking in a tournament, might indicate a representative performance score in the case of a golfer. The term learning is used quite specifically to refer to a stable improvement in skill over time—an improvement that has specifically occurred as the result of practice. The problem illustrated by the criticisms leveled at Michelle Wie is that the critics fail to consider the difference between performance and learning; they mistakenly confuse her failures to improve her performance in men's PGA events as a failure to learn from these experiences. Let's use another example to illustrate this distinction between performance and learning.
Suppose you like to bowl, play regularly in a league, and also practice occasionally. Two years ago your average was 110, last year it was 120, and this year it is 130. Those averages probably indicate that you have become a better bowler as a result of learning: the scores indicate improvements that appear to be rather stable and that have resulted from practice.
So, does this mean that you will score a 130 the next time you bowl? Not necessarily, because there are many reasons for the fluctuations in scores that occur from game to game. Next time you may score well below your average because you are not feeling well, your shoes are too tight, or the crowd is very noisy that night. But this does not mean that you have suddenly lost some of your learned skill. You may score well above your average because you try extra hard to impress someone, or because everything just seems to be well focused (your mojo is working). As before, this does not mean that you have suddenly had a change in learned skill, because there is no indication that the sudden improvement in performance reflects a stable and permanent change in your capability to perform.
At any one time we have a theoretical capability for attaining a certain level of performance. When that theoretical capability changes to a higher level as a result of practice, we can say with confidence that we have learned. The confusion lies in the fact that individual performances may sometimes exceed or not live up to these theoretical capabilities. These fluctuations are expected and in no way diminish or detract from the performer's theoretical capability to perform at a certain skill level.
What I have described is the typical distinction between learning and performance. But, there is another side to this issue, the one that can be applied to Michelle Wie. The issue concerns the situation in which performances do not appear to change, or appear to be getting worse. Does that mean that learning is not occurring? Let's go back to the bowling example.
Suppose that this year your average was 120, which is the same as it was last year. You may take the absence of a stable improvement, despite all of your practice, as evidence that your learning has stalled. But, as it turns out, last year your league bowled Tuesday nights, which is your day off. This year you bowl on Wednesday nights, one hour after completing your 10-hour work shift. You are typically tired and hungry on league nights, and your bowling scores suffered this year as a result. What does all this suggest? What is your theoretical capability to bowl, and has this changed from last year, despite the change in your bowling night? Could it be that you actually have improved (i.e., learned), but the feeling of fatigue and other factors directly related to your work schedule prevented you from actually performing up to these expectations?
The point here is that making inferences about learning requires evidence that goes beyond a single performance or sometimes even a set of performances. Under what circumstances were these performances observed? Are there mitigating factors that might explain why the theoretical true score differs from the observed score? The absence of any observable change in performance does not mean that the unobservable (that theoretical capability to perform at a certain level of skill) has not improved.
Michelle Wie took a very different route in the development of her golfing skill. Her decision to compete in PGA events placed enormous pressures on her to perform, and only an exceptional result would have convinced the public that she was benefiting from these extreme challenges. All we could observe were her scores in these events. The unobservable, however, what she has learned by playing in these events, can never be truly understood. I suspect that if she achieves the rank of one of the greatest players of all time, it will be due in no small measure to these experiences competing against the very best golfers in the world.
Self-Directed Learning Activities
1. In your own words, explain the distinction between performance and learning.
2. What is the difference between a theoretical true score and an observed score?
3. Suggest a reason a person's observed score might exceed the true score, and a reason the observed score might fail to achieve the level of the true score.
4. Identify another sport, athlete, or situation in which the failure to account for the learning-performance distinction has resulted in an inappropriate conclusion.
Notes
- Babe Zaharias (see “The Babe”) was the first woman to compete in a men's PGA event.
- Here is a sampling of some of the criticism of Michele Wie's decision to compete in PGA Tour events, from two sports commentary sites:
- Michelle Wie won her second LPGA Tour title in Winnipeg, Manitoba, at the 2010 CN Canadian Women's Open.
Suggested Readings
Schmidt, R.A. (1972). The case against learning and forgetting scores. Journal of Motor Behavior, 4, 79-88.
Schmidt, R.A., & Lee, T.D. (2011). Motor learning concepts and research methods. In Motor control and learning: A behavioral emphasis (5th ed., pp. 327-346). Champaign, IL: Human Kinetics.
Cutting wood and missing putts
How are constant error, variable error, and absolute error useful for understanding motor control?
Two things I like to do—cutting wood and golfing—remind me often of the importance of measures of central tendency and variability in motor control performance. For example, I use my chainsaw to cut up fallen trees to burn in the woodstove. Cutting logs into burnable lengths is not an exact science. In the middle of a Canadian winter, when the temperature is cold and plenty of snow blankets the ground, it is generally considered a good idea to keep moving. I do not take time to measure the lengths of the logs I saw; rather, I make a quick visual estimate and then start to make my cut. But, in doing so, I must remember two important things related to the mean and the standard deviation of the logs I cut. First, because our woodstove is only large enough to hold pieces of wood that are 50 centimeters (20 in.) or shorter, any pieces that are longer than that are useless. And second, wood is easier to stack when the pieces are all roughly of the same length. Essentially, these features remind me that the mean and standard deviation are both important; I must have a mean that is less than 50 centimeters and a standard deviation that is as small as possible.
When considering data in terms of a specific goal or standard, it is often better to express the mean and standard deviation in terms of error measures—in my case, relative to a goal wood length of 50 centimeters. For example, wood lengths of 45, 42, 48, 47, and 52 centimeters would be expressed as lengths that are -5, -8, -2, -3, and +2 centimeters, respectively, relative to my goal wood length of 50 centimeters. If we calculated the mean of these lengths of wood, then we could express the stack of wood as having either an average length of 46.8 centimeters, or the error scores as having a mean constant error of -3.2 centimeters. Having a negative constant error is good in this case, because the average piece of wood that I have cut will easily fit in the woodstove (i.e., the mean is less than 50 centimeters).
The careful reader will have noted, however, that even though my mean constant error is negative, there is still one block of wood that will not fit (the 52 cm log). Thus, by itself, the mean constant error score (-3.2 cm) is misleading, because it does not provide any indication of how many logs might not actually fit in the woodstove. Our measure of variability, the standard deviation, gives us some indication of this information. The measure of variability of these error scores (called the variable error) represents the deviation of the error score for each log relative to the mean constant error and is calculated in the same way as is a standard deviation. In this case, the variable error is 3.3 centimeters. In general, when I am cutting wood, I want to achieve a negative mean constant error with a variable error that is as small as possible.
Although constant error provides a useful heuristic measure of average performance in cutting wood with a chainsaw, it is a quite misleading measure of central tendency in another activity I like to do. Let's say that I have struck 10 golf putts—5 of these putts go past the hole by 5, 10, 15, 20, and 25 centimeters, respectively, and the other five putts come up short of the hole by equal amounts (-5, -10, -15, -20 and -25 cm). On average, by how much have I missed the hole with these putts? If you were to calculate the average as in the previous wood-cutting example (the sum of the 10 individual error scores, divided by 10) the answer would be a mean constant error of 0. In other words, my “average” putt ended up in the hole. However, we know that this answer surely must be wrong because none of the individual putts actually went in the hole. So, we must use a different way to express these error scores to avoid an answer that makes no sense.
The problem that we sometimes run into with constant error is that it provides a measure of average bias, the tendency to err in a specific way (e.g., by too much or too little; too far left versus too far right). In some cases a specific bias is desirable, such as the tendency to undercut a wood length of 50 centimeters so that all of my wood pieces fit in the woodstove. In the case of my putts that have no consistent bias, we are much better off using a measure of central tendency that removes the bias from each score prior to calculating the mean. Such a procedure is called using the absolute (unsigned) scores. Hence, this measure of central tendency is called the mean absolute error. Here, our measure of central tendency is the mean of the unsigned scores (5, 5, 10, 10, 15, 15, 20, 20, 25, 25), which is 15 centimeters. That is, the golf putts tended to miss the hole by an average of 15 centimeters. In this particular case, the mean absolute error score (15 cm from the hole) represents a more accurate picture of the entire set of individual scores than what is represented in the mean constant error (in the hole).
Averages are convenient ways to express how we tend to perform. But, numbers that represent central tendencies can be misleading if the tendency itself is not a strong one. Measures of variability are one way to identify the strength of a tendency. Depending on the context, different methods of calculating central tendencies may characterize the data in more representative ways.
Self-Directed Learning Activities
1. Define constant error, absolute error, and variable error in your own words.
2. What do the terms algebraic error and total error (or Henry's E) refer to? How are they calculated, and how are they similar to the terms constant error and absolute error, respectively?
3. Look up and briefly describe a research investigation that uses at least two error measures. Why do you think the researchers used these measures in particular?
4. Conduct a brief study on yourself. Close your eyes and, using a pencil, draw five lines as close to 4 inches (10 cm) as possible, one below the previous one, without opening your eyes until you have drawn the fifth line. Measure and record the length of each line you drew; then calculate all of the error measures that have been discussed.
Notes
- Here is a web-based standard deviation calculator:
www.tinyurl.com/standarddeviationcalculator
Suggested Readings
Chapanis, A. (1951). Theory and methods for analyzing errors in man-machine systems. Annals of the New York Academy of Sciences, 51, 1179-1203.
Schmidt, R.A., & Lee, T.D. (2011). Methodology for studying motor performance. In Motor control and learning: A behavioral emphasis (5th ed., pp. 21-56). Champaign, IL: Human Kinetics.
Forensic motor control
What are generalized motor programs, and what do keystroke dynamics reveal about them?
Most of what I know about forensic identification comes from crime shows that I've seen on TV. The suspect is caught because a fingerprint was left on the murder weapon. The evidence is used to convict the suspect because no two fingerprints have ever been found to be identical. Crime shows tell us that other forensic methods can be used as well, such as DNA and eye retina data, but the fingerprint is the oldest biometric tool used for identification. However, another method that reveals a great deal about your identity is how you write or even type your name. For example, the password that you type to log on to your e-mail account may be just as identifiable as your fingerprint.
Try the following task as an example: go into any word processing program and type your name 10 times, once on each line, as I have done here:
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
I am not a skilled typist. I use my right middle finger to hold down the shift key and press the letter T with my left middle finger, then use my right middle finger to press i and my right index finger to press m, and then my right thumb to press the space bar. It is not very efficient, but I do it the same way every time I type my name. Regardless of how skilled a typist you are, you most likely repeat much the same process each of the 10 times you type your name.
Now, let's try to unravel the temporal “fingerprint” that you left behind when you typed your name. Suppose we conducted a simple analysis of the keystrokes and the time between each of the keystrokes that you made. The total time taken to type your name once is simply the time from the first keystroke to the last keystroke. At a more fine-grained level of analysis, the total time constitutes the time each key is held down (the dwell time) plus the time between the release of one key and the depression of the next (the transition time). All of the individual dwell times plus all of the transition times will sum to the total time.
In figure 7.5, I have plotted a hypothetical example of the 10 trials to type my name. The light gray bands denote the dwell times, and the darker gray bands indicate the transition times. I have ordered the lines of bands that represent the 10 trials from the fastest (least total time) at the top to the slowest trial at the bottom (most total time).
If you were to analyze each of the 10 repetitions of your name, you would probably find that all of the total times would be similar, but probably never exactly the same each time, just as in figure 7.5. The “noise” in our central nervous systems, plus other factors, causes the results to vary a little bit each time, resulting in some repetitions to be slower, and others faster, than the average total time.
But, take a closer look at each individual band in figure 7.5 as it changes over the 10 trials. What you will notice is that as the total time increases, each band gets proportionally longer too. We could express these numbers another way by dividing the time for each individual band by the total time for that trial to obtain a relative proportion of time represented by each band. Given the hypothetical evidence in figure 7.5, what we would discover is that the relative time for each band, expressed as a percentage, would stay roughly the same across all of the repetitions. Applying temporal and other methods to deconstruct how we type (e.g., key press forces) is called the study of keystroke dynamics.
According to generalized motor program theory, relative time is one of the key features of movement that is controlled by the central nervous system, especially for brief, rapid actions. Relative time is an expression of our motor control fingerprint. For well-learned tasks such as typing and handwriting, our central nervous system regulates the relative timing of impulses that are sent to the muscles that carry out these tasks. Various factors influence the real time on any particular instance. For example, using a keyboard that requires more force to depress the keys might result in overall slower times than would result using a keyboard that has a light touch, but the relative timing would likely remain the same.
Research in keystroke dynamics may result in the ability to identify people who carry out Internet fraud. In many ways it is similar to the use of handwriting dynamics to analyze the timing of the cursive expressions of a signature. It is fairly easy to forge the spatial representation of someone's signature, but very difficult to forge the temporal dynamics that result in that signature. These expressions of timing behavior represent our motor control fingerprints.
Self-Directed Learning Activities
1. Define the term generalized motor program in your own words.
2. How does the concept of a generalized motor program differ from the concept of a motor program as it was used in stories such as “Antilock Brakes” (in chapter 5) and “Point of No Return” (earlier in this
chapter)?
3. Suggest a method by which the handwriting dynamics of signatures could be used to detect fraud.
4. Some people contend that there is one generalized motor program for the full swing in golf, regardless of which club is used. How could a temporal dynamics analysis be used to assess this contention?
Notes
- An excellent review of fingerprint analysis by David Ashbaugh of the Royal Canadian Mounted Police is available here:
www.onin.com/fp/ridgeology.pdf
- A lot of controversy remains about invariances in motor performance; how they are measured; and what invariance, or lack of invariance, means in terms of motor programs. The following articles provide good arguments for the debate:
Gentner, D.R. (1987). Timing of skilled motor performance: Tests of the proportional duration model. Psychological Review, 94, 255-276.
Heuer, H. (1988). Testing the invariance of relative timing: Comment on Gentner (1987). Psychological Review, 95, 552-557.
Suggested Readings
Schmidt, R.A. (1975). A schema theory of discrete motor skill learning. Psychological Review, 82, 225-260.
Schmidt, R.A. (1985). The search for invariance in skilled movement behavior. Research Quarterly for Exercise and Sport, 56, 188-200.
Schmidt, R.A., & Lee, T.D. (2011). Central contributions to motor control. In Motor control and learning: A behavioral emphasis (5th ed., pp. 177-222) Champaign, IL: Human Kinetics.
The Farmers' Market
What are the roles of motor error and hypervigilance in unintended acceleration accidents?
The prosecution, at Weller's trial several years later, claimed it was no accident—that Weller deliberately drove his car through the crowded market. The reason, they said, was that Weller had been involved in a minor fender bender just moments before he entered the farmers' market. His response to the fender bender was to flee the scene of the accident. Witnesses for the prosecution painted Weller as a cold-blooded killer, commenting on the determined look on his face during the ordeal and his relatively calm demeanor afterward. It probably also didn't help his case when he emerged from the car immediately afterward and wondered aloud why the people he had hit had not jumped out of his way. Adding his age into the mix, Weller's actions were painted as rather pathetic.
Richard Schmidt, a renowned motor control scholar and human factors expert, testified on behalf of the defense team and argued that the facts of the case shared many similarities to accidents caused by errors of pedal misapplication, or unintended acceleration. Accidents of this type, which, thankfully, are quite rare, occur when the driver intends to apply pressure to the brake pedal, but misses the brake and pushes down on the accelerator instead.
Unintended acceleration accidents had been investigated for many years prior to the Weller case. These accidents were more common when the driver first got into the car, started the engine, and engaged the automatic transmission from the Park position into either Drive or Reverse. Indeed, such frequent episodes of “runaway cars” were the primary reason auto manufacturers added the brake-transmission shift interlock system in the 1980s so that an automatic transmission lever could not be moved from the Park position until the car sensed that a certain amount of pressure had been applied to the brake. However, the brake lock system only prevented this particular type of pedal misapplication; moving the foot to the accelerator instead of the brake would not be prevented once the transmission was successfully engaged out of Park and the car was in motion.
It is important to note that reaching for the brake requires that we steer the foot from a comfortable seated position to a target (the brake) in the absence of any visual guidance. We do this all the time without making any errors. We know where the brake is located from experience, and we also know that the brake feels different underfoot than does the accelerator. So then why would we suddenly miss the brake, push down on the accelerator instead, and then keep the foot there?
Schmidt (1989) presented evidence that unintended acceleration cases frequently involved accidents in which the driver had less experience than usual with that particular vehicle. Therefore, in some of these cases, the exact location and feel of the brake might not have been as familiar to the driver as normally could have been expected. These accidents also occurred more frequently on start-up, compared to later in the driving cycle, perhaps due to temporary factors associated with preparing an action (see “Shooting Two From the Line” in chapter 11). Driver inattention has also been linked with these cases, so it may not come as a surprise that drivers would not immediately notice the difference between the brake and gas pedal if engaged in a distracting activity at the same time (see “Gumbo” in chapter 6).
But for Weller, none of the common profiles for these accidents fit the case: he had already been driving before the supposed pedal misapplication error, so it was not a matter of missing the brake on initial start-up. Weller was quite familiar with his own vehicle, an 11-year-old Buick LeSabre. And he was not talking on a cell phone. Instead, Schmidt argued that Weller's pedal misapplication error was likely triggered by a catastrophic case of panic, termed hypervigilance, which could have been initiated when Weller had been involved in the fender bender just prior to the episode.
But one last issue seemed particularly problematic, according to the prosecution. Once the pedal misapplication error had occurred and the car started to accelerate wildly out of control, why didn't Weller simply remove his foot from the pedal or turn off the engine—actions that would have brought the car quickly to a stop? Again, failure to carry out corrective actions is typical of unintended acceleration cases, and some reasonable accounts have been offered to explain why drivers do not perform them. First, the driver probably does not realize that the foot is on the accelerator rather than the brake. The intention was to press the brake, and the fact that the pedal has gone all the way to the floor could reasonably be interpreted as brake failure rather than human error. And second, once the driver enters into this catastrophic state of panic, all normal modes of thinking cease. Reasoning and problem solving, the kinds of activities that are easy to do when unflustered, become unlikely, if not impossible, to carry out when in this state of hypervigilance.
Thomas Shelton, a member of the California Highway Patrol, testified at Weller's trial that he once investigated an unintended acceleration case in which an elderly woman ended up driving her car onto the top of another vehicle. The woman was in such a panicked state that when Shelton arrived at the accident scene, he had to climb up into the car to shut off the racing engine, at which time he noticed that the woman was still seated, very much alive, staring straight ahead with a death grip on the steering wheel, and with her foot still pushing the accelerator all the way to the floor.
Unfortunately, all of these arguments can only be used to speculate about what may have occurred in George Russell Weller's Buick LeSabre on that fateful day. On October 20, 2006, the jury convicted him of vehicular manslaughter in the 10 deaths resulting from the Santa Monica farmers' market crash. Nobody will ever know whether the verdict was the right one.
Self-Directed Learning Activities
1. In your own words describe the phenomenon known as unintended acceleration.
2. Describe a situation in which you made an action error that you were able to correct. How did you know that you had made the error, and what did you do to correct it?
3. Using our feet to manipulate car pedals involves aiming without visual feedback. What factors influence our ability to make these aiming movements accurately?
4. Propose a research methodology that examines one's ability to (a) move to a target without visual feedback and (b) estimate the accuracy of those aimed movements (again, without vision).
Notes
- Evidence from George Weller's trial during September and October of 2006 was summarized in the Santa Monica Daily Press, which can be accessed through its archives:
- Not all cases of unintended acceleration are generally agreed to be the result of a pedal misapplication. A segment of the television show 60 Minutes, hosted by Ed Bradley and which aired November 22, 1986, claimed that accidents of similar etiology involving the Audi 5000 were the result of a faulty idle stabilizer, which caused the car to accelerate wildly out of control when put into gear. An investigation by the U.S. NHTSA (National Highway Traffic Safety Administration) failed to support 60 Minutes' claim.
- More recently, runaway Toyotas have been a topic of concern. Once again, a government investigation failed to support a claim that these cases of unintended acceleration were due to an electronic fault in the engine. Again, this leaves open the very real possibility that driver error is to blame, as suggested by Richard Schmidt in the New York Times:
Suggested Readings
Castelli, J., Nash, C., Ditlow, C., & Pecht, M. (2003). Sudden acceleration: The myth of driver error. University of Maryland: CALCE EPSC Press.
Pollard, J., & Sussman, E.D. (1989). An examination of sudden acceleration. National Highway Traffic Safety Administration Final Report # DOT-TSC-NHTSA-89-1.
Schmidt, R.A. (1989). Unintended acceleration: A review of human factors contributions. Human Factors, 31, 345-364.
Schmidt, R.A. (1993). Unintended acceleration: Human performance considerations. In B. Peacock & W. Karwowski (Eds.), Automotive ergonomics (pp. 431-451). London: Taylor & Francis.
Schmidt, R.A., & Lee, T.D. (2011). Central contributions to motor control. In Motor control and learning: A behavioral emphasis (5th ed., pp. 177-222) Champaign, IL: Human Kinetics.
Push or Pull?
How do product designs influence people to perform specific actions?
It seems hard to believe, but brand-new buildings are still being constructed with very silly design features. For example, a new building on our campus opened last year and featured many modern technologies, including very nice-looking glass doors through which you enter and exit the building. The problem is that the doors have identical handles on either side. The handles themselves are very attractive, but they don't tell me what to do when I want to enter the building—should I push or pull the door? The handles on these particular doors suggest to me that they are made for grasping and pulling. However, the door goes in only one direction, and the very same handles appear on both sides of the door. Therefore, to enter the building, I have to use the handle to pull the door, and to exit, I have to use the identical type of handle to push the door. How would someone possibly know which way the door swings when approaching it? After a year of using these doors on a daily basis, I still just guess about whether to push or pull. Why would someone design a door like that?
Michael Darnell has created a wonderful website that illustrates many instances of products that have been designed without the user in mind. One of his web pages describes a door handle problem similar to the one I just described, in which a building has a short walkway with one set of doors on either end. Each door has identical handles on either side of the door (the same type of handle is used for pushing and pulling the door). However, according to Darnell, these particular doors were installed such that the problem was magnified even more—the doors at either end of the walkway both swung out from the walkway. He tells the story of a friend who entered the walkway by pulling on the handle of one door; then went to the second door and found that the door would not open when she pulled on the handle. So, concluding that the door was locked, she returned to the first door to exit and found herself trapped in the walkway when the door wouldn't move when she pulled on its handle. After trying to get the attention of others on the outside to tell them that she was trapped in the walkway, she finally realized, with much embarrassment, that the handles had to be pushed when inside the walkway. Darnell reminds us that this funny story could have had tragic consequences if, for example, a panicked person were trying to flee the building in case of a fire.
There are many ways to solve this problem of having identical handles used for opposite actions. One solution would be to put signs on the door that say “push” or “pull.” Although these signs would probably work, a well-designed door (i.e., designed for the user rather than for its looks) should not need a sign to tell people how to use it. Pushing and pulling a door are not rocket science, after all. A better solution would be to install a flat plate on one side of the door, which would make the correct choice obvious: Because there is no handle to grab to pull toward you, there is only one other option—push! A sign would be unnecessary in this case because there would be no question about which action to take.
Signs and symbols that tell us how to use things should be the last option in technology design. Signs are sometimes small and hard to read, they fall off, they wear out, other things cover them up, and they are often presented in only one language, making them less than desirable in a multicultural society. Symbols avoid being unilingual but introduce other problems, such as ambiguity. For example, how would one create simple, unambiguous, and instantly recognizable symbols for “push” and “pull”? As in the door handle example, a better design would be one that made the correct action as obvious as possible, one that would negate the need for a sign or symbol.
A classic example of the failure of technology to exploit this simple idea is a standard topic in ergonomics textbooks and on websites: the arrangement of burners on the stove top. A typical stove top layout is shown in figure 1.5.
The four knobs used to turn on the elements are located at the top of the figure. Quick, say out loud which one is used to turn on the right front burner. My guess is that it is one of the two rightmost knobs. But which one? It is impossible to know for certain without turning one of the knobs and waiting to see which element begins to heat up. So, designers add little labels to tell us which knobs are mapped to which burners. Do they work?
My personal experience is that these labels, if they have not fallen off, worn out, been covered by last night's pasta sauce, or been printed too small to read, are still hard to interpret. More times than I wish to remember, I have returned to a stove expecting to find a pot of boiling water for my pasta but instead found a glowing burner with nothing on it. The relative failure rate of these labels leads me to believe that they are only slightly more helpful than if I had just turned a knob at random.
Have a look at the two alternative stove top designs in figure 1.6. The layout of the knobs in figure 1.6a is the same as before; only the burner layout has been altered. In figure 1.6b, the burner layout is the same as it was before, but now the arrangement of the knobs has been altered.
In both cases, the mapping of each knob to its associated burner has been made obvious by a simple rearrangement of either the knobs or the burner layout. No symbols or labels are needed, and the chances of turning on the wrong burner are minimal. Moreover, the amount of space occupied by the burners and knobs in these more compatible layouts is identical to the space occupied in the incompatible layout. It is interesting to note that these alternative layouts have been available for many years. But have a look at any appliance store flyer or website, and you will see that the classic design still continues to dominate the market.
Sometimes it is simply not possible to avoid the use of labels or signs. In such cases the designer's goal should be to make them as informative as possible. The typical ceiling fan is another good example of a failure to consider the needs of the user when designing the product. The ceiling fan that we have in our home has a wall switch that turns the fan off and on. But the ceiling fan has three additional controls, located on the base of the fan itself (where the fan connects to the ceiling), that are operated by pulling on long strings that hang down from the base. These three strings can be pulled to adjust the speed of the fan, to reverse its direction, or to turn off or on a light located on the bottom of the fan. The problem is that the strings are identical in appearance; I have no way of knowing which control I am activating when I pull on any of the strings.
Actually, that is not completely true. If I stand on a stepladder and get up really close to the ceiling, I can see little labels beside the holes where the strings exit the electrical base of the fan that say “speed,” “light,” and “direction.” I could also stand on the floor and use binoculars to read these labels, I suppose. But, you get the point. Because these labels are next to useless to me, I really just have a one-in-three chance of getting it right each time I pull one of the strings.
How could each control be designed to communicate its function easily to the user? I have used two strategies to modify my own fan. Because reversing the fan's direction is the control that I use least often, I have shortened the string associated with this option to be the most difficult of the three to reach. However, I use the light and speed options about equally often. So, to make these as distinct as possible I borrowed two charms from an old charm bracelet and attached them to the ends of each string. A charm of a rabbit is attached to the string that controls the speed of the fan, and a charm of a book (for which I turn on the light to read) is attached to the string that controls the light. The fan may not look the way the manufacturer had intended it to look, but I have not made the mistake of pulling the wrong string since I attached these charms.
Motor skills such as pushing doors, turning knobs, and pulling strings are simple to learn, and we have all mastered these simple skills over our lifetimes. So why do we have such trouble using them? One reason is that the manufacturer is often paying more attention to aesthetics than to the needs of the user.
Self-Directed Learning Activities
1. Describe, in general, the principle of how product design influences human performance.
2. Search the literature for the term stimulus-response compatibility; then define it in your own words with specific reference to an example presented in this story.
3. Over the next 24 hours, keep a diary of all the objects or things you encounter that you think represent stimulus-response incompatibility. Propose ways each of these things could be made more compatible.
4. Find three research articles in which stimulus-response compatibility has been investigated. What are the similarities and differences in the stimuli and responses used in these studies?
Notes
- Michael Darnell has a wonderful website that features many examples of poorly designed products, with suggestions about simple ways to greatly improve usability: www.baddesigns.com
Suggested Readings
Proctor, R.W., & Van Zandt, T. (2008). Human factors in simple and complex systems (2nd ed.). Boca Raton, FL: CRC Press.
Proctor, R.W., & Vu, K.P.L. (2006). Stimulus-response compatibility principles: Data, theory, and application. Boca Raton, FL: CRC Press.
Schmidt, R.A., & Lee, T.D. (2011). Human information processing. In Motor control and learning: A behavioral emphasis (5th ed., pp. 57-96) Champaign, IL: Human Kinetics.
Learning to win from losing
Why is the learning–performance distinction important?
One day, Michelle Wie might be remembered as one of the greatest golfers of all time. At the age of 14 she was pounding out 300-yard (274 m) drives and not only competing in Ladies Professional (LPGA) Tour events, but also performing quite well in them. Her talent is undeniable. But controversy surrounded this exceptional athlete because of her desire to compete in PGA Tour events, which traditionally have included only male competitors. Wie has performed admirably in some of these events. For example, in the 2004 Sony Open, she shot rounds of 72 and 68 and missed the 36-hole cut by just a single stroke. Her two-round total of 140 placed her in a tie for 80th, which was better than 53 male professionals, including PGA stars Zach Johnson, Hunter Mahan, and Adam Scott. However, Wie has not performed well in most of the other PGA events she has entered, including a 139th-place finish (out of 141 players) at the 2007 Sony Open.
Many argued that Wie should stop competing in PGA Tour events and concentrate her efforts on the LPGA Tour. The typical argument was that her skill development as a golfer was being stalled by repeated failures in PGA Tour events, and that if she concentrated on LPGA events instead, the greater chance for success would escalate her skill development.
I argue that there is a fundamental flaw in this theory, because it falls into a trap known as the learning-performance distinction. The trap underlies one of the most misunderstood concepts in motor learning research, so let's start by defining the two terms. Motor skills researchers use the term performance to refer to a single observation. It could be a score or outcome that reflects the value of a single attempt at a motor skill, or perhaps an average score that statistically summarizes a number of attempts. An 18- or 36-hole score, or a final placement ranking in a tournament, might indicate a representative performance score in the case of a golfer. The term learning is used quite specifically to refer to a stable improvement in skill over time—an improvement that has specifically occurred as the result of practice. The problem illustrated by the criticisms leveled at Michelle Wie is that the critics fail to consider the difference between performance and learning; they mistakenly confuse her failures to improve her performance in men's PGA events as a failure to learn from these experiences. Let's use another example to illustrate this distinction between performance and learning.
Suppose you like to bowl, play regularly in a league, and also practice occasionally. Two years ago your average was 110, last year it was 120, and this year it is 130. Those averages probably indicate that you have become a better bowler as a result of learning: the scores indicate improvements that appear to be rather stable and that have resulted from practice.
So, does this mean that you will score a 130 the next time you bowl? Not necessarily, because there are many reasons for the fluctuations in scores that occur from game to game. Next time you may score well below your average because you are not feeling well, your shoes are too tight, or the crowd is very noisy that night. But this does not mean that you have suddenly lost some of your learned skill. You may score well above your average because you try extra hard to impress someone, or because everything just seems to be well focused (your mojo is working). As before, this does not mean that you have suddenly had a change in learned skill, because there is no indication that the sudden improvement in performance reflects a stable and permanent change in your capability to perform.
At any one time we have a theoretical capability for attaining a certain level of performance. When that theoretical capability changes to a higher level as a result of practice, we can say with confidence that we have learned. The confusion lies in the fact that individual performances may sometimes exceed or not live up to these theoretical capabilities. These fluctuations are expected and in no way diminish or detract from the performer's theoretical capability to perform at a certain skill level.
What I have described is the typical distinction between learning and performance. But, there is another side to this issue, the one that can be applied to Michelle Wie. The issue concerns the situation in which performances do not appear to change, or appear to be getting worse. Does that mean that learning is not occurring? Let's go back to the bowling example.
Suppose that this year your average was 120, which is the same as it was last year. You may take the absence of a stable improvement, despite all of your practice, as evidence that your learning has stalled. But, as it turns out, last year your league bowled Tuesday nights, which is your day off. This year you bowl on Wednesday nights, one hour after completing your 10-hour work shift. You are typically tired and hungry on league nights, and your bowling scores suffered this year as a result. What does all this suggest? What is your theoretical capability to bowl, and has this changed from last year, despite the change in your bowling night? Could it be that you actually have improved (i.e., learned), but the feeling of fatigue and other factors directly related to your work schedule prevented you from actually performing up to these expectations?
The point here is that making inferences about learning requires evidence that goes beyond a single performance or sometimes even a set of performances. Under what circumstances were these performances observed? Are there mitigating factors that might explain why the theoretical true score differs from the observed score? The absence of any observable change in performance does not mean that the unobservable (that theoretical capability to perform at a certain level of skill) has not improved.
Michelle Wie took a very different route in the development of her golfing skill. Her decision to compete in PGA events placed enormous pressures on her to perform, and only an exceptional result would have convinced the public that she was benefiting from these extreme challenges. All we could observe were her scores in these events. The unobservable, however, what she has learned by playing in these events, can never be truly understood. I suspect that if she achieves the rank of one of the greatest players of all time, it will be due in no small measure to these experiences competing against the very best golfers in the world.
Self-Directed Learning Activities
1. In your own words, explain the distinction between performance and learning.
2. What is the difference between a theoretical true score and an observed score?
3. Suggest a reason a person's observed score might exceed the true score, and a reason the observed score might fail to achieve the level of the true score.
4. Identify another sport, athlete, or situation in which the failure to account for the learning-performance distinction has resulted in an inappropriate conclusion.
Notes
- Babe Zaharias (see “The Babe”) was the first woman to compete in a men's PGA event.
- Here is a sampling of some of the criticism of Michele Wie's decision to compete in PGA Tour events, from two sports commentary sites:
- Michelle Wie won her second LPGA Tour title in Winnipeg, Manitoba, at the 2010 CN Canadian Women's Open.
Suggested Readings
Schmidt, R.A. (1972). The case against learning and forgetting scores. Journal of Motor Behavior, 4, 79-88.
Schmidt, R.A., & Lee, T.D. (2011). Motor learning concepts and research methods. In Motor control and learning: A behavioral emphasis (5th ed., pp. 327-346). Champaign, IL: Human Kinetics.
Cutting wood and missing putts
How are constant error, variable error, and absolute error useful for understanding motor control?
Two things I like to do—cutting wood and golfing—remind me often of the importance of measures of central tendency and variability in motor control performance. For example, I use my chainsaw to cut up fallen trees to burn in the woodstove. Cutting logs into burnable lengths is not an exact science. In the middle of a Canadian winter, when the temperature is cold and plenty of snow blankets the ground, it is generally considered a good idea to keep moving. I do not take time to measure the lengths of the logs I saw; rather, I make a quick visual estimate and then start to make my cut. But, in doing so, I must remember two important things related to the mean and the standard deviation of the logs I cut. First, because our woodstove is only large enough to hold pieces of wood that are 50 centimeters (20 in.) or shorter, any pieces that are longer than that are useless. And second, wood is easier to stack when the pieces are all roughly of the same length. Essentially, these features remind me that the mean and standard deviation are both important; I must have a mean that is less than 50 centimeters and a standard deviation that is as small as possible.
When considering data in terms of a specific goal or standard, it is often better to express the mean and standard deviation in terms of error measures—in my case, relative to a goal wood length of 50 centimeters. For example, wood lengths of 45, 42, 48, 47, and 52 centimeters would be expressed as lengths that are -5, -8, -2, -3, and +2 centimeters, respectively, relative to my goal wood length of 50 centimeters. If we calculated the mean of these lengths of wood, then we could express the stack of wood as having either an average length of 46.8 centimeters, or the error scores as having a mean constant error of -3.2 centimeters. Having a negative constant error is good in this case, because the average piece of wood that I have cut will easily fit in the woodstove (i.e., the mean is less than 50 centimeters).
The careful reader will have noted, however, that even though my mean constant error is negative, there is still one block of wood that will not fit (the 52 cm log). Thus, by itself, the mean constant error score (-3.2 cm) is misleading, because it does not provide any indication of how many logs might not actually fit in the woodstove. Our measure of variability, the standard deviation, gives us some indication of this information. The measure of variability of these error scores (called the variable error) represents the deviation of the error score for each log relative to the mean constant error and is calculated in the same way as is a standard deviation. In this case, the variable error is 3.3 centimeters. In general, when I am cutting wood, I want to achieve a negative mean constant error with a variable error that is as small as possible.
Although constant error provides a useful heuristic measure of average performance in cutting wood with a chainsaw, it is a quite misleading measure of central tendency in another activity I like to do. Let's say that I have struck 10 golf putts—5 of these putts go past the hole by 5, 10, 15, 20, and 25 centimeters, respectively, and the other five putts come up short of the hole by equal amounts (-5, -10, -15, -20 and -25 cm). On average, by how much have I missed the hole with these putts? If you were to calculate the average as in the previous wood-cutting example (the sum of the 10 individual error scores, divided by 10) the answer would be a mean constant error of 0. In other words, my “average” putt ended up in the hole. However, we know that this answer surely must be wrong because none of the individual putts actually went in the hole. So, we must use a different way to express these error scores to avoid an answer that makes no sense.
The problem that we sometimes run into with constant error is that it provides a measure of average bias, the tendency to err in a specific way (e.g., by too much or too little; too far left versus too far right). In some cases a specific bias is desirable, such as the tendency to undercut a wood length of 50 centimeters so that all of my wood pieces fit in the woodstove. In the case of my putts that have no consistent bias, we are much better off using a measure of central tendency that removes the bias from each score prior to calculating the mean. Such a procedure is called using the absolute (unsigned) scores. Hence, this measure of central tendency is called the mean absolute error. Here, our measure of central tendency is the mean of the unsigned scores (5, 5, 10, 10, 15, 15, 20, 20, 25, 25), which is 15 centimeters. That is, the golf putts tended to miss the hole by an average of 15 centimeters. In this particular case, the mean absolute error score (15 cm from the hole) represents a more accurate picture of the entire set of individual scores than what is represented in the mean constant error (in the hole).
Averages are convenient ways to express how we tend to perform. But, numbers that represent central tendencies can be misleading if the tendency itself is not a strong one. Measures of variability are one way to identify the strength of a tendency. Depending on the context, different methods of calculating central tendencies may characterize the data in more representative ways.
Self-Directed Learning Activities
1. Define constant error, absolute error, and variable error in your own words.
2. What do the terms algebraic error and total error (or Henry's E) refer to? How are they calculated, and how are they similar to the terms constant error and absolute error, respectively?
3. Look up and briefly describe a research investigation that uses at least two error measures. Why do you think the researchers used these measures in particular?
4. Conduct a brief study on yourself. Close your eyes and, using a pencil, draw five lines as close to 4 inches (10 cm) as possible, one below the previous one, without opening your eyes until you have drawn the fifth line. Measure and record the length of each line you drew; then calculate all of the error measures that have been discussed.
Notes
- Here is a web-based standard deviation calculator:
www.tinyurl.com/standarddeviationcalculator
Suggested Readings
Chapanis, A. (1951). Theory and methods for analyzing errors in man-machine systems. Annals of the New York Academy of Sciences, 51, 1179-1203.
Schmidt, R.A., & Lee, T.D. (2011). Methodology for studying motor performance. In Motor control and learning: A behavioral emphasis (5th ed., pp. 21-56). Champaign, IL: Human Kinetics.
Forensic motor control
What are generalized motor programs, and what do keystroke dynamics reveal about them?
Most of what I know about forensic identification comes from crime shows that I've seen on TV. The suspect is caught because a fingerprint was left on the murder weapon. The evidence is used to convict the suspect because no two fingerprints have ever been found to be identical. Crime shows tell us that other forensic methods can be used as well, such as DNA and eye retina data, but the fingerprint is the oldest biometric tool used for identification. However, another method that reveals a great deal about your identity is how you write or even type your name. For example, the password that you type to log on to your e-mail account may be just as identifiable as your fingerprint.
Try the following task as an example: go into any word processing program and type your name 10 times, once on each line, as I have done here:
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
I am not a skilled typist. I use my right middle finger to hold down the shift key and press the letter T with my left middle finger, then use my right middle finger to press i and my right index finger to press m, and then my right thumb to press the space bar. It is not very efficient, but I do it the same way every time I type my name. Regardless of how skilled a typist you are, you most likely repeat much the same process each of the 10 times you type your name.
Now, let's try to unravel the temporal “fingerprint” that you left behind when you typed your name. Suppose we conducted a simple analysis of the keystrokes and the time between each of the keystrokes that you made. The total time taken to type your name once is simply the time from the first keystroke to the last keystroke. At a more fine-grained level of analysis, the total time constitutes the time each key is held down (the dwell time) plus the time between the release of one key and the depression of the next (the transition time). All of the individual dwell times plus all of the transition times will sum to the total time.
In figure 7.5, I have plotted a hypothetical example of the 10 trials to type my name. The light gray bands denote the dwell times, and the darker gray bands indicate the transition times. I have ordered the lines of bands that represent the 10 trials from the fastest (least total time) at the top to the slowest trial at the bottom (most total time).
If you were to analyze each of the 10 repetitions of your name, you would probably find that all of the total times would be similar, but probably never exactly the same each time, just as in figure 7.5. The “noise” in our central nervous systems, plus other factors, causes the results to vary a little bit each time, resulting in some repetitions to be slower, and others faster, than the average total time.
But, take a closer look at each individual band in figure 7.5 as it changes over the 10 trials. What you will notice is that as the total time increases, each band gets proportionally longer too. We could express these numbers another way by dividing the time for each individual band by the total time for that trial to obtain a relative proportion of time represented by each band. Given the hypothetical evidence in figure 7.5, what we would discover is that the relative time for each band, expressed as a percentage, would stay roughly the same across all of the repetitions. Applying temporal and other methods to deconstruct how we type (e.g., key press forces) is called the study of keystroke dynamics.
According to generalized motor program theory, relative time is one of the key features of movement that is controlled by the central nervous system, especially for brief, rapid actions. Relative time is an expression of our motor control fingerprint. For well-learned tasks such as typing and handwriting, our central nervous system regulates the relative timing of impulses that are sent to the muscles that carry out these tasks. Various factors influence the real time on any particular instance. For example, using a keyboard that requires more force to depress the keys might result in overall slower times than would result using a keyboard that has a light touch, but the relative timing would likely remain the same.
Research in keystroke dynamics may result in the ability to identify people who carry out Internet fraud. In many ways it is similar to the use of handwriting dynamics to analyze the timing of the cursive expressions of a signature. It is fairly easy to forge the spatial representation of someone's signature, but very difficult to forge the temporal dynamics that result in that signature. These expressions of timing behavior represent our motor control fingerprints.
Self-Directed Learning Activities
1. Define the term generalized motor program in your own words.
2. How does the concept of a generalized motor program differ from the concept of a motor program as it was used in stories such as “Antilock Brakes” (in chapter 5) and “Point of No Return” (earlier in this
chapter)?
3. Suggest a method by which the handwriting dynamics of signatures could be used to detect fraud.
4. Some people contend that there is one generalized motor program for the full swing in golf, regardless of which club is used. How could a temporal dynamics analysis be used to assess this contention?
Notes
- An excellent review of fingerprint analysis by David Ashbaugh of the Royal Canadian Mounted Police is available here:
www.onin.com/fp/ridgeology.pdf
- A lot of controversy remains about invariances in motor performance; how they are measured; and what invariance, or lack of invariance, means in terms of motor programs. The following articles provide good arguments for the debate:
Gentner, D.R. (1987). Timing of skilled motor performance: Tests of the proportional duration model. Psychological Review, 94, 255-276.
Heuer, H. (1988). Testing the invariance of relative timing: Comment on Gentner (1987). Psychological Review, 95, 552-557.
Suggested Readings
Schmidt, R.A. (1975). A schema theory of discrete motor skill learning. Psychological Review, 82, 225-260.
Schmidt, R.A. (1985). The search for invariance in skilled movement behavior. Research Quarterly for Exercise and Sport, 56, 188-200.
Schmidt, R.A., & Lee, T.D. (2011). Central contributions to motor control. In Motor control and learning: A behavioral emphasis (5th ed., pp. 177-222) Champaign, IL: Human Kinetics.
The Farmers' Market
What are the roles of motor error and hypervigilance in unintended acceleration accidents?
The prosecution, at Weller's trial several years later, claimed it was no accident—that Weller deliberately drove his car through the crowded market. The reason, they said, was that Weller had been involved in a minor fender bender just moments before he entered the farmers' market. His response to the fender bender was to flee the scene of the accident. Witnesses for the prosecution painted Weller as a cold-blooded killer, commenting on the determined look on his face during the ordeal and his relatively calm demeanor afterward. It probably also didn't help his case when he emerged from the car immediately afterward and wondered aloud why the people he had hit had not jumped out of his way. Adding his age into the mix, Weller's actions were painted as rather pathetic.
Richard Schmidt, a renowned motor control scholar and human factors expert, testified on behalf of the defense team and argued that the facts of the case shared many similarities to accidents caused by errors of pedal misapplication, or unintended acceleration. Accidents of this type, which, thankfully, are quite rare, occur when the driver intends to apply pressure to the brake pedal, but misses the brake and pushes down on the accelerator instead.
Unintended acceleration accidents had been investigated for many years prior to the Weller case. These accidents were more common when the driver first got into the car, started the engine, and engaged the automatic transmission from the Park position into either Drive or Reverse. Indeed, such frequent episodes of “runaway cars” were the primary reason auto manufacturers added the brake-transmission shift interlock system in the 1980s so that an automatic transmission lever could not be moved from the Park position until the car sensed that a certain amount of pressure had been applied to the brake. However, the brake lock system only prevented this particular type of pedal misapplication; moving the foot to the accelerator instead of the brake would not be prevented once the transmission was successfully engaged out of Park and the car was in motion.
It is important to note that reaching for the brake requires that we steer the foot from a comfortable seated position to a target (the brake) in the absence of any visual guidance. We do this all the time without making any errors. We know where the brake is located from experience, and we also know that the brake feels different underfoot than does the accelerator. So then why would we suddenly miss the brake, push down on the accelerator instead, and then keep the foot there?
Schmidt (1989) presented evidence that unintended acceleration cases frequently involved accidents in which the driver had less experience than usual with that particular vehicle. Therefore, in some of these cases, the exact location and feel of the brake might not have been as familiar to the driver as normally could have been expected. These accidents also occurred more frequently on start-up, compared to later in the driving cycle, perhaps due to temporary factors associated with preparing an action (see “Shooting Two From the Line” in chapter 11). Driver inattention has also been linked with these cases, so it may not come as a surprise that drivers would not immediately notice the difference between the brake and gas pedal if engaged in a distracting activity at the same time (see “Gumbo” in chapter 6).
But for Weller, none of the common profiles for these accidents fit the case: he had already been driving before the supposed pedal misapplication error, so it was not a matter of missing the brake on initial start-up. Weller was quite familiar with his own vehicle, an 11-year-old Buick LeSabre. And he was not talking on a cell phone. Instead, Schmidt argued that Weller's pedal misapplication error was likely triggered by a catastrophic case of panic, termed hypervigilance, which could have been initiated when Weller had been involved in the fender bender just prior to the episode.
But one last issue seemed particularly problematic, according to the prosecution. Once the pedal misapplication error had occurred and the car started to accelerate wildly out of control, why didn't Weller simply remove his foot from the pedal or turn off the engine—actions that would have brought the car quickly to a stop? Again, failure to carry out corrective actions is typical of unintended acceleration cases, and some reasonable accounts have been offered to explain why drivers do not perform them. First, the driver probably does not realize that the foot is on the accelerator rather than the brake. The intention was to press the brake, and the fact that the pedal has gone all the way to the floor could reasonably be interpreted as brake failure rather than human error. And second, once the driver enters into this catastrophic state of panic, all normal modes of thinking cease. Reasoning and problem solving, the kinds of activities that are easy to do when unflustered, become unlikely, if not impossible, to carry out when in this state of hypervigilance.
Thomas Shelton, a member of the California Highway Patrol, testified at Weller's trial that he once investigated an unintended acceleration case in which an elderly woman ended up driving her car onto the top of another vehicle. The woman was in such a panicked state that when Shelton arrived at the accident scene, he had to climb up into the car to shut off the racing engine, at which time he noticed that the woman was still seated, very much alive, staring straight ahead with a death grip on the steering wheel, and with her foot still pushing the accelerator all the way to the floor.
Unfortunately, all of these arguments can only be used to speculate about what may have occurred in George Russell Weller's Buick LeSabre on that fateful day. On October 20, 2006, the jury convicted him of vehicular manslaughter in the 10 deaths resulting from the Santa Monica farmers' market crash. Nobody will ever know whether the verdict was the right one.
Self-Directed Learning Activities
1. In your own words describe the phenomenon known as unintended acceleration.
2. Describe a situation in which you made an action error that you were able to correct. How did you know that you had made the error, and what did you do to correct it?
3. Using our feet to manipulate car pedals involves aiming without visual feedback. What factors influence our ability to make these aiming movements accurately?
4. Propose a research methodology that examines one's ability to (a) move to a target without visual feedback and (b) estimate the accuracy of those aimed movements (again, without vision).
Notes
- Evidence from George Weller's trial during September and October of 2006 was summarized in the Santa Monica Daily Press, which can be accessed through its archives:
- Not all cases of unintended acceleration are generally agreed to be the result of a pedal misapplication. A segment of the television show 60 Minutes, hosted by Ed Bradley and which aired November 22, 1986, claimed that accidents of similar etiology involving the Audi 5000 were the result of a faulty idle stabilizer, which caused the car to accelerate wildly out of control when put into gear. An investigation by the U.S. NHTSA (National Highway Traffic Safety Administration) failed to support 60 Minutes' claim.
- More recently, runaway Toyotas have been a topic of concern. Once again, a government investigation failed to support a claim that these cases of unintended acceleration were due to an electronic fault in the engine. Again, this leaves open the very real possibility that driver error is to blame, as suggested by Richard Schmidt in the New York Times:
Suggested Readings
Castelli, J., Nash, C., Ditlow, C., & Pecht, M. (2003). Sudden acceleration: The myth of driver error. University of Maryland: CALCE EPSC Press.
Pollard, J., & Sussman, E.D. (1989). An examination of sudden acceleration. National Highway Traffic Safety Administration Final Report # DOT-TSC-NHTSA-89-1.
Schmidt, R.A. (1989). Unintended acceleration: A review of human factors contributions. Human Factors, 31, 345-364.
Schmidt, R.A. (1993). Unintended acceleration: Human performance considerations. In B. Peacock & W. Karwowski (Eds.), Automotive ergonomics (pp. 431-451). London: Taylor & Francis.
Schmidt, R.A., & Lee, T.D. (2011). Central contributions to motor control. In Motor control and learning: A behavioral emphasis (5th ed., pp. 177-222) Champaign, IL: Human Kinetics.
Push or Pull?
How do product designs influence people to perform specific actions?
It seems hard to believe, but brand-new buildings are still being constructed with very silly design features. For example, a new building on our campus opened last year and featured many modern technologies, including very nice-looking glass doors through which you enter and exit the building. The problem is that the doors have identical handles on either side. The handles themselves are very attractive, but they don't tell me what to do when I want to enter the building—should I push or pull the door? The handles on these particular doors suggest to me that they are made for grasping and pulling. However, the door goes in only one direction, and the very same handles appear on both sides of the door. Therefore, to enter the building, I have to use the handle to pull the door, and to exit, I have to use the identical type of handle to push the door. How would someone possibly know which way the door swings when approaching it? After a year of using these doors on a daily basis, I still just guess about whether to push or pull. Why would someone design a door like that?
Michael Darnell has created a wonderful website that illustrates many instances of products that have been designed without the user in mind. One of his web pages describes a door handle problem similar to the one I just described, in which a building has a short walkway with one set of doors on either end. Each door has identical handles on either side of the door (the same type of handle is used for pushing and pulling the door). However, according to Darnell, these particular doors were installed such that the problem was magnified even more—the doors at either end of the walkway both swung out from the walkway. He tells the story of a friend who entered the walkway by pulling on the handle of one door; then went to the second door and found that the door would not open when she pulled on the handle. So, concluding that the door was locked, she returned to the first door to exit and found herself trapped in the walkway when the door wouldn't move when she pulled on its handle. After trying to get the attention of others on the outside to tell them that she was trapped in the walkway, she finally realized, with much embarrassment, that the handles had to be pushed when inside the walkway. Darnell reminds us that this funny story could have had tragic consequences if, for example, a panicked person were trying to flee the building in case of a fire.
There are many ways to solve this problem of having identical handles used for opposite actions. One solution would be to put signs on the door that say “push” or “pull.” Although these signs would probably work, a well-designed door (i.e., designed for the user rather than for its looks) should not need a sign to tell people how to use it. Pushing and pulling a door are not rocket science, after all. A better solution would be to install a flat plate on one side of the door, which would make the correct choice obvious: Because there is no handle to grab to pull toward you, there is only one other option—push! A sign would be unnecessary in this case because there would be no question about which action to take.
Signs and symbols that tell us how to use things should be the last option in technology design. Signs are sometimes small and hard to read, they fall off, they wear out, other things cover them up, and they are often presented in only one language, making them less than desirable in a multicultural society. Symbols avoid being unilingual but introduce other problems, such as ambiguity. For example, how would one create simple, unambiguous, and instantly recognizable symbols for “push” and “pull”? As in the door handle example, a better design would be one that made the correct action as obvious as possible, one that would negate the need for a sign or symbol.
A classic example of the failure of technology to exploit this simple idea is a standard topic in ergonomics textbooks and on websites: the arrangement of burners on the stove top. A typical stove top layout is shown in figure 1.5.
The four knobs used to turn on the elements are located at the top of the figure. Quick, say out loud which one is used to turn on the right front burner. My guess is that it is one of the two rightmost knobs. But which one? It is impossible to know for certain without turning one of the knobs and waiting to see which element begins to heat up. So, designers add little labels to tell us which knobs are mapped to which burners. Do they work?
My personal experience is that these labels, if they have not fallen off, worn out, been covered by last night's pasta sauce, or been printed too small to read, are still hard to interpret. More times than I wish to remember, I have returned to a stove expecting to find a pot of boiling water for my pasta but instead found a glowing burner with nothing on it. The relative failure rate of these labels leads me to believe that they are only slightly more helpful than if I had just turned a knob at random.
Have a look at the two alternative stove top designs in figure 1.6. The layout of the knobs in figure 1.6a is the same as before; only the burner layout has been altered. In figure 1.6b, the burner layout is the same as it was before, but now the arrangement of the knobs has been altered.
In both cases, the mapping of each knob to its associated burner has been made obvious by a simple rearrangement of either the knobs or the burner layout. No symbols or labels are needed, and the chances of turning on the wrong burner are minimal. Moreover, the amount of space occupied by the burners and knobs in these more compatible layouts is identical to the space occupied in the incompatible layout. It is interesting to note that these alternative layouts have been available for many years. But have a look at any appliance store flyer or website, and you will see that the classic design still continues to dominate the market.
Sometimes it is simply not possible to avoid the use of labels or signs. In such cases the designer's goal should be to make them as informative as possible. The typical ceiling fan is another good example of a failure to consider the needs of the user when designing the product. The ceiling fan that we have in our home has a wall switch that turns the fan off and on. But the ceiling fan has three additional controls, located on the base of the fan itself (where the fan connects to the ceiling), that are operated by pulling on long strings that hang down from the base. These three strings can be pulled to adjust the speed of the fan, to reverse its direction, or to turn off or on a light located on the bottom of the fan. The problem is that the strings are identical in appearance; I have no way of knowing which control I am activating when I pull on any of the strings.
Actually, that is not completely true. If I stand on a stepladder and get up really close to the ceiling, I can see little labels beside the holes where the strings exit the electrical base of the fan that say “speed,” “light,” and “direction.” I could also stand on the floor and use binoculars to read these labels, I suppose. But, you get the point. Because these labels are next to useless to me, I really just have a one-in-three chance of getting it right each time I pull one of the strings.
How could each control be designed to communicate its function easily to the user? I have used two strategies to modify my own fan. Because reversing the fan's direction is the control that I use least often, I have shortened the string associated with this option to be the most difficult of the three to reach. However, I use the light and speed options about equally often. So, to make these as distinct as possible I borrowed two charms from an old charm bracelet and attached them to the ends of each string. A charm of a rabbit is attached to the string that controls the speed of the fan, and a charm of a book (for which I turn on the light to read) is attached to the string that controls the light. The fan may not look the way the manufacturer had intended it to look, but I have not made the mistake of pulling the wrong string since I attached these charms.
Motor skills such as pushing doors, turning knobs, and pulling strings are simple to learn, and we have all mastered these simple skills over our lifetimes. So why do we have such trouble using them? One reason is that the manufacturer is often paying more attention to aesthetics than to the needs of the user.
Self-Directed Learning Activities
1. Describe, in general, the principle of how product design influences human performance.
2. Search the literature for the term stimulus-response compatibility; then define it in your own words with specific reference to an example presented in this story.
3. Over the next 24 hours, keep a diary of all the objects or things you encounter that you think represent stimulus-response incompatibility. Propose ways each of these things could be made more compatible.
4. Find three research articles in which stimulus-response compatibility has been investigated. What are the similarities and differences in the stimuli and responses used in these studies?
Notes
- Michael Darnell has a wonderful website that features many examples of poorly designed products, with suggestions about simple ways to greatly improve usability: www.baddesigns.com
Suggested Readings
Proctor, R.W., & Van Zandt, T. (2008). Human factors in simple and complex systems (2nd ed.). Boca Raton, FL: CRC Press.
Proctor, R.W., & Vu, K.P.L. (2006). Stimulus-response compatibility principles: Data, theory, and application. Boca Raton, FL: CRC Press.
Schmidt, R.A., & Lee, T.D. (2011). Human information processing. In Motor control and learning: A behavioral emphasis (5th ed., pp. 57-96) Champaign, IL: Human Kinetics.
Learning to win from losing
Why is the learning–performance distinction important?
One day, Michelle Wie might be remembered as one of the greatest golfers of all time. At the age of 14 she was pounding out 300-yard (274 m) drives and not only competing in Ladies Professional (LPGA) Tour events, but also performing quite well in them. Her talent is undeniable. But controversy surrounded this exceptional athlete because of her desire to compete in PGA Tour events, which traditionally have included only male competitors. Wie has performed admirably in some of these events. For example, in the 2004 Sony Open, she shot rounds of 72 and 68 and missed the 36-hole cut by just a single stroke. Her two-round total of 140 placed her in a tie for 80th, which was better than 53 male professionals, including PGA stars Zach Johnson, Hunter Mahan, and Adam Scott. However, Wie has not performed well in most of the other PGA events she has entered, including a 139th-place finish (out of 141 players) at the 2007 Sony Open.
Many argued that Wie should stop competing in PGA Tour events and concentrate her efforts on the LPGA Tour. The typical argument was that her skill development as a golfer was being stalled by repeated failures in PGA Tour events, and that if she concentrated on LPGA events instead, the greater chance for success would escalate her skill development.
I argue that there is a fundamental flaw in this theory, because it falls into a trap known as the learning-performance distinction. The trap underlies one of the most misunderstood concepts in motor learning research, so let's start by defining the two terms. Motor skills researchers use the term performance to refer to a single observation. It could be a score or outcome that reflects the value of a single attempt at a motor skill, or perhaps an average score that statistically summarizes a number of attempts. An 18- or 36-hole score, or a final placement ranking in a tournament, might indicate a representative performance score in the case of a golfer. The term learning is used quite specifically to refer to a stable improvement in skill over time—an improvement that has specifically occurred as the result of practice. The problem illustrated by the criticisms leveled at Michelle Wie is that the critics fail to consider the difference between performance and learning; they mistakenly confuse her failures to improve her performance in men's PGA events as a failure to learn from these experiences. Let's use another example to illustrate this distinction between performance and learning.
Suppose you like to bowl, play regularly in a league, and also practice occasionally. Two years ago your average was 110, last year it was 120, and this year it is 130. Those averages probably indicate that you have become a better bowler as a result of learning: the scores indicate improvements that appear to be rather stable and that have resulted from practice.
So, does this mean that you will score a 130 the next time you bowl? Not necessarily, because there are many reasons for the fluctuations in scores that occur from game to game. Next time you may score well below your average because you are not feeling well, your shoes are too tight, or the crowd is very noisy that night. But this does not mean that you have suddenly lost some of your learned skill. You may score well above your average because you try extra hard to impress someone, or because everything just seems to be well focused (your mojo is working). As before, this does not mean that you have suddenly had a change in learned skill, because there is no indication that the sudden improvement in performance reflects a stable and permanent change in your capability to perform.
At any one time we have a theoretical capability for attaining a certain level of performance. When that theoretical capability changes to a higher level as a result of practice, we can say with confidence that we have learned. The confusion lies in the fact that individual performances may sometimes exceed or not live up to these theoretical capabilities. These fluctuations are expected and in no way diminish or detract from the performer's theoretical capability to perform at a certain skill level.
What I have described is the typical distinction between learning and performance. But, there is another side to this issue, the one that can be applied to Michelle Wie. The issue concerns the situation in which performances do not appear to change, or appear to be getting worse. Does that mean that learning is not occurring? Let's go back to the bowling example.
Suppose that this year your average was 120, which is the same as it was last year. You may take the absence of a stable improvement, despite all of your practice, as evidence that your learning has stalled. But, as it turns out, last year your league bowled Tuesday nights, which is your day off. This year you bowl on Wednesday nights, one hour after completing your 10-hour work shift. You are typically tired and hungry on league nights, and your bowling scores suffered this year as a result. What does all this suggest? What is your theoretical capability to bowl, and has this changed from last year, despite the change in your bowling night? Could it be that you actually have improved (i.e., learned), but the feeling of fatigue and other factors directly related to your work schedule prevented you from actually performing up to these expectations?
The point here is that making inferences about learning requires evidence that goes beyond a single performance or sometimes even a set of performances. Under what circumstances were these performances observed? Are there mitigating factors that might explain why the theoretical true score differs from the observed score? The absence of any observable change in performance does not mean that the unobservable (that theoretical capability to perform at a certain level of skill) has not improved.
Michelle Wie took a very different route in the development of her golfing skill. Her decision to compete in PGA events placed enormous pressures on her to perform, and only an exceptional result would have convinced the public that she was benefiting from these extreme challenges. All we could observe were her scores in these events. The unobservable, however, what she has learned by playing in these events, can never be truly understood. I suspect that if she achieves the rank of one of the greatest players of all time, it will be due in no small measure to these experiences competing against the very best golfers in the world.
Self-Directed Learning Activities
1. In your own words, explain the distinction between performance and learning.
2. What is the difference between a theoretical true score and an observed score?
3. Suggest a reason a person's observed score might exceed the true score, and a reason the observed score might fail to achieve the level of the true score.
4. Identify another sport, athlete, or situation in which the failure to account for the learning-performance distinction has resulted in an inappropriate conclusion.
Notes
- Babe Zaharias (see “The Babe”) was the first woman to compete in a men's PGA event.
- Here is a sampling of some of the criticism of Michele Wie's decision to compete in PGA Tour events, from two sports commentary sites:
- Michelle Wie won her second LPGA Tour title in Winnipeg, Manitoba, at the 2010 CN Canadian Women's Open.
Suggested Readings
Schmidt, R.A. (1972). The case against learning and forgetting scores. Journal of Motor Behavior, 4, 79-88.
Schmidt, R.A., & Lee, T.D. (2011). Motor learning concepts and research methods. In Motor control and learning: A behavioral emphasis (5th ed., pp. 327-346). Champaign, IL: Human Kinetics.
Cutting wood and missing putts
How are constant error, variable error, and absolute error useful for understanding motor control?
Two things I like to do—cutting wood and golfing—remind me often of the importance of measures of central tendency and variability in motor control performance. For example, I use my chainsaw to cut up fallen trees to burn in the woodstove. Cutting logs into burnable lengths is not an exact science. In the middle of a Canadian winter, when the temperature is cold and plenty of snow blankets the ground, it is generally considered a good idea to keep moving. I do not take time to measure the lengths of the logs I saw; rather, I make a quick visual estimate and then start to make my cut. But, in doing so, I must remember two important things related to the mean and the standard deviation of the logs I cut. First, because our woodstove is only large enough to hold pieces of wood that are 50 centimeters (20 in.) or shorter, any pieces that are longer than that are useless. And second, wood is easier to stack when the pieces are all roughly of the same length. Essentially, these features remind me that the mean and standard deviation are both important; I must have a mean that is less than 50 centimeters and a standard deviation that is as small as possible.
When considering data in terms of a specific goal or standard, it is often better to express the mean and standard deviation in terms of error measures—in my case, relative to a goal wood length of 50 centimeters. For example, wood lengths of 45, 42, 48, 47, and 52 centimeters would be expressed as lengths that are -5, -8, -2, -3, and +2 centimeters, respectively, relative to my goal wood length of 50 centimeters. If we calculated the mean of these lengths of wood, then we could express the stack of wood as having either an average length of 46.8 centimeters, or the error scores as having a mean constant error of -3.2 centimeters. Having a negative constant error is good in this case, because the average piece of wood that I have cut will easily fit in the woodstove (i.e., the mean is less than 50 centimeters).
The careful reader will have noted, however, that even though my mean constant error is negative, there is still one block of wood that will not fit (the 52 cm log). Thus, by itself, the mean constant error score (-3.2 cm) is misleading, because it does not provide any indication of how many logs might not actually fit in the woodstove. Our measure of variability, the standard deviation, gives us some indication of this information. The measure of variability of these error scores (called the variable error) represents the deviation of the error score for each log relative to the mean constant error and is calculated in the same way as is a standard deviation. In this case, the variable error is 3.3 centimeters. In general, when I am cutting wood, I want to achieve a negative mean constant error with a variable error that is as small as possible.
Although constant error provides a useful heuristic measure of average performance in cutting wood with a chainsaw, it is a quite misleading measure of central tendency in another activity I like to do. Let's say that I have struck 10 golf putts—5 of these putts go past the hole by 5, 10, 15, 20, and 25 centimeters, respectively, and the other five putts come up short of the hole by equal amounts (-5, -10, -15, -20 and -25 cm). On average, by how much have I missed the hole with these putts? If you were to calculate the average as in the previous wood-cutting example (the sum of the 10 individual error scores, divided by 10) the answer would be a mean constant error of 0. In other words, my “average” putt ended up in the hole. However, we know that this answer surely must be wrong because none of the individual putts actually went in the hole. So, we must use a different way to express these error scores to avoid an answer that makes no sense.
The problem that we sometimes run into with constant error is that it provides a measure of average bias, the tendency to err in a specific way (e.g., by too much or too little; too far left versus too far right). In some cases a specific bias is desirable, such as the tendency to undercut a wood length of 50 centimeters so that all of my wood pieces fit in the woodstove. In the case of my putts that have no consistent bias, we are much better off using a measure of central tendency that removes the bias from each score prior to calculating the mean. Such a procedure is called using the absolute (unsigned) scores. Hence, this measure of central tendency is called the mean absolute error. Here, our measure of central tendency is the mean of the unsigned scores (5, 5, 10, 10, 15, 15, 20, 20, 25, 25), which is 15 centimeters. That is, the golf putts tended to miss the hole by an average of 15 centimeters. In this particular case, the mean absolute error score (15 cm from the hole) represents a more accurate picture of the entire set of individual scores than what is represented in the mean constant error (in the hole).
Averages are convenient ways to express how we tend to perform. But, numbers that represent central tendencies can be misleading if the tendency itself is not a strong one. Measures of variability are one way to identify the strength of a tendency. Depending on the context, different methods of calculating central tendencies may characterize the data in more representative ways.
Self-Directed Learning Activities
1. Define constant error, absolute error, and variable error in your own words.
2. What do the terms algebraic error and total error (or Henry's E) refer to? How are they calculated, and how are they similar to the terms constant error and absolute error, respectively?
3. Look up and briefly describe a research investigation that uses at least two error measures. Why do you think the researchers used these measures in particular?
4. Conduct a brief study on yourself. Close your eyes and, using a pencil, draw five lines as close to 4 inches (10 cm) as possible, one below the previous one, without opening your eyes until you have drawn the fifth line. Measure and record the length of each line you drew; then calculate all of the error measures that have been discussed.
Notes
- Here is a web-based standard deviation calculator:
www.tinyurl.com/standarddeviationcalculator
Suggested Readings
Chapanis, A. (1951). Theory and methods for analyzing errors in man-machine systems. Annals of the New York Academy of Sciences, 51, 1179-1203.
Schmidt, R.A., & Lee, T.D. (2011). Methodology for studying motor performance. In Motor control and learning: A behavioral emphasis (5th ed., pp. 21-56). Champaign, IL: Human Kinetics.
Forensic motor control
What are generalized motor programs, and what do keystroke dynamics reveal about them?
Most of what I know about forensic identification comes from crime shows that I've seen on TV. The suspect is caught because a fingerprint was left on the murder weapon. The evidence is used to convict the suspect because no two fingerprints have ever been found to be identical. Crime shows tell us that other forensic methods can be used as well, such as DNA and eye retina data, but the fingerprint is the oldest biometric tool used for identification. However, another method that reveals a great deal about your identity is how you write or even type your name. For example, the password that you type to log on to your e-mail account may be just as identifiable as your fingerprint.
Try the following task as an example: go into any word processing program and type your name 10 times, once on each line, as I have done here:
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
I am not a skilled typist. I use my right middle finger to hold down the shift key and press the letter T with my left middle finger, then use my right middle finger to press i and my right index finger to press m, and then my right thumb to press the space bar. It is not very efficient, but I do it the same way every time I type my name. Regardless of how skilled a typist you are, you most likely repeat much the same process each of the 10 times you type your name.
Now, let's try to unravel the temporal “fingerprint” that you left behind when you typed your name. Suppose we conducted a simple analysis of the keystrokes and the time between each of the keystrokes that you made. The total time taken to type your name once is simply the time from the first keystroke to the last keystroke. At a more fine-grained level of analysis, the total time constitutes the time each key is held down (the dwell time) plus the time between the release of one key and the depression of the next (the transition time). All of the individual dwell times plus all of the transition times will sum to the total time.
In figure 7.5, I have plotted a hypothetical example of the 10 trials to type my name. The light gray bands denote the dwell times, and the darker gray bands indicate the transition times. I have ordered the lines of bands that represent the 10 trials from the fastest (least total time) at the top to the slowest trial at the bottom (most total time).
If you were to analyze each of the 10 repetitions of your name, you would probably find that all of the total times would be similar, but probably never exactly the same each time, just as in figure 7.5. The “noise” in our central nervous systems, plus other factors, causes the results to vary a little bit each time, resulting in some repetitions to be slower, and others faster, than the average total time.
But, take a closer look at each individual band in figure 7.5 as it changes over the 10 trials. What you will notice is that as the total time increases, each band gets proportionally longer too. We could express these numbers another way by dividing the time for each individual band by the total time for that trial to obtain a relative proportion of time represented by each band. Given the hypothetical evidence in figure 7.5, what we would discover is that the relative time for each band, expressed as a percentage, would stay roughly the same across all of the repetitions. Applying temporal and other methods to deconstruct how we type (e.g., key press forces) is called the study of keystroke dynamics.
According to generalized motor program theory, relative time is one of the key features of movement that is controlled by the central nervous system, especially for brief, rapid actions. Relative time is an expression of our motor control fingerprint. For well-learned tasks such as typing and handwriting, our central nervous system regulates the relative timing of impulses that are sent to the muscles that carry out these tasks. Various factors influence the real time on any particular instance. For example, using a keyboard that requires more force to depress the keys might result in overall slower times than would result using a keyboard that has a light touch, but the relative timing would likely remain the same.
Research in keystroke dynamics may result in the ability to identify people who carry out Internet fraud. In many ways it is similar to the use of handwriting dynamics to analyze the timing of the cursive expressions of a signature. It is fairly easy to forge the spatial representation of someone's signature, but very difficult to forge the temporal dynamics that result in that signature. These expressions of timing behavior represent our motor control fingerprints.
Self-Directed Learning Activities
1. Define the term generalized motor program in your own words.
2. How does the concept of a generalized motor program differ from the concept of a motor program as it was used in stories such as “Antilock Brakes” (in chapter 5) and “Point of No Return” (earlier in this
chapter)?
3. Suggest a method by which the handwriting dynamics of signatures could be used to detect fraud.
4. Some people contend that there is one generalized motor program for the full swing in golf, regardless of which club is used. How could a temporal dynamics analysis be used to assess this contention?
Notes
- An excellent review of fingerprint analysis by David Ashbaugh of the Royal Canadian Mounted Police is available here:
www.onin.com/fp/ridgeology.pdf
- A lot of controversy remains about invariances in motor performance; how they are measured; and what invariance, or lack of invariance, means in terms of motor programs. The following articles provide good arguments for the debate:
Gentner, D.R. (1987). Timing of skilled motor performance: Tests of the proportional duration model. Psychological Review, 94, 255-276.
Heuer, H. (1988). Testing the invariance of relative timing: Comment on Gentner (1987). Psychological Review, 95, 552-557.
Suggested Readings
Schmidt, R.A. (1975). A schema theory of discrete motor skill learning. Psychological Review, 82, 225-260.
Schmidt, R.A. (1985). The search for invariance in skilled movement behavior. Research Quarterly for Exercise and Sport, 56, 188-200.
Schmidt, R.A., & Lee, T.D. (2011). Central contributions to motor control. In Motor control and learning: A behavioral emphasis (5th ed., pp. 177-222) Champaign, IL: Human Kinetics.
The Farmers' Market
What are the roles of motor error and hypervigilance in unintended acceleration accidents?
The prosecution, at Weller's trial several years later, claimed it was no accident—that Weller deliberately drove his car through the crowded market. The reason, they said, was that Weller had been involved in a minor fender bender just moments before he entered the farmers' market. His response to the fender bender was to flee the scene of the accident. Witnesses for the prosecution painted Weller as a cold-blooded killer, commenting on the determined look on his face during the ordeal and his relatively calm demeanor afterward. It probably also didn't help his case when he emerged from the car immediately afterward and wondered aloud why the people he had hit had not jumped out of his way. Adding his age into the mix, Weller's actions were painted as rather pathetic.
Richard Schmidt, a renowned motor control scholar and human factors expert, testified on behalf of the defense team and argued that the facts of the case shared many similarities to accidents caused by errors of pedal misapplication, or unintended acceleration. Accidents of this type, which, thankfully, are quite rare, occur when the driver intends to apply pressure to the brake pedal, but misses the brake and pushes down on the accelerator instead.
Unintended acceleration accidents had been investigated for many years prior to the Weller case. These accidents were more common when the driver first got into the car, started the engine, and engaged the automatic transmission from the Park position into either Drive or Reverse. Indeed, such frequent episodes of “runaway cars” were the primary reason auto manufacturers added the brake-transmission shift interlock system in the 1980s so that an automatic transmission lever could not be moved from the Park position until the car sensed that a certain amount of pressure had been applied to the brake. However, the brake lock system only prevented this particular type of pedal misapplication; moving the foot to the accelerator instead of the brake would not be prevented once the transmission was successfully engaged out of Park and the car was in motion.
It is important to note that reaching for the brake requires that we steer the foot from a comfortable seated position to a target (the brake) in the absence of any visual guidance. We do this all the time without making any errors. We know where the brake is located from experience, and we also know that the brake feels different underfoot than does the accelerator. So then why would we suddenly miss the brake, push down on the accelerator instead, and then keep the foot there?
Schmidt (1989) presented evidence that unintended acceleration cases frequently involved accidents in which the driver had less experience than usual with that particular vehicle. Therefore, in some of these cases, the exact location and feel of the brake might not have been as familiar to the driver as normally could have been expected. These accidents also occurred more frequently on start-up, compared to later in the driving cycle, perhaps due to temporary factors associated with preparing an action (see “Shooting Two From the Line” in chapter 11). Driver inattention has also been linked with these cases, so it may not come as a surprise that drivers would not immediately notice the difference between the brake and gas pedal if engaged in a distracting activity at the same time (see “Gumbo” in chapter 6).
But for Weller, none of the common profiles for these accidents fit the case: he had already been driving before the supposed pedal misapplication error, so it was not a matter of missing the brake on initial start-up. Weller was quite familiar with his own vehicle, an 11-year-old Buick LeSabre. And he was not talking on a cell phone. Instead, Schmidt argued that Weller's pedal misapplication error was likely triggered by a catastrophic case of panic, termed hypervigilance, which could have been initiated when Weller had been involved in the fender bender just prior to the episode.
But one last issue seemed particularly problematic, according to the prosecution. Once the pedal misapplication error had occurred and the car started to accelerate wildly out of control, why didn't Weller simply remove his foot from the pedal or turn off the engine—actions that would have brought the car quickly to a stop? Again, failure to carry out corrective actions is typical of unintended acceleration cases, and some reasonable accounts have been offered to explain why drivers do not perform them. First, the driver probably does not realize that the foot is on the accelerator rather than the brake. The intention was to press the brake, and the fact that the pedal has gone all the way to the floor could reasonably be interpreted as brake failure rather than human error. And second, once the driver enters into this catastrophic state of panic, all normal modes of thinking cease. Reasoning and problem solving, the kinds of activities that are easy to do when unflustered, become unlikely, if not impossible, to carry out when in this state of hypervigilance.
Thomas Shelton, a member of the California Highway Patrol, testified at Weller's trial that he once investigated an unintended acceleration case in which an elderly woman ended up driving her car onto the top of another vehicle. The woman was in such a panicked state that when Shelton arrived at the accident scene, he had to climb up into the car to shut off the racing engine, at which time he noticed that the woman was still seated, very much alive, staring straight ahead with a death grip on the steering wheel, and with her foot still pushing the accelerator all the way to the floor.
Unfortunately, all of these arguments can only be used to speculate about what may have occurred in George Russell Weller's Buick LeSabre on that fateful day. On October 20, 2006, the jury convicted him of vehicular manslaughter in the 10 deaths resulting from the Santa Monica farmers' market crash. Nobody will ever know whether the verdict was the right one.
Self-Directed Learning Activities
1. In your own words describe the phenomenon known as unintended acceleration.
2. Describe a situation in which you made an action error that you were able to correct. How did you know that you had made the error, and what did you do to correct it?
3. Using our feet to manipulate car pedals involves aiming without visual feedback. What factors influence our ability to make these aiming movements accurately?
4. Propose a research methodology that examines one's ability to (a) move to a target without visual feedback and (b) estimate the accuracy of those aimed movements (again, without vision).
Notes
- Evidence from George Weller's trial during September and October of 2006 was summarized in the Santa Monica Daily Press, which can be accessed through its archives:
- Not all cases of unintended acceleration are generally agreed to be the result of a pedal misapplication. A segment of the television show 60 Minutes, hosted by Ed Bradley and which aired November 22, 1986, claimed that accidents of similar etiology involving the Audi 5000 were the result of a faulty idle stabilizer, which caused the car to accelerate wildly out of control when put into gear. An investigation by the U.S. NHTSA (National Highway Traffic Safety Administration) failed to support 60 Minutes' claim.
- More recently, runaway Toyotas have been a topic of concern. Once again, a government investigation failed to support a claim that these cases of unintended acceleration were due to an electronic fault in the engine. Again, this leaves open the very real possibility that driver error is to blame, as suggested by Richard Schmidt in the New York Times:
Suggested Readings
Castelli, J., Nash, C., Ditlow, C., & Pecht, M. (2003). Sudden acceleration: The myth of driver error. University of Maryland: CALCE EPSC Press.
Pollard, J., & Sussman, E.D. (1989). An examination of sudden acceleration. National Highway Traffic Safety Administration Final Report # DOT-TSC-NHTSA-89-1.
Schmidt, R.A. (1989). Unintended acceleration: A review of human factors contributions. Human Factors, 31, 345-364.
Schmidt, R.A. (1993). Unintended acceleration: Human performance considerations. In B. Peacock & W. Karwowski (Eds.), Automotive ergonomics (pp. 431-451). London: Taylor & Francis.
Schmidt, R.A., & Lee, T.D. (2011). Central contributions to motor control. In Motor control and learning: A behavioral emphasis (5th ed., pp. 177-222) Champaign, IL: Human Kinetics.
Push or Pull?
How do product designs influence people to perform specific actions?
It seems hard to believe, but brand-new buildings are still being constructed with very silly design features. For example, a new building on our campus opened last year and featured many modern technologies, including very nice-looking glass doors through which you enter and exit the building. The problem is that the doors have identical handles on either side. The handles themselves are very attractive, but they don't tell me what to do when I want to enter the building—should I push or pull the door? The handles on these particular doors suggest to me that they are made for grasping and pulling. However, the door goes in only one direction, and the very same handles appear on both sides of the door. Therefore, to enter the building, I have to use the handle to pull the door, and to exit, I have to use the identical type of handle to push the door. How would someone possibly know which way the door swings when approaching it? After a year of using these doors on a daily basis, I still just guess about whether to push or pull. Why would someone design a door like that?
Michael Darnell has created a wonderful website that illustrates many instances of products that have been designed without the user in mind. One of his web pages describes a door handle problem similar to the one I just described, in which a building has a short walkway with one set of doors on either end. Each door has identical handles on either side of the door (the same type of handle is used for pushing and pulling the door). However, according to Darnell, these particular doors were installed such that the problem was magnified even more—the doors at either end of the walkway both swung out from the walkway. He tells the story of a friend who entered the walkway by pulling on the handle of one door; then went to the second door and found that the door would not open when she pulled on the handle. So, concluding that the door was locked, she returned to the first door to exit and found herself trapped in the walkway when the door wouldn't move when she pulled on its handle. After trying to get the attention of others on the outside to tell them that she was trapped in the walkway, she finally realized, with much embarrassment, that the handles had to be pushed when inside the walkway. Darnell reminds us that this funny story could have had tragic consequences if, for example, a panicked person were trying to flee the building in case of a fire.
There are many ways to solve this problem of having identical handles used for opposite actions. One solution would be to put signs on the door that say “push” or “pull.” Although these signs would probably work, a well-designed door (i.e., designed for the user rather than for its looks) should not need a sign to tell people how to use it. Pushing and pulling a door are not rocket science, after all. A better solution would be to install a flat plate on one side of the door, which would make the correct choice obvious: Because there is no handle to grab to pull toward you, there is only one other option—push! A sign would be unnecessary in this case because there would be no question about which action to take.
Signs and symbols that tell us how to use things should be the last option in technology design. Signs are sometimes small and hard to read, they fall off, they wear out, other things cover them up, and they are often presented in only one language, making them less than desirable in a multicultural society. Symbols avoid being unilingual but introduce other problems, such as ambiguity. For example, how would one create simple, unambiguous, and instantly recognizable symbols for “push” and “pull”? As in the door handle example, a better design would be one that made the correct action as obvious as possible, one that would negate the need for a sign or symbol.
A classic example of the failure of technology to exploit this simple idea is a standard topic in ergonomics textbooks and on websites: the arrangement of burners on the stove top. A typical stove top layout is shown in figure 1.5.
The four knobs used to turn on the elements are located at the top of the figure. Quick, say out loud which one is used to turn on the right front burner. My guess is that it is one of the two rightmost knobs. But which one? It is impossible to know for certain without turning one of the knobs and waiting to see which element begins to heat up. So, designers add little labels to tell us which knobs are mapped to which burners. Do they work?
My personal experience is that these labels, if they have not fallen off, worn out, been covered by last night's pasta sauce, or been printed too small to read, are still hard to interpret. More times than I wish to remember, I have returned to a stove expecting to find a pot of boiling water for my pasta but instead found a glowing burner with nothing on it. The relative failure rate of these labels leads me to believe that they are only slightly more helpful than if I had just turned a knob at random.
Have a look at the two alternative stove top designs in figure 1.6. The layout of the knobs in figure 1.6a is the same as before; only the burner layout has been altered. In figure 1.6b, the burner layout is the same as it was before, but now the arrangement of the knobs has been altered.
In both cases, the mapping of each knob to its associated burner has been made obvious by a simple rearrangement of either the knobs or the burner layout. No symbols or labels are needed, and the chances of turning on the wrong burner are minimal. Moreover, the amount of space occupied by the burners and knobs in these more compatible layouts is identical to the space occupied in the incompatible layout. It is interesting to note that these alternative layouts have been available for many years. But have a look at any appliance store flyer or website, and you will see that the classic design still continues to dominate the market.
Sometimes it is simply not possible to avoid the use of labels or signs. In such cases the designer's goal should be to make them as informative as possible. The typical ceiling fan is another good example of a failure to consider the needs of the user when designing the product. The ceiling fan that we have in our home has a wall switch that turns the fan off and on. But the ceiling fan has three additional controls, located on the base of the fan itself (where the fan connects to the ceiling), that are operated by pulling on long strings that hang down from the base. These three strings can be pulled to adjust the speed of the fan, to reverse its direction, or to turn off or on a light located on the bottom of the fan. The problem is that the strings are identical in appearance; I have no way of knowing which control I am activating when I pull on any of the strings.
Actually, that is not completely true. If I stand on a stepladder and get up really close to the ceiling, I can see little labels beside the holes where the strings exit the electrical base of the fan that say “speed,” “light,” and “direction.” I could also stand on the floor and use binoculars to read these labels, I suppose. But, you get the point. Because these labels are next to useless to me, I really just have a one-in-three chance of getting it right each time I pull one of the strings.
How could each control be designed to communicate its function easily to the user? I have used two strategies to modify my own fan. Because reversing the fan's direction is the control that I use least often, I have shortened the string associated with this option to be the most difficult of the three to reach. However, I use the light and speed options about equally often. So, to make these as distinct as possible I borrowed two charms from an old charm bracelet and attached them to the ends of each string. A charm of a rabbit is attached to the string that controls the speed of the fan, and a charm of a book (for which I turn on the light to read) is attached to the string that controls the light. The fan may not look the way the manufacturer had intended it to look, but I have not made the mistake of pulling the wrong string since I attached these charms.
Motor skills such as pushing doors, turning knobs, and pulling strings are simple to learn, and we have all mastered these simple skills over our lifetimes. So why do we have such trouble using them? One reason is that the manufacturer is often paying more attention to aesthetics than to the needs of the user.
Self-Directed Learning Activities
1. Describe, in general, the principle of how product design influences human performance.
2. Search the literature for the term stimulus-response compatibility; then define it in your own words with specific reference to an example presented in this story.
3. Over the next 24 hours, keep a diary of all the objects or things you encounter that you think represent stimulus-response incompatibility. Propose ways each of these things could be made more compatible.
4. Find three research articles in which stimulus-response compatibility has been investigated. What are the similarities and differences in the stimuli and responses used in these studies?
Notes
- Michael Darnell has a wonderful website that features many examples of poorly designed products, with suggestions about simple ways to greatly improve usability: www.baddesigns.com
Suggested Readings
Proctor, R.W., & Van Zandt, T. (2008). Human factors in simple and complex systems (2nd ed.). Boca Raton, FL: CRC Press.
Proctor, R.W., & Vu, K.P.L. (2006). Stimulus-response compatibility principles: Data, theory, and application. Boca Raton, FL: CRC Press.
Schmidt, R.A., & Lee, T.D. (2011). Human information processing. In Motor control and learning: A behavioral emphasis (5th ed., pp. 57-96) Champaign, IL: Human Kinetics.
Learning to win from losing
Why is the learning–performance distinction important?
One day, Michelle Wie might be remembered as one of the greatest golfers of all time. At the age of 14 she was pounding out 300-yard (274 m) drives and not only competing in Ladies Professional (LPGA) Tour events, but also performing quite well in them. Her talent is undeniable. But controversy surrounded this exceptional athlete because of her desire to compete in PGA Tour events, which traditionally have included only male competitors. Wie has performed admirably in some of these events. For example, in the 2004 Sony Open, she shot rounds of 72 and 68 and missed the 36-hole cut by just a single stroke. Her two-round total of 140 placed her in a tie for 80th, which was better than 53 male professionals, including PGA stars Zach Johnson, Hunter Mahan, and Adam Scott. However, Wie has not performed well in most of the other PGA events she has entered, including a 139th-place finish (out of 141 players) at the 2007 Sony Open.
Many argued that Wie should stop competing in PGA Tour events and concentrate her efforts on the LPGA Tour. The typical argument was that her skill development as a golfer was being stalled by repeated failures in PGA Tour events, and that if she concentrated on LPGA events instead, the greater chance for success would escalate her skill development.
I argue that there is a fundamental flaw in this theory, because it falls into a trap known as the learning-performance distinction. The trap underlies one of the most misunderstood concepts in motor learning research, so let's start by defining the two terms. Motor skills researchers use the term performance to refer to a single observation. It could be a score or outcome that reflects the value of a single attempt at a motor skill, or perhaps an average score that statistically summarizes a number of attempts. An 18- or 36-hole score, or a final placement ranking in a tournament, might indicate a representative performance score in the case of a golfer. The term learning is used quite specifically to refer to a stable improvement in skill over time—an improvement that has specifically occurred as the result of practice. The problem illustrated by the criticisms leveled at Michelle Wie is that the critics fail to consider the difference between performance and learning; they mistakenly confuse her failures to improve her performance in men's PGA events as a failure to learn from these experiences. Let's use another example to illustrate this distinction between performance and learning.
Suppose you like to bowl, play regularly in a league, and also practice occasionally. Two years ago your average was 110, last year it was 120, and this year it is 130. Those averages probably indicate that you have become a better bowler as a result of learning: the scores indicate improvements that appear to be rather stable and that have resulted from practice.
So, does this mean that you will score a 130 the next time you bowl? Not necessarily, because there are many reasons for the fluctuations in scores that occur from game to game. Next time you may score well below your average because you are not feeling well, your shoes are too tight, or the crowd is very noisy that night. But this does not mean that you have suddenly lost some of your learned skill. You may score well above your average because you try extra hard to impress someone, or because everything just seems to be well focused (your mojo is working). As before, this does not mean that you have suddenly had a change in learned skill, because there is no indication that the sudden improvement in performance reflects a stable and permanent change in your capability to perform.
At any one time we have a theoretical capability for attaining a certain level of performance. When that theoretical capability changes to a higher level as a result of practice, we can say with confidence that we have learned. The confusion lies in the fact that individual performances may sometimes exceed or not live up to these theoretical capabilities. These fluctuations are expected and in no way diminish or detract from the performer's theoretical capability to perform at a certain skill level.
What I have described is the typical distinction between learning and performance. But, there is another side to this issue, the one that can be applied to Michelle Wie. The issue concerns the situation in which performances do not appear to change, or appear to be getting worse. Does that mean that learning is not occurring? Let's go back to the bowling example.
Suppose that this year your average was 120, which is the same as it was last year. You may take the absence of a stable improvement, despite all of your practice, as evidence that your learning has stalled. But, as it turns out, last year your league bowled Tuesday nights, which is your day off. This year you bowl on Wednesday nights, one hour after completing your 10-hour work shift. You are typically tired and hungry on league nights, and your bowling scores suffered this year as a result. What does all this suggest? What is your theoretical capability to bowl, and has this changed from last year, despite the change in your bowling night? Could it be that you actually have improved (i.e., learned), but the feeling of fatigue and other factors directly related to your work schedule prevented you from actually performing up to these expectations?
The point here is that making inferences about learning requires evidence that goes beyond a single performance or sometimes even a set of performances. Under what circumstances were these performances observed? Are there mitigating factors that might explain why the theoretical true score differs from the observed score? The absence of any observable change in performance does not mean that the unobservable (that theoretical capability to perform at a certain level of skill) has not improved.
Michelle Wie took a very different route in the development of her golfing skill. Her decision to compete in PGA events placed enormous pressures on her to perform, and only an exceptional result would have convinced the public that she was benefiting from these extreme challenges. All we could observe were her scores in these events. The unobservable, however, what she has learned by playing in these events, can never be truly understood. I suspect that if she achieves the rank of one of the greatest players of all time, it will be due in no small measure to these experiences competing against the very best golfers in the world.
Self-Directed Learning Activities
1. In your own words, explain the distinction between performance and learning.
2. What is the difference between a theoretical true score and an observed score?
3. Suggest a reason a person's observed score might exceed the true score, and a reason the observed score might fail to achieve the level of the true score.
4. Identify another sport, athlete, or situation in which the failure to account for the learning-performance distinction has resulted in an inappropriate conclusion.
Notes
- Babe Zaharias (see “The Babe”) was the first woman to compete in a men's PGA event.
- Here is a sampling of some of the criticism of Michele Wie's decision to compete in PGA Tour events, from two sports commentary sites:
- Michelle Wie won her second LPGA Tour title in Winnipeg, Manitoba, at the 2010 CN Canadian Women's Open.
Suggested Readings
Schmidt, R.A. (1972). The case against learning and forgetting scores. Journal of Motor Behavior, 4, 79-88.
Schmidt, R.A., & Lee, T.D. (2011). Motor learning concepts and research methods. In Motor control and learning: A behavioral emphasis (5th ed., pp. 327-346). Champaign, IL: Human Kinetics.
Cutting wood and missing putts
How are constant error, variable error, and absolute error useful for understanding motor control?
Two things I like to do—cutting wood and golfing—remind me often of the importance of measures of central tendency and variability in motor control performance. For example, I use my chainsaw to cut up fallen trees to burn in the woodstove. Cutting logs into burnable lengths is not an exact science. In the middle of a Canadian winter, when the temperature is cold and plenty of snow blankets the ground, it is generally considered a good idea to keep moving. I do not take time to measure the lengths of the logs I saw; rather, I make a quick visual estimate and then start to make my cut. But, in doing so, I must remember two important things related to the mean and the standard deviation of the logs I cut. First, because our woodstove is only large enough to hold pieces of wood that are 50 centimeters (20 in.) or shorter, any pieces that are longer than that are useless. And second, wood is easier to stack when the pieces are all roughly of the same length. Essentially, these features remind me that the mean and standard deviation are both important; I must have a mean that is less than 50 centimeters and a standard deviation that is as small as possible.
When considering data in terms of a specific goal or standard, it is often better to express the mean and standard deviation in terms of error measures—in my case, relative to a goal wood length of 50 centimeters. For example, wood lengths of 45, 42, 48, 47, and 52 centimeters would be expressed as lengths that are -5, -8, -2, -3, and +2 centimeters, respectively, relative to my goal wood length of 50 centimeters. If we calculated the mean of these lengths of wood, then we could express the stack of wood as having either an average length of 46.8 centimeters, or the error scores as having a mean constant error of -3.2 centimeters. Having a negative constant error is good in this case, because the average piece of wood that I have cut will easily fit in the woodstove (i.e., the mean is less than 50 centimeters).
The careful reader will have noted, however, that even though my mean constant error is negative, there is still one block of wood that will not fit (the 52 cm log). Thus, by itself, the mean constant error score (-3.2 cm) is misleading, because it does not provide any indication of how many logs might not actually fit in the woodstove. Our measure of variability, the standard deviation, gives us some indication of this information. The measure of variability of these error scores (called the variable error) represents the deviation of the error score for each log relative to the mean constant error and is calculated in the same way as is a standard deviation. In this case, the variable error is 3.3 centimeters. In general, when I am cutting wood, I want to achieve a negative mean constant error with a variable error that is as small as possible.
Although constant error provides a useful heuristic measure of average performance in cutting wood with a chainsaw, it is a quite misleading measure of central tendency in another activity I like to do. Let's say that I have struck 10 golf putts—5 of these putts go past the hole by 5, 10, 15, 20, and 25 centimeters, respectively, and the other five putts come up short of the hole by equal amounts (-5, -10, -15, -20 and -25 cm). On average, by how much have I missed the hole with these putts? If you were to calculate the average as in the previous wood-cutting example (the sum of the 10 individual error scores, divided by 10) the answer would be a mean constant error of 0. In other words, my “average” putt ended up in the hole. However, we know that this answer surely must be wrong because none of the individual putts actually went in the hole. So, we must use a different way to express these error scores to avoid an answer that makes no sense.
The problem that we sometimes run into with constant error is that it provides a measure of average bias, the tendency to err in a specific way (e.g., by too much or too little; too far left versus too far right). In some cases a specific bias is desirable, such as the tendency to undercut a wood length of 50 centimeters so that all of my wood pieces fit in the woodstove. In the case of my putts that have no consistent bias, we are much better off using a measure of central tendency that removes the bias from each score prior to calculating the mean. Such a procedure is called using the absolute (unsigned) scores. Hence, this measure of central tendency is called the mean absolute error. Here, our measure of central tendency is the mean of the unsigned scores (5, 5, 10, 10, 15, 15, 20, 20, 25, 25), which is 15 centimeters. That is, the golf putts tended to miss the hole by an average of 15 centimeters. In this particular case, the mean absolute error score (15 cm from the hole) represents a more accurate picture of the entire set of individual scores than what is represented in the mean constant error (in the hole).
Averages are convenient ways to express how we tend to perform. But, numbers that represent central tendencies can be misleading if the tendency itself is not a strong one. Measures of variability are one way to identify the strength of a tendency. Depending on the context, different methods of calculating central tendencies may characterize the data in more representative ways.
Self-Directed Learning Activities
1. Define constant error, absolute error, and variable error in your own words.
2. What do the terms algebraic error and total error (or Henry's E) refer to? How are they calculated, and how are they similar to the terms constant error and absolute error, respectively?
3. Look up and briefly describe a research investigation that uses at least two error measures. Why do you think the researchers used these measures in particular?
4. Conduct a brief study on yourself. Close your eyes and, using a pencil, draw five lines as close to 4 inches (10 cm) as possible, one below the previous one, without opening your eyes until you have drawn the fifth line. Measure and record the length of each line you drew; then calculate all of the error measures that have been discussed.
Notes
- Here is a web-based standard deviation calculator:
www.tinyurl.com/standarddeviationcalculator
Suggested Readings
Chapanis, A. (1951). Theory and methods for analyzing errors in man-machine systems. Annals of the New York Academy of Sciences, 51, 1179-1203.
Schmidt, R.A., & Lee, T.D. (2011). Methodology for studying motor performance. In Motor control and learning: A behavioral emphasis (5th ed., pp. 21-56). Champaign, IL: Human Kinetics.
Forensic motor control
What are generalized motor programs, and what do keystroke dynamics reveal about them?
Most of what I know about forensic identification comes from crime shows that I've seen on TV. The suspect is caught because a fingerprint was left on the murder weapon. The evidence is used to convict the suspect because no two fingerprints have ever been found to be identical. Crime shows tell us that other forensic methods can be used as well, such as DNA and eye retina data, but the fingerprint is the oldest biometric tool used for identification. However, another method that reveals a great deal about your identity is how you write or even type your name. For example, the password that you type to log on to your e-mail account may be just as identifiable as your fingerprint.
Try the following task as an example: go into any word processing program and type your name 10 times, once on each line, as I have done here:
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
I am not a skilled typist. I use my right middle finger to hold down the shift key and press the letter T with my left middle finger, then use my right middle finger to press i and my right index finger to press m, and then my right thumb to press the space bar. It is not very efficient, but I do it the same way every time I type my name. Regardless of how skilled a typist you are, you most likely repeat much the same process each of the 10 times you type your name.
Now, let's try to unravel the temporal “fingerprint” that you left behind when you typed your name. Suppose we conducted a simple analysis of the keystrokes and the time between each of the keystrokes that you made. The total time taken to type your name once is simply the time from the first keystroke to the last keystroke. At a more fine-grained level of analysis, the total time constitutes the time each key is held down (the dwell time) plus the time between the release of one key and the depression of the next (the transition time). All of the individual dwell times plus all of the transition times will sum to the total time.
In figure 7.5, I have plotted a hypothetical example of the 10 trials to type my name. The light gray bands denote the dwell times, and the darker gray bands indicate the transition times. I have ordered the lines of bands that represent the 10 trials from the fastest (least total time) at the top to the slowest trial at the bottom (most total time).
If you were to analyze each of the 10 repetitions of your name, you would probably find that all of the total times would be similar, but probably never exactly the same each time, just as in figure 7.5. The “noise” in our central nervous systems, plus other factors, causes the results to vary a little bit each time, resulting in some repetitions to be slower, and others faster, than the average total time.
But, take a closer look at each individual band in figure 7.5 as it changes over the 10 trials. What you will notice is that as the total time increases, each band gets proportionally longer too. We could express these numbers another way by dividing the time for each individual band by the total time for that trial to obtain a relative proportion of time represented by each band. Given the hypothetical evidence in figure 7.5, what we would discover is that the relative time for each band, expressed as a percentage, would stay roughly the same across all of the repetitions. Applying temporal and other methods to deconstruct how we type (e.g., key press forces) is called the study of keystroke dynamics.
According to generalized motor program theory, relative time is one of the key features of movement that is controlled by the central nervous system, especially for brief, rapid actions. Relative time is an expression of our motor control fingerprint. For well-learned tasks such as typing and handwriting, our central nervous system regulates the relative timing of impulses that are sent to the muscles that carry out these tasks. Various factors influence the real time on any particular instance. For example, using a keyboard that requires more force to depress the keys might result in overall slower times than would result using a keyboard that has a light touch, but the relative timing would likely remain the same.
Research in keystroke dynamics may result in the ability to identify people who carry out Internet fraud. In many ways it is similar to the use of handwriting dynamics to analyze the timing of the cursive expressions of a signature. It is fairly easy to forge the spatial representation of someone's signature, but very difficult to forge the temporal dynamics that result in that signature. These expressions of timing behavior represent our motor control fingerprints.
Self-Directed Learning Activities
1. Define the term generalized motor program in your own words.
2. How does the concept of a generalized motor program differ from the concept of a motor program as it was used in stories such as “Antilock Brakes” (in chapter 5) and “Point of No Return” (earlier in this
chapter)?
3. Suggest a method by which the handwriting dynamics of signatures could be used to detect fraud.
4. Some people contend that there is one generalized motor program for the full swing in golf, regardless of which club is used. How could a temporal dynamics analysis be used to assess this contention?
Notes
- An excellent review of fingerprint analysis by David Ashbaugh of the Royal Canadian Mounted Police is available here:
www.onin.com/fp/ridgeology.pdf
- A lot of controversy remains about invariances in motor performance; how they are measured; and what invariance, or lack of invariance, means in terms of motor programs. The following articles provide good arguments for the debate:
Gentner, D.R. (1987). Timing of skilled motor performance: Tests of the proportional duration model. Psychological Review, 94, 255-276.
Heuer, H. (1988). Testing the invariance of relative timing: Comment on Gentner (1987). Psychological Review, 95, 552-557.
Suggested Readings
Schmidt, R.A. (1975). A schema theory of discrete motor skill learning. Psychological Review, 82, 225-260.
Schmidt, R.A. (1985). The search for invariance in skilled movement behavior. Research Quarterly for Exercise and Sport, 56, 188-200.
Schmidt, R.A., & Lee, T.D. (2011). Central contributions to motor control. In Motor control and learning: A behavioral emphasis (5th ed., pp. 177-222) Champaign, IL: Human Kinetics.
The Farmers' Market
What are the roles of motor error and hypervigilance in unintended acceleration accidents?
The prosecution, at Weller's trial several years later, claimed it was no accident—that Weller deliberately drove his car through the crowded market. The reason, they said, was that Weller had been involved in a minor fender bender just moments before he entered the farmers' market. His response to the fender bender was to flee the scene of the accident. Witnesses for the prosecution painted Weller as a cold-blooded killer, commenting on the determined look on his face during the ordeal and his relatively calm demeanor afterward. It probably also didn't help his case when he emerged from the car immediately afterward and wondered aloud why the people he had hit had not jumped out of his way. Adding his age into the mix, Weller's actions were painted as rather pathetic.
Richard Schmidt, a renowned motor control scholar and human factors expert, testified on behalf of the defense team and argued that the facts of the case shared many similarities to accidents caused by errors of pedal misapplication, or unintended acceleration. Accidents of this type, which, thankfully, are quite rare, occur when the driver intends to apply pressure to the brake pedal, but misses the brake and pushes down on the accelerator instead.
Unintended acceleration accidents had been investigated for many years prior to the Weller case. These accidents were more common when the driver first got into the car, started the engine, and engaged the automatic transmission from the Park position into either Drive or Reverse. Indeed, such frequent episodes of “runaway cars” were the primary reason auto manufacturers added the brake-transmission shift interlock system in the 1980s so that an automatic transmission lever could not be moved from the Park position until the car sensed that a certain amount of pressure had been applied to the brake. However, the brake lock system only prevented this particular type of pedal misapplication; moving the foot to the accelerator instead of the brake would not be prevented once the transmission was successfully engaged out of Park and the car was in motion.
It is important to note that reaching for the brake requires that we steer the foot from a comfortable seated position to a target (the brake) in the absence of any visual guidance. We do this all the time without making any errors. We know where the brake is located from experience, and we also know that the brake feels different underfoot than does the accelerator. So then why would we suddenly miss the brake, push down on the accelerator instead, and then keep the foot there?
Schmidt (1989) presented evidence that unintended acceleration cases frequently involved accidents in which the driver had less experience than usual with that particular vehicle. Therefore, in some of these cases, the exact location and feel of the brake might not have been as familiar to the driver as normally could have been expected. These accidents also occurred more frequently on start-up, compared to later in the driving cycle, perhaps due to temporary factors associated with preparing an action (see “Shooting Two From the Line” in chapter 11). Driver inattention has also been linked with these cases, so it may not come as a surprise that drivers would not immediately notice the difference between the brake and gas pedal if engaged in a distracting activity at the same time (see “Gumbo” in chapter 6).
But for Weller, none of the common profiles for these accidents fit the case: he had already been driving before the supposed pedal misapplication error, so it was not a matter of missing the brake on initial start-up. Weller was quite familiar with his own vehicle, an 11-year-old Buick LeSabre. And he was not talking on a cell phone. Instead, Schmidt argued that Weller's pedal misapplication error was likely triggered by a catastrophic case of panic, termed hypervigilance, which could have been initiated when Weller had been involved in the fender bender just prior to the episode.
But one last issue seemed particularly problematic, according to the prosecution. Once the pedal misapplication error had occurred and the car started to accelerate wildly out of control, why didn't Weller simply remove his foot from the pedal or turn off the engine—actions that would have brought the car quickly to a stop? Again, failure to carry out corrective actions is typical of unintended acceleration cases, and some reasonable accounts have been offered to explain why drivers do not perform them. First, the driver probably does not realize that the foot is on the accelerator rather than the brake. The intention was to press the brake, and the fact that the pedal has gone all the way to the floor could reasonably be interpreted as brake failure rather than human error. And second, once the driver enters into this catastrophic state of panic, all normal modes of thinking cease. Reasoning and problem solving, the kinds of activities that are easy to do when unflustered, become unlikely, if not impossible, to carry out when in this state of hypervigilance.
Thomas Shelton, a member of the California Highway Patrol, testified at Weller's trial that he once investigated an unintended acceleration case in which an elderly woman ended up driving her car onto the top of another vehicle. The woman was in such a panicked state that when Shelton arrived at the accident scene, he had to climb up into the car to shut off the racing engine, at which time he noticed that the woman was still seated, very much alive, staring straight ahead with a death grip on the steering wheel, and with her foot still pushing the accelerator all the way to the floor.
Unfortunately, all of these arguments can only be used to speculate about what may have occurred in George Russell Weller's Buick LeSabre on that fateful day. On October 20, 2006, the jury convicted him of vehicular manslaughter in the 10 deaths resulting from the Santa Monica farmers' market crash. Nobody will ever know whether the verdict was the right one.
Self-Directed Learning Activities
1. In your own words describe the phenomenon known as unintended acceleration.
2. Describe a situation in which you made an action error that you were able to correct. How did you know that you had made the error, and what did you do to correct it?
3. Using our feet to manipulate car pedals involves aiming without visual feedback. What factors influence our ability to make these aiming movements accurately?
4. Propose a research methodology that examines one's ability to (a) move to a target without visual feedback and (b) estimate the accuracy of those aimed movements (again, without vision).
Notes
- Evidence from George Weller's trial during September and October of 2006 was summarized in the Santa Monica Daily Press, which can be accessed through its archives:
- Not all cases of unintended acceleration are generally agreed to be the result of a pedal misapplication. A segment of the television show 60 Minutes, hosted by Ed Bradley and which aired November 22, 1986, claimed that accidents of similar etiology involving the Audi 5000 were the result of a faulty idle stabilizer, which caused the car to accelerate wildly out of control when put into gear. An investigation by the U.S. NHTSA (National Highway Traffic Safety Administration) failed to support 60 Minutes' claim.
- More recently, runaway Toyotas have been a topic of concern. Once again, a government investigation failed to support a claim that these cases of unintended acceleration were due to an electronic fault in the engine. Again, this leaves open the very real possibility that driver error is to blame, as suggested by Richard Schmidt in the New York Times:
Suggested Readings
Castelli, J., Nash, C., Ditlow, C., & Pecht, M. (2003). Sudden acceleration: The myth of driver error. University of Maryland: CALCE EPSC Press.
Pollard, J., & Sussman, E.D. (1989). An examination of sudden acceleration. National Highway Traffic Safety Administration Final Report # DOT-TSC-NHTSA-89-1.
Schmidt, R.A. (1989). Unintended acceleration: A review of human factors contributions. Human Factors, 31, 345-364.
Schmidt, R.A. (1993). Unintended acceleration: Human performance considerations. In B. Peacock & W. Karwowski (Eds.), Automotive ergonomics (pp. 431-451). London: Taylor & Francis.
Schmidt, R.A., & Lee, T.D. (2011). Central contributions to motor control. In Motor control and learning: A behavioral emphasis (5th ed., pp. 177-222) Champaign, IL: Human Kinetics.
Push or Pull?
How do product designs influence people to perform specific actions?
It seems hard to believe, but brand-new buildings are still being constructed with very silly design features. For example, a new building on our campus opened last year and featured many modern technologies, including very nice-looking glass doors through which you enter and exit the building. The problem is that the doors have identical handles on either side. The handles themselves are very attractive, but they don't tell me what to do when I want to enter the building—should I push or pull the door? The handles on these particular doors suggest to me that they are made for grasping and pulling. However, the door goes in only one direction, and the very same handles appear on both sides of the door. Therefore, to enter the building, I have to use the handle to pull the door, and to exit, I have to use the identical type of handle to push the door. How would someone possibly know which way the door swings when approaching it? After a year of using these doors on a daily basis, I still just guess about whether to push or pull. Why would someone design a door like that?
Michael Darnell has created a wonderful website that illustrates many instances of products that have been designed without the user in mind. One of his web pages describes a door handle problem similar to the one I just described, in which a building has a short walkway with one set of doors on either end. Each door has identical handles on either side of the door (the same type of handle is used for pushing and pulling the door). However, according to Darnell, these particular doors were installed such that the problem was magnified even more—the doors at either end of the walkway both swung out from the walkway. He tells the story of a friend who entered the walkway by pulling on the handle of one door; then went to the second door and found that the door would not open when she pulled on the handle. So, concluding that the door was locked, she returned to the first door to exit and found herself trapped in the walkway when the door wouldn't move when she pulled on its handle. After trying to get the attention of others on the outside to tell them that she was trapped in the walkway, she finally realized, with much embarrassment, that the handles had to be pushed when inside the walkway. Darnell reminds us that this funny story could have had tragic consequences if, for example, a panicked person were trying to flee the building in case of a fire.
There are many ways to solve this problem of having identical handles used for opposite actions. One solution would be to put signs on the door that say “push” or “pull.” Although these signs would probably work, a well-designed door (i.e., designed for the user rather than for its looks) should not need a sign to tell people how to use it. Pushing and pulling a door are not rocket science, after all. A better solution would be to install a flat plate on one side of the door, which would make the correct choice obvious: Because there is no handle to grab to pull toward you, there is only one other option—push! A sign would be unnecessary in this case because there would be no question about which action to take.
Signs and symbols that tell us how to use things should be the last option in technology design. Signs are sometimes small and hard to read, they fall off, they wear out, other things cover them up, and they are often presented in only one language, making them less than desirable in a multicultural society. Symbols avoid being unilingual but introduce other problems, such as ambiguity. For example, how would one create simple, unambiguous, and instantly recognizable symbols for “push” and “pull”? As in the door handle example, a better design would be one that made the correct action as obvious as possible, one that would negate the need for a sign or symbol.
A classic example of the failure of technology to exploit this simple idea is a standard topic in ergonomics textbooks and on websites: the arrangement of burners on the stove top. A typical stove top layout is shown in figure 1.5.
The four knobs used to turn on the elements are located at the top of the figure. Quick, say out loud which one is used to turn on the right front burner. My guess is that it is one of the two rightmost knobs. But which one? It is impossible to know for certain without turning one of the knobs and waiting to see which element begins to heat up. So, designers add little labels to tell us which knobs are mapped to which burners. Do they work?
My personal experience is that these labels, if they have not fallen off, worn out, been covered by last night's pasta sauce, or been printed too small to read, are still hard to interpret. More times than I wish to remember, I have returned to a stove expecting to find a pot of boiling water for my pasta but instead found a glowing burner with nothing on it. The relative failure rate of these labels leads me to believe that they are only slightly more helpful than if I had just turned a knob at random.
Have a look at the two alternative stove top designs in figure 1.6. The layout of the knobs in figure 1.6a is the same as before; only the burner layout has been altered. In figure 1.6b, the burner layout is the same as it was before, but now the arrangement of the knobs has been altered.
In both cases, the mapping of each knob to its associated burner has been made obvious by a simple rearrangement of either the knobs or the burner layout. No symbols or labels are needed, and the chances of turning on the wrong burner are minimal. Moreover, the amount of space occupied by the burners and knobs in these more compatible layouts is identical to the space occupied in the incompatible layout. It is interesting to note that these alternative layouts have been available for many years. But have a look at any appliance store flyer or website, and you will see that the classic design still continues to dominate the market.
Sometimes it is simply not possible to avoid the use of labels or signs. In such cases the designer's goal should be to make them as informative as possible. The typical ceiling fan is another good example of a failure to consider the needs of the user when designing the product. The ceiling fan that we have in our home has a wall switch that turns the fan off and on. But the ceiling fan has three additional controls, located on the base of the fan itself (where the fan connects to the ceiling), that are operated by pulling on long strings that hang down from the base. These three strings can be pulled to adjust the speed of the fan, to reverse its direction, or to turn off or on a light located on the bottom of the fan. The problem is that the strings are identical in appearance; I have no way of knowing which control I am activating when I pull on any of the strings.
Actually, that is not completely true. If I stand on a stepladder and get up really close to the ceiling, I can see little labels beside the holes where the strings exit the electrical base of the fan that say “speed,” “light,” and “direction.” I could also stand on the floor and use binoculars to read these labels, I suppose. But, you get the point. Because these labels are next to useless to me, I really just have a one-in-three chance of getting it right each time I pull one of the strings.
How could each control be designed to communicate its function easily to the user? I have used two strategies to modify my own fan. Because reversing the fan's direction is the control that I use least often, I have shortened the string associated with this option to be the most difficult of the three to reach. However, I use the light and speed options about equally often. So, to make these as distinct as possible I borrowed two charms from an old charm bracelet and attached them to the ends of each string. A charm of a rabbit is attached to the string that controls the speed of the fan, and a charm of a book (for which I turn on the light to read) is attached to the string that controls the light. The fan may not look the way the manufacturer had intended it to look, but I have not made the mistake of pulling the wrong string since I attached these charms.
Motor skills such as pushing doors, turning knobs, and pulling strings are simple to learn, and we have all mastered these simple skills over our lifetimes. So why do we have such trouble using them? One reason is that the manufacturer is often paying more attention to aesthetics than to the needs of the user.
Self-Directed Learning Activities
1. Describe, in general, the principle of how product design influences human performance.
2. Search the literature for the term stimulus-response compatibility; then define it in your own words with specific reference to an example presented in this story.
3. Over the next 24 hours, keep a diary of all the objects or things you encounter that you think represent stimulus-response incompatibility. Propose ways each of these things could be made more compatible.
4. Find three research articles in which stimulus-response compatibility has been investigated. What are the similarities and differences in the stimuli and responses used in these studies?
Notes
- Michael Darnell has a wonderful website that features many examples of poorly designed products, with suggestions about simple ways to greatly improve usability: www.baddesigns.com
Suggested Readings
Proctor, R.W., & Van Zandt, T. (2008). Human factors in simple and complex systems (2nd ed.). Boca Raton, FL: CRC Press.
Proctor, R.W., & Vu, K.P.L. (2006). Stimulus-response compatibility principles: Data, theory, and application. Boca Raton, FL: CRC Press.
Schmidt, R.A., & Lee, T.D. (2011). Human information processing. In Motor control and learning: A behavioral emphasis (5th ed., pp. 57-96) Champaign, IL: Human Kinetics.
Learning to win from losing
Why is the learning–performance distinction important?
One day, Michelle Wie might be remembered as one of the greatest golfers of all time. At the age of 14 she was pounding out 300-yard (274 m) drives and not only competing in Ladies Professional (LPGA) Tour events, but also performing quite well in them. Her talent is undeniable. But controversy surrounded this exceptional athlete because of her desire to compete in PGA Tour events, which traditionally have included only male competitors. Wie has performed admirably in some of these events. For example, in the 2004 Sony Open, she shot rounds of 72 and 68 and missed the 36-hole cut by just a single stroke. Her two-round total of 140 placed her in a tie for 80th, which was better than 53 male professionals, including PGA stars Zach Johnson, Hunter Mahan, and Adam Scott. However, Wie has not performed well in most of the other PGA events she has entered, including a 139th-place finish (out of 141 players) at the 2007 Sony Open.
Many argued that Wie should stop competing in PGA Tour events and concentrate her efforts on the LPGA Tour. The typical argument was that her skill development as a golfer was being stalled by repeated failures in PGA Tour events, and that if she concentrated on LPGA events instead, the greater chance for success would escalate her skill development.
I argue that there is a fundamental flaw in this theory, because it falls into a trap known as the learning-performance distinction. The trap underlies one of the most misunderstood concepts in motor learning research, so let's start by defining the two terms. Motor skills researchers use the term performance to refer to a single observation. It could be a score or outcome that reflects the value of a single attempt at a motor skill, or perhaps an average score that statistically summarizes a number of attempts. An 18- or 36-hole score, or a final placement ranking in a tournament, might indicate a representative performance score in the case of a golfer. The term learning is used quite specifically to refer to a stable improvement in skill over time—an improvement that has specifically occurred as the result of practice. The problem illustrated by the criticisms leveled at Michelle Wie is that the critics fail to consider the difference between performance and learning; they mistakenly confuse her failures to improve her performance in men's PGA events as a failure to learn from these experiences. Let's use another example to illustrate this distinction between performance and learning.
Suppose you like to bowl, play regularly in a league, and also practice occasionally. Two years ago your average was 110, last year it was 120, and this year it is 130. Those averages probably indicate that you have become a better bowler as a result of learning: the scores indicate improvements that appear to be rather stable and that have resulted from practice.
So, does this mean that you will score a 130 the next time you bowl? Not necessarily, because there are many reasons for the fluctuations in scores that occur from game to game. Next time you may score well below your average because you are not feeling well, your shoes are too tight, or the crowd is very noisy that night. But this does not mean that you have suddenly lost some of your learned skill. You may score well above your average because you try extra hard to impress someone, or because everything just seems to be well focused (your mojo is working). As before, this does not mean that you have suddenly had a change in learned skill, because there is no indication that the sudden improvement in performance reflects a stable and permanent change in your capability to perform.
At any one time we have a theoretical capability for attaining a certain level of performance. When that theoretical capability changes to a higher level as a result of practice, we can say with confidence that we have learned. The confusion lies in the fact that individual performances may sometimes exceed or not live up to these theoretical capabilities. These fluctuations are expected and in no way diminish or detract from the performer's theoretical capability to perform at a certain skill level.
What I have described is the typical distinction between learning and performance. But, there is another side to this issue, the one that can be applied to Michelle Wie. The issue concerns the situation in which performances do not appear to change, or appear to be getting worse. Does that mean that learning is not occurring? Let's go back to the bowling example.
Suppose that this year your average was 120, which is the same as it was last year. You may take the absence of a stable improvement, despite all of your practice, as evidence that your learning has stalled. But, as it turns out, last year your league bowled Tuesday nights, which is your day off. This year you bowl on Wednesday nights, one hour after completing your 10-hour work shift. You are typically tired and hungry on league nights, and your bowling scores suffered this year as a result. What does all this suggest? What is your theoretical capability to bowl, and has this changed from last year, despite the change in your bowling night? Could it be that you actually have improved (i.e., learned), but the feeling of fatigue and other factors directly related to your work schedule prevented you from actually performing up to these expectations?
The point here is that making inferences about learning requires evidence that goes beyond a single performance or sometimes even a set of performances. Under what circumstances were these performances observed? Are there mitigating factors that might explain why the theoretical true score differs from the observed score? The absence of any observable change in performance does not mean that the unobservable (that theoretical capability to perform at a certain level of skill) has not improved.
Michelle Wie took a very different route in the development of her golfing skill. Her decision to compete in PGA events placed enormous pressures on her to perform, and only an exceptional result would have convinced the public that she was benefiting from these extreme challenges. All we could observe were her scores in these events. The unobservable, however, what she has learned by playing in these events, can never be truly understood. I suspect that if she achieves the rank of one of the greatest players of all time, it will be due in no small measure to these experiences competing against the very best golfers in the world.
Self-Directed Learning Activities
1. In your own words, explain the distinction between performance and learning.
2. What is the difference between a theoretical true score and an observed score?
3. Suggest a reason a person's observed score might exceed the true score, and a reason the observed score might fail to achieve the level of the true score.
4. Identify another sport, athlete, or situation in which the failure to account for the learning-performance distinction has resulted in an inappropriate conclusion.
Notes
- Babe Zaharias (see “The Babe”) was the first woman to compete in a men's PGA event.
- Here is a sampling of some of the criticism of Michele Wie's decision to compete in PGA Tour events, from two sports commentary sites:
- Michelle Wie won her second LPGA Tour title in Winnipeg, Manitoba, at the 2010 CN Canadian Women's Open.
Suggested Readings
Schmidt, R.A. (1972). The case against learning and forgetting scores. Journal of Motor Behavior, 4, 79-88.
Schmidt, R.A., & Lee, T.D. (2011). Motor learning concepts and research methods. In Motor control and learning: A behavioral emphasis (5th ed., pp. 327-346). Champaign, IL: Human Kinetics.
Cutting wood and missing putts
How are constant error, variable error, and absolute error useful for understanding motor control?
Two things I like to do—cutting wood and golfing—remind me often of the importance of measures of central tendency and variability in motor control performance. For example, I use my chainsaw to cut up fallen trees to burn in the woodstove. Cutting logs into burnable lengths is not an exact science. In the middle of a Canadian winter, when the temperature is cold and plenty of snow blankets the ground, it is generally considered a good idea to keep moving. I do not take time to measure the lengths of the logs I saw; rather, I make a quick visual estimate and then start to make my cut. But, in doing so, I must remember two important things related to the mean and the standard deviation of the logs I cut. First, because our woodstove is only large enough to hold pieces of wood that are 50 centimeters (20 in.) or shorter, any pieces that are longer than that are useless. And second, wood is easier to stack when the pieces are all roughly of the same length. Essentially, these features remind me that the mean and standard deviation are both important; I must have a mean that is less than 50 centimeters and a standard deviation that is as small as possible.
When considering data in terms of a specific goal or standard, it is often better to express the mean and standard deviation in terms of error measures—in my case, relative to a goal wood length of 50 centimeters. For example, wood lengths of 45, 42, 48, 47, and 52 centimeters would be expressed as lengths that are -5, -8, -2, -3, and +2 centimeters, respectively, relative to my goal wood length of 50 centimeters. If we calculated the mean of these lengths of wood, then we could express the stack of wood as having either an average length of 46.8 centimeters, or the error scores as having a mean constant error of -3.2 centimeters. Having a negative constant error is good in this case, because the average piece of wood that I have cut will easily fit in the woodstove (i.e., the mean is less than 50 centimeters).
The careful reader will have noted, however, that even though my mean constant error is negative, there is still one block of wood that will not fit (the 52 cm log). Thus, by itself, the mean constant error score (-3.2 cm) is misleading, because it does not provide any indication of how many logs might not actually fit in the woodstove. Our measure of variability, the standard deviation, gives us some indication of this information. The measure of variability of these error scores (called the variable error) represents the deviation of the error score for each log relative to the mean constant error and is calculated in the same way as is a standard deviation. In this case, the variable error is 3.3 centimeters. In general, when I am cutting wood, I want to achieve a negative mean constant error with a variable error that is as small as possible.
Although constant error provides a useful heuristic measure of average performance in cutting wood with a chainsaw, it is a quite misleading measure of central tendency in another activity I like to do. Let's say that I have struck 10 golf putts—5 of these putts go past the hole by 5, 10, 15, 20, and 25 centimeters, respectively, and the other five putts come up short of the hole by equal amounts (-5, -10, -15, -20 and -25 cm). On average, by how much have I missed the hole with these putts? If you were to calculate the average as in the previous wood-cutting example (the sum of the 10 individual error scores, divided by 10) the answer would be a mean constant error of 0. In other words, my “average” putt ended up in the hole. However, we know that this answer surely must be wrong because none of the individual putts actually went in the hole. So, we must use a different way to express these error scores to avoid an answer that makes no sense.
The problem that we sometimes run into with constant error is that it provides a measure of average bias, the tendency to err in a specific way (e.g., by too much or too little; too far left versus too far right). In some cases a specific bias is desirable, such as the tendency to undercut a wood length of 50 centimeters so that all of my wood pieces fit in the woodstove. In the case of my putts that have no consistent bias, we are much better off using a measure of central tendency that removes the bias from each score prior to calculating the mean. Such a procedure is called using the absolute (unsigned) scores. Hence, this measure of central tendency is called the mean absolute error. Here, our measure of central tendency is the mean of the unsigned scores (5, 5, 10, 10, 15, 15, 20, 20, 25, 25), which is 15 centimeters. That is, the golf putts tended to miss the hole by an average of 15 centimeters. In this particular case, the mean absolute error score (15 cm from the hole) represents a more accurate picture of the entire set of individual scores than what is represented in the mean constant error (in the hole).
Averages are convenient ways to express how we tend to perform. But, numbers that represent central tendencies can be misleading if the tendency itself is not a strong one. Measures of variability are one way to identify the strength of a tendency. Depending on the context, different methods of calculating central tendencies may characterize the data in more representative ways.
Self-Directed Learning Activities
1. Define constant error, absolute error, and variable error in your own words.
2. What do the terms algebraic error and total error (or Henry's E) refer to? How are they calculated, and how are they similar to the terms constant error and absolute error, respectively?
3. Look up and briefly describe a research investigation that uses at least two error measures. Why do you think the researchers used these measures in particular?
4. Conduct a brief study on yourself. Close your eyes and, using a pencil, draw five lines as close to 4 inches (10 cm) as possible, one below the previous one, without opening your eyes until you have drawn the fifth line. Measure and record the length of each line you drew; then calculate all of the error measures that have been discussed.
Notes
- Here is a web-based standard deviation calculator:
www.tinyurl.com/standarddeviationcalculator
Suggested Readings
Chapanis, A. (1951). Theory and methods for analyzing errors in man-machine systems. Annals of the New York Academy of Sciences, 51, 1179-1203.
Schmidt, R.A., & Lee, T.D. (2011). Methodology for studying motor performance. In Motor control and learning: A behavioral emphasis (5th ed., pp. 21-56). Champaign, IL: Human Kinetics.
Forensic motor control
What are generalized motor programs, and what do keystroke dynamics reveal about them?
Most of what I know about forensic identification comes from crime shows that I've seen on TV. The suspect is caught because a fingerprint was left on the murder weapon. The evidence is used to convict the suspect because no two fingerprints have ever been found to be identical. Crime shows tell us that other forensic methods can be used as well, such as DNA and eye retina data, but the fingerprint is the oldest biometric tool used for identification. However, another method that reveals a great deal about your identity is how you write or even type your name. For example, the password that you type to log on to your e-mail account may be just as identifiable as your fingerprint.
Try the following task as an example: go into any word processing program and type your name 10 times, once on each line, as I have done here:
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
I am not a skilled typist. I use my right middle finger to hold down the shift key and press the letter T with my left middle finger, then use my right middle finger to press i and my right index finger to press m, and then my right thumb to press the space bar. It is not very efficient, but I do it the same way every time I type my name. Regardless of how skilled a typist you are, you most likely repeat much the same process each of the 10 times you type your name.
Now, let's try to unravel the temporal “fingerprint” that you left behind when you typed your name. Suppose we conducted a simple analysis of the keystrokes and the time between each of the keystrokes that you made. The total time taken to type your name once is simply the time from the first keystroke to the last keystroke. At a more fine-grained level of analysis, the total time constitutes the time each key is held down (the dwell time) plus the time between the release of one key and the depression of the next (the transition time). All of the individual dwell times plus all of the transition times will sum to the total time.
In figure 7.5, I have plotted a hypothetical example of the 10 trials to type my name. The light gray bands denote the dwell times, and the darker gray bands indicate the transition times. I have ordered the lines of bands that represent the 10 trials from the fastest (least total time) at the top to the slowest trial at the bottom (most total time).
If you were to analyze each of the 10 repetitions of your name, you would probably find that all of the total times would be similar, but probably never exactly the same each time, just as in figure 7.5. The “noise” in our central nervous systems, plus other factors, causes the results to vary a little bit each time, resulting in some repetitions to be slower, and others faster, than the average total time.
But, take a closer look at each individual band in figure 7.5 as it changes over the 10 trials. What you will notice is that as the total time increases, each band gets proportionally longer too. We could express these numbers another way by dividing the time for each individual band by the total time for that trial to obtain a relative proportion of time represented by each band. Given the hypothetical evidence in figure 7.5, what we would discover is that the relative time for each band, expressed as a percentage, would stay roughly the same across all of the repetitions. Applying temporal and other methods to deconstruct how we type (e.g., key press forces) is called the study of keystroke dynamics.
According to generalized motor program theory, relative time is one of the key features of movement that is controlled by the central nervous system, especially for brief, rapid actions. Relative time is an expression of our motor control fingerprint. For well-learned tasks such as typing and handwriting, our central nervous system regulates the relative timing of impulses that are sent to the muscles that carry out these tasks. Various factors influence the real time on any particular instance. For example, using a keyboard that requires more force to depress the keys might result in overall slower times than would result using a keyboard that has a light touch, but the relative timing would likely remain the same.
Research in keystroke dynamics may result in the ability to identify people who carry out Internet fraud. In many ways it is similar to the use of handwriting dynamics to analyze the timing of the cursive expressions of a signature. It is fairly easy to forge the spatial representation of someone's signature, but very difficult to forge the temporal dynamics that result in that signature. These expressions of timing behavior represent our motor control fingerprints.
Self-Directed Learning Activities
1. Define the term generalized motor program in your own words.
2. How does the concept of a generalized motor program differ from the concept of a motor program as it was used in stories such as “Antilock Brakes” (in chapter 5) and “Point of No Return” (earlier in this
chapter)?
3. Suggest a method by which the handwriting dynamics of signatures could be used to detect fraud.
4. Some people contend that there is one generalized motor program for the full swing in golf, regardless of which club is used. How could a temporal dynamics analysis be used to assess this contention?
Notes
- An excellent review of fingerprint analysis by David Ashbaugh of the Royal Canadian Mounted Police is available here:
www.onin.com/fp/ridgeology.pdf
- A lot of controversy remains about invariances in motor performance; how they are measured; and what invariance, or lack of invariance, means in terms of motor programs. The following articles provide good arguments for the debate:
Gentner, D.R. (1987). Timing of skilled motor performance: Tests of the proportional duration model. Psychological Review, 94, 255-276.
Heuer, H. (1988). Testing the invariance of relative timing: Comment on Gentner (1987). Psychological Review, 95, 552-557.
Suggested Readings
Schmidt, R.A. (1975). A schema theory of discrete motor skill learning. Psychological Review, 82, 225-260.
Schmidt, R.A. (1985). The search for invariance in skilled movement behavior. Research Quarterly for Exercise and Sport, 56, 188-200.
Schmidt, R.A., & Lee, T.D. (2011). Central contributions to motor control. In Motor control and learning: A behavioral emphasis (5th ed., pp. 177-222) Champaign, IL: Human Kinetics.
The Farmers' Market
What are the roles of motor error and hypervigilance in unintended acceleration accidents?
The prosecution, at Weller's trial several years later, claimed it was no accident—that Weller deliberately drove his car through the crowded market. The reason, they said, was that Weller had been involved in a minor fender bender just moments before he entered the farmers' market. His response to the fender bender was to flee the scene of the accident. Witnesses for the prosecution painted Weller as a cold-blooded killer, commenting on the determined look on his face during the ordeal and his relatively calm demeanor afterward. It probably also didn't help his case when he emerged from the car immediately afterward and wondered aloud why the people he had hit had not jumped out of his way. Adding his age into the mix, Weller's actions were painted as rather pathetic.
Richard Schmidt, a renowned motor control scholar and human factors expert, testified on behalf of the defense team and argued that the facts of the case shared many similarities to accidents caused by errors of pedal misapplication, or unintended acceleration. Accidents of this type, which, thankfully, are quite rare, occur when the driver intends to apply pressure to the brake pedal, but misses the brake and pushes down on the accelerator instead.
Unintended acceleration accidents had been investigated for many years prior to the Weller case. These accidents were more common when the driver first got into the car, started the engine, and engaged the automatic transmission from the Park position into either Drive or Reverse. Indeed, such frequent episodes of “runaway cars” were the primary reason auto manufacturers added the brake-transmission shift interlock system in the 1980s so that an automatic transmission lever could not be moved from the Park position until the car sensed that a certain amount of pressure had been applied to the brake. However, the brake lock system only prevented this particular type of pedal misapplication; moving the foot to the accelerator instead of the brake would not be prevented once the transmission was successfully engaged out of Park and the car was in motion.
It is important to note that reaching for the brake requires that we steer the foot from a comfortable seated position to a target (the brake) in the absence of any visual guidance. We do this all the time without making any errors. We know where the brake is located from experience, and we also know that the brake feels different underfoot than does the accelerator. So then why would we suddenly miss the brake, push down on the accelerator instead, and then keep the foot there?
Schmidt (1989) presented evidence that unintended acceleration cases frequently involved accidents in which the driver had less experience than usual with that particular vehicle. Therefore, in some of these cases, the exact location and feel of the brake might not have been as familiar to the driver as normally could have been expected. These accidents also occurred more frequently on start-up, compared to later in the driving cycle, perhaps due to temporary factors associated with preparing an action (see “Shooting Two From the Line” in chapter 11). Driver inattention has also been linked with these cases, so it may not come as a surprise that drivers would not immediately notice the difference between the brake and gas pedal if engaged in a distracting activity at the same time (see “Gumbo” in chapter 6).
But for Weller, none of the common profiles for these accidents fit the case: he had already been driving before the supposed pedal misapplication error, so it was not a matter of missing the brake on initial start-up. Weller was quite familiar with his own vehicle, an 11-year-old Buick LeSabre. And he was not talking on a cell phone. Instead, Schmidt argued that Weller's pedal misapplication error was likely triggered by a catastrophic case of panic, termed hypervigilance, which could have been initiated when Weller had been involved in the fender bender just prior to the episode.
But one last issue seemed particularly problematic, according to the prosecution. Once the pedal misapplication error had occurred and the car started to accelerate wildly out of control, why didn't Weller simply remove his foot from the pedal or turn off the engine—actions that would have brought the car quickly to a stop? Again, failure to carry out corrective actions is typical of unintended acceleration cases, and some reasonable accounts have been offered to explain why drivers do not perform them. First, the driver probably does not realize that the foot is on the accelerator rather than the brake. The intention was to press the brake, and the fact that the pedal has gone all the way to the floor could reasonably be interpreted as brake failure rather than human error. And second, once the driver enters into this catastrophic state of panic, all normal modes of thinking cease. Reasoning and problem solving, the kinds of activities that are easy to do when unflustered, become unlikely, if not impossible, to carry out when in this state of hypervigilance.
Thomas Shelton, a member of the California Highway Patrol, testified at Weller's trial that he once investigated an unintended acceleration case in which an elderly woman ended up driving her car onto the top of another vehicle. The woman was in such a panicked state that when Shelton arrived at the accident scene, he had to climb up into the car to shut off the racing engine, at which time he noticed that the woman was still seated, very much alive, staring straight ahead with a death grip on the steering wheel, and with her foot still pushing the accelerator all the way to the floor.
Unfortunately, all of these arguments can only be used to speculate about what may have occurred in George Russell Weller's Buick LeSabre on that fateful day. On October 20, 2006, the jury convicted him of vehicular manslaughter in the 10 deaths resulting from the Santa Monica farmers' market crash. Nobody will ever know whether the verdict was the right one.
Self-Directed Learning Activities
1. In your own words describe the phenomenon known as unintended acceleration.
2. Describe a situation in which you made an action error that you were able to correct. How did you know that you had made the error, and what did you do to correct it?
3. Using our feet to manipulate car pedals involves aiming without visual feedback. What factors influence our ability to make these aiming movements accurately?
4. Propose a research methodology that examines one's ability to (a) move to a target without visual feedback and (b) estimate the accuracy of those aimed movements (again, without vision).
Notes
- Evidence from George Weller's trial during September and October of 2006 was summarized in the Santa Monica Daily Press, which can be accessed through its archives:
- Not all cases of unintended acceleration are generally agreed to be the result of a pedal misapplication. A segment of the television show 60 Minutes, hosted by Ed Bradley and which aired November 22, 1986, claimed that accidents of similar etiology involving the Audi 5000 were the result of a faulty idle stabilizer, which caused the car to accelerate wildly out of control when put into gear. An investigation by the U.S. NHTSA (National Highway Traffic Safety Administration) failed to support 60 Minutes' claim.
- More recently, runaway Toyotas have been a topic of concern. Once again, a government investigation failed to support a claim that these cases of unintended acceleration were due to an electronic fault in the engine. Again, this leaves open the very real possibility that driver error is to blame, as suggested by Richard Schmidt in the New York Times:
Suggested Readings
Castelli, J., Nash, C., Ditlow, C., & Pecht, M. (2003). Sudden acceleration: The myth of driver error. University of Maryland: CALCE EPSC Press.
Pollard, J., & Sussman, E.D. (1989). An examination of sudden acceleration. National Highway Traffic Safety Administration Final Report # DOT-TSC-NHTSA-89-1.
Schmidt, R.A. (1989). Unintended acceleration: A review of human factors contributions. Human Factors, 31, 345-364.
Schmidt, R.A. (1993). Unintended acceleration: Human performance considerations. In B. Peacock & W. Karwowski (Eds.), Automotive ergonomics (pp. 431-451). London: Taylor & Francis.
Schmidt, R.A., & Lee, T.D. (2011). Central contributions to motor control. In Motor control and learning: A behavioral emphasis (5th ed., pp. 177-222) Champaign, IL: Human Kinetics.
Push or Pull?
How do product designs influence people to perform specific actions?
It seems hard to believe, but brand-new buildings are still being constructed with very silly design features. For example, a new building on our campus opened last year and featured many modern technologies, including very nice-looking glass doors through which you enter and exit the building. The problem is that the doors have identical handles on either side. The handles themselves are very attractive, but they don't tell me what to do when I want to enter the building—should I push or pull the door? The handles on these particular doors suggest to me that they are made for grasping and pulling. However, the door goes in only one direction, and the very same handles appear on both sides of the door. Therefore, to enter the building, I have to use the handle to pull the door, and to exit, I have to use the identical type of handle to push the door. How would someone possibly know which way the door swings when approaching it? After a year of using these doors on a daily basis, I still just guess about whether to push or pull. Why would someone design a door like that?
Michael Darnell has created a wonderful website that illustrates many instances of products that have been designed without the user in mind. One of his web pages describes a door handle problem similar to the one I just described, in which a building has a short walkway with one set of doors on either end. Each door has identical handles on either side of the door (the same type of handle is used for pushing and pulling the door). However, according to Darnell, these particular doors were installed such that the problem was magnified even more—the doors at either end of the walkway both swung out from the walkway. He tells the story of a friend who entered the walkway by pulling on the handle of one door; then went to the second door and found that the door would not open when she pulled on the handle. So, concluding that the door was locked, she returned to the first door to exit and found herself trapped in the walkway when the door wouldn't move when she pulled on its handle. After trying to get the attention of others on the outside to tell them that she was trapped in the walkway, she finally realized, with much embarrassment, that the handles had to be pushed when inside the walkway. Darnell reminds us that this funny story could have had tragic consequences if, for example, a panicked person were trying to flee the building in case of a fire.
There are many ways to solve this problem of having identical handles used for opposite actions. One solution would be to put signs on the door that say “push” or “pull.” Although these signs would probably work, a well-designed door (i.e., designed for the user rather than for its looks) should not need a sign to tell people how to use it. Pushing and pulling a door are not rocket science, after all. A better solution would be to install a flat plate on one side of the door, which would make the correct choice obvious: Because there is no handle to grab to pull toward you, there is only one other option—push! A sign would be unnecessary in this case because there would be no question about which action to take.
Signs and symbols that tell us how to use things should be the last option in technology design. Signs are sometimes small and hard to read, they fall off, they wear out, other things cover them up, and they are often presented in only one language, making them less than desirable in a multicultural society. Symbols avoid being unilingual but introduce other problems, such as ambiguity. For example, how would one create simple, unambiguous, and instantly recognizable symbols for “push” and “pull”? As in the door handle example, a better design would be one that made the correct action as obvious as possible, one that would negate the need for a sign or symbol.
A classic example of the failure of technology to exploit this simple idea is a standard topic in ergonomics textbooks and on websites: the arrangement of burners on the stove top. A typical stove top layout is shown in figure 1.5.
The four knobs used to turn on the elements are located at the top of the figure. Quick, say out loud which one is used to turn on the right front burner. My guess is that it is one of the two rightmost knobs. But which one? It is impossible to know for certain without turning one of the knobs and waiting to see which element begins to heat up. So, designers add little labels to tell us which knobs are mapped to which burners. Do they work?
My personal experience is that these labels, if they have not fallen off, worn out, been covered by last night's pasta sauce, or been printed too small to read, are still hard to interpret. More times than I wish to remember, I have returned to a stove expecting to find a pot of boiling water for my pasta but instead found a glowing burner with nothing on it. The relative failure rate of these labels leads me to believe that they are only slightly more helpful than if I had just turned a knob at random.
Have a look at the two alternative stove top designs in figure 1.6. The layout of the knobs in figure 1.6a is the same as before; only the burner layout has been altered. In figure 1.6b, the burner layout is the same as it was before, but now the arrangement of the knobs has been altered.
In both cases, the mapping of each knob to its associated burner has been made obvious by a simple rearrangement of either the knobs or the burner layout. No symbols or labels are needed, and the chances of turning on the wrong burner are minimal. Moreover, the amount of space occupied by the burners and knobs in these more compatible layouts is identical to the space occupied in the incompatible layout. It is interesting to note that these alternative layouts have been available for many years. But have a look at any appliance store flyer or website, and you will see that the classic design still continues to dominate the market.
Sometimes it is simply not possible to avoid the use of labels or signs. In such cases the designer's goal should be to make them as informative as possible. The typical ceiling fan is another good example of a failure to consider the needs of the user when designing the product. The ceiling fan that we have in our home has a wall switch that turns the fan off and on. But the ceiling fan has three additional controls, located on the base of the fan itself (where the fan connects to the ceiling), that are operated by pulling on long strings that hang down from the base. These three strings can be pulled to adjust the speed of the fan, to reverse its direction, or to turn off or on a light located on the bottom of the fan. The problem is that the strings are identical in appearance; I have no way of knowing which control I am activating when I pull on any of the strings.
Actually, that is not completely true. If I stand on a stepladder and get up really close to the ceiling, I can see little labels beside the holes where the strings exit the electrical base of the fan that say “speed,” “light,” and “direction.” I could also stand on the floor and use binoculars to read these labels, I suppose. But, you get the point. Because these labels are next to useless to me, I really just have a one-in-three chance of getting it right each time I pull one of the strings.
How could each control be designed to communicate its function easily to the user? I have used two strategies to modify my own fan. Because reversing the fan's direction is the control that I use least often, I have shortened the string associated with this option to be the most difficult of the three to reach. However, I use the light and speed options about equally often. So, to make these as distinct as possible I borrowed two charms from an old charm bracelet and attached them to the ends of each string. A charm of a rabbit is attached to the string that controls the speed of the fan, and a charm of a book (for which I turn on the light to read) is attached to the string that controls the light. The fan may not look the way the manufacturer had intended it to look, but I have not made the mistake of pulling the wrong string since I attached these charms.
Motor skills such as pushing doors, turning knobs, and pulling strings are simple to learn, and we have all mastered these simple skills over our lifetimes. So why do we have such trouble using them? One reason is that the manufacturer is often paying more attention to aesthetics than to the needs of the user.
Self-Directed Learning Activities
1. Describe, in general, the principle of how product design influences human performance.
2. Search the literature for the term stimulus-response compatibility; then define it in your own words with specific reference to an example presented in this story.
3. Over the next 24 hours, keep a diary of all the objects or things you encounter that you think represent stimulus-response incompatibility. Propose ways each of these things could be made more compatible.
4. Find three research articles in which stimulus-response compatibility has been investigated. What are the similarities and differences in the stimuli and responses used in these studies?
Notes
- Michael Darnell has a wonderful website that features many examples of poorly designed products, with suggestions about simple ways to greatly improve usability: www.baddesigns.com
Suggested Readings
Proctor, R.W., & Van Zandt, T. (2008). Human factors in simple and complex systems (2nd ed.). Boca Raton, FL: CRC Press.
Proctor, R.W., & Vu, K.P.L. (2006). Stimulus-response compatibility principles: Data, theory, and application. Boca Raton, FL: CRC Press.
Schmidt, R.A., & Lee, T.D. (2011). Human information processing. In Motor control and learning: A behavioral emphasis (5th ed., pp. 57-96) Champaign, IL: Human Kinetics.
Learning to win from losing
Why is the learning–performance distinction important?
One day, Michelle Wie might be remembered as one of the greatest golfers of all time. At the age of 14 she was pounding out 300-yard (274 m) drives and not only competing in Ladies Professional (LPGA) Tour events, but also performing quite well in them. Her talent is undeniable. But controversy surrounded this exceptional athlete because of her desire to compete in PGA Tour events, which traditionally have included only male competitors. Wie has performed admirably in some of these events. For example, in the 2004 Sony Open, she shot rounds of 72 and 68 and missed the 36-hole cut by just a single stroke. Her two-round total of 140 placed her in a tie for 80th, which was better than 53 male professionals, including PGA stars Zach Johnson, Hunter Mahan, and Adam Scott. However, Wie has not performed well in most of the other PGA events she has entered, including a 139th-place finish (out of 141 players) at the 2007 Sony Open.
Many argued that Wie should stop competing in PGA Tour events and concentrate her efforts on the LPGA Tour. The typical argument was that her skill development as a golfer was being stalled by repeated failures in PGA Tour events, and that if she concentrated on LPGA events instead, the greater chance for success would escalate her skill development.
I argue that there is a fundamental flaw in this theory, because it falls into a trap known as the learning-performance distinction. The trap underlies one of the most misunderstood concepts in motor learning research, so let's start by defining the two terms. Motor skills researchers use the term performance to refer to a single observation. It could be a score or outcome that reflects the value of a single attempt at a motor skill, or perhaps an average score that statistically summarizes a number of attempts. An 18- or 36-hole score, or a final placement ranking in a tournament, might indicate a representative performance score in the case of a golfer. The term learning is used quite specifically to refer to a stable improvement in skill over time—an improvement that has specifically occurred as the result of practice. The problem illustrated by the criticisms leveled at Michelle Wie is that the critics fail to consider the difference between performance and learning; they mistakenly confuse her failures to improve her performance in men's PGA events as a failure to learn from these experiences. Let's use another example to illustrate this distinction between performance and learning.
Suppose you like to bowl, play regularly in a league, and also practice occasionally. Two years ago your average was 110, last year it was 120, and this year it is 130. Those averages probably indicate that you have become a better bowler as a result of learning: the scores indicate improvements that appear to be rather stable and that have resulted from practice.
So, does this mean that you will score a 130 the next time you bowl? Not necessarily, because there are many reasons for the fluctuations in scores that occur from game to game. Next time you may score well below your average because you are not feeling well, your shoes are too tight, or the crowd is very noisy that night. But this does not mean that you have suddenly lost some of your learned skill. You may score well above your average because you try extra hard to impress someone, or because everything just seems to be well focused (your mojo is working). As before, this does not mean that you have suddenly had a change in learned skill, because there is no indication that the sudden improvement in performance reflects a stable and permanent change in your capability to perform.
At any one time we have a theoretical capability for attaining a certain level of performance. When that theoretical capability changes to a higher level as a result of practice, we can say with confidence that we have learned. The confusion lies in the fact that individual performances may sometimes exceed or not live up to these theoretical capabilities. These fluctuations are expected and in no way diminish or detract from the performer's theoretical capability to perform at a certain skill level.
What I have described is the typical distinction between learning and performance. But, there is another side to this issue, the one that can be applied to Michelle Wie. The issue concerns the situation in which performances do not appear to change, or appear to be getting worse. Does that mean that learning is not occurring? Let's go back to the bowling example.
Suppose that this year your average was 120, which is the same as it was last year. You may take the absence of a stable improvement, despite all of your practice, as evidence that your learning has stalled. But, as it turns out, last year your league bowled Tuesday nights, which is your day off. This year you bowl on Wednesday nights, one hour after completing your 10-hour work shift. You are typically tired and hungry on league nights, and your bowling scores suffered this year as a result. What does all this suggest? What is your theoretical capability to bowl, and has this changed from last year, despite the change in your bowling night? Could it be that you actually have improved (i.e., learned), but the feeling of fatigue and other factors directly related to your work schedule prevented you from actually performing up to these expectations?
The point here is that making inferences about learning requires evidence that goes beyond a single performance or sometimes even a set of performances. Under what circumstances were these performances observed? Are there mitigating factors that might explain why the theoretical true score differs from the observed score? The absence of any observable change in performance does not mean that the unobservable (that theoretical capability to perform at a certain level of skill) has not improved.
Michelle Wie took a very different route in the development of her golfing skill. Her decision to compete in PGA events placed enormous pressures on her to perform, and only an exceptional result would have convinced the public that she was benefiting from these extreme challenges. All we could observe were her scores in these events. The unobservable, however, what she has learned by playing in these events, can never be truly understood. I suspect that if she achieves the rank of one of the greatest players of all time, it will be due in no small measure to these experiences competing against the very best golfers in the world.
Self-Directed Learning Activities
1. In your own words, explain the distinction between performance and learning.
2. What is the difference between a theoretical true score and an observed score?
3. Suggest a reason a person's observed score might exceed the true score, and a reason the observed score might fail to achieve the level of the true score.
4. Identify another sport, athlete, or situation in which the failure to account for the learning-performance distinction has resulted in an inappropriate conclusion.
Notes
- Babe Zaharias (see “The Babe”) was the first woman to compete in a men's PGA event.
- Here is a sampling of some of the criticism of Michele Wie's decision to compete in PGA Tour events, from two sports commentary sites:
- Michelle Wie won her second LPGA Tour title in Winnipeg, Manitoba, at the 2010 CN Canadian Women's Open.
Suggested Readings
Schmidt, R.A. (1972). The case against learning and forgetting scores. Journal of Motor Behavior, 4, 79-88.
Schmidt, R.A., & Lee, T.D. (2011). Motor learning concepts and research methods. In Motor control and learning: A behavioral emphasis (5th ed., pp. 327-346). Champaign, IL: Human Kinetics.
Cutting wood and missing putts
How are constant error, variable error, and absolute error useful for understanding motor control?
Two things I like to do—cutting wood and golfing—remind me often of the importance of measures of central tendency and variability in motor control performance. For example, I use my chainsaw to cut up fallen trees to burn in the woodstove. Cutting logs into burnable lengths is not an exact science. In the middle of a Canadian winter, when the temperature is cold and plenty of snow blankets the ground, it is generally considered a good idea to keep moving. I do not take time to measure the lengths of the logs I saw; rather, I make a quick visual estimate and then start to make my cut. But, in doing so, I must remember two important things related to the mean and the standard deviation of the logs I cut. First, because our woodstove is only large enough to hold pieces of wood that are 50 centimeters (20 in.) or shorter, any pieces that are longer than that are useless. And second, wood is easier to stack when the pieces are all roughly of the same length. Essentially, these features remind me that the mean and standard deviation are both important; I must have a mean that is less than 50 centimeters and a standard deviation that is as small as possible.
When considering data in terms of a specific goal or standard, it is often better to express the mean and standard deviation in terms of error measures—in my case, relative to a goal wood length of 50 centimeters. For example, wood lengths of 45, 42, 48, 47, and 52 centimeters would be expressed as lengths that are -5, -8, -2, -3, and +2 centimeters, respectively, relative to my goal wood length of 50 centimeters. If we calculated the mean of these lengths of wood, then we could express the stack of wood as having either an average length of 46.8 centimeters, or the error scores as having a mean constant error of -3.2 centimeters. Having a negative constant error is good in this case, because the average piece of wood that I have cut will easily fit in the woodstove (i.e., the mean is less than 50 centimeters).
The careful reader will have noted, however, that even though my mean constant error is negative, there is still one block of wood that will not fit (the 52 cm log). Thus, by itself, the mean constant error score (-3.2 cm) is misleading, because it does not provide any indication of how many logs might not actually fit in the woodstove. Our measure of variability, the standard deviation, gives us some indication of this information. The measure of variability of these error scores (called the variable error) represents the deviation of the error score for each log relative to the mean constant error and is calculated in the same way as is a standard deviation. In this case, the variable error is 3.3 centimeters. In general, when I am cutting wood, I want to achieve a negative mean constant error with a variable error that is as small as possible.
Although constant error provides a useful heuristic measure of average performance in cutting wood with a chainsaw, it is a quite misleading measure of central tendency in another activity I like to do. Let's say that I have struck 10 golf putts—5 of these putts go past the hole by 5, 10, 15, 20, and 25 centimeters, respectively, and the other five putts come up short of the hole by equal amounts (-5, -10, -15, -20 and -25 cm). On average, by how much have I missed the hole with these putts? If you were to calculate the average as in the previous wood-cutting example (the sum of the 10 individual error scores, divided by 10) the answer would be a mean constant error of 0. In other words, my “average” putt ended up in the hole. However, we know that this answer surely must be wrong because none of the individual putts actually went in the hole. So, we must use a different way to express these error scores to avoid an answer that makes no sense.
The problem that we sometimes run into with constant error is that it provides a measure of average bias, the tendency to err in a specific way (e.g., by too much or too little; too far left versus too far right). In some cases a specific bias is desirable, such as the tendency to undercut a wood length of 50 centimeters so that all of my wood pieces fit in the woodstove. In the case of my putts that have no consistent bias, we are much better off using a measure of central tendency that removes the bias from each score prior to calculating the mean. Such a procedure is called using the absolute (unsigned) scores. Hence, this measure of central tendency is called the mean absolute error. Here, our measure of central tendency is the mean of the unsigned scores (5, 5, 10, 10, 15, 15, 20, 20, 25, 25), which is 15 centimeters. That is, the golf putts tended to miss the hole by an average of 15 centimeters. In this particular case, the mean absolute error score (15 cm from the hole) represents a more accurate picture of the entire set of individual scores than what is represented in the mean constant error (in the hole).
Averages are convenient ways to express how we tend to perform. But, numbers that represent central tendencies can be misleading if the tendency itself is not a strong one. Measures of variability are one way to identify the strength of a tendency. Depending on the context, different methods of calculating central tendencies may characterize the data in more representative ways.
Self-Directed Learning Activities
1. Define constant error, absolute error, and variable error in your own words.
2. What do the terms algebraic error and total error (or Henry's E) refer to? How are they calculated, and how are they similar to the terms constant error and absolute error, respectively?
3. Look up and briefly describe a research investigation that uses at least two error measures. Why do you think the researchers used these measures in particular?
4. Conduct a brief study on yourself. Close your eyes and, using a pencil, draw five lines as close to 4 inches (10 cm) as possible, one below the previous one, without opening your eyes until you have drawn the fifth line. Measure and record the length of each line you drew; then calculate all of the error measures that have been discussed.
Notes
- Here is a web-based standard deviation calculator:
www.tinyurl.com/standarddeviationcalculator
Suggested Readings
Chapanis, A. (1951). Theory and methods for analyzing errors in man-machine systems. Annals of the New York Academy of Sciences, 51, 1179-1203.
Schmidt, R.A., & Lee, T.D. (2011). Methodology for studying motor performance. In Motor control and learning: A behavioral emphasis (5th ed., pp. 21-56). Champaign, IL: Human Kinetics.
Forensic motor control
What are generalized motor programs, and what do keystroke dynamics reveal about them?
Most of what I know about forensic identification comes from crime shows that I've seen on TV. The suspect is caught because a fingerprint was left on the murder weapon. The evidence is used to convict the suspect because no two fingerprints have ever been found to be identical. Crime shows tell us that other forensic methods can be used as well, such as DNA and eye retina data, but the fingerprint is the oldest biometric tool used for identification. However, another method that reveals a great deal about your identity is how you write or even type your name. For example, the password that you type to log on to your e-mail account may be just as identifiable as your fingerprint.
Try the following task as an example: go into any word processing program and type your name 10 times, once on each line, as I have done here:
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
I am not a skilled typist. I use my right middle finger to hold down the shift key and press the letter T with my left middle finger, then use my right middle finger to press i and my right index finger to press m, and then my right thumb to press the space bar. It is not very efficient, but I do it the same way every time I type my name. Regardless of how skilled a typist you are, you most likely repeat much the same process each of the 10 times you type your name.
Now, let's try to unravel the temporal “fingerprint” that you left behind when you typed your name. Suppose we conducted a simple analysis of the keystrokes and the time between each of the keystrokes that you made. The total time taken to type your name once is simply the time from the first keystroke to the last keystroke. At a more fine-grained level of analysis, the total time constitutes the time each key is held down (the dwell time) plus the time between the release of one key and the depression of the next (the transition time). All of the individual dwell times plus all of the transition times will sum to the total time.
In figure 7.5, I have plotted a hypothetical example of the 10 trials to type my name. The light gray bands denote the dwell times, and the darker gray bands indicate the transition times. I have ordered the lines of bands that represent the 10 trials from the fastest (least total time) at the top to the slowest trial at the bottom (most total time).
If you were to analyze each of the 10 repetitions of your name, you would probably find that all of the total times would be similar, but probably never exactly the same each time, just as in figure 7.5. The “noise” in our central nervous systems, plus other factors, causes the results to vary a little bit each time, resulting in some repetitions to be slower, and others faster, than the average total time.
But, take a closer look at each individual band in figure 7.5 as it changes over the 10 trials. What you will notice is that as the total time increases, each band gets proportionally longer too. We could express these numbers another way by dividing the time for each individual band by the total time for that trial to obtain a relative proportion of time represented by each band. Given the hypothetical evidence in figure 7.5, what we would discover is that the relative time for each band, expressed as a percentage, would stay roughly the same across all of the repetitions. Applying temporal and other methods to deconstruct how we type (e.g., key press forces) is called the study of keystroke dynamics.
According to generalized motor program theory, relative time is one of the key features of movement that is controlled by the central nervous system, especially for brief, rapid actions. Relative time is an expression of our motor control fingerprint. For well-learned tasks such as typing and handwriting, our central nervous system regulates the relative timing of impulses that are sent to the muscles that carry out these tasks. Various factors influence the real time on any particular instance. For example, using a keyboard that requires more force to depress the keys might result in overall slower times than would result using a keyboard that has a light touch, but the relative timing would likely remain the same.
Research in keystroke dynamics may result in the ability to identify people who carry out Internet fraud. In many ways it is similar to the use of handwriting dynamics to analyze the timing of the cursive expressions of a signature. It is fairly easy to forge the spatial representation of someone's signature, but very difficult to forge the temporal dynamics that result in that signature. These expressions of timing behavior represent our motor control fingerprints.
Self-Directed Learning Activities
1. Define the term generalized motor program in your own words.
2. How does the concept of a generalized motor program differ from the concept of a motor program as it was used in stories such as “Antilock Brakes” (in chapter 5) and “Point of No Return” (earlier in this
chapter)?
3. Suggest a method by which the handwriting dynamics of signatures could be used to detect fraud.
4. Some people contend that there is one generalized motor program for the full swing in golf, regardless of which club is used. How could a temporal dynamics analysis be used to assess this contention?
Notes
- An excellent review of fingerprint analysis by David Ashbaugh of the Royal Canadian Mounted Police is available here:
www.onin.com/fp/ridgeology.pdf
- A lot of controversy remains about invariances in motor performance; how they are measured; and what invariance, or lack of invariance, means in terms of motor programs. The following articles provide good arguments for the debate:
Gentner, D.R. (1987). Timing of skilled motor performance: Tests of the proportional duration model. Psychological Review, 94, 255-276.
Heuer, H. (1988). Testing the invariance of relative timing: Comment on Gentner (1987). Psychological Review, 95, 552-557.
Suggested Readings
Schmidt, R.A. (1975). A schema theory of discrete motor skill learning. Psychological Review, 82, 225-260.
Schmidt, R.A. (1985). The search for invariance in skilled movement behavior. Research Quarterly for Exercise and Sport, 56, 188-200.
Schmidt, R.A., & Lee, T.D. (2011). Central contributions to motor control. In Motor control and learning: A behavioral emphasis (5th ed., pp. 177-222) Champaign, IL: Human Kinetics.
The Farmers' Market
What are the roles of motor error and hypervigilance in unintended acceleration accidents?
The prosecution, at Weller's trial several years later, claimed it was no accident—that Weller deliberately drove his car through the crowded market. The reason, they said, was that Weller had been involved in a minor fender bender just moments before he entered the farmers' market. His response to the fender bender was to flee the scene of the accident. Witnesses for the prosecution painted Weller as a cold-blooded killer, commenting on the determined look on his face during the ordeal and his relatively calm demeanor afterward. It probably also didn't help his case when he emerged from the car immediately afterward and wondered aloud why the people he had hit had not jumped out of his way. Adding his age into the mix, Weller's actions were painted as rather pathetic.
Richard Schmidt, a renowned motor control scholar and human factors expert, testified on behalf of the defense team and argued that the facts of the case shared many similarities to accidents caused by errors of pedal misapplication, or unintended acceleration. Accidents of this type, which, thankfully, are quite rare, occur when the driver intends to apply pressure to the brake pedal, but misses the brake and pushes down on the accelerator instead.
Unintended acceleration accidents had been investigated for many years prior to the Weller case. These accidents were more common when the driver first got into the car, started the engine, and engaged the automatic transmission from the Park position into either Drive or Reverse. Indeed, such frequent episodes of “runaway cars” were the primary reason auto manufacturers added the brake-transmission shift interlock system in the 1980s so that an automatic transmission lever could not be moved from the Park position until the car sensed that a certain amount of pressure had been applied to the brake. However, the brake lock system only prevented this particular type of pedal misapplication; moving the foot to the accelerator instead of the brake would not be prevented once the transmission was successfully engaged out of Park and the car was in motion.
It is important to note that reaching for the brake requires that we steer the foot from a comfortable seated position to a target (the brake) in the absence of any visual guidance. We do this all the time without making any errors. We know where the brake is located from experience, and we also know that the brake feels different underfoot than does the accelerator. So then why would we suddenly miss the brake, push down on the accelerator instead, and then keep the foot there?
Schmidt (1989) presented evidence that unintended acceleration cases frequently involved accidents in which the driver had less experience than usual with that particular vehicle. Therefore, in some of these cases, the exact location and feel of the brake might not have been as familiar to the driver as normally could have been expected. These accidents also occurred more frequently on start-up, compared to later in the driving cycle, perhaps due to temporary factors associated with preparing an action (see “Shooting Two From the Line” in chapter 11). Driver inattention has also been linked with these cases, so it may not come as a surprise that drivers would not immediately notice the difference between the brake and gas pedal if engaged in a distracting activity at the same time (see “Gumbo” in chapter 6).
But for Weller, none of the common profiles for these accidents fit the case: he had already been driving before the supposed pedal misapplication error, so it was not a matter of missing the brake on initial start-up. Weller was quite familiar with his own vehicle, an 11-year-old Buick LeSabre. And he was not talking on a cell phone. Instead, Schmidt argued that Weller's pedal misapplication error was likely triggered by a catastrophic case of panic, termed hypervigilance, which could have been initiated when Weller had been involved in the fender bender just prior to the episode.
But one last issue seemed particularly problematic, according to the prosecution. Once the pedal misapplication error had occurred and the car started to accelerate wildly out of control, why didn't Weller simply remove his foot from the pedal or turn off the engine—actions that would have brought the car quickly to a stop? Again, failure to carry out corrective actions is typical of unintended acceleration cases, and some reasonable accounts have been offered to explain why drivers do not perform them. First, the driver probably does not realize that the foot is on the accelerator rather than the brake. The intention was to press the brake, and the fact that the pedal has gone all the way to the floor could reasonably be interpreted as brake failure rather than human error. And second, once the driver enters into this catastrophic state of panic, all normal modes of thinking cease. Reasoning and problem solving, the kinds of activities that are easy to do when unflustered, become unlikely, if not impossible, to carry out when in this state of hypervigilance.
Thomas Shelton, a member of the California Highway Patrol, testified at Weller's trial that he once investigated an unintended acceleration case in which an elderly woman ended up driving her car onto the top of another vehicle. The woman was in such a panicked state that when Shelton arrived at the accident scene, he had to climb up into the car to shut off the racing engine, at which time he noticed that the woman was still seated, very much alive, staring straight ahead with a death grip on the steering wheel, and with her foot still pushing the accelerator all the way to the floor.
Unfortunately, all of these arguments can only be used to speculate about what may have occurred in George Russell Weller's Buick LeSabre on that fateful day. On October 20, 2006, the jury convicted him of vehicular manslaughter in the 10 deaths resulting from the Santa Monica farmers' market crash. Nobody will ever know whether the verdict was the right one.
Self-Directed Learning Activities
1. In your own words describe the phenomenon known as unintended acceleration.
2. Describe a situation in which you made an action error that you were able to correct. How did you know that you had made the error, and what did you do to correct it?
3. Using our feet to manipulate car pedals involves aiming without visual feedback. What factors influence our ability to make these aiming movements accurately?
4. Propose a research methodology that examines one's ability to (a) move to a target without visual feedback and (b) estimate the accuracy of those aimed movements (again, without vision).
Notes
- Evidence from George Weller's trial during September and October of 2006 was summarized in the Santa Monica Daily Press, which can be accessed through its archives:
- Not all cases of unintended acceleration are generally agreed to be the result of a pedal misapplication. A segment of the television show 60 Minutes, hosted by Ed Bradley and which aired November 22, 1986, claimed that accidents of similar etiology involving the Audi 5000 were the result of a faulty idle stabilizer, which caused the car to accelerate wildly out of control when put into gear. An investigation by the U.S. NHTSA (National Highway Traffic Safety Administration) failed to support 60 Minutes' claim.
- More recently, runaway Toyotas have been a topic of concern. Once again, a government investigation failed to support a claim that these cases of unintended acceleration were due to an electronic fault in the engine. Again, this leaves open the very real possibility that driver error is to blame, as suggested by Richard Schmidt in the New York Times:
Suggested Readings
Castelli, J., Nash, C., Ditlow, C., & Pecht, M. (2003). Sudden acceleration: The myth of driver error. University of Maryland: CALCE EPSC Press.
Pollard, J., & Sussman, E.D. (1989). An examination of sudden acceleration. National Highway Traffic Safety Administration Final Report # DOT-TSC-NHTSA-89-1.
Schmidt, R.A. (1989). Unintended acceleration: A review of human factors contributions. Human Factors, 31, 345-364.
Schmidt, R.A. (1993). Unintended acceleration: Human performance considerations. In B. Peacock & W. Karwowski (Eds.), Automotive ergonomics (pp. 431-451). London: Taylor & Francis.
Schmidt, R.A., & Lee, T.D. (2011). Central contributions to motor control. In Motor control and learning: A behavioral emphasis (5th ed., pp. 177-222) Champaign, IL: Human Kinetics.
Push or Pull?
How do product designs influence people to perform specific actions?
It seems hard to believe, but brand-new buildings are still being constructed with very silly design features. For example, a new building on our campus opened last year and featured many modern technologies, including very nice-looking glass doors through which you enter and exit the building. The problem is that the doors have identical handles on either side. The handles themselves are very attractive, but they don't tell me what to do when I want to enter the building—should I push or pull the door? The handles on these particular doors suggest to me that they are made for grasping and pulling. However, the door goes in only one direction, and the very same handles appear on both sides of the door. Therefore, to enter the building, I have to use the handle to pull the door, and to exit, I have to use the identical type of handle to push the door. How would someone possibly know which way the door swings when approaching it? After a year of using these doors on a daily basis, I still just guess about whether to push or pull. Why would someone design a door like that?
Michael Darnell has created a wonderful website that illustrates many instances of products that have been designed without the user in mind. One of his web pages describes a door handle problem similar to the one I just described, in which a building has a short walkway with one set of doors on either end. Each door has identical handles on either side of the door (the same type of handle is used for pushing and pulling the door). However, according to Darnell, these particular doors were installed such that the problem was magnified even more—the doors at either end of the walkway both swung out from the walkway. He tells the story of a friend who entered the walkway by pulling on the handle of one door; then went to the second door and found that the door would not open when she pulled on the handle. So, concluding that the door was locked, she returned to the first door to exit and found herself trapped in the walkway when the door wouldn't move when she pulled on its handle. After trying to get the attention of others on the outside to tell them that she was trapped in the walkway, she finally realized, with much embarrassment, that the handles had to be pushed when inside the walkway. Darnell reminds us that this funny story could have had tragic consequences if, for example, a panicked person were trying to flee the building in case of a fire.
There are many ways to solve this problem of having identical handles used for opposite actions. One solution would be to put signs on the door that say “push” or “pull.” Although these signs would probably work, a well-designed door (i.e., designed for the user rather than for its looks) should not need a sign to tell people how to use it. Pushing and pulling a door are not rocket science, after all. A better solution would be to install a flat plate on one side of the door, which would make the correct choice obvious: Because there is no handle to grab to pull toward you, there is only one other option—push! A sign would be unnecessary in this case because there would be no question about which action to take.
Signs and symbols that tell us how to use things should be the last option in technology design. Signs are sometimes small and hard to read, they fall off, they wear out, other things cover them up, and they are often presented in only one language, making them less than desirable in a multicultural society. Symbols avoid being unilingual but introduce other problems, such as ambiguity. For example, how would one create simple, unambiguous, and instantly recognizable symbols for “push” and “pull”? As in the door handle example, a better design would be one that made the correct action as obvious as possible, one that would negate the need for a sign or symbol.
A classic example of the failure of technology to exploit this simple idea is a standard topic in ergonomics textbooks and on websites: the arrangement of burners on the stove top. A typical stove top layout is shown in figure 1.5.
The four knobs used to turn on the elements are located at the top of the figure. Quick, say out loud which one is used to turn on the right front burner. My guess is that it is one of the two rightmost knobs. But which one? It is impossible to know for certain without turning one of the knobs and waiting to see which element begins to heat up. So, designers add little labels to tell us which knobs are mapped to which burners. Do they work?
My personal experience is that these labels, if they have not fallen off, worn out, been covered by last night's pasta sauce, or been printed too small to read, are still hard to interpret. More times than I wish to remember, I have returned to a stove expecting to find a pot of boiling water for my pasta but instead found a glowing burner with nothing on it. The relative failure rate of these labels leads me to believe that they are only slightly more helpful than if I had just turned a knob at random.
Have a look at the two alternative stove top designs in figure 1.6. The layout of the knobs in figure 1.6a is the same as before; only the burner layout has been altered. In figure 1.6b, the burner layout is the same as it was before, but now the arrangement of the knobs has been altered.
In both cases, the mapping of each knob to its associated burner has been made obvious by a simple rearrangement of either the knobs or the burner layout. No symbols or labels are needed, and the chances of turning on the wrong burner are minimal. Moreover, the amount of space occupied by the burners and knobs in these more compatible layouts is identical to the space occupied in the incompatible layout. It is interesting to note that these alternative layouts have been available for many years. But have a look at any appliance store flyer or website, and you will see that the classic design still continues to dominate the market.
Sometimes it is simply not possible to avoid the use of labels or signs. In such cases the designer's goal should be to make them as informative as possible. The typical ceiling fan is another good example of a failure to consider the needs of the user when designing the product. The ceiling fan that we have in our home has a wall switch that turns the fan off and on. But the ceiling fan has three additional controls, located on the base of the fan itself (where the fan connects to the ceiling), that are operated by pulling on long strings that hang down from the base. These three strings can be pulled to adjust the speed of the fan, to reverse its direction, or to turn off or on a light located on the bottom of the fan. The problem is that the strings are identical in appearance; I have no way of knowing which control I am activating when I pull on any of the strings.
Actually, that is not completely true. If I stand on a stepladder and get up really close to the ceiling, I can see little labels beside the holes where the strings exit the electrical base of the fan that say “speed,” “light,” and “direction.” I could also stand on the floor and use binoculars to read these labels, I suppose. But, you get the point. Because these labels are next to useless to me, I really just have a one-in-three chance of getting it right each time I pull one of the strings.
How could each control be designed to communicate its function easily to the user? I have used two strategies to modify my own fan. Because reversing the fan's direction is the control that I use least often, I have shortened the string associated with this option to be the most difficult of the three to reach. However, I use the light and speed options about equally often. So, to make these as distinct as possible I borrowed two charms from an old charm bracelet and attached them to the ends of each string. A charm of a rabbit is attached to the string that controls the speed of the fan, and a charm of a book (for which I turn on the light to read) is attached to the string that controls the light. The fan may not look the way the manufacturer had intended it to look, but I have not made the mistake of pulling the wrong string since I attached these charms.
Motor skills such as pushing doors, turning knobs, and pulling strings are simple to learn, and we have all mastered these simple skills over our lifetimes. So why do we have such trouble using them? One reason is that the manufacturer is often paying more attention to aesthetics than to the needs of the user.
Self-Directed Learning Activities
1. Describe, in general, the principle of how product design influences human performance.
2. Search the literature for the term stimulus-response compatibility; then define it in your own words with specific reference to an example presented in this story.
3. Over the next 24 hours, keep a diary of all the objects or things you encounter that you think represent stimulus-response incompatibility. Propose ways each of these things could be made more compatible.
4. Find three research articles in which stimulus-response compatibility has been investigated. What are the similarities and differences in the stimuli and responses used in these studies?
Notes
- Michael Darnell has a wonderful website that features many examples of poorly designed products, with suggestions about simple ways to greatly improve usability: www.baddesigns.com
Suggested Readings
Proctor, R.W., & Van Zandt, T. (2008). Human factors in simple and complex systems (2nd ed.). Boca Raton, FL: CRC Press.
Proctor, R.W., & Vu, K.P.L. (2006). Stimulus-response compatibility principles: Data, theory, and application. Boca Raton, FL: CRC Press.
Schmidt, R.A., & Lee, T.D. (2011). Human information processing. In Motor control and learning: A behavioral emphasis (5th ed., pp. 57-96) Champaign, IL: Human Kinetics.
Learning to win from losing
Why is the learning–performance distinction important?
One day, Michelle Wie might be remembered as one of the greatest golfers of all time. At the age of 14 she was pounding out 300-yard (274 m) drives and not only competing in Ladies Professional (LPGA) Tour events, but also performing quite well in them. Her talent is undeniable. But controversy surrounded this exceptional athlete because of her desire to compete in PGA Tour events, which traditionally have included only male competitors. Wie has performed admirably in some of these events. For example, in the 2004 Sony Open, she shot rounds of 72 and 68 and missed the 36-hole cut by just a single stroke. Her two-round total of 140 placed her in a tie for 80th, which was better than 53 male professionals, including PGA stars Zach Johnson, Hunter Mahan, and Adam Scott. However, Wie has not performed well in most of the other PGA events she has entered, including a 139th-place finish (out of 141 players) at the 2007 Sony Open.
Many argued that Wie should stop competing in PGA Tour events and concentrate her efforts on the LPGA Tour. The typical argument was that her skill development as a golfer was being stalled by repeated failures in PGA Tour events, and that if she concentrated on LPGA events instead, the greater chance for success would escalate her skill development.
I argue that there is a fundamental flaw in this theory, because it falls into a trap known as the learning-performance distinction. The trap underlies one of the most misunderstood concepts in motor learning research, so let's start by defining the two terms. Motor skills researchers use the term performance to refer to a single observation. It could be a score or outcome that reflects the value of a single attempt at a motor skill, or perhaps an average score that statistically summarizes a number of attempts. An 18- or 36-hole score, or a final placement ranking in a tournament, might indicate a representative performance score in the case of a golfer. The term learning is used quite specifically to refer to a stable improvement in skill over time—an improvement that has specifically occurred as the result of practice. The problem illustrated by the criticisms leveled at Michelle Wie is that the critics fail to consider the difference between performance and learning; they mistakenly confuse her failures to improve her performance in men's PGA events as a failure to learn from these experiences. Let's use another example to illustrate this distinction between performance and learning.
Suppose you like to bowl, play regularly in a league, and also practice occasionally. Two years ago your average was 110, last year it was 120, and this year it is 130. Those averages probably indicate that you have become a better bowler as a result of learning: the scores indicate improvements that appear to be rather stable and that have resulted from practice.
So, does this mean that you will score a 130 the next time you bowl? Not necessarily, because there are many reasons for the fluctuations in scores that occur from game to game. Next time you may score well below your average because you are not feeling well, your shoes are too tight, or the crowd is very noisy that night. But this does not mean that you have suddenly lost some of your learned skill. You may score well above your average because you try extra hard to impress someone, or because everything just seems to be well focused (your mojo is working). As before, this does not mean that you have suddenly had a change in learned skill, because there is no indication that the sudden improvement in performance reflects a stable and permanent change in your capability to perform.
At any one time we have a theoretical capability for attaining a certain level of performance. When that theoretical capability changes to a higher level as a result of practice, we can say with confidence that we have learned. The confusion lies in the fact that individual performances may sometimes exceed or not live up to these theoretical capabilities. These fluctuations are expected and in no way diminish or detract from the performer's theoretical capability to perform at a certain skill level.
What I have described is the typical distinction between learning and performance. But, there is another side to this issue, the one that can be applied to Michelle Wie. The issue concerns the situation in which performances do not appear to change, or appear to be getting worse. Does that mean that learning is not occurring? Let's go back to the bowling example.
Suppose that this year your average was 120, which is the same as it was last year. You may take the absence of a stable improvement, despite all of your practice, as evidence that your learning has stalled. But, as it turns out, last year your league bowled Tuesday nights, which is your day off. This year you bowl on Wednesday nights, one hour after completing your 10-hour work shift. You are typically tired and hungry on league nights, and your bowling scores suffered this year as a result. What does all this suggest? What is your theoretical capability to bowl, and has this changed from last year, despite the change in your bowling night? Could it be that you actually have improved (i.e., learned), but the feeling of fatigue and other factors directly related to your work schedule prevented you from actually performing up to these expectations?
The point here is that making inferences about learning requires evidence that goes beyond a single performance or sometimes even a set of performances. Under what circumstances were these performances observed? Are there mitigating factors that might explain why the theoretical true score differs from the observed score? The absence of any observable change in performance does not mean that the unobservable (that theoretical capability to perform at a certain level of skill) has not improved.
Michelle Wie took a very different route in the development of her golfing skill. Her decision to compete in PGA events placed enormous pressures on her to perform, and only an exceptional result would have convinced the public that she was benefiting from these extreme challenges. All we could observe were her scores in these events. The unobservable, however, what she has learned by playing in these events, can never be truly understood. I suspect that if she achieves the rank of one of the greatest players of all time, it will be due in no small measure to these experiences competing against the very best golfers in the world.
Self-Directed Learning Activities
1. In your own words, explain the distinction between performance and learning.
2. What is the difference between a theoretical true score and an observed score?
3. Suggest a reason a person's observed score might exceed the true score, and a reason the observed score might fail to achieve the level of the true score.
4. Identify another sport, athlete, or situation in which the failure to account for the learning-performance distinction has resulted in an inappropriate conclusion.
Notes
- Babe Zaharias (see “The Babe”) was the first woman to compete in a men's PGA event.
- Here is a sampling of some of the criticism of Michele Wie's decision to compete in PGA Tour events, from two sports commentary sites:
- Michelle Wie won her second LPGA Tour title in Winnipeg, Manitoba, at the 2010 CN Canadian Women's Open.
Suggested Readings
Schmidt, R.A. (1972). The case against learning and forgetting scores. Journal of Motor Behavior, 4, 79-88.
Schmidt, R.A., & Lee, T.D. (2011). Motor learning concepts and research methods. In Motor control and learning: A behavioral emphasis (5th ed., pp. 327-346). Champaign, IL: Human Kinetics.
Cutting wood and missing putts
How are constant error, variable error, and absolute error useful for understanding motor control?
Two things I like to do—cutting wood and golfing—remind me often of the importance of measures of central tendency and variability in motor control performance. For example, I use my chainsaw to cut up fallen trees to burn in the woodstove. Cutting logs into burnable lengths is not an exact science. In the middle of a Canadian winter, when the temperature is cold and plenty of snow blankets the ground, it is generally considered a good idea to keep moving. I do not take time to measure the lengths of the logs I saw; rather, I make a quick visual estimate and then start to make my cut. But, in doing so, I must remember two important things related to the mean and the standard deviation of the logs I cut. First, because our woodstove is only large enough to hold pieces of wood that are 50 centimeters (20 in.) or shorter, any pieces that are longer than that are useless. And second, wood is easier to stack when the pieces are all roughly of the same length. Essentially, these features remind me that the mean and standard deviation are both important; I must have a mean that is less than 50 centimeters and a standard deviation that is as small as possible.
When considering data in terms of a specific goal or standard, it is often better to express the mean and standard deviation in terms of error measures—in my case, relative to a goal wood length of 50 centimeters. For example, wood lengths of 45, 42, 48, 47, and 52 centimeters would be expressed as lengths that are -5, -8, -2, -3, and +2 centimeters, respectively, relative to my goal wood length of 50 centimeters. If we calculated the mean of these lengths of wood, then we could express the stack of wood as having either an average length of 46.8 centimeters, or the error scores as having a mean constant error of -3.2 centimeters. Having a negative constant error is good in this case, because the average piece of wood that I have cut will easily fit in the woodstove (i.e., the mean is less than 50 centimeters).
The careful reader will have noted, however, that even though my mean constant error is negative, there is still one block of wood that will not fit (the 52 cm log). Thus, by itself, the mean constant error score (-3.2 cm) is misleading, because it does not provide any indication of how many logs might not actually fit in the woodstove. Our measure of variability, the standard deviation, gives us some indication of this information. The measure of variability of these error scores (called the variable error) represents the deviation of the error score for each log relative to the mean constant error and is calculated in the same way as is a standard deviation. In this case, the variable error is 3.3 centimeters. In general, when I am cutting wood, I want to achieve a negative mean constant error with a variable error that is as small as possible.
Although constant error provides a useful heuristic measure of average performance in cutting wood with a chainsaw, it is a quite misleading measure of central tendency in another activity I like to do. Let's say that I have struck 10 golf putts—5 of these putts go past the hole by 5, 10, 15, 20, and 25 centimeters, respectively, and the other five putts come up short of the hole by equal amounts (-5, -10, -15, -20 and -25 cm). On average, by how much have I missed the hole with these putts? If you were to calculate the average as in the previous wood-cutting example (the sum of the 10 individual error scores, divided by 10) the answer would be a mean constant error of 0. In other words, my “average” putt ended up in the hole. However, we know that this answer surely must be wrong because none of the individual putts actually went in the hole. So, we must use a different way to express these error scores to avoid an answer that makes no sense.
The problem that we sometimes run into with constant error is that it provides a measure of average bias, the tendency to err in a specific way (e.g., by too much or too little; too far left versus too far right). In some cases a specific bias is desirable, such as the tendency to undercut a wood length of 50 centimeters so that all of my wood pieces fit in the woodstove. In the case of my putts that have no consistent bias, we are much better off using a measure of central tendency that removes the bias from each score prior to calculating the mean. Such a procedure is called using the absolute (unsigned) scores. Hence, this measure of central tendency is called the mean absolute error. Here, our measure of central tendency is the mean of the unsigned scores (5, 5, 10, 10, 15, 15, 20, 20, 25, 25), which is 15 centimeters. That is, the golf putts tended to miss the hole by an average of 15 centimeters. In this particular case, the mean absolute error score (15 cm from the hole) represents a more accurate picture of the entire set of individual scores than what is represented in the mean constant error (in the hole).
Averages are convenient ways to express how we tend to perform. But, numbers that represent central tendencies can be misleading if the tendency itself is not a strong one. Measures of variability are one way to identify the strength of a tendency. Depending on the context, different methods of calculating central tendencies may characterize the data in more representative ways.
Self-Directed Learning Activities
1. Define constant error, absolute error, and variable error in your own words.
2. What do the terms algebraic error and total error (or Henry's E) refer to? How are they calculated, and how are they similar to the terms constant error and absolute error, respectively?
3. Look up and briefly describe a research investigation that uses at least two error measures. Why do you think the researchers used these measures in particular?
4. Conduct a brief study on yourself. Close your eyes and, using a pencil, draw five lines as close to 4 inches (10 cm) as possible, one below the previous one, without opening your eyes until you have drawn the fifth line. Measure and record the length of each line you drew; then calculate all of the error measures that have been discussed.
Notes
- Here is a web-based standard deviation calculator:
www.tinyurl.com/standarddeviationcalculator
Suggested Readings
Chapanis, A. (1951). Theory and methods for analyzing errors in man-machine systems. Annals of the New York Academy of Sciences, 51, 1179-1203.
Schmidt, R.A., & Lee, T.D. (2011). Methodology for studying motor performance. In Motor control and learning: A behavioral emphasis (5th ed., pp. 21-56). Champaign, IL: Human Kinetics.
Forensic motor control
What are generalized motor programs, and what do keystroke dynamics reveal about them?
Most of what I know about forensic identification comes from crime shows that I've seen on TV. The suspect is caught because a fingerprint was left on the murder weapon. The evidence is used to convict the suspect because no two fingerprints have ever been found to be identical. Crime shows tell us that other forensic methods can be used as well, such as DNA and eye retina data, but the fingerprint is the oldest biometric tool used for identification. However, another method that reveals a great deal about your identity is how you write or even type your name. For example, the password that you type to log on to your e-mail account may be just as identifiable as your fingerprint.
Try the following task as an example: go into any word processing program and type your name 10 times, once on each line, as I have done here:
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
I am not a skilled typist. I use my right middle finger to hold down the shift key and press the letter T with my left middle finger, then use my right middle finger to press i and my right index finger to press m, and then my right thumb to press the space bar. It is not very efficient, but I do it the same way every time I type my name. Regardless of how skilled a typist you are, you most likely repeat much the same process each of the 10 times you type your name.
Now, let's try to unravel the temporal “fingerprint” that you left behind when you typed your name. Suppose we conducted a simple analysis of the keystrokes and the time between each of the keystrokes that you made. The total time taken to type your name once is simply the time from the first keystroke to the last keystroke. At a more fine-grained level of analysis, the total time constitutes the time each key is held down (the dwell time) plus the time between the release of one key and the depression of the next (the transition time). All of the individual dwell times plus all of the transition times will sum to the total time.
In figure 7.5, I have plotted a hypothetical example of the 10 trials to type my name. The light gray bands denote the dwell times, and the darker gray bands indicate the transition times. I have ordered the lines of bands that represent the 10 trials from the fastest (least total time) at the top to the slowest trial at the bottom (most total time).
If you were to analyze each of the 10 repetitions of your name, you would probably find that all of the total times would be similar, but probably never exactly the same each time, just as in figure 7.5. The “noise” in our central nervous systems, plus other factors, causes the results to vary a little bit each time, resulting in some repetitions to be slower, and others faster, than the average total time.
But, take a closer look at each individual band in figure 7.5 as it changes over the 10 trials. What you will notice is that as the total time increases, each band gets proportionally longer too. We could express these numbers another way by dividing the time for each individual band by the total time for that trial to obtain a relative proportion of time represented by each band. Given the hypothetical evidence in figure 7.5, what we would discover is that the relative time for each band, expressed as a percentage, would stay roughly the same across all of the repetitions. Applying temporal and other methods to deconstruct how we type (e.g., key press forces) is called the study of keystroke dynamics.
According to generalized motor program theory, relative time is one of the key features of movement that is controlled by the central nervous system, especially for brief, rapid actions. Relative time is an expression of our motor control fingerprint. For well-learned tasks such as typing and handwriting, our central nervous system regulates the relative timing of impulses that are sent to the muscles that carry out these tasks. Various factors influence the real time on any particular instance. For example, using a keyboard that requires more force to depress the keys might result in overall slower times than would result using a keyboard that has a light touch, but the relative timing would likely remain the same.
Research in keystroke dynamics may result in the ability to identify people who carry out Internet fraud. In many ways it is similar to the use of handwriting dynamics to analyze the timing of the cursive expressions of a signature. It is fairly easy to forge the spatial representation of someone's signature, but very difficult to forge the temporal dynamics that result in that signature. These expressions of timing behavior represent our motor control fingerprints.
Self-Directed Learning Activities
1. Define the term generalized motor program in your own words.
2. How does the concept of a generalized motor program differ from the concept of a motor program as it was used in stories such as “Antilock Brakes” (in chapter 5) and “Point of No Return” (earlier in this
chapter)?
3. Suggest a method by which the handwriting dynamics of signatures could be used to detect fraud.
4. Some people contend that there is one generalized motor program for the full swing in golf, regardless of which club is used. How could a temporal dynamics analysis be used to assess this contention?
Notes
- An excellent review of fingerprint analysis by David Ashbaugh of the Royal Canadian Mounted Police is available here:
www.onin.com/fp/ridgeology.pdf
- A lot of controversy remains about invariances in motor performance; how they are measured; and what invariance, or lack of invariance, means in terms of motor programs. The following articles provide good arguments for the debate:
Gentner, D.R. (1987). Timing of skilled motor performance: Tests of the proportional duration model. Psychological Review, 94, 255-276.
Heuer, H. (1988). Testing the invariance of relative timing: Comment on Gentner (1987). Psychological Review, 95, 552-557.
Suggested Readings
Schmidt, R.A. (1975). A schema theory of discrete motor skill learning. Psychological Review, 82, 225-260.
Schmidt, R.A. (1985). The search for invariance in skilled movement behavior. Research Quarterly for Exercise and Sport, 56, 188-200.
Schmidt, R.A., & Lee, T.D. (2011). Central contributions to motor control. In Motor control and learning: A behavioral emphasis (5th ed., pp. 177-222) Champaign, IL: Human Kinetics.
The Farmers' Market
What are the roles of motor error and hypervigilance in unintended acceleration accidents?
The prosecution, at Weller's trial several years later, claimed it was no accident—that Weller deliberately drove his car through the crowded market. The reason, they said, was that Weller had been involved in a minor fender bender just moments before he entered the farmers' market. His response to the fender bender was to flee the scene of the accident. Witnesses for the prosecution painted Weller as a cold-blooded killer, commenting on the determined look on his face during the ordeal and his relatively calm demeanor afterward. It probably also didn't help his case when he emerged from the car immediately afterward and wondered aloud why the people he had hit had not jumped out of his way. Adding his age into the mix, Weller's actions were painted as rather pathetic.
Richard Schmidt, a renowned motor control scholar and human factors expert, testified on behalf of the defense team and argued that the facts of the case shared many similarities to accidents caused by errors of pedal misapplication, or unintended acceleration. Accidents of this type, which, thankfully, are quite rare, occur when the driver intends to apply pressure to the brake pedal, but misses the brake and pushes down on the accelerator instead.
Unintended acceleration accidents had been investigated for many years prior to the Weller case. These accidents were more common when the driver first got into the car, started the engine, and engaged the automatic transmission from the Park position into either Drive or Reverse. Indeed, such frequent episodes of “runaway cars” were the primary reason auto manufacturers added the brake-transmission shift interlock system in the 1980s so that an automatic transmission lever could not be moved from the Park position until the car sensed that a certain amount of pressure had been applied to the brake. However, the brake lock system only prevented this particular type of pedal misapplication; moving the foot to the accelerator instead of the brake would not be prevented once the transmission was successfully engaged out of Park and the car was in motion.
It is important to note that reaching for the brake requires that we steer the foot from a comfortable seated position to a target (the brake) in the absence of any visual guidance. We do this all the time without making any errors. We know where the brake is located from experience, and we also know that the brake feels different underfoot than does the accelerator. So then why would we suddenly miss the brake, push down on the accelerator instead, and then keep the foot there?
Schmidt (1989) presented evidence that unintended acceleration cases frequently involved accidents in which the driver had less experience than usual with that particular vehicle. Therefore, in some of these cases, the exact location and feel of the brake might not have been as familiar to the driver as normally could have been expected. These accidents also occurred more frequently on start-up, compared to later in the driving cycle, perhaps due to temporary factors associated with preparing an action (see “Shooting Two From the Line” in chapter 11). Driver inattention has also been linked with these cases, so it may not come as a surprise that drivers would not immediately notice the difference between the brake and gas pedal if engaged in a distracting activity at the same time (see “Gumbo” in chapter 6).
But for Weller, none of the common profiles for these accidents fit the case: he had already been driving before the supposed pedal misapplication error, so it was not a matter of missing the brake on initial start-up. Weller was quite familiar with his own vehicle, an 11-year-old Buick LeSabre. And he was not talking on a cell phone. Instead, Schmidt argued that Weller's pedal misapplication error was likely triggered by a catastrophic case of panic, termed hypervigilance, which could have been initiated when Weller had been involved in the fender bender just prior to the episode.
But one last issue seemed particularly problematic, according to the prosecution. Once the pedal misapplication error had occurred and the car started to accelerate wildly out of control, why didn't Weller simply remove his foot from the pedal or turn off the engine—actions that would have brought the car quickly to a stop? Again, failure to carry out corrective actions is typical of unintended acceleration cases, and some reasonable accounts have been offered to explain why drivers do not perform them. First, the driver probably does not realize that the foot is on the accelerator rather than the brake. The intention was to press the brake, and the fact that the pedal has gone all the way to the floor could reasonably be interpreted as brake failure rather than human error. And second, once the driver enters into this catastrophic state of panic, all normal modes of thinking cease. Reasoning and problem solving, the kinds of activities that are easy to do when unflustered, become unlikely, if not impossible, to carry out when in this state of hypervigilance.
Thomas Shelton, a member of the California Highway Patrol, testified at Weller's trial that he once investigated an unintended acceleration case in which an elderly woman ended up driving her car onto the top of another vehicle. The woman was in such a panicked state that when Shelton arrived at the accident scene, he had to climb up into the car to shut off the racing engine, at which time he noticed that the woman was still seated, very much alive, staring straight ahead with a death grip on the steering wheel, and with her foot still pushing the accelerator all the way to the floor.
Unfortunately, all of these arguments can only be used to speculate about what may have occurred in George Russell Weller's Buick LeSabre on that fateful day. On October 20, 2006, the jury convicted him of vehicular manslaughter in the 10 deaths resulting from the Santa Monica farmers' market crash. Nobody will ever know whether the verdict was the right one.
Self-Directed Learning Activities
1. In your own words describe the phenomenon known as unintended acceleration.
2. Describe a situation in which you made an action error that you were able to correct. How did you know that you had made the error, and what did you do to correct it?
3. Using our feet to manipulate car pedals involves aiming without visual feedback. What factors influence our ability to make these aiming movements accurately?
4. Propose a research methodology that examines one's ability to (a) move to a target without visual feedback and (b) estimate the accuracy of those aimed movements (again, without vision).
Notes
- Evidence from George Weller's trial during September and October of 2006 was summarized in the Santa Monica Daily Press, which can be accessed through its archives:
- Not all cases of unintended acceleration are generally agreed to be the result of a pedal misapplication. A segment of the television show 60 Minutes, hosted by Ed Bradley and which aired November 22, 1986, claimed that accidents of similar etiology involving the Audi 5000 were the result of a faulty idle stabilizer, which caused the car to accelerate wildly out of control when put into gear. An investigation by the U.S. NHTSA (National Highway Traffic Safety Administration) failed to support 60 Minutes' claim.
- More recently, runaway Toyotas have been a topic of concern. Once again, a government investigation failed to support a claim that these cases of unintended acceleration were due to an electronic fault in the engine. Again, this leaves open the very real possibility that driver error is to blame, as suggested by Richard Schmidt in the New York Times:
Suggested Readings
Castelli, J., Nash, C., Ditlow, C., & Pecht, M. (2003). Sudden acceleration: The myth of driver error. University of Maryland: CALCE EPSC Press.
Pollard, J., & Sussman, E.D. (1989). An examination of sudden acceleration. National Highway Traffic Safety Administration Final Report # DOT-TSC-NHTSA-89-1.
Schmidt, R.A. (1989). Unintended acceleration: A review of human factors contributions. Human Factors, 31, 345-364.
Schmidt, R.A. (1993). Unintended acceleration: Human performance considerations. In B. Peacock & W. Karwowski (Eds.), Automotive ergonomics (pp. 431-451). London: Taylor & Francis.
Schmidt, R.A., & Lee, T.D. (2011). Central contributions to motor control. In Motor control and learning: A behavioral emphasis (5th ed., pp. 177-222) Champaign, IL: Human Kinetics.
Push or Pull?
How do product designs influence people to perform specific actions?
It seems hard to believe, but brand-new buildings are still being constructed with very silly design features. For example, a new building on our campus opened last year and featured many modern technologies, including very nice-looking glass doors through which you enter and exit the building. The problem is that the doors have identical handles on either side. The handles themselves are very attractive, but they don't tell me what to do when I want to enter the building—should I push or pull the door? The handles on these particular doors suggest to me that they are made for grasping and pulling. However, the door goes in only one direction, and the very same handles appear on both sides of the door. Therefore, to enter the building, I have to use the handle to pull the door, and to exit, I have to use the identical type of handle to push the door. How would someone possibly know which way the door swings when approaching it? After a year of using these doors on a daily basis, I still just guess about whether to push or pull. Why would someone design a door like that?
Michael Darnell has created a wonderful website that illustrates many instances of products that have been designed without the user in mind. One of his web pages describes a door handle problem similar to the one I just described, in which a building has a short walkway with one set of doors on either end. Each door has identical handles on either side of the door (the same type of handle is used for pushing and pulling the door). However, according to Darnell, these particular doors were installed such that the problem was magnified even more—the doors at either end of the walkway both swung out from the walkway. He tells the story of a friend who entered the walkway by pulling on the handle of one door; then went to the second door and found that the door would not open when she pulled on the handle. So, concluding that the door was locked, she returned to the first door to exit and found herself trapped in the walkway when the door wouldn't move when she pulled on its handle. After trying to get the attention of others on the outside to tell them that she was trapped in the walkway, she finally realized, with much embarrassment, that the handles had to be pushed when inside the walkway. Darnell reminds us that this funny story could have had tragic consequences if, for example, a panicked person were trying to flee the building in case of a fire.
There are many ways to solve this problem of having identical handles used for opposite actions. One solution would be to put signs on the door that say “push” or “pull.” Although these signs would probably work, a well-designed door (i.e., designed for the user rather than for its looks) should not need a sign to tell people how to use it. Pushing and pulling a door are not rocket science, after all. A better solution would be to install a flat plate on one side of the door, which would make the correct choice obvious: Because there is no handle to grab to pull toward you, there is only one other option—push! A sign would be unnecessary in this case because there would be no question about which action to take.
Signs and symbols that tell us how to use things should be the last option in technology design. Signs are sometimes small and hard to read, they fall off, they wear out, other things cover them up, and they are often presented in only one language, making them less than desirable in a multicultural society. Symbols avoid being unilingual but introduce other problems, such as ambiguity. For example, how would one create simple, unambiguous, and instantly recognizable symbols for “push” and “pull”? As in the door handle example, a better design would be one that made the correct action as obvious as possible, one that would negate the need for a sign or symbol.
A classic example of the failure of technology to exploit this simple idea is a standard topic in ergonomics textbooks and on websites: the arrangement of burners on the stove top. A typical stove top layout is shown in figure 1.5.
The four knobs used to turn on the elements are located at the top of the figure. Quick, say out loud which one is used to turn on the right front burner. My guess is that it is one of the two rightmost knobs. But which one? It is impossible to know for certain without turning one of the knobs and waiting to see which element begins to heat up. So, designers add little labels to tell us which knobs are mapped to which burners. Do they work?
My personal experience is that these labels, if they have not fallen off, worn out, been covered by last night's pasta sauce, or been printed too small to read, are still hard to interpret. More times than I wish to remember, I have returned to a stove expecting to find a pot of boiling water for my pasta but instead found a glowing burner with nothing on it. The relative failure rate of these labels leads me to believe that they are only slightly more helpful than if I had just turned a knob at random.
Have a look at the two alternative stove top designs in figure 1.6. The layout of the knobs in figure 1.6a is the same as before; only the burner layout has been altered. In figure 1.6b, the burner layout is the same as it was before, but now the arrangement of the knobs has been altered.
In both cases, the mapping of each knob to its associated burner has been made obvious by a simple rearrangement of either the knobs or the burner layout. No symbols or labels are needed, and the chances of turning on the wrong burner are minimal. Moreover, the amount of space occupied by the burners and knobs in these more compatible layouts is identical to the space occupied in the incompatible layout. It is interesting to note that these alternative layouts have been available for many years. But have a look at any appliance store flyer or website, and you will see that the classic design still continues to dominate the market.
Sometimes it is simply not possible to avoid the use of labels or signs. In such cases the designer's goal should be to make them as informative as possible. The typical ceiling fan is another good example of a failure to consider the needs of the user when designing the product. The ceiling fan that we have in our home has a wall switch that turns the fan off and on. But the ceiling fan has three additional controls, located on the base of the fan itself (where the fan connects to the ceiling), that are operated by pulling on long strings that hang down from the base. These three strings can be pulled to adjust the speed of the fan, to reverse its direction, or to turn off or on a light located on the bottom of the fan. The problem is that the strings are identical in appearance; I have no way of knowing which control I am activating when I pull on any of the strings.
Actually, that is not completely true. If I stand on a stepladder and get up really close to the ceiling, I can see little labels beside the holes where the strings exit the electrical base of the fan that say “speed,” “light,” and “direction.” I could also stand on the floor and use binoculars to read these labels, I suppose. But, you get the point. Because these labels are next to useless to me, I really just have a one-in-three chance of getting it right each time I pull one of the strings.
How could each control be designed to communicate its function easily to the user? I have used two strategies to modify my own fan. Because reversing the fan's direction is the control that I use least often, I have shortened the string associated with this option to be the most difficult of the three to reach. However, I use the light and speed options about equally often. So, to make these as distinct as possible I borrowed two charms from an old charm bracelet and attached them to the ends of each string. A charm of a rabbit is attached to the string that controls the speed of the fan, and a charm of a book (for which I turn on the light to read) is attached to the string that controls the light. The fan may not look the way the manufacturer had intended it to look, but I have not made the mistake of pulling the wrong string since I attached these charms.
Motor skills such as pushing doors, turning knobs, and pulling strings are simple to learn, and we have all mastered these simple skills over our lifetimes. So why do we have such trouble using them? One reason is that the manufacturer is often paying more attention to aesthetics than to the needs of the user.
Self-Directed Learning Activities
1. Describe, in general, the principle of how product design influences human performance.
2. Search the literature for the term stimulus-response compatibility; then define it in your own words with specific reference to an example presented in this story.
3. Over the next 24 hours, keep a diary of all the objects or things you encounter that you think represent stimulus-response incompatibility. Propose ways each of these things could be made more compatible.
4. Find three research articles in which stimulus-response compatibility has been investigated. What are the similarities and differences in the stimuli and responses used in these studies?
Notes
- Michael Darnell has a wonderful website that features many examples of poorly designed products, with suggestions about simple ways to greatly improve usability: www.baddesigns.com
Suggested Readings
Proctor, R.W., & Van Zandt, T. (2008). Human factors in simple and complex systems (2nd ed.). Boca Raton, FL: CRC Press.
Proctor, R.W., & Vu, K.P.L. (2006). Stimulus-response compatibility principles: Data, theory, and application. Boca Raton, FL: CRC Press.
Schmidt, R.A., & Lee, T.D. (2011). Human information processing. In Motor control and learning: A behavioral emphasis (5th ed., pp. 57-96) Champaign, IL: Human Kinetics.
Learning to win from losing
Why is the learning–performance distinction important?
One day, Michelle Wie might be remembered as one of the greatest golfers of all time. At the age of 14 she was pounding out 300-yard (274 m) drives and not only competing in Ladies Professional (LPGA) Tour events, but also performing quite well in them. Her talent is undeniable. But controversy surrounded this exceptional athlete because of her desire to compete in PGA Tour events, which traditionally have included only male competitors. Wie has performed admirably in some of these events. For example, in the 2004 Sony Open, she shot rounds of 72 and 68 and missed the 36-hole cut by just a single stroke. Her two-round total of 140 placed her in a tie for 80th, which was better than 53 male professionals, including PGA stars Zach Johnson, Hunter Mahan, and Adam Scott. However, Wie has not performed well in most of the other PGA events she has entered, including a 139th-place finish (out of 141 players) at the 2007 Sony Open.
Many argued that Wie should stop competing in PGA Tour events and concentrate her efforts on the LPGA Tour. The typical argument was that her skill development as a golfer was being stalled by repeated failures in PGA Tour events, and that if she concentrated on LPGA events instead, the greater chance for success would escalate her skill development.
I argue that there is a fundamental flaw in this theory, because it falls into a trap known as the learning-performance distinction. The trap underlies one of the most misunderstood concepts in motor learning research, so let's start by defining the two terms. Motor skills researchers use the term performance to refer to a single observation. It could be a score or outcome that reflects the value of a single attempt at a motor skill, or perhaps an average score that statistically summarizes a number of attempts. An 18- or 36-hole score, or a final placement ranking in a tournament, might indicate a representative performance score in the case of a golfer. The term learning is used quite specifically to refer to a stable improvement in skill over time—an improvement that has specifically occurred as the result of practice. The problem illustrated by the criticisms leveled at Michelle Wie is that the critics fail to consider the difference between performance and learning; they mistakenly confuse her failures to improve her performance in men's PGA events as a failure to learn from these experiences. Let's use another example to illustrate this distinction between performance and learning.
Suppose you like to bowl, play regularly in a league, and also practice occasionally. Two years ago your average was 110, last year it was 120, and this year it is 130. Those averages probably indicate that you have become a better bowler as a result of learning: the scores indicate improvements that appear to be rather stable and that have resulted from practice.
So, does this mean that you will score a 130 the next time you bowl? Not necessarily, because there are many reasons for the fluctuations in scores that occur from game to game. Next time you may score well below your average because you are not feeling well, your shoes are too tight, or the crowd is very noisy that night. But this does not mean that you have suddenly lost some of your learned skill. You may score well above your average because you try extra hard to impress someone, or because everything just seems to be well focused (your mojo is working). As before, this does not mean that you have suddenly had a change in learned skill, because there is no indication that the sudden improvement in performance reflects a stable and permanent change in your capability to perform.
At any one time we have a theoretical capability for attaining a certain level of performance. When that theoretical capability changes to a higher level as a result of practice, we can say with confidence that we have learned. The confusion lies in the fact that individual performances may sometimes exceed or not live up to these theoretical capabilities. These fluctuations are expected and in no way diminish or detract from the performer's theoretical capability to perform at a certain skill level.
What I have described is the typical distinction between learning and performance. But, there is another side to this issue, the one that can be applied to Michelle Wie. The issue concerns the situation in which performances do not appear to change, or appear to be getting worse. Does that mean that learning is not occurring? Let's go back to the bowling example.
Suppose that this year your average was 120, which is the same as it was last year. You may take the absence of a stable improvement, despite all of your practice, as evidence that your learning has stalled. But, as it turns out, last year your league bowled Tuesday nights, which is your day off. This year you bowl on Wednesday nights, one hour after completing your 10-hour work shift. You are typically tired and hungry on league nights, and your bowling scores suffered this year as a result. What does all this suggest? What is your theoretical capability to bowl, and has this changed from last year, despite the change in your bowling night? Could it be that you actually have improved (i.e., learned), but the feeling of fatigue and other factors directly related to your work schedule prevented you from actually performing up to these expectations?
The point here is that making inferences about learning requires evidence that goes beyond a single performance or sometimes even a set of performances. Under what circumstances were these performances observed? Are there mitigating factors that might explain why the theoretical true score differs from the observed score? The absence of any observable change in performance does not mean that the unobservable (that theoretical capability to perform at a certain level of skill) has not improved.
Michelle Wie took a very different route in the development of her golfing skill. Her decision to compete in PGA events placed enormous pressures on her to perform, and only an exceptional result would have convinced the public that she was benefiting from these extreme challenges. All we could observe were her scores in these events. The unobservable, however, what she has learned by playing in these events, can never be truly understood. I suspect that if she achieves the rank of one of the greatest players of all time, it will be due in no small measure to these experiences competing against the very best golfers in the world.
Self-Directed Learning Activities
1. In your own words, explain the distinction between performance and learning.
2. What is the difference between a theoretical true score and an observed score?
3. Suggest a reason a person's observed score might exceed the true score, and a reason the observed score might fail to achieve the level of the true score.
4. Identify another sport, athlete, or situation in which the failure to account for the learning-performance distinction has resulted in an inappropriate conclusion.
Notes
- Babe Zaharias (see “The Babe”) was the first woman to compete in a men's PGA event.
- Here is a sampling of some of the criticism of Michele Wie's decision to compete in PGA Tour events, from two sports commentary sites:
- Michelle Wie won her second LPGA Tour title in Winnipeg, Manitoba, at the 2010 CN Canadian Women's Open.
Suggested Readings
Schmidt, R.A. (1972). The case against learning and forgetting scores. Journal of Motor Behavior, 4, 79-88.
Schmidt, R.A., & Lee, T.D. (2011). Motor learning concepts and research methods. In Motor control and learning: A behavioral emphasis (5th ed., pp. 327-346). Champaign, IL: Human Kinetics.
Cutting wood and missing putts
How are constant error, variable error, and absolute error useful for understanding motor control?
Two things I like to do—cutting wood and golfing—remind me often of the importance of measures of central tendency and variability in motor control performance. For example, I use my chainsaw to cut up fallen trees to burn in the woodstove. Cutting logs into burnable lengths is not an exact science. In the middle of a Canadian winter, when the temperature is cold and plenty of snow blankets the ground, it is generally considered a good idea to keep moving. I do not take time to measure the lengths of the logs I saw; rather, I make a quick visual estimate and then start to make my cut. But, in doing so, I must remember two important things related to the mean and the standard deviation of the logs I cut. First, because our woodstove is only large enough to hold pieces of wood that are 50 centimeters (20 in.) or shorter, any pieces that are longer than that are useless. And second, wood is easier to stack when the pieces are all roughly of the same length. Essentially, these features remind me that the mean and standard deviation are both important; I must have a mean that is less than 50 centimeters and a standard deviation that is as small as possible.
When considering data in terms of a specific goal or standard, it is often better to express the mean and standard deviation in terms of error measures—in my case, relative to a goal wood length of 50 centimeters. For example, wood lengths of 45, 42, 48, 47, and 52 centimeters would be expressed as lengths that are -5, -8, -2, -3, and +2 centimeters, respectively, relative to my goal wood length of 50 centimeters. If we calculated the mean of these lengths of wood, then we could express the stack of wood as having either an average length of 46.8 centimeters, or the error scores as having a mean constant error of -3.2 centimeters. Having a negative constant error is good in this case, because the average piece of wood that I have cut will easily fit in the woodstove (i.e., the mean is less than 50 centimeters).
The careful reader will have noted, however, that even though my mean constant error is negative, there is still one block of wood that will not fit (the 52 cm log). Thus, by itself, the mean constant error score (-3.2 cm) is misleading, because it does not provide any indication of how many logs might not actually fit in the woodstove. Our measure of variability, the standard deviation, gives us some indication of this information. The measure of variability of these error scores (called the variable error) represents the deviation of the error score for each log relative to the mean constant error and is calculated in the same way as is a standard deviation. In this case, the variable error is 3.3 centimeters. In general, when I am cutting wood, I want to achieve a negative mean constant error with a variable error that is as small as possible.
Although constant error provides a useful heuristic measure of average performance in cutting wood with a chainsaw, it is a quite misleading measure of central tendency in another activity I like to do. Let's say that I have struck 10 golf putts—5 of these putts go past the hole by 5, 10, 15, 20, and 25 centimeters, respectively, and the other five putts come up short of the hole by equal amounts (-5, -10, -15, -20 and -25 cm). On average, by how much have I missed the hole with these putts? If you were to calculate the average as in the previous wood-cutting example (the sum of the 10 individual error scores, divided by 10) the answer would be a mean constant error of 0. In other words, my “average” putt ended up in the hole. However, we know that this answer surely must be wrong because none of the individual putts actually went in the hole. So, we must use a different way to express these error scores to avoid an answer that makes no sense.
The problem that we sometimes run into with constant error is that it provides a measure of average bias, the tendency to err in a specific way (e.g., by too much or too little; too far left versus too far right). In some cases a specific bias is desirable, such as the tendency to undercut a wood length of 50 centimeters so that all of my wood pieces fit in the woodstove. In the case of my putts that have no consistent bias, we are much better off using a measure of central tendency that removes the bias from each score prior to calculating the mean. Such a procedure is called using the absolute (unsigned) scores. Hence, this measure of central tendency is called the mean absolute error. Here, our measure of central tendency is the mean of the unsigned scores (5, 5, 10, 10, 15, 15, 20, 20, 25, 25), which is 15 centimeters. That is, the golf putts tended to miss the hole by an average of 15 centimeters. In this particular case, the mean absolute error score (15 cm from the hole) represents a more accurate picture of the entire set of individual scores than what is represented in the mean constant error (in the hole).
Averages are convenient ways to express how we tend to perform. But, numbers that represent central tendencies can be misleading if the tendency itself is not a strong one. Measures of variability are one way to identify the strength of a tendency. Depending on the context, different methods of calculating central tendencies may characterize the data in more representative ways.
Self-Directed Learning Activities
1. Define constant error, absolute error, and variable error in your own words.
2. What do the terms algebraic error and total error (or Henry's E) refer to? How are they calculated, and how are they similar to the terms constant error and absolute error, respectively?
3. Look up and briefly describe a research investigation that uses at least two error measures. Why do you think the researchers used these measures in particular?
4. Conduct a brief study on yourself. Close your eyes and, using a pencil, draw five lines as close to 4 inches (10 cm) as possible, one below the previous one, without opening your eyes until you have drawn the fifth line. Measure and record the length of each line you drew; then calculate all of the error measures that have been discussed.
Notes
- Here is a web-based standard deviation calculator:
www.tinyurl.com/standarddeviationcalculator
Suggested Readings
Chapanis, A. (1951). Theory and methods for analyzing errors in man-machine systems. Annals of the New York Academy of Sciences, 51, 1179-1203.
Schmidt, R.A., & Lee, T.D. (2011). Methodology for studying motor performance. In Motor control and learning: A behavioral emphasis (5th ed., pp. 21-56). Champaign, IL: Human Kinetics.
Forensic motor control
What are generalized motor programs, and what do keystroke dynamics reveal about them?
Most of what I know about forensic identification comes from crime shows that I've seen on TV. The suspect is caught because a fingerprint was left on the murder weapon. The evidence is used to convict the suspect because no two fingerprints have ever been found to be identical. Crime shows tell us that other forensic methods can be used as well, such as DNA and eye retina data, but the fingerprint is the oldest biometric tool used for identification. However, another method that reveals a great deal about your identity is how you write or even type your name. For example, the password that you type to log on to your e-mail account may be just as identifiable as your fingerprint.
Try the following task as an example: go into any word processing program and type your name 10 times, once on each line, as I have done here:
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
I am not a skilled typist. I use my right middle finger to hold down the shift key and press the letter T with my left middle finger, then use my right middle finger to press i and my right index finger to press m, and then my right thumb to press the space bar. It is not very efficient, but I do it the same way every time I type my name. Regardless of how skilled a typist you are, you most likely repeat much the same process each of the 10 times you type your name.
Now, let's try to unravel the temporal “fingerprint” that you left behind when you typed your name. Suppose we conducted a simple analysis of the keystrokes and the time between each of the keystrokes that you made. The total time taken to type your name once is simply the time from the first keystroke to the last keystroke. At a more fine-grained level of analysis, the total time constitutes the time each key is held down (the dwell time) plus the time between the release of one key and the depression of the next (the transition time). All of the individual dwell times plus all of the transition times will sum to the total time.
In figure 7.5, I have plotted a hypothetical example of the 10 trials to type my name. The light gray bands denote the dwell times, and the darker gray bands indicate the transition times. I have ordered the lines of bands that represent the 10 trials from the fastest (least total time) at the top to the slowest trial at the bottom (most total time).
If you were to analyze each of the 10 repetitions of your name, you would probably find that all of the total times would be similar, but probably never exactly the same each time, just as in figure 7.5. The “noise” in our central nervous systems, plus other factors, causes the results to vary a little bit each time, resulting in some repetitions to be slower, and others faster, than the average total time.
But, take a closer look at each individual band in figure 7.5 as it changes over the 10 trials. What you will notice is that as the total time increases, each band gets proportionally longer too. We could express these numbers another way by dividing the time for each individual band by the total time for that trial to obtain a relative proportion of time represented by each band. Given the hypothetical evidence in figure 7.5, what we would discover is that the relative time for each band, expressed as a percentage, would stay roughly the same across all of the repetitions. Applying temporal and other methods to deconstruct how we type (e.g., key press forces) is called the study of keystroke dynamics.
According to generalized motor program theory, relative time is one of the key features of movement that is controlled by the central nervous system, especially for brief, rapid actions. Relative time is an expression of our motor control fingerprint. For well-learned tasks such as typing and handwriting, our central nervous system regulates the relative timing of impulses that are sent to the muscles that carry out these tasks. Various factors influence the real time on any particular instance. For example, using a keyboard that requires more force to depress the keys might result in overall slower times than would result using a keyboard that has a light touch, but the relative timing would likely remain the same.
Research in keystroke dynamics may result in the ability to identify people who carry out Internet fraud. In many ways it is similar to the use of handwriting dynamics to analyze the timing of the cursive expressions of a signature. It is fairly easy to forge the spatial representation of someone's signature, but very difficult to forge the temporal dynamics that result in that signature. These expressions of timing behavior represent our motor control fingerprints.
Self-Directed Learning Activities
1. Define the term generalized motor program in your own words.
2. How does the concept of a generalized motor program differ from the concept of a motor program as it was used in stories such as “Antilock Brakes” (in chapter 5) and “Point of No Return” (earlier in this
chapter)?
3. Suggest a method by which the handwriting dynamics of signatures could be used to detect fraud.
4. Some people contend that there is one generalized motor program for the full swing in golf, regardless of which club is used. How could a temporal dynamics analysis be used to assess this contention?
Notes
- An excellent review of fingerprint analysis by David Ashbaugh of the Royal Canadian Mounted Police is available here:
www.onin.com/fp/ridgeology.pdf
- A lot of controversy remains about invariances in motor performance; how they are measured; and what invariance, or lack of invariance, means in terms of motor programs. The following articles provide good arguments for the debate:
Gentner, D.R. (1987). Timing of skilled motor performance: Tests of the proportional duration model. Psychological Review, 94, 255-276.
Heuer, H. (1988). Testing the invariance of relative timing: Comment on Gentner (1987). Psychological Review, 95, 552-557.
Suggested Readings
Schmidt, R.A. (1975). A schema theory of discrete motor skill learning. Psychological Review, 82, 225-260.
Schmidt, R.A. (1985). The search for invariance in skilled movement behavior. Research Quarterly for Exercise and Sport, 56, 188-200.
Schmidt, R.A., & Lee, T.D. (2011). Central contributions to motor control. In Motor control and learning: A behavioral emphasis (5th ed., pp. 177-222) Champaign, IL: Human Kinetics.
The Farmers' Market
What are the roles of motor error and hypervigilance in unintended acceleration accidents?
The prosecution, at Weller's trial several years later, claimed it was no accident—that Weller deliberately drove his car through the crowded market. The reason, they said, was that Weller had been involved in a minor fender bender just moments before he entered the farmers' market. His response to the fender bender was to flee the scene of the accident. Witnesses for the prosecution painted Weller as a cold-blooded killer, commenting on the determined look on his face during the ordeal and his relatively calm demeanor afterward. It probably also didn't help his case when he emerged from the car immediately afterward and wondered aloud why the people he had hit had not jumped out of his way. Adding his age into the mix, Weller's actions were painted as rather pathetic.
Richard Schmidt, a renowned motor control scholar and human factors expert, testified on behalf of the defense team and argued that the facts of the case shared many similarities to accidents caused by errors of pedal misapplication, or unintended acceleration. Accidents of this type, which, thankfully, are quite rare, occur when the driver intends to apply pressure to the brake pedal, but misses the brake and pushes down on the accelerator instead.
Unintended acceleration accidents had been investigated for many years prior to the Weller case. These accidents were more common when the driver first got into the car, started the engine, and engaged the automatic transmission from the Park position into either Drive or Reverse. Indeed, such frequent episodes of “runaway cars” were the primary reason auto manufacturers added the brake-transmission shift interlock system in the 1980s so that an automatic transmission lever could not be moved from the Park position until the car sensed that a certain amount of pressure had been applied to the brake. However, the brake lock system only prevented this particular type of pedal misapplication; moving the foot to the accelerator instead of the brake would not be prevented once the transmission was successfully engaged out of Park and the car was in motion.
It is important to note that reaching for the brake requires that we steer the foot from a comfortable seated position to a target (the brake) in the absence of any visual guidance. We do this all the time without making any errors. We know where the brake is located from experience, and we also know that the brake feels different underfoot than does the accelerator. So then why would we suddenly miss the brake, push down on the accelerator instead, and then keep the foot there?
Schmidt (1989) presented evidence that unintended acceleration cases frequently involved accidents in which the driver had less experience than usual with that particular vehicle. Therefore, in some of these cases, the exact location and feel of the brake might not have been as familiar to the driver as normally could have been expected. These accidents also occurred more frequently on start-up, compared to later in the driving cycle, perhaps due to temporary factors associated with preparing an action (see “Shooting Two From the Line” in chapter 11). Driver inattention has also been linked with these cases, so it may not come as a surprise that drivers would not immediately notice the difference between the brake and gas pedal if engaged in a distracting activity at the same time (see “Gumbo” in chapter 6).
But for Weller, none of the common profiles for these accidents fit the case: he had already been driving before the supposed pedal misapplication error, so it was not a matter of missing the brake on initial start-up. Weller was quite familiar with his own vehicle, an 11-year-old Buick LeSabre. And he was not talking on a cell phone. Instead, Schmidt argued that Weller's pedal misapplication error was likely triggered by a catastrophic case of panic, termed hypervigilance, which could have been initiated when Weller had been involved in the fender bender just prior to the episode.
But one last issue seemed particularly problematic, according to the prosecution. Once the pedal misapplication error had occurred and the car started to accelerate wildly out of control, why didn't Weller simply remove his foot from the pedal or turn off the engine—actions that would have brought the car quickly to a stop? Again, failure to carry out corrective actions is typical of unintended acceleration cases, and some reasonable accounts have been offered to explain why drivers do not perform them. First, the driver probably does not realize that the foot is on the accelerator rather than the brake. The intention was to press the brake, and the fact that the pedal has gone all the way to the floor could reasonably be interpreted as brake failure rather than human error. And second, once the driver enters into this catastrophic state of panic, all normal modes of thinking cease. Reasoning and problem solving, the kinds of activities that are easy to do when unflustered, become unlikely, if not impossible, to carry out when in this state of hypervigilance.
Thomas Shelton, a member of the California Highway Patrol, testified at Weller's trial that he once investigated an unintended acceleration case in which an elderly woman ended up driving her car onto the top of another vehicle. The woman was in such a panicked state that when Shelton arrived at the accident scene, he had to climb up into the car to shut off the racing engine, at which time he noticed that the woman was still seated, very much alive, staring straight ahead with a death grip on the steering wheel, and with her foot still pushing the accelerator all the way to the floor.
Unfortunately, all of these arguments can only be used to speculate about what may have occurred in George Russell Weller's Buick LeSabre on that fateful day. On October 20, 2006, the jury convicted him of vehicular manslaughter in the 10 deaths resulting from the Santa Monica farmers' market crash. Nobody will ever know whether the verdict was the right one.
Self-Directed Learning Activities
1. In your own words describe the phenomenon known as unintended acceleration.
2. Describe a situation in which you made an action error that you were able to correct. How did you know that you had made the error, and what did you do to correct it?
3. Using our feet to manipulate car pedals involves aiming without visual feedback. What factors influence our ability to make these aiming movements accurately?
4. Propose a research methodology that examines one's ability to (a) move to a target without visual feedback and (b) estimate the accuracy of those aimed movements (again, without vision).
Notes
- Evidence from George Weller's trial during September and October of 2006 was summarized in the Santa Monica Daily Press, which can be accessed through its archives:
- Not all cases of unintended acceleration are generally agreed to be the result of a pedal misapplication. A segment of the television show 60 Minutes, hosted by Ed Bradley and which aired November 22, 1986, claimed that accidents of similar etiology involving the Audi 5000 were the result of a faulty idle stabilizer, which caused the car to accelerate wildly out of control when put into gear. An investigation by the U.S. NHTSA (National Highway Traffic Safety Administration) failed to support 60 Minutes' claim.
- More recently, runaway Toyotas have been a topic of concern. Once again, a government investigation failed to support a claim that these cases of unintended acceleration were due to an electronic fault in the engine. Again, this leaves open the very real possibility that driver error is to blame, as suggested by Richard Schmidt in the New York Times:
Suggested Readings
Castelli, J., Nash, C., Ditlow, C., & Pecht, M. (2003). Sudden acceleration: The myth of driver error. University of Maryland: CALCE EPSC Press.
Pollard, J., & Sussman, E.D. (1989). An examination of sudden acceleration. National Highway Traffic Safety Administration Final Report # DOT-TSC-NHTSA-89-1.
Schmidt, R.A. (1989). Unintended acceleration: A review of human factors contributions. Human Factors, 31, 345-364.
Schmidt, R.A. (1993). Unintended acceleration: Human performance considerations. In B. Peacock & W. Karwowski (Eds.), Automotive ergonomics (pp. 431-451). London: Taylor & Francis.
Schmidt, R.A., & Lee, T.D. (2011). Central contributions to motor control. In Motor control and learning: A behavioral emphasis (5th ed., pp. 177-222) Champaign, IL: Human Kinetics.
Push or Pull?
How do product designs influence people to perform specific actions?
It seems hard to believe, but brand-new buildings are still being constructed with very silly design features. For example, a new building on our campus opened last year and featured many modern technologies, including very nice-looking glass doors through which you enter and exit the building. The problem is that the doors have identical handles on either side. The handles themselves are very attractive, but they don't tell me what to do when I want to enter the building—should I push or pull the door? The handles on these particular doors suggest to me that they are made for grasping and pulling. However, the door goes in only one direction, and the very same handles appear on both sides of the door. Therefore, to enter the building, I have to use the handle to pull the door, and to exit, I have to use the identical type of handle to push the door. How would someone possibly know which way the door swings when approaching it? After a year of using these doors on a daily basis, I still just guess about whether to push or pull. Why would someone design a door like that?
Michael Darnell has created a wonderful website that illustrates many instances of products that have been designed without the user in mind. One of his web pages describes a door handle problem similar to the one I just described, in which a building has a short walkway with one set of doors on either end. Each door has identical handles on either side of the door (the same type of handle is used for pushing and pulling the door). However, according to Darnell, these particular doors were installed such that the problem was magnified even more—the doors at either end of the walkway both swung out from the walkway. He tells the story of a friend who entered the walkway by pulling on the handle of one door; then went to the second door and found that the door would not open when she pulled on the handle. So, concluding that the door was locked, she returned to the first door to exit and found herself trapped in the walkway when the door wouldn't move when she pulled on its handle. After trying to get the attention of others on the outside to tell them that she was trapped in the walkway, she finally realized, with much embarrassment, that the handles had to be pushed when inside the walkway. Darnell reminds us that this funny story could have had tragic consequences if, for example, a panicked person were trying to flee the building in case of a fire.
There are many ways to solve this problem of having identical handles used for opposite actions. One solution would be to put signs on the door that say “push” or “pull.” Although these signs would probably work, a well-designed door (i.e., designed for the user rather than for its looks) should not need a sign to tell people how to use it. Pushing and pulling a door are not rocket science, after all. A better solution would be to install a flat plate on one side of the door, which would make the correct choice obvious: Because there is no handle to grab to pull toward you, there is only one other option—push! A sign would be unnecessary in this case because there would be no question about which action to take.
Signs and symbols that tell us how to use things should be the last option in technology design. Signs are sometimes small and hard to read, they fall off, they wear out, other things cover them up, and they are often presented in only one language, making them less than desirable in a multicultural society. Symbols avoid being unilingual but introduce other problems, such as ambiguity. For example, how would one create simple, unambiguous, and instantly recognizable symbols for “push” and “pull”? As in the door handle example, a better design would be one that made the correct action as obvious as possible, one that would negate the need for a sign or symbol.
A classic example of the failure of technology to exploit this simple idea is a standard topic in ergonomics textbooks and on websites: the arrangement of burners on the stove top. A typical stove top layout is shown in figure 1.5.
The four knobs used to turn on the elements are located at the top of the figure. Quick, say out loud which one is used to turn on the right front burner. My guess is that it is one of the two rightmost knobs. But which one? It is impossible to know for certain without turning one of the knobs and waiting to see which element begins to heat up. So, designers add little labels to tell us which knobs are mapped to which burners. Do they work?
My personal experience is that these labels, if they have not fallen off, worn out, been covered by last night's pasta sauce, or been printed too small to read, are still hard to interpret. More times than I wish to remember, I have returned to a stove expecting to find a pot of boiling water for my pasta but instead found a glowing burner with nothing on it. The relative failure rate of these labels leads me to believe that they are only slightly more helpful than if I had just turned a knob at random.
Have a look at the two alternative stove top designs in figure 1.6. The layout of the knobs in figure 1.6a is the same as before; only the burner layout has been altered. In figure 1.6b, the burner layout is the same as it was before, but now the arrangement of the knobs has been altered.
In both cases, the mapping of each knob to its associated burner has been made obvious by a simple rearrangement of either the knobs or the burner layout. No symbols or labels are needed, and the chances of turning on the wrong burner are minimal. Moreover, the amount of space occupied by the burners and knobs in these more compatible layouts is identical to the space occupied in the incompatible layout. It is interesting to note that these alternative layouts have been available for many years. But have a look at any appliance store flyer or website, and you will see that the classic design still continues to dominate the market.
Sometimes it is simply not possible to avoid the use of labels or signs. In such cases the designer's goal should be to make them as informative as possible. The typical ceiling fan is another good example of a failure to consider the needs of the user when designing the product. The ceiling fan that we have in our home has a wall switch that turns the fan off and on. But the ceiling fan has three additional controls, located on the base of the fan itself (where the fan connects to the ceiling), that are operated by pulling on long strings that hang down from the base. These three strings can be pulled to adjust the speed of the fan, to reverse its direction, or to turn off or on a light located on the bottom of the fan. The problem is that the strings are identical in appearance; I have no way of knowing which control I am activating when I pull on any of the strings.
Actually, that is not completely true. If I stand on a stepladder and get up really close to the ceiling, I can see little labels beside the holes where the strings exit the electrical base of the fan that say “speed,” “light,” and “direction.” I could also stand on the floor and use binoculars to read these labels, I suppose. But, you get the point. Because these labels are next to useless to me, I really just have a one-in-three chance of getting it right each time I pull one of the strings.
How could each control be designed to communicate its function easily to the user? I have used two strategies to modify my own fan. Because reversing the fan's direction is the control that I use least often, I have shortened the string associated with this option to be the most difficult of the three to reach. However, I use the light and speed options about equally often. So, to make these as distinct as possible I borrowed two charms from an old charm bracelet and attached them to the ends of each string. A charm of a rabbit is attached to the string that controls the speed of the fan, and a charm of a book (for which I turn on the light to read) is attached to the string that controls the light. The fan may not look the way the manufacturer had intended it to look, but I have not made the mistake of pulling the wrong string since I attached these charms.
Motor skills such as pushing doors, turning knobs, and pulling strings are simple to learn, and we have all mastered these simple skills over our lifetimes. So why do we have such trouble using them? One reason is that the manufacturer is often paying more attention to aesthetics than to the needs of the user.
Self-Directed Learning Activities
1. Describe, in general, the principle of how product design influences human performance.
2. Search the literature for the term stimulus-response compatibility; then define it in your own words with specific reference to an example presented in this story.
3. Over the next 24 hours, keep a diary of all the objects or things you encounter that you think represent stimulus-response incompatibility. Propose ways each of these things could be made more compatible.
4. Find three research articles in which stimulus-response compatibility has been investigated. What are the similarities and differences in the stimuli and responses used in these studies?
Notes
- Michael Darnell has a wonderful website that features many examples of poorly designed products, with suggestions about simple ways to greatly improve usability: www.baddesigns.com
Suggested Readings
Proctor, R.W., & Van Zandt, T. (2008). Human factors in simple and complex systems (2nd ed.). Boca Raton, FL: CRC Press.
Proctor, R.W., & Vu, K.P.L. (2006). Stimulus-response compatibility principles: Data, theory, and application. Boca Raton, FL: CRC Press.
Schmidt, R.A., & Lee, T.D. (2011). Human information processing. In Motor control and learning: A behavioral emphasis (5th ed., pp. 57-96) Champaign, IL: Human Kinetics.
Learning to win from losing
Why is the learning–performance distinction important?
One day, Michelle Wie might be remembered as one of the greatest golfers of all time. At the age of 14 she was pounding out 300-yard (274 m) drives and not only competing in Ladies Professional (LPGA) Tour events, but also performing quite well in them. Her talent is undeniable. But controversy surrounded this exceptional athlete because of her desire to compete in PGA Tour events, which traditionally have included only male competitors. Wie has performed admirably in some of these events. For example, in the 2004 Sony Open, she shot rounds of 72 and 68 and missed the 36-hole cut by just a single stroke. Her two-round total of 140 placed her in a tie for 80th, which was better than 53 male professionals, including PGA stars Zach Johnson, Hunter Mahan, and Adam Scott. However, Wie has not performed well in most of the other PGA events she has entered, including a 139th-place finish (out of 141 players) at the 2007 Sony Open.
Many argued that Wie should stop competing in PGA Tour events and concentrate her efforts on the LPGA Tour. The typical argument was that her skill development as a golfer was being stalled by repeated failures in PGA Tour events, and that if she concentrated on LPGA events instead, the greater chance for success would escalate her skill development.
I argue that there is a fundamental flaw in this theory, because it falls into a trap known as the learning-performance distinction. The trap underlies one of the most misunderstood concepts in motor learning research, so let's start by defining the two terms. Motor skills researchers use the term performance to refer to a single observation. It could be a score or outcome that reflects the value of a single attempt at a motor skill, or perhaps an average score that statistically summarizes a number of attempts. An 18- or 36-hole score, or a final placement ranking in a tournament, might indicate a representative performance score in the case of a golfer. The term learning is used quite specifically to refer to a stable improvement in skill over time—an improvement that has specifically occurred as the result of practice. The problem illustrated by the criticisms leveled at Michelle Wie is that the critics fail to consider the difference between performance and learning; they mistakenly confuse her failures to improve her performance in men's PGA events as a failure to learn from these experiences. Let's use another example to illustrate this distinction between performance and learning.
Suppose you like to bowl, play regularly in a league, and also practice occasionally. Two years ago your average was 110, last year it was 120, and this year it is 130. Those averages probably indicate that you have become a better bowler as a result of learning: the scores indicate improvements that appear to be rather stable and that have resulted from practice.
So, does this mean that you will score a 130 the next time you bowl? Not necessarily, because there are many reasons for the fluctuations in scores that occur from game to game. Next time you may score well below your average because you are not feeling well, your shoes are too tight, or the crowd is very noisy that night. But this does not mean that you have suddenly lost some of your learned skill. You may score well above your average because you try extra hard to impress someone, or because everything just seems to be well focused (your mojo is working). As before, this does not mean that you have suddenly had a change in learned skill, because there is no indication that the sudden improvement in performance reflects a stable and permanent change in your capability to perform.
At any one time we have a theoretical capability for attaining a certain level of performance. When that theoretical capability changes to a higher level as a result of practice, we can say with confidence that we have learned. The confusion lies in the fact that individual performances may sometimes exceed or not live up to these theoretical capabilities. These fluctuations are expected and in no way diminish or detract from the performer's theoretical capability to perform at a certain skill level.
What I have described is the typical distinction between learning and performance. But, there is another side to this issue, the one that can be applied to Michelle Wie. The issue concerns the situation in which performances do not appear to change, or appear to be getting worse. Does that mean that learning is not occurring? Let's go back to the bowling example.
Suppose that this year your average was 120, which is the same as it was last year. You may take the absence of a stable improvement, despite all of your practice, as evidence that your learning has stalled. But, as it turns out, last year your league bowled Tuesday nights, which is your day off. This year you bowl on Wednesday nights, one hour after completing your 10-hour work shift. You are typically tired and hungry on league nights, and your bowling scores suffered this year as a result. What does all this suggest? What is your theoretical capability to bowl, and has this changed from last year, despite the change in your bowling night? Could it be that you actually have improved (i.e., learned), but the feeling of fatigue and other factors directly related to your work schedule prevented you from actually performing up to these expectations?
The point here is that making inferences about learning requires evidence that goes beyond a single performance or sometimes even a set of performances. Under what circumstances were these performances observed? Are there mitigating factors that might explain why the theoretical true score differs from the observed score? The absence of any observable change in performance does not mean that the unobservable (that theoretical capability to perform at a certain level of skill) has not improved.
Michelle Wie took a very different route in the development of her golfing skill. Her decision to compete in PGA events placed enormous pressures on her to perform, and only an exceptional result would have convinced the public that she was benefiting from these extreme challenges. All we could observe were her scores in these events. The unobservable, however, what she has learned by playing in these events, can never be truly understood. I suspect that if she achieves the rank of one of the greatest players of all time, it will be due in no small measure to these experiences competing against the very best golfers in the world.
Self-Directed Learning Activities
1. In your own words, explain the distinction between performance and learning.
2. What is the difference between a theoretical true score and an observed score?
3. Suggest a reason a person's observed score might exceed the true score, and a reason the observed score might fail to achieve the level of the true score.
4. Identify another sport, athlete, or situation in which the failure to account for the learning-performance distinction has resulted in an inappropriate conclusion.
Notes
- Babe Zaharias (see “The Babe”) was the first woman to compete in a men's PGA event.
- Here is a sampling of some of the criticism of Michele Wie's decision to compete in PGA Tour events, from two sports commentary sites:
- Michelle Wie won her second LPGA Tour title in Winnipeg, Manitoba, at the 2010 CN Canadian Women's Open.
Suggested Readings
Schmidt, R.A. (1972). The case against learning and forgetting scores. Journal of Motor Behavior, 4, 79-88.
Schmidt, R.A., & Lee, T.D. (2011). Motor learning concepts and research methods. In Motor control and learning: A behavioral emphasis (5th ed., pp. 327-346). Champaign, IL: Human Kinetics.
Cutting wood and missing putts
How are constant error, variable error, and absolute error useful for understanding motor control?
Two things I like to do—cutting wood and golfing—remind me often of the importance of measures of central tendency and variability in motor control performance. For example, I use my chainsaw to cut up fallen trees to burn in the woodstove. Cutting logs into burnable lengths is not an exact science. In the middle of a Canadian winter, when the temperature is cold and plenty of snow blankets the ground, it is generally considered a good idea to keep moving. I do not take time to measure the lengths of the logs I saw; rather, I make a quick visual estimate and then start to make my cut. But, in doing so, I must remember two important things related to the mean and the standard deviation of the logs I cut. First, because our woodstove is only large enough to hold pieces of wood that are 50 centimeters (20 in.) or shorter, any pieces that are longer than that are useless. And second, wood is easier to stack when the pieces are all roughly of the same length. Essentially, these features remind me that the mean and standard deviation are both important; I must have a mean that is less than 50 centimeters and a standard deviation that is as small as possible.
When considering data in terms of a specific goal or standard, it is often better to express the mean and standard deviation in terms of error measures—in my case, relative to a goal wood length of 50 centimeters. For example, wood lengths of 45, 42, 48, 47, and 52 centimeters would be expressed as lengths that are -5, -8, -2, -3, and +2 centimeters, respectively, relative to my goal wood length of 50 centimeters. If we calculated the mean of these lengths of wood, then we could express the stack of wood as having either an average length of 46.8 centimeters, or the error scores as having a mean constant error of -3.2 centimeters. Having a negative constant error is good in this case, because the average piece of wood that I have cut will easily fit in the woodstove (i.e., the mean is less than 50 centimeters).
The careful reader will have noted, however, that even though my mean constant error is negative, there is still one block of wood that will not fit (the 52 cm log). Thus, by itself, the mean constant error score (-3.2 cm) is misleading, because it does not provide any indication of how many logs might not actually fit in the woodstove. Our measure of variability, the standard deviation, gives us some indication of this information. The measure of variability of these error scores (called the variable error) represents the deviation of the error score for each log relative to the mean constant error and is calculated in the same way as is a standard deviation. In this case, the variable error is 3.3 centimeters. In general, when I am cutting wood, I want to achieve a negative mean constant error with a variable error that is as small as possible.
Although constant error provides a useful heuristic measure of average performance in cutting wood with a chainsaw, it is a quite misleading measure of central tendency in another activity I like to do. Let's say that I have struck 10 golf putts—5 of these putts go past the hole by 5, 10, 15, 20, and 25 centimeters, respectively, and the other five putts come up short of the hole by equal amounts (-5, -10, -15, -20 and -25 cm). On average, by how much have I missed the hole with these putts? If you were to calculate the average as in the previous wood-cutting example (the sum of the 10 individual error scores, divided by 10) the answer would be a mean constant error of 0. In other words, my “average” putt ended up in the hole. However, we know that this answer surely must be wrong because none of the individual putts actually went in the hole. So, we must use a different way to express these error scores to avoid an answer that makes no sense.
The problem that we sometimes run into with constant error is that it provides a measure of average bias, the tendency to err in a specific way (e.g., by too much or too little; too far left versus too far right). In some cases a specific bias is desirable, such as the tendency to undercut a wood length of 50 centimeters so that all of my wood pieces fit in the woodstove. In the case of my putts that have no consistent bias, we are much better off using a measure of central tendency that removes the bias from each score prior to calculating the mean. Such a procedure is called using the absolute (unsigned) scores. Hence, this measure of central tendency is called the mean absolute error. Here, our measure of central tendency is the mean of the unsigned scores (5, 5, 10, 10, 15, 15, 20, 20, 25, 25), which is 15 centimeters. That is, the golf putts tended to miss the hole by an average of 15 centimeters. In this particular case, the mean absolute error score (15 cm from the hole) represents a more accurate picture of the entire set of individual scores than what is represented in the mean constant error (in the hole).
Averages are convenient ways to express how we tend to perform. But, numbers that represent central tendencies can be misleading if the tendency itself is not a strong one. Measures of variability are one way to identify the strength of a tendency. Depending on the context, different methods of calculating central tendencies may characterize the data in more representative ways.
Self-Directed Learning Activities
1. Define constant error, absolute error, and variable error in your own words.
2. What do the terms algebraic error and total error (or Henry's E) refer to? How are they calculated, and how are they similar to the terms constant error and absolute error, respectively?
3. Look up and briefly describe a research investigation that uses at least two error measures. Why do you think the researchers used these measures in particular?
4. Conduct a brief study on yourself. Close your eyes and, using a pencil, draw five lines as close to 4 inches (10 cm) as possible, one below the previous one, without opening your eyes until you have drawn the fifth line. Measure and record the length of each line you drew; then calculate all of the error measures that have been discussed.
Notes
- Here is a web-based standard deviation calculator:
www.tinyurl.com/standarddeviationcalculator
Suggested Readings
Chapanis, A. (1951). Theory and methods for analyzing errors in man-machine systems. Annals of the New York Academy of Sciences, 51, 1179-1203.
Schmidt, R.A., & Lee, T.D. (2011). Methodology for studying motor performance. In Motor control and learning: A behavioral emphasis (5th ed., pp. 21-56). Champaign, IL: Human Kinetics.
Forensic motor control
What are generalized motor programs, and what do keystroke dynamics reveal about them?
Most of what I know about forensic identification comes from crime shows that I've seen on TV. The suspect is caught because a fingerprint was left on the murder weapon. The evidence is used to convict the suspect because no two fingerprints have ever been found to be identical. Crime shows tell us that other forensic methods can be used as well, such as DNA and eye retina data, but the fingerprint is the oldest biometric tool used for identification. However, another method that reveals a great deal about your identity is how you write or even type your name. For example, the password that you type to log on to your e-mail account may be just as identifiable as your fingerprint.
Try the following task as an example: go into any word processing program and type your name 10 times, once on each line, as I have done here:
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
I am not a skilled typist. I use my right middle finger to hold down the shift key and press the letter T with my left middle finger, then use my right middle finger to press i and my right index finger to press m, and then my right thumb to press the space bar. It is not very efficient, but I do it the same way every time I type my name. Regardless of how skilled a typist you are, you most likely repeat much the same process each of the 10 times you type your name.
Now, let's try to unravel the temporal “fingerprint” that you left behind when you typed your name. Suppose we conducted a simple analysis of the keystrokes and the time between each of the keystrokes that you made. The total time taken to type your name once is simply the time from the first keystroke to the last keystroke. At a more fine-grained level of analysis, the total time constitutes the time each key is held down (the dwell time) plus the time between the release of one key and the depression of the next (the transition time). All of the individual dwell times plus all of the transition times will sum to the total time.
In figure 7.5, I have plotted a hypothetical example of the 10 trials to type my name. The light gray bands denote the dwell times, and the darker gray bands indicate the transition times. I have ordered the lines of bands that represent the 10 trials from the fastest (least total time) at the top to the slowest trial at the bottom (most total time).
If you were to analyze each of the 10 repetitions of your name, you would probably find that all of the total times would be similar, but probably never exactly the same each time, just as in figure 7.5. The “noise” in our central nervous systems, plus other factors, causes the results to vary a little bit each time, resulting in some repetitions to be slower, and others faster, than the average total time.
But, take a closer look at each individual band in figure 7.5 as it changes over the 10 trials. What you will notice is that as the total time increases, each band gets proportionally longer too. We could express these numbers another way by dividing the time for each individual band by the total time for that trial to obtain a relative proportion of time represented by each band. Given the hypothetical evidence in figure 7.5, what we would discover is that the relative time for each band, expressed as a percentage, would stay roughly the same across all of the repetitions. Applying temporal and other methods to deconstruct how we type (e.g., key press forces) is called the study of keystroke dynamics.
According to generalized motor program theory, relative time is one of the key features of movement that is controlled by the central nervous system, especially for brief, rapid actions. Relative time is an expression of our motor control fingerprint. For well-learned tasks such as typing and handwriting, our central nervous system regulates the relative timing of impulses that are sent to the muscles that carry out these tasks. Various factors influence the real time on any particular instance. For example, using a keyboard that requires more force to depress the keys might result in overall slower times than would result using a keyboard that has a light touch, but the relative timing would likely remain the same.
Research in keystroke dynamics may result in the ability to identify people who carry out Internet fraud. In many ways it is similar to the use of handwriting dynamics to analyze the timing of the cursive expressions of a signature. It is fairly easy to forge the spatial representation of someone's signature, but very difficult to forge the temporal dynamics that result in that signature. These expressions of timing behavior represent our motor control fingerprints.
Self-Directed Learning Activities
1. Define the term generalized motor program in your own words.
2. How does the concept of a generalized motor program differ from the concept of a motor program as it was used in stories such as “Antilock Brakes” (in chapter 5) and “Point of No Return” (earlier in this
chapter)?
3. Suggest a method by which the handwriting dynamics of signatures could be used to detect fraud.
4. Some people contend that there is one generalized motor program for the full swing in golf, regardless of which club is used. How could a temporal dynamics analysis be used to assess this contention?
Notes
- An excellent review of fingerprint analysis by David Ashbaugh of the Royal Canadian Mounted Police is available here:
www.onin.com/fp/ridgeology.pdf
- A lot of controversy remains about invariances in motor performance; how they are measured; and what invariance, or lack of invariance, means in terms of motor programs. The following articles provide good arguments for the debate:
Gentner, D.R. (1987). Timing of skilled motor performance: Tests of the proportional duration model. Psychological Review, 94, 255-276.
Heuer, H. (1988). Testing the invariance of relative timing: Comment on Gentner (1987). Psychological Review, 95, 552-557.
Suggested Readings
Schmidt, R.A. (1975). A schema theory of discrete motor skill learning. Psychological Review, 82, 225-260.
Schmidt, R.A. (1985). The search for invariance in skilled movement behavior. Research Quarterly for Exercise and Sport, 56, 188-200.
Schmidt, R.A., & Lee, T.D. (2011). Central contributions to motor control. In Motor control and learning: A behavioral emphasis (5th ed., pp. 177-222) Champaign, IL: Human Kinetics.
The Farmers' Market
What are the roles of motor error and hypervigilance in unintended acceleration accidents?
The prosecution, at Weller's trial several years later, claimed it was no accident—that Weller deliberately drove his car through the crowded market. The reason, they said, was that Weller had been involved in a minor fender bender just moments before he entered the farmers' market. His response to the fender bender was to flee the scene of the accident. Witnesses for the prosecution painted Weller as a cold-blooded killer, commenting on the determined look on his face during the ordeal and his relatively calm demeanor afterward. It probably also didn't help his case when he emerged from the car immediately afterward and wondered aloud why the people he had hit had not jumped out of his way. Adding his age into the mix, Weller's actions were painted as rather pathetic.
Richard Schmidt, a renowned motor control scholar and human factors expert, testified on behalf of the defense team and argued that the facts of the case shared many similarities to accidents caused by errors of pedal misapplication, or unintended acceleration. Accidents of this type, which, thankfully, are quite rare, occur when the driver intends to apply pressure to the brake pedal, but misses the brake and pushes down on the accelerator instead.
Unintended acceleration accidents had been investigated for many years prior to the Weller case. These accidents were more common when the driver first got into the car, started the engine, and engaged the automatic transmission from the Park position into either Drive or Reverse. Indeed, such frequent episodes of “runaway cars” were the primary reason auto manufacturers added the brake-transmission shift interlock system in the 1980s so that an automatic transmission lever could not be moved from the Park position until the car sensed that a certain amount of pressure had been applied to the brake. However, the brake lock system only prevented this particular type of pedal misapplication; moving the foot to the accelerator instead of the brake would not be prevented once the transmission was successfully engaged out of Park and the car was in motion.
It is important to note that reaching for the brake requires that we steer the foot from a comfortable seated position to a target (the brake) in the absence of any visual guidance. We do this all the time without making any errors. We know where the brake is located from experience, and we also know that the brake feels different underfoot than does the accelerator. So then why would we suddenly miss the brake, push down on the accelerator instead, and then keep the foot there?
Schmidt (1989) presented evidence that unintended acceleration cases frequently involved accidents in which the driver had less experience than usual with that particular vehicle. Therefore, in some of these cases, the exact location and feel of the brake might not have been as familiar to the driver as normally could have been expected. These accidents also occurred more frequently on start-up, compared to later in the driving cycle, perhaps due to temporary factors associated with preparing an action (see “Shooting Two From the Line” in chapter 11). Driver inattention has also been linked with these cases, so it may not come as a surprise that drivers would not immediately notice the difference between the brake and gas pedal if engaged in a distracting activity at the same time (see “Gumbo” in chapter 6).
But for Weller, none of the common profiles for these accidents fit the case: he had already been driving before the supposed pedal misapplication error, so it was not a matter of missing the brake on initial start-up. Weller was quite familiar with his own vehicle, an 11-year-old Buick LeSabre. And he was not talking on a cell phone. Instead, Schmidt argued that Weller's pedal misapplication error was likely triggered by a catastrophic case of panic, termed hypervigilance, which could have been initiated when Weller had been involved in the fender bender just prior to the episode.
But one last issue seemed particularly problematic, according to the prosecution. Once the pedal misapplication error had occurred and the car started to accelerate wildly out of control, why didn't Weller simply remove his foot from the pedal or turn off the engine—actions that would have brought the car quickly to a stop? Again, failure to carry out corrective actions is typical of unintended acceleration cases, and some reasonable accounts have been offered to explain why drivers do not perform them. First, the driver probably does not realize that the foot is on the accelerator rather than the brake. The intention was to press the brake, and the fact that the pedal has gone all the way to the floor could reasonably be interpreted as brake failure rather than human error. And second, once the driver enters into this catastrophic state of panic, all normal modes of thinking cease. Reasoning and problem solving, the kinds of activities that are easy to do when unflustered, become unlikely, if not impossible, to carry out when in this state of hypervigilance.
Thomas Shelton, a member of the California Highway Patrol, testified at Weller's trial that he once investigated an unintended acceleration case in which an elderly woman ended up driving her car onto the top of another vehicle. The woman was in such a panicked state that when Shelton arrived at the accident scene, he had to climb up into the car to shut off the racing engine, at which time he noticed that the woman was still seated, very much alive, staring straight ahead with a death grip on the steering wheel, and with her foot still pushing the accelerator all the way to the floor.
Unfortunately, all of these arguments can only be used to speculate about what may have occurred in George Russell Weller's Buick LeSabre on that fateful day. On October 20, 2006, the jury convicted him of vehicular manslaughter in the 10 deaths resulting from the Santa Monica farmers' market crash. Nobody will ever know whether the verdict was the right one.
Self-Directed Learning Activities
1. In your own words describe the phenomenon known as unintended acceleration.
2. Describe a situation in which you made an action error that you were able to correct. How did you know that you had made the error, and what did you do to correct it?
3. Using our feet to manipulate car pedals involves aiming without visual feedback. What factors influence our ability to make these aiming movements accurately?
4. Propose a research methodology that examines one's ability to (a) move to a target without visual feedback and (b) estimate the accuracy of those aimed movements (again, without vision).
Notes
- Evidence from George Weller's trial during September and October of 2006 was summarized in the Santa Monica Daily Press, which can be accessed through its archives:
- Not all cases of unintended acceleration are generally agreed to be the result of a pedal misapplication. A segment of the television show 60 Minutes, hosted by Ed Bradley and which aired November 22, 1986, claimed that accidents of similar etiology involving the Audi 5000 were the result of a faulty idle stabilizer, which caused the car to accelerate wildly out of control when put into gear. An investigation by the U.S. NHTSA (National Highway Traffic Safety Administration) failed to support 60 Minutes' claim.
- More recently, runaway Toyotas have been a topic of concern. Once again, a government investigation failed to support a claim that these cases of unintended acceleration were due to an electronic fault in the engine. Again, this leaves open the very real possibility that driver error is to blame, as suggested by Richard Schmidt in the New York Times:
Suggested Readings
Castelli, J., Nash, C., Ditlow, C., & Pecht, M. (2003). Sudden acceleration: The myth of driver error. University of Maryland: CALCE EPSC Press.
Pollard, J., & Sussman, E.D. (1989). An examination of sudden acceleration. National Highway Traffic Safety Administration Final Report # DOT-TSC-NHTSA-89-1.
Schmidt, R.A. (1989). Unintended acceleration: A review of human factors contributions. Human Factors, 31, 345-364.
Schmidt, R.A. (1993). Unintended acceleration: Human performance considerations. In B. Peacock & W. Karwowski (Eds.), Automotive ergonomics (pp. 431-451). London: Taylor & Francis.
Schmidt, R.A., & Lee, T.D. (2011). Central contributions to motor control. In Motor control and learning: A behavioral emphasis (5th ed., pp. 177-222) Champaign, IL: Human Kinetics.
Push or Pull?
How do product designs influence people to perform specific actions?
It seems hard to believe, but brand-new buildings are still being constructed with very silly design features. For example, a new building on our campus opened last year and featured many modern technologies, including very nice-looking glass doors through which you enter and exit the building. The problem is that the doors have identical handles on either side. The handles themselves are very attractive, but they don't tell me what to do when I want to enter the building—should I push or pull the door? The handles on these particular doors suggest to me that they are made for grasping and pulling. However, the door goes in only one direction, and the very same handles appear on both sides of the door. Therefore, to enter the building, I have to use the handle to pull the door, and to exit, I have to use the identical type of handle to push the door. How would someone possibly know which way the door swings when approaching it? After a year of using these doors on a daily basis, I still just guess about whether to push or pull. Why would someone design a door like that?
Michael Darnell has created a wonderful website that illustrates many instances of products that have been designed without the user in mind. One of his web pages describes a door handle problem similar to the one I just described, in which a building has a short walkway with one set of doors on either end. Each door has identical handles on either side of the door (the same type of handle is used for pushing and pulling the door). However, according to Darnell, these particular doors were installed such that the problem was magnified even more—the doors at either end of the walkway both swung out from the walkway. He tells the story of a friend who entered the walkway by pulling on the handle of one door; then went to the second door and found that the door would not open when she pulled on the handle. So, concluding that the door was locked, she returned to the first door to exit and found herself trapped in the walkway when the door wouldn't move when she pulled on its handle. After trying to get the attention of others on the outside to tell them that she was trapped in the walkway, she finally realized, with much embarrassment, that the handles had to be pushed when inside the walkway. Darnell reminds us that this funny story could have had tragic consequences if, for example, a panicked person were trying to flee the building in case of a fire.
There are many ways to solve this problem of having identical handles used for opposite actions. One solution would be to put signs on the door that say “push” or “pull.” Although these signs would probably work, a well-designed door (i.e., designed for the user rather than for its looks) should not need a sign to tell people how to use it. Pushing and pulling a door are not rocket science, after all. A better solution would be to install a flat plate on one side of the door, which would make the correct choice obvious: Because there is no handle to grab to pull toward you, there is only one other option—push! A sign would be unnecessary in this case because there would be no question about which action to take.
Signs and symbols that tell us how to use things should be the last option in technology design. Signs are sometimes small and hard to read, they fall off, they wear out, other things cover them up, and they are often presented in only one language, making them less than desirable in a multicultural society. Symbols avoid being unilingual but introduce other problems, such as ambiguity. For example, how would one create simple, unambiguous, and instantly recognizable symbols for “push” and “pull”? As in the door handle example, a better design would be one that made the correct action as obvious as possible, one that would negate the need for a sign or symbol.
A classic example of the failure of technology to exploit this simple idea is a standard topic in ergonomics textbooks and on websites: the arrangement of burners on the stove top. A typical stove top layout is shown in figure 1.5.
The four knobs used to turn on the elements are located at the top of the figure. Quick, say out loud which one is used to turn on the right front burner. My guess is that it is one of the two rightmost knobs. But which one? It is impossible to know for certain without turning one of the knobs and waiting to see which element begins to heat up. So, designers add little labels to tell us which knobs are mapped to which burners. Do they work?
My personal experience is that these labels, if they have not fallen off, worn out, been covered by last night's pasta sauce, or been printed too small to read, are still hard to interpret. More times than I wish to remember, I have returned to a stove expecting to find a pot of boiling water for my pasta but instead found a glowing burner with nothing on it. The relative failure rate of these labels leads me to believe that they are only slightly more helpful than if I had just turned a knob at random.
Have a look at the two alternative stove top designs in figure 1.6. The layout of the knobs in figure 1.6a is the same as before; only the burner layout has been altered. In figure 1.6b, the burner layout is the same as it was before, but now the arrangement of the knobs has been altered.
In both cases, the mapping of each knob to its associated burner has been made obvious by a simple rearrangement of either the knobs or the burner layout. No symbols or labels are needed, and the chances of turning on the wrong burner are minimal. Moreover, the amount of space occupied by the burners and knobs in these more compatible layouts is identical to the space occupied in the incompatible layout. It is interesting to note that these alternative layouts have been available for many years. But have a look at any appliance store flyer or website, and you will see that the classic design still continues to dominate the market.
Sometimes it is simply not possible to avoid the use of labels or signs. In such cases the designer's goal should be to make them as informative as possible. The typical ceiling fan is another good example of a failure to consider the needs of the user when designing the product. The ceiling fan that we have in our home has a wall switch that turns the fan off and on. But the ceiling fan has three additional controls, located on the base of the fan itself (where the fan connects to the ceiling), that are operated by pulling on long strings that hang down from the base. These three strings can be pulled to adjust the speed of the fan, to reverse its direction, or to turn off or on a light located on the bottom of the fan. The problem is that the strings are identical in appearance; I have no way of knowing which control I am activating when I pull on any of the strings.
Actually, that is not completely true. If I stand on a stepladder and get up really close to the ceiling, I can see little labels beside the holes where the strings exit the electrical base of the fan that say “speed,” “light,” and “direction.” I could also stand on the floor and use binoculars to read these labels, I suppose. But, you get the point. Because these labels are next to useless to me, I really just have a one-in-three chance of getting it right each time I pull one of the strings.
How could each control be designed to communicate its function easily to the user? I have used two strategies to modify my own fan. Because reversing the fan's direction is the control that I use least often, I have shortened the string associated with this option to be the most difficult of the three to reach. However, I use the light and speed options about equally often. So, to make these as distinct as possible I borrowed two charms from an old charm bracelet and attached them to the ends of each string. A charm of a rabbit is attached to the string that controls the speed of the fan, and a charm of a book (for which I turn on the light to read) is attached to the string that controls the light. The fan may not look the way the manufacturer had intended it to look, but I have not made the mistake of pulling the wrong string since I attached these charms.
Motor skills such as pushing doors, turning knobs, and pulling strings are simple to learn, and we have all mastered these simple skills over our lifetimes. So why do we have such trouble using them? One reason is that the manufacturer is often paying more attention to aesthetics than to the needs of the user.
Self-Directed Learning Activities
1. Describe, in general, the principle of how product design influences human performance.
2. Search the literature for the term stimulus-response compatibility; then define it in your own words with specific reference to an example presented in this story.
3. Over the next 24 hours, keep a diary of all the objects or things you encounter that you think represent stimulus-response incompatibility. Propose ways each of these things could be made more compatible.
4. Find three research articles in which stimulus-response compatibility has been investigated. What are the similarities and differences in the stimuli and responses used in these studies?
Notes
- Michael Darnell has a wonderful website that features many examples of poorly designed products, with suggestions about simple ways to greatly improve usability: www.baddesigns.com
Suggested Readings
Proctor, R.W., & Van Zandt, T. (2008). Human factors in simple and complex systems (2nd ed.). Boca Raton, FL: CRC Press.
Proctor, R.W., & Vu, K.P.L. (2006). Stimulus-response compatibility principles: Data, theory, and application. Boca Raton, FL: CRC Press.
Schmidt, R.A., & Lee, T.D. (2011). Human information processing. In Motor control and learning: A behavioral emphasis (5th ed., pp. 57-96) Champaign, IL: Human Kinetics.
Learning to win from losing
Why is the learning–performance distinction important?
One day, Michelle Wie might be remembered as one of the greatest golfers of all time. At the age of 14 she was pounding out 300-yard (274 m) drives and not only competing in Ladies Professional (LPGA) Tour events, but also performing quite well in them. Her talent is undeniable. But controversy surrounded this exceptional athlete because of her desire to compete in PGA Tour events, which traditionally have included only male competitors. Wie has performed admirably in some of these events. For example, in the 2004 Sony Open, she shot rounds of 72 and 68 and missed the 36-hole cut by just a single stroke. Her two-round total of 140 placed her in a tie for 80th, which was better than 53 male professionals, including PGA stars Zach Johnson, Hunter Mahan, and Adam Scott. However, Wie has not performed well in most of the other PGA events she has entered, including a 139th-place finish (out of 141 players) at the 2007 Sony Open.
Many argued that Wie should stop competing in PGA Tour events and concentrate her efforts on the LPGA Tour. The typical argument was that her skill development as a golfer was being stalled by repeated failures in PGA Tour events, and that if she concentrated on LPGA events instead, the greater chance for success would escalate her skill development.
I argue that there is a fundamental flaw in this theory, because it falls into a trap known as the learning-performance distinction. The trap underlies one of the most misunderstood concepts in motor learning research, so let's start by defining the two terms. Motor skills researchers use the term performance to refer to a single observation. It could be a score or outcome that reflects the value of a single attempt at a motor skill, or perhaps an average score that statistically summarizes a number of attempts. An 18- or 36-hole score, or a final placement ranking in a tournament, might indicate a representative performance score in the case of a golfer. The term learning is used quite specifically to refer to a stable improvement in skill over time—an improvement that has specifically occurred as the result of practice. The problem illustrated by the criticisms leveled at Michelle Wie is that the critics fail to consider the difference between performance and learning; they mistakenly confuse her failures to improve her performance in men's PGA events as a failure to learn from these experiences. Let's use another example to illustrate this distinction between performance and learning.
Suppose you like to bowl, play regularly in a league, and also practice occasionally. Two years ago your average was 110, last year it was 120, and this year it is 130. Those averages probably indicate that you have become a better bowler as a result of learning: the scores indicate improvements that appear to be rather stable and that have resulted from practice.
So, does this mean that you will score a 130 the next time you bowl? Not necessarily, because there are many reasons for the fluctuations in scores that occur from game to game. Next time you may score well below your average because you are not feeling well, your shoes are too tight, or the crowd is very noisy that night. But this does not mean that you have suddenly lost some of your learned skill. You may score well above your average because you try extra hard to impress someone, or because everything just seems to be well focused (your mojo is working). As before, this does not mean that you have suddenly had a change in learned skill, because there is no indication that the sudden improvement in performance reflects a stable and permanent change in your capability to perform.
At any one time we have a theoretical capability for attaining a certain level of performance. When that theoretical capability changes to a higher level as a result of practice, we can say with confidence that we have learned. The confusion lies in the fact that individual performances may sometimes exceed or not live up to these theoretical capabilities. These fluctuations are expected and in no way diminish or detract from the performer's theoretical capability to perform at a certain skill level.
What I have described is the typical distinction between learning and performance. But, there is another side to this issue, the one that can be applied to Michelle Wie. The issue concerns the situation in which performances do not appear to change, or appear to be getting worse. Does that mean that learning is not occurring? Let's go back to the bowling example.
Suppose that this year your average was 120, which is the same as it was last year. You may take the absence of a stable improvement, despite all of your practice, as evidence that your learning has stalled. But, as it turns out, last year your league bowled Tuesday nights, which is your day off. This year you bowl on Wednesday nights, one hour after completing your 10-hour work shift. You are typically tired and hungry on league nights, and your bowling scores suffered this year as a result. What does all this suggest? What is your theoretical capability to bowl, and has this changed from last year, despite the change in your bowling night? Could it be that you actually have improved (i.e., learned), but the feeling of fatigue and other factors directly related to your work schedule prevented you from actually performing up to these expectations?
The point here is that making inferences about learning requires evidence that goes beyond a single performance or sometimes even a set of performances. Under what circumstances were these performances observed? Are there mitigating factors that might explain why the theoretical true score differs from the observed score? The absence of any observable change in performance does not mean that the unobservable (that theoretical capability to perform at a certain level of skill) has not improved.
Michelle Wie took a very different route in the development of her golfing skill. Her decision to compete in PGA events placed enormous pressures on her to perform, and only an exceptional result would have convinced the public that she was benefiting from these extreme challenges. All we could observe were her scores in these events. The unobservable, however, what she has learned by playing in these events, can never be truly understood. I suspect that if she achieves the rank of one of the greatest players of all time, it will be due in no small measure to these experiences competing against the very best golfers in the world.
Self-Directed Learning Activities
1. In your own words, explain the distinction between performance and learning.
2. What is the difference between a theoretical true score and an observed score?
3. Suggest a reason a person's observed score might exceed the true score, and a reason the observed score might fail to achieve the level of the true score.
4. Identify another sport, athlete, or situation in which the failure to account for the learning-performance distinction has resulted in an inappropriate conclusion.
Notes
- Babe Zaharias (see “The Babe”) was the first woman to compete in a men's PGA event.
- Here is a sampling of some of the criticism of Michele Wie's decision to compete in PGA Tour events, from two sports commentary sites:
- Michelle Wie won her second LPGA Tour title in Winnipeg, Manitoba, at the 2010 CN Canadian Women's Open.
Suggested Readings
Schmidt, R.A. (1972). The case against learning and forgetting scores. Journal of Motor Behavior, 4, 79-88.
Schmidt, R.A., & Lee, T.D. (2011). Motor learning concepts and research methods. In Motor control and learning: A behavioral emphasis (5th ed., pp. 327-346). Champaign, IL: Human Kinetics.
Cutting wood and missing putts
How are constant error, variable error, and absolute error useful for understanding motor control?
Two things I like to do—cutting wood and golfing—remind me often of the importance of measures of central tendency and variability in motor control performance. For example, I use my chainsaw to cut up fallen trees to burn in the woodstove. Cutting logs into burnable lengths is not an exact science. In the middle of a Canadian winter, when the temperature is cold and plenty of snow blankets the ground, it is generally considered a good idea to keep moving. I do not take time to measure the lengths of the logs I saw; rather, I make a quick visual estimate and then start to make my cut. But, in doing so, I must remember two important things related to the mean and the standard deviation of the logs I cut. First, because our woodstove is only large enough to hold pieces of wood that are 50 centimeters (20 in.) or shorter, any pieces that are longer than that are useless. And second, wood is easier to stack when the pieces are all roughly of the same length. Essentially, these features remind me that the mean and standard deviation are both important; I must have a mean that is less than 50 centimeters and a standard deviation that is as small as possible.
When considering data in terms of a specific goal or standard, it is often better to express the mean and standard deviation in terms of error measures—in my case, relative to a goal wood length of 50 centimeters. For example, wood lengths of 45, 42, 48, 47, and 52 centimeters would be expressed as lengths that are -5, -8, -2, -3, and +2 centimeters, respectively, relative to my goal wood length of 50 centimeters. If we calculated the mean of these lengths of wood, then we could express the stack of wood as having either an average length of 46.8 centimeters, or the error scores as having a mean constant error of -3.2 centimeters. Having a negative constant error is good in this case, because the average piece of wood that I have cut will easily fit in the woodstove (i.e., the mean is less than 50 centimeters).
The careful reader will have noted, however, that even though my mean constant error is negative, there is still one block of wood that will not fit (the 52 cm log). Thus, by itself, the mean constant error score (-3.2 cm) is misleading, because it does not provide any indication of how many logs might not actually fit in the woodstove. Our measure of variability, the standard deviation, gives us some indication of this information. The measure of variability of these error scores (called the variable error) represents the deviation of the error score for each log relative to the mean constant error and is calculated in the same way as is a standard deviation. In this case, the variable error is 3.3 centimeters. In general, when I am cutting wood, I want to achieve a negative mean constant error with a variable error that is as small as possible.
Although constant error provides a useful heuristic measure of average performance in cutting wood with a chainsaw, it is a quite misleading measure of central tendency in another activity I like to do. Let's say that I have struck 10 golf putts—5 of these putts go past the hole by 5, 10, 15, 20, and 25 centimeters, respectively, and the other five putts come up short of the hole by equal amounts (-5, -10, -15, -20 and -25 cm). On average, by how much have I missed the hole with these putts? If you were to calculate the average as in the previous wood-cutting example (the sum of the 10 individual error scores, divided by 10) the answer would be a mean constant error of 0. In other words, my “average” putt ended up in the hole. However, we know that this answer surely must be wrong because none of the individual putts actually went in the hole. So, we must use a different way to express these error scores to avoid an answer that makes no sense.
The problem that we sometimes run into with constant error is that it provides a measure of average bias, the tendency to err in a specific way (e.g., by too much or too little; too far left versus too far right). In some cases a specific bias is desirable, such as the tendency to undercut a wood length of 50 centimeters so that all of my wood pieces fit in the woodstove. In the case of my putts that have no consistent bias, we are much better off using a measure of central tendency that removes the bias from each score prior to calculating the mean. Such a procedure is called using the absolute (unsigned) scores. Hence, this measure of central tendency is called the mean absolute error. Here, our measure of central tendency is the mean of the unsigned scores (5, 5, 10, 10, 15, 15, 20, 20, 25, 25), which is 15 centimeters. That is, the golf putts tended to miss the hole by an average of 15 centimeters. In this particular case, the mean absolute error score (15 cm from the hole) represents a more accurate picture of the entire set of individual scores than what is represented in the mean constant error (in the hole).
Averages are convenient ways to express how we tend to perform. But, numbers that represent central tendencies can be misleading if the tendency itself is not a strong one. Measures of variability are one way to identify the strength of a tendency. Depending on the context, different methods of calculating central tendencies may characterize the data in more representative ways.
Self-Directed Learning Activities
1. Define constant error, absolute error, and variable error in your own words.
2. What do the terms algebraic error and total error (or Henry's E) refer to? How are they calculated, and how are they similar to the terms constant error and absolute error, respectively?
3. Look up and briefly describe a research investigation that uses at least two error measures. Why do you think the researchers used these measures in particular?
4. Conduct a brief study on yourself. Close your eyes and, using a pencil, draw five lines as close to 4 inches (10 cm) as possible, one below the previous one, without opening your eyes until you have drawn the fifth line. Measure and record the length of each line you drew; then calculate all of the error measures that have been discussed.
Notes
- Here is a web-based standard deviation calculator:
www.tinyurl.com/standarddeviationcalculator
Suggested Readings
Chapanis, A. (1951). Theory and methods for analyzing errors in man-machine systems. Annals of the New York Academy of Sciences, 51, 1179-1203.
Schmidt, R.A., & Lee, T.D. (2011). Methodology for studying motor performance. In Motor control and learning: A behavioral emphasis (5th ed., pp. 21-56). Champaign, IL: Human Kinetics.
Forensic motor control
What are generalized motor programs, and what do keystroke dynamics reveal about them?
Most of what I know about forensic identification comes from crime shows that I've seen on TV. The suspect is caught because a fingerprint was left on the murder weapon. The evidence is used to convict the suspect because no two fingerprints have ever been found to be identical. Crime shows tell us that other forensic methods can be used as well, such as DNA and eye retina data, but the fingerprint is the oldest biometric tool used for identification. However, another method that reveals a great deal about your identity is how you write or even type your name. For example, the password that you type to log on to your e-mail account may be just as identifiable as your fingerprint.
Try the following task as an example: go into any word processing program and type your name 10 times, once on each line, as I have done here:
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
Tim Lee
I am not a skilled typist. I use my right middle finger to hold down the shift key and press the letter T with my left middle finger, then use my right middle finger to press i and my right index finger to press m, and then my right thumb to press the space bar. It is not very efficient, but I do it the same way every time I type my name. Regardless of how skilled a typist you are, you most likely repeat much the same process each of the 10 times you type your name.
Now, let's try to unravel the temporal “fingerprint” that you left behind when you typed your name. Suppose we conducted a simple analysis of the keystrokes and the time between each of the keystrokes that you made. The total time taken to type your name once is simply the time from the first keystroke to the last keystroke. At a more fine-grained level of analysis, the total time constitutes the time each key is held down (the dwell time) plus the time between the release of one key and the depression of the next (the transition time). All of the individual dwell times plus all of the transition times will sum to the total time.
In figure 7.5, I have plotted a hypothetical example of the 10 trials to type my name. The light gray bands denote the dwell times, and the darker gray bands indicate the transition times. I have ordered the lines of bands that represent the 10 trials from the fastest (least total time) at the top to the slowest trial at the bottom (most total time).
If you were to analyze each of the 10 repetitions of your name, you would probably find that all of the total times would be similar, but probably never exactly the same each time, just as in figure 7.5. The “noise” in our central nervous systems, plus other factors, causes the results to vary a little bit each time, resulting in some repetitions to be slower, and others faster, than the average total time.
But, take a closer look at each individual band in figure 7.5 as it changes over the 10 trials. What you will notice is that as the total time increases, each band gets proportionally longer too. We could express these numbers another way by dividing the time for each individual band by the total time for that trial to obtain a relative proportion of time represented by each band. Given the hypothetical evidence in figure 7.5, what we would discover is that the relative time for each band, expressed as a percentage, would stay roughly the same across all of the repetitions. Applying temporal and other methods to deconstruct how we type (e.g., key press forces) is called the study of keystroke dynamics.
According to generalized motor program theory, relative time is one of the key features of movement that is controlled by the central nervous system, especially for brief, rapid actions. Relative time is an expression of our motor control fingerprint. For well-learned tasks such as typing and handwriting, our central nervous system regulates the relative timing of impulses that are sent to the muscles that carry out these tasks. Various factors influence the real time on any particular instance. For example, using a keyboard that requires more force to depress the keys might result in overall slower times than would result using a keyboard that has a light touch, but the relative timing would likely remain the same.
Research in keystroke dynamics may result in the ability to identify people who carry out Internet fraud. In many ways it is similar to the use of handwriting dynamics to analyze the timing of the cursive expressions of a signature. It is fairly easy to forge the spatial representation of someone's signature, but very difficult to forge the temporal dynamics that result in that signature. These expressions of timing behavior represent our motor control fingerprints.
Self-Directed Learning Activities
1. Define the term generalized motor program in your own words.
2. How does the concept of a generalized motor program differ from the concept of a motor program as it was used in stories such as “Antilock Brakes” (in chapter 5) and “Point of No Return” (earlier in this
chapter)?
3. Suggest a method by which the handwriting dynamics of signatures could be used to detect fraud.
4. Some people contend that there is one generalized motor program for the full swing in golf, regardless of which club is used. How could a temporal dynamics analysis be used to assess this contention?
Notes
- An excellent review of fingerprint analysis by David Ashbaugh of the Royal Canadian Mounted Police is available here:
www.onin.com/fp/ridgeology.pdf
- A lot of controversy remains about invariances in motor performance; how they are measured; and what invariance, or lack of invariance, means in terms of motor programs. The following articles provide good arguments for the debate:
Gentner, D.R. (1987). Timing of skilled motor performance: Tests of the proportional duration model. Psychological Review, 94, 255-276.
Heuer, H. (1988). Testing the invariance of relative timing: Comment on Gentner (1987). Psychological Review, 95, 552-557.
Suggested Readings
Schmidt, R.A. (1975). A schema theory of discrete motor skill learning. Psychological Review, 82, 225-260.
Schmidt, R.A. (1985). The search for invariance in skilled movement behavior. Research Quarterly for Exercise and Sport, 56, 188-200.
Schmidt, R.A., & Lee, T.D. (2011). Central contributions to motor control. In Motor control and learning: A behavioral emphasis (5th ed., pp. 177-222) Champaign, IL: Human Kinetics.