Sylvester
03-31-2010, 10:13 AM
Here's an interesting article relating to "human error" and sudden acceleration.
http://www.canadiandriver.com/2010/03/22/feature-not-so-fast-toyota%e2%80%99s-unintended-acceleration.htm?page=all
March 22, 2010
Feature: Not so fast – Toyota’s unintended acceleration
By Richard A. Schmidt, Ph.D.
One can hardly miss all of the interest and concern about Toyota’s recent issues with unintended acceleration (or sudden acceleration, as it is often termed in the Toyota debate) present in the media recently. All of this reminds me strongly of a very similar set of circumstances dating to the mid- to late-1980s concerning unintended acceleration in the Audi 5000 automobile (among many other models and manufacturers). A broader examination of Toyota’s situation in view of the lessons we learned from the analogous situation with Audi seems warranted to add some perspective to the present debate.
Some may recall that the Audi 5000 (and other vehicles as well) was also the subject of intense debate in the late-1980s over unintended-acceleration allegations. The typical phenomenon occurred like this: The driver enters the vehicle, starts the engine, and then he/she intends to press lightly on the brake pedal (to prevent the car from creeping) while shifting from Park to a drive gear (either Drive or Reverse, depending on the car’s location). The driver then reports sudden, unintended, full-throttle acceleration, and also reports that the brakes would not stop the vehicle until a crash occurred. The driver insists that his/her right foot was on the brake pedal – and not on the accelerator – and that the vehicle malfunctioned in some way to cause the event. There were approximately 10,000 such events reported in all makes and models of cars (even busses, tractors, and golf carts), rendering these events relatively rare among car crashes generally, but nevertheless frequent enough, and severe enough, to raise concerns with the automakers, the driving public, and the media.
The problem is that, when the vehicle was examined by engineers post-crash, nothing was found wrong with the car (except for crash damage, of course) that would account for what the driver had reported. These events occurred in vehicles with various forms of fuel delivery (fuel injection, carburetors, and even diesels), with and without cruise control, with and without electronic engine controls, large and small, expensive or inexpensive, but only in vehicles with automatic transmissions. Yet no malfunction in any of the accident-vehicles was ever found, despite a systematic, 10-year, effort by the governments of the U.S. (i.e., NHTSA), Canada, and Japan, intense study by private consulting firms, or by the automakers themselves.
All of this is eerily similar to the situation facing Toyota today. (Of course, some have suggested that unintended acceleration in Toyotas is caused by floor mats, and/or sticking throttles; but there is considerable disbelief that this is the problem, and the search for the “smoking gun” continues.) It is interesting to note that unintended acceleration in the 1980s (and even before) was occurring before the arrival of drive-by-wire controls and computerized engine-management systems – one of the leading set of suggestions about what might be going on to cause unintended acceleration in today’s Toyotas.
At the time (i.e., the 1980s), many of us who worked in fields such as human factors (or ergonomics), human performance, psychology, and kinesiology suspected that these unintended-acceleration events might have a human component. These suspicions were bolstered by several facts about these events: (1) the complaints were far more frequent among older drivers than among younger drivers (the 60-70 year-olds had six times the rate of complaints as compared to 20-30 year-olds in a GM study); (2) drivers with little experience with the specific accident vehicle (but not with little driving experience, generally) such as parking-lot attendants, car-wash workers, rental-car patrons, car-borrowers, etc., were over-represented in terms of the number of complaints; (3) women and people with shorter statures were over-represented slightly. Faced with these findings, many of us asked, “OK, if these events are caused by some electro-mechanical malfunction, why would the car “care” if the driver is old or young, is experienced or inexperienced with this car, or is a short female?” The answer, of course, is that the car simply cannot “care.” This is the sort of evidence that gives credence to the problem having a human cause of some kind.
The human-based explanation for unintended acceleration goes as follows: When the driver intends to apply the foot brake (so as to keep the car from “creeping”), he/she actually strikes the accelerator instead. Then, when, the transmission becomes engaged in a drive gear, the car moves more than expected. This is at least mildly startling. Given that the driver has intended to apply the brake, and given that the car is moving unexpectedly, his/her solution is to press the brake harder. But the right foot is actually on the accelerator, so the throttle is opened and car actually moves faster. This then leads to pressing the “brake” harder and, because the foot is on the accelerator, to even more acceleration. Eventually, the car is at full throttle, which lasts until the car crashes. The right foot goes to the floor, giving the driver the impression that the brakes have failed, which is what the driver reports later.
In my experience with over 150 such cases, many of which have gone to court in a lawsuit against the automakers, in order to believe this account, the jurors seem to want the answers to three questions (the reader of this piece will also ask the same questions): (1) Why would a driver with (sometimes) 50 years of driving experience suddenly “mistake” the accelerator for the brake? (2) If the driver did somehow make such an error, why did he/she not correct it? After all, the brake and accelerator are in different locations, they “feel different” (have different force/displacement relationships) which would seem to provide ample basis for detecting the pedal error. (3) Why does the episode last so long – often 6 to 10 seconds or more? Isn’t that ample time to shut off the ignition, shift to Neutral, and/or apply the parking brake, none of which are typically done by the driver in unintended acceleration?
There are research-based answers to all three questions, and I’ll give a brief description here. (For more detail, the interested reader could consult my 1989 article in the journal Human Factors, referenced at the end of this note.)
For the first question, the driver does not really “confuse” the accelerator and the brake. Rather, it is well known that the limbs do not do quite what the brain “orders” them to do. Noisy neuromuscular processes intervene in the processes between the brain and muscles to make the action slightly different from the one intended. The driver intends to press the brake, but once in a while these noisy processes cause the foot to deviate from the intended trajectory – just as the free-throw 90 per cent shooter in basketball intends to make the basket, but sometimes the ball misses its target. This effect can be enhanced by the driver being slightly misaligned in the seat when he/she first enters the car. If the misalignment is in the clockwise direction (as viewed from above) then the foot can be directed too far to the right, toward the accelerator. Such misalignment could be maximal when the driver first enters the car, which is consistent with the high frequency of unintended-acceleration reported at initial start-up.
The answer to the second question is based on the idea that, during the foot- and shift-lever movement, the driver is busy attending to many things (his/her view out the windshield, the presence of pedestrians nearby, movements of the shift lever, etc.) but he/she is typically not attending to the sensory information from the leg/foot that might signal that the foot was on the accelerator rather than the brake. These unattended signals are simply not processed, and are missed, so that the driver does not detect the error. Indeed, because the driver intended to press the brake, he/she is confident (after the crash) that the foot was on the brake.
The answer to the third question is that, when the car accelerates unexpectedly, the driver’s reaction is one of panic, termed “hypervigilance.” This temporary (“primitive”) panic-state renders the driver incapable of reasoning creatively and/or logically about the situation, with the only goal being to stop the car, typically by pressing the brake harder and harder. These panic-states have been used to explain seemingly irrational behaviour seen in situations like theatre fires, attacks in warfare, etc.
This view of unintended acceleration being caused by a pedal misapplication (error) was the position adopted by NHTSA in their closing report about unintended acceleration in 1989. (See Pollard & Sussman, 1989, cited at the end here.) And, because no vehicle defect was ever found that could explain these unintended acceleration events, many take the position that most, if not all, of the unintended acceleration episodes associated with Audi and others were actually caused by pedal errors – not some electro-mechanical defect in the vehicle. If the recent situation with Toyota is the same situation as Audi experienced 23 years ago, it stands to reason that most, if not all, of the unintended acceleration episodes with Toyota are caused by pedal error as well. If this is so, there may not even be a defect in Toyota’s vehicles.
If so, why are there so many complaints of unintended acceleration in Toyotas today? Isn’t this, by itself, evidence that something is wrong with Toyotas? Not necessarily.
First, this high complaint-rate is analogous to the situation with Audi 23 years ago, where unintended-acceleration complaints for the Audi 5000 were far more numerous than those for any other make of vehicle. One way to understand this heightened complaint-rate for Audi (and perhaps also for Toyota in the present situation) is through publicity effects. For Audi, there was a New York Public Interest Research Group (NYPIRG, a “Victims Network”) which was quite active in publicizing Audi unintended-acceleration episodes, encouraging customers to complain; it appeared to me that this group had the overall goal of abolishing Audi from the road in the U.S. Also, many articles appeared in automotive magazines about unintended acceleration and Audi. There were several TV productions about the phenomenon, including an extremely biased and unfair segment on “60 Minutes” which I show to my UCLA students. In 1986, engineer William Rosenbluth appeared on the “60 Minutes” segment, showing that, if you drill a hole into the side of the Audi 5000 transmission and pressurize it via an external source, you can get the throttle to open by itself on some years of the Audi 5000. This is analogous to ABC’s Brian Ross reporting recently on Prof. David Gilbert’s finding that, if you short-circuit the Toyota’s computer in a certain way, one can get the throttle to open. But these are highly contrived demonstrations that are far from the reality of unintended acceleration – yet they are highly influential in having the viewers believe that there really is a vehicle defect. Although it is hard to quantify, publicity like this undoubtedly leads to increased complaint rates.
Second, a customer-complaint is not evidence of a vehicle defect – only that the customer complained. If the customer complained of an unwanted acceleration, and the event was caused by an undetected pedal error, this would provide a misleading statistic about the magnitude of the “defect” problem, which increases publicity, etc.
It is easy to imagine that a customer might watch “60 Minutes” and see the unintended-acceleration phenomenon described. Then, the next morning in cold weather, the customer starts his/her Audi and notes an (uncommanded) high idle after start up – perfectly normal behaviour for a car in wintertime. The customer then might say, “Aha, there it is, unintended acceleration, just like on 60 Minutes!” and then lodge a complaint to Audi and/or NHTSA.
This kind of explanation for unintended acceleration is tricky for the automakers, Toyota included. In 1986, Ed Bradley on “60 Minutes” interviewed the head of Audi’s North American operation, who believed that the problem was caused by pedal errors. Mr. Bradley said, essentially, “OK, so you can’t find the mechanical defect. So then you want to blame it on the driver?” Similarly, Mr. Toyoda of Toyota said the other day that he would never blame the driver for these phenomena. Yet, if the phenomenon is caused by a pedal error, isn’t the driver “to blame” for it? If not, who is to blame?
Such a question raises the idea that the automakers should design their products so that such errors cannot occur. Audi did this 23 years ago. In response to mounting criticism of its Audi 5000, and with their internal belief that these events were caused by pedal misapplications, Audi designed an interlock to block this kind of error. Termed the Automatic Shift Lock (ASL) – sometimes referred to as the Brake-Transmission Shift Interlock (BTSI, or “Bitsy”) – this device locks the transmission in Park unless and until the brake pedal is depressed. So, if the driver does make a pedal error (foot on the accelerator when the brake was intended), then the vehicle will not move at all, as it will still be in Park. Audi ultimately gave away (to the world’s other automakers) the rights to their patent on the ASL. By 1995, virtually all new cars had adopted this feature, or some variant of it.
The ASL issue is actually an interesting line of evidence for the pedal-misapplication theory of unintended acceleration. The pedal-error theory would demand (predict) that the ASL should have very large effects on reducing unintended acceleration, whereas some other defect theory would predict no advantage. Studying Audis that were (a) factory-outfitted with ASLs, (b) retrofitted with ASLs, and (c) cars without any ASL, NHTSA investigators found massive reductions in unintended acceleration with the ASL-equipped cars (a and b) relative to the non-ASL-equipped vehicles (c). Besides providing a way to reduce unintended accelerations, this exercise comprised a strong naturalistic experiment whose results supported the pedal-error view and contradicted the idea of some vehicle defect as the cause.
However, Mr. Bradley pointed out during the “60 Minutes” segment that a few Audis with ASLs were also known to have unintended acceleration episodes. He’s right; the ASL drastically reduced unintended accelerations, but did not eliminate them completely. Why? One reason is that, once the vehicle is out of Park and underway, the ASL is no longer functional to prevent a pedal misapplication. So, pedal errors could occur during a driving cycle -not just at the start of it. Our research group analyzed pedal errors via accident reports, finding a relatively large number of pedal errors that – by the driver’s own admission – occurred well into the driving cycle (where the ASL could no longer act to prevent it).
Toyota has suggested interesting new software such that, if the brake pedal is activated at all, the throttle is returned to idle. This, unlike the ASL, will not prevent an unintended-acceleration episode caused by a pedal error because, with a pedal error, the driver’s foot does not reach the brake pedal at all. But, if there is some actual vehicle defect – and the driver has not made a pedal error – this software could be effective in bringing the car to a stop if the driver reaches the brake. It would also provide the basis for a strong naturalistic experiment about the causes of unintended acceleration.
Overall, the present situation experienced by Toyota today is almost exactly like the situation 23 years ago with Audi and other brands. In my view, what today’s critics of Toyota must do is to show that a pedal-misapplication cannot explain a certain accident, or set of accidents; then, and only then, should we start looking for hypothetical, mysterious vehicle defects. As was the case 23 years ago when no vehicle defect was ever found, all (or perhaps nearly all) of Toyota’s unintended acceleration episodes might be nothing more complicated than pedal misapplication, which is far different from today’s wild speculation about drive-by-wire throttle controls, computer glitches, and the like, coupled with the suggestion, voiced by many, that Toyota is covering up some devious defect.
References:
Pollard, J., & Sussman, E.D. (1989). An examination of sudden acceleration. (Tech Report No DOT HS-806-509). Washington, DC: U.S. Department of Transportation.
Schmidt, R.A. (1989). Unintended acceleration: A review of human factors contributions. Human Factors, 31, 345-364.
Richard A. Schmidt, Ph.D. is Professor (Emeritus) of Psychology at University of California, Los Angeles, where he has done research and teaching in the areas of human motor performance, motor learning and human factors (or ergonomics). He is the author of four books and over 150 research articles in these areas. Since 1994, he has been a consultant in human factors and human performance through Exponent Failure Analysis, Inc. and, since 2001, his own firm, Human Performance Research (HPR). Lately, he has been studying human contributions to various kinds of accidents, and has been consulting with professionals in speech and physical therapy, medicine, police work, sport (golf, tennis, skiing) and, recently, high-level music.
A list of publications and other information can be found on the HPR website.
A version of this article first appeared in the New York Times
http://www.canadiandriver.com/2010/03/22/feature-not-so-fast-toyota%e2%80%99s-unintended-acceleration.htm?page=all
March 22, 2010
Feature: Not so fast – Toyota’s unintended acceleration
By Richard A. Schmidt, Ph.D.
One can hardly miss all of the interest and concern about Toyota’s recent issues with unintended acceleration (or sudden acceleration, as it is often termed in the Toyota debate) present in the media recently. All of this reminds me strongly of a very similar set of circumstances dating to the mid- to late-1980s concerning unintended acceleration in the Audi 5000 automobile (among many other models and manufacturers). A broader examination of Toyota’s situation in view of the lessons we learned from the analogous situation with Audi seems warranted to add some perspective to the present debate.
Some may recall that the Audi 5000 (and other vehicles as well) was also the subject of intense debate in the late-1980s over unintended-acceleration allegations. The typical phenomenon occurred like this: The driver enters the vehicle, starts the engine, and then he/she intends to press lightly on the brake pedal (to prevent the car from creeping) while shifting from Park to a drive gear (either Drive or Reverse, depending on the car’s location). The driver then reports sudden, unintended, full-throttle acceleration, and also reports that the brakes would not stop the vehicle until a crash occurred. The driver insists that his/her right foot was on the brake pedal – and not on the accelerator – and that the vehicle malfunctioned in some way to cause the event. There were approximately 10,000 such events reported in all makes and models of cars (even busses, tractors, and golf carts), rendering these events relatively rare among car crashes generally, but nevertheless frequent enough, and severe enough, to raise concerns with the automakers, the driving public, and the media.
The problem is that, when the vehicle was examined by engineers post-crash, nothing was found wrong with the car (except for crash damage, of course) that would account for what the driver had reported. These events occurred in vehicles with various forms of fuel delivery (fuel injection, carburetors, and even diesels), with and without cruise control, with and without electronic engine controls, large and small, expensive or inexpensive, but only in vehicles with automatic transmissions. Yet no malfunction in any of the accident-vehicles was ever found, despite a systematic, 10-year, effort by the governments of the U.S. (i.e., NHTSA), Canada, and Japan, intense study by private consulting firms, or by the automakers themselves.
All of this is eerily similar to the situation facing Toyota today. (Of course, some have suggested that unintended acceleration in Toyotas is caused by floor mats, and/or sticking throttles; but there is considerable disbelief that this is the problem, and the search for the “smoking gun” continues.) It is interesting to note that unintended acceleration in the 1980s (and even before) was occurring before the arrival of drive-by-wire controls and computerized engine-management systems – one of the leading set of suggestions about what might be going on to cause unintended acceleration in today’s Toyotas.
At the time (i.e., the 1980s), many of us who worked in fields such as human factors (or ergonomics), human performance, psychology, and kinesiology suspected that these unintended-acceleration events might have a human component. These suspicions were bolstered by several facts about these events: (1) the complaints were far more frequent among older drivers than among younger drivers (the 60-70 year-olds had six times the rate of complaints as compared to 20-30 year-olds in a GM study); (2) drivers with little experience with the specific accident vehicle (but not with little driving experience, generally) such as parking-lot attendants, car-wash workers, rental-car patrons, car-borrowers, etc., were over-represented in terms of the number of complaints; (3) women and people with shorter statures were over-represented slightly. Faced with these findings, many of us asked, “OK, if these events are caused by some electro-mechanical malfunction, why would the car “care” if the driver is old or young, is experienced or inexperienced with this car, or is a short female?” The answer, of course, is that the car simply cannot “care.” This is the sort of evidence that gives credence to the problem having a human cause of some kind.
The human-based explanation for unintended acceleration goes as follows: When the driver intends to apply the foot brake (so as to keep the car from “creeping”), he/she actually strikes the accelerator instead. Then, when, the transmission becomes engaged in a drive gear, the car moves more than expected. This is at least mildly startling. Given that the driver has intended to apply the brake, and given that the car is moving unexpectedly, his/her solution is to press the brake harder. But the right foot is actually on the accelerator, so the throttle is opened and car actually moves faster. This then leads to pressing the “brake” harder and, because the foot is on the accelerator, to even more acceleration. Eventually, the car is at full throttle, which lasts until the car crashes. The right foot goes to the floor, giving the driver the impression that the brakes have failed, which is what the driver reports later.
In my experience with over 150 such cases, many of which have gone to court in a lawsuit against the automakers, in order to believe this account, the jurors seem to want the answers to three questions (the reader of this piece will also ask the same questions): (1) Why would a driver with (sometimes) 50 years of driving experience suddenly “mistake” the accelerator for the brake? (2) If the driver did somehow make such an error, why did he/she not correct it? After all, the brake and accelerator are in different locations, they “feel different” (have different force/displacement relationships) which would seem to provide ample basis for detecting the pedal error. (3) Why does the episode last so long – often 6 to 10 seconds or more? Isn’t that ample time to shut off the ignition, shift to Neutral, and/or apply the parking brake, none of which are typically done by the driver in unintended acceleration?
There are research-based answers to all three questions, and I’ll give a brief description here. (For more detail, the interested reader could consult my 1989 article in the journal Human Factors, referenced at the end of this note.)
For the first question, the driver does not really “confuse” the accelerator and the brake. Rather, it is well known that the limbs do not do quite what the brain “orders” them to do. Noisy neuromuscular processes intervene in the processes between the brain and muscles to make the action slightly different from the one intended. The driver intends to press the brake, but once in a while these noisy processes cause the foot to deviate from the intended trajectory – just as the free-throw 90 per cent shooter in basketball intends to make the basket, but sometimes the ball misses its target. This effect can be enhanced by the driver being slightly misaligned in the seat when he/she first enters the car. If the misalignment is in the clockwise direction (as viewed from above) then the foot can be directed too far to the right, toward the accelerator. Such misalignment could be maximal when the driver first enters the car, which is consistent with the high frequency of unintended-acceleration reported at initial start-up.
The answer to the second question is based on the idea that, during the foot- and shift-lever movement, the driver is busy attending to many things (his/her view out the windshield, the presence of pedestrians nearby, movements of the shift lever, etc.) but he/she is typically not attending to the sensory information from the leg/foot that might signal that the foot was on the accelerator rather than the brake. These unattended signals are simply not processed, and are missed, so that the driver does not detect the error. Indeed, because the driver intended to press the brake, he/she is confident (after the crash) that the foot was on the brake.
The answer to the third question is that, when the car accelerates unexpectedly, the driver’s reaction is one of panic, termed “hypervigilance.” This temporary (“primitive”) panic-state renders the driver incapable of reasoning creatively and/or logically about the situation, with the only goal being to stop the car, typically by pressing the brake harder and harder. These panic-states have been used to explain seemingly irrational behaviour seen in situations like theatre fires, attacks in warfare, etc.
This view of unintended acceleration being caused by a pedal misapplication (error) was the position adopted by NHTSA in their closing report about unintended acceleration in 1989. (See Pollard & Sussman, 1989, cited at the end here.) And, because no vehicle defect was ever found that could explain these unintended acceleration events, many take the position that most, if not all, of the unintended acceleration episodes associated with Audi and others were actually caused by pedal errors – not some electro-mechanical defect in the vehicle. If the recent situation with Toyota is the same situation as Audi experienced 23 years ago, it stands to reason that most, if not all, of the unintended acceleration episodes with Toyota are caused by pedal error as well. If this is so, there may not even be a defect in Toyota’s vehicles.
If so, why are there so many complaints of unintended acceleration in Toyotas today? Isn’t this, by itself, evidence that something is wrong with Toyotas? Not necessarily.
First, this high complaint-rate is analogous to the situation with Audi 23 years ago, where unintended-acceleration complaints for the Audi 5000 were far more numerous than those for any other make of vehicle. One way to understand this heightened complaint-rate for Audi (and perhaps also for Toyota in the present situation) is through publicity effects. For Audi, there was a New York Public Interest Research Group (NYPIRG, a “Victims Network”) which was quite active in publicizing Audi unintended-acceleration episodes, encouraging customers to complain; it appeared to me that this group had the overall goal of abolishing Audi from the road in the U.S. Also, many articles appeared in automotive magazines about unintended acceleration and Audi. There were several TV productions about the phenomenon, including an extremely biased and unfair segment on “60 Minutes” which I show to my UCLA students. In 1986, engineer William Rosenbluth appeared on the “60 Minutes” segment, showing that, if you drill a hole into the side of the Audi 5000 transmission and pressurize it via an external source, you can get the throttle to open by itself on some years of the Audi 5000. This is analogous to ABC’s Brian Ross reporting recently on Prof. David Gilbert’s finding that, if you short-circuit the Toyota’s computer in a certain way, one can get the throttle to open. But these are highly contrived demonstrations that are far from the reality of unintended acceleration – yet they are highly influential in having the viewers believe that there really is a vehicle defect. Although it is hard to quantify, publicity like this undoubtedly leads to increased complaint rates.
Second, a customer-complaint is not evidence of a vehicle defect – only that the customer complained. If the customer complained of an unwanted acceleration, and the event was caused by an undetected pedal error, this would provide a misleading statistic about the magnitude of the “defect” problem, which increases publicity, etc.
It is easy to imagine that a customer might watch “60 Minutes” and see the unintended-acceleration phenomenon described. Then, the next morning in cold weather, the customer starts his/her Audi and notes an (uncommanded) high idle after start up – perfectly normal behaviour for a car in wintertime. The customer then might say, “Aha, there it is, unintended acceleration, just like on 60 Minutes!” and then lodge a complaint to Audi and/or NHTSA.
This kind of explanation for unintended acceleration is tricky for the automakers, Toyota included. In 1986, Ed Bradley on “60 Minutes” interviewed the head of Audi’s North American operation, who believed that the problem was caused by pedal errors. Mr. Bradley said, essentially, “OK, so you can’t find the mechanical defect. So then you want to blame it on the driver?” Similarly, Mr. Toyoda of Toyota said the other day that he would never blame the driver for these phenomena. Yet, if the phenomenon is caused by a pedal error, isn’t the driver “to blame” for it? If not, who is to blame?
Such a question raises the idea that the automakers should design their products so that such errors cannot occur. Audi did this 23 years ago. In response to mounting criticism of its Audi 5000, and with their internal belief that these events were caused by pedal misapplications, Audi designed an interlock to block this kind of error. Termed the Automatic Shift Lock (ASL) – sometimes referred to as the Brake-Transmission Shift Interlock (BTSI, or “Bitsy”) – this device locks the transmission in Park unless and until the brake pedal is depressed. So, if the driver does make a pedal error (foot on the accelerator when the brake was intended), then the vehicle will not move at all, as it will still be in Park. Audi ultimately gave away (to the world’s other automakers) the rights to their patent on the ASL. By 1995, virtually all new cars had adopted this feature, or some variant of it.
The ASL issue is actually an interesting line of evidence for the pedal-misapplication theory of unintended acceleration. The pedal-error theory would demand (predict) that the ASL should have very large effects on reducing unintended acceleration, whereas some other defect theory would predict no advantage. Studying Audis that were (a) factory-outfitted with ASLs, (b) retrofitted with ASLs, and (c) cars without any ASL, NHTSA investigators found massive reductions in unintended acceleration with the ASL-equipped cars (a and b) relative to the non-ASL-equipped vehicles (c). Besides providing a way to reduce unintended accelerations, this exercise comprised a strong naturalistic experiment whose results supported the pedal-error view and contradicted the idea of some vehicle defect as the cause.
However, Mr. Bradley pointed out during the “60 Minutes” segment that a few Audis with ASLs were also known to have unintended acceleration episodes. He’s right; the ASL drastically reduced unintended accelerations, but did not eliminate them completely. Why? One reason is that, once the vehicle is out of Park and underway, the ASL is no longer functional to prevent a pedal misapplication. So, pedal errors could occur during a driving cycle -not just at the start of it. Our research group analyzed pedal errors via accident reports, finding a relatively large number of pedal errors that – by the driver’s own admission – occurred well into the driving cycle (where the ASL could no longer act to prevent it).
Toyota has suggested interesting new software such that, if the brake pedal is activated at all, the throttle is returned to idle. This, unlike the ASL, will not prevent an unintended-acceleration episode caused by a pedal error because, with a pedal error, the driver’s foot does not reach the brake pedal at all. But, if there is some actual vehicle defect – and the driver has not made a pedal error – this software could be effective in bringing the car to a stop if the driver reaches the brake. It would also provide the basis for a strong naturalistic experiment about the causes of unintended acceleration.
Overall, the present situation experienced by Toyota today is almost exactly like the situation 23 years ago with Audi and other brands. In my view, what today’s critics of Toyota must do is to show that a pedal-misapplication cannot explain a certain accident, or set of accidents; then, and only then, should we start looking for hypothetical, mysterious vehicle defects. As was the case 23 years ago when no vehicle defect was ever found, all (or perhaps nearly all) of Toyota’s unintended acceleration episodes might be nothing more complicated than pedal misapplication, which is far different from today’s wild speculation about drive-by-wire throttle controls, computer glitches, and the like, coupled with the suggestion, voiced by many, that Toyota is covering up some devious defect.
References:
Pollard, J., & Sussman, E.D. (1989). An examination of sudden acceleration. (Tech Report No DOT HS-806-509). Washington, DC: U.S. Department of Transportation.
Schmidt, R.A. (1989). Unintended acceleration: A review of human factors contributions. Human Factors, 31, 345-364.
Richard A. Schmidt, Ph.D. is Professor (Emeritus) of Psychology at University of California, Los Angeles, where he has done research and teaching in the areas of human motor performance, motor learning and human factors (or ergonomics). He is the author of four books and over 150 research articles in these areas. Since 1994, he has been a consultant in human factors and human performance through Exponent Failure Analysis, Inc. and, since 2001, his own firm, Human Performance Research (HPR). Lately, he has been studying human contributions to various kinds of accidents, and has been consulting with professionals in speech and physical therapy, medicine, police work, sport (golf, tennis, skiing) and, recently, high-level music.
A list of publications and other information can be found on the HPR website.
A version of this article first appeared in the New York Times