Navigation
Topics
Last Day
Last Week
Tree View

Documentation
Getting Started
Formatting
Troubleshooting
Program Credits

Utilities
New Messages
Keyword Search
Contact Moderators
Edit Profile

ABSTRACT: Novice drivers are at greater risk of crashing on the road than more experienced drivers. One reason for this is that they have not yet developed, through experience, certain perceptual and cognitive skills that are critical in reducing their crash risk. In late 1995 the Monash University Accident Research Centre commenced a program of simulator research which recently culminated in the development of a CD-ROM based training product for accelerating in young novice drivers the development of risk perception, attentional control and time-sharing skills. This paper reviews the outcomes of this research program and describes the development of the CD-ROM product deriving from the research.

Dear colleague. I welcome any questions you have about our paper and the CD ROM product. I am an Australian delegate. We are about 9 hours ahead of Germany and 15 hours ahead of Chicago. Michael Regan, PhD.

This product is a very exciting development. The developers have incorporated many of the factors which I also considered to be important in my Drive Alert For Life programs. Driver inattention is just beginning to receive attention in  media campaigns here in Australia, and so I am
delighted that specific aspects of attention and
concentration are the focus of this CD ROM. It will be interesting to see how effective the training program will be at improving young drivers on the road performance with regard to
inattention and difficulties with sustained attention.

To the authors: I agree with Bea. The focus on attention-dividing is very interesting. However, I'm a little unclear on the role of the Attentional Control Training (ACT) in relation to risk. As I understand from the paper, the CD-ROM provides video clips which are used to instruct in risk perception. The ACT, on the other hand, is more like a video game with virtual reality environments. Practising dividing attention between following distance with the car ahead and watching for numeric mental tasks in the environment improves attention-dividing ability, but how does this relate to the perception of risk? You mention that it can be "indirectly enhanced".

To Dan Keegan. Hi Dan. The full answer to your question can be found in a conference paper that Tom Triggs and I presented to the Ergonomics Society of Australia Annual Conference recently. It is referred to in the present conference paper as Regan, Triggs and Deery 1998c. If you provide me with your email address I'll send you an electronic version. The "short" answer to your question is as follows. The attentional control training paradigm we have employed in the training product is based on a series of experiments we carried out in a mid-range simulator (180 degree front views, 60 degree rear views, full car body, motion platform etc) here at the Monash University Accident Research Centre in Melbourne. In one of the experiments, 70 subjects participated. Half had an average of 41 hours driving experience and the other half had an average of 360 hours. Half the subjects in each group received attentional control training and the other half served as controls. Both groups performed concurrently in the simulator the two tasks described in the present conference paper - maintaining a constant headway distance behind the car in front and performing mathematical calculations on a continuous series of numbers appearing in different areas of the visual field, in front and behind. Control and treatment subjects received 26 trials of practice performing these tasks. The control group was instructed to simply perform the two tasks as well as possible. The treatment group was instructed to change the relative priority given to the tasks over the 26 trials in the manner described by Danny Gopher and his colleagues eg "give twice as much attention to the headway task" or "give equal attention to both" etc. Following training, all subjects completed 4 transfer of training drives, again in the simulator. In two of these drives, subjects were exposed to three traffic events which had the potential to result in a crash unless they anticipated the potential risk and responded in a safe manner. During these drives, subjects also performed a subsidiary task in which they responded vocally to a series of letter/number pairs which appeared in a constant location in front of them. When we analysed the data from these transfer of training drives, we found that subjects who had completed the attentional control training performed significantly better than control subjects on both of the sub-tasks in each drive - they were quicker to detect and respond safely to emerging traffic risks and, to a lesser extent, performed better on the arithmetic subsidiary task. Basically, it seems that the training enhanced both arithmetic performance and risk perception. The effect was stronger for the less experienced drivers. Thus, in an indirect way, the training facilitated risk perception, implying that attentional control is an important component of risk perception. To our knowledge, this has not been demonstrated before and suggested to us the need to integrate attentional control training into the CD ROM product. I hope that answers your question. Regards, Michael.

To Dan Keegan: Dan, I should qualify my last comment by saying that the purpose of attentional control training is not per se to indirectly enhance risk perception. This is a subsidiary benefit. The primary purpose is to develop in subjects the ability to prioritise their attention in graded amounts between different simultaneous activities and to learn to give a task only as much attention as is needed to maintain a moderate level of performance. In one of our simulator experiments (as yet unpublished) we found that, under very high workload, novice drivers tend to "protect" performance of the more automated driving task and let go of the more safety critical risk perception task going on outside the car. This was not the case with experienced drivers. The attentional control training is designed, among other things, to prevent this from happening.

It does answer my question about attentional control. I see attentional control training as having all kind of possibilities.
Now I'm trying to relate all of this to experiences with in-car instruction. Some novices arrive in the car with powerful predispositions towards risk...e.g. the student who had been a passenger in a car that was hit by another vehicle coming from the right (that'd be the left in Australia) was supersensitive to vehicles appearing on that side and almost seemed to have extra-sensory perception with regard to them (saw them way before her instructor). On the other hand, she seemed less concerned about other situations that had potentially more risk - such as a car preparing to turn left across her path ahead. If drivers have biases such as these, based on personal experience, can training overcome them? And will personal experiences after licensing quickly overcome any trained response to situations they've been told are risky?

Re: Dan Keegan - The issue you raise here is an important one, and I have argued for some time that any reliance on training approaches to the young/inexperienced driver problem fails to recognise the powerful influence of experience on the cognitive skills that underlie safe driving behaviours (see, for example, my 1997 conference paper Limits of Driver Training). The problem is that the development of automaticity in many of the cognitive and behavioural skills that underlie driving behaviour (and the development of a complex mental model or representation of the driving environment that directs attentional processes in hazard detection) is directly influenced by the experience of cues and behaviours in the driving environment. Expecting driver training to have anything other than a short-term influence may be optimistic - although I would be happy to be convinced! Simulation-based training may provide some advantages over real-world training in terms of the control that can be exercised over driver-experiences in the training process, but the risk here is that the learner driver will develop a mental model of the simulated driving environment that may not transfer to driving in the real world. Again, I would be very happy to be convinced, but I think there is a need to show that any driver training programs influence BOTH the skills of drivers (as an intermediate measure) and, more importantly, the crash risk of novice drivers. Given the general failure of driver-training in the empirical literature to this point, the onus for this demonstration clearly falls on the protagonists.

Have you any data in general or specifically referring to your CD ROM product relating to novice drivers who were diagnosed as suffering from attention deficity disorder (ADD or ADHD) in the past ?

In a small study I presented some years ago I looked at test results from the ART90 battery of safe drivers, accident prone drivers and ADD youngsters and found many similarities between the second and third groups. We are speaking of a significantly sized group who have a hard time listening to a lesson, choosing to disregard an unimportant stimulus, of finish a mental task which demands more than a few minutes' intense concentration. I do think that any empirical studies on novice drivers re attention, perception and cognitive skills should take past ADD as a controled variable into account.

P.S. How can one obtain your CD ROM ?

Re Michael Cale: I recall your paper as one of the few examples in the literature concerning the potential role of ADHD characteristics in novice (and other) crash problems. It had an impact on my own thinking in this area, and combined with some of Russell Barkley's work on cognitive aspects of this problem and attentional processes in general it made me wonder about the possibility that inhibitory processes might play an important role in safe driving, and that a failure of inhibition (either as a result of ADHD problems, youth, or alcohol impairment) might be an important contributor to less-safe driving behaviour. Barkley has published some material on the crash risk of ADHD drivers as well, with results that are generally consistent with some sort of role for this problem in the road toll. This could be a promising area for future work, and the combination of cognitive psych, theory, and road safety makes it quite attractive. Having said this, though, I've had no success gaining the interest of potential funders here in Australia!

Re Michael Cale and Michael Harrison’s entries:
ADDers do not choose to ignore information.
ADD brains function differently - depending on
which areas of brain and which wave patterns
predominate. Terminology is an issue.
Inattention is a failure of the brain to respond to
stimuli. Distractibility occurs as a number of
stimuli are attended to and irrelevant stimuli are
not suppressed. ADDers require a huge amount
of effort to stay focused on a task. In contrast a
person who is worried, anxious, depressed etc
may be preoccupied but is able to focus attention
and concentrate on the task. In ADD circles in
America in particular there is growing concern
about driving and ADD. It will become
increasingly important in Australia with the
recent refusal of insurance pay out for a woman
who was depressed. My Drive Alert For Life
programs start with the challenge of not touching
your keys unless you are fit to drive. In terms of
ADD/ADHD this means being adequately
treated, and having a strategy to monitor your
level of alertness and attention. Am I right in
guessing that I am the only person involved in
this discussion who is an educational and clinical
neuropsychologist in private practice and an
adult ADD? Although I have worked in
educational and clinical settings without research
funds, I have been collecting data on every client
since 1979 re their driving performance. It
started with questions about effects of
medication and psychiatric condition and driving
decisions. Then when I lectured in Special
Education and Ageing, or worked with patients
with depression or early dementia it became
obvious that people assumed they had a right to
drive, even though for various reasons their
attention, perception and decision making were
impaired. ADDers are an obvious target in the
current context but many others have impaired
attentional processes, and concentration.
Research needs to cover sleep deprivation, and
depression in novice drivers. High risk suicidal
young people are preoccupied, and they are
often reacting to specific situations. Inattention
has many facets.

Re: Bea Pullar. I think we agree more than disagree, Bea. I did not suggest, I don't think, that there is an element of conscious choice in the attentional processes evident in ADHD. My comment related to a failure of inhibition in general as a problem for attentional processes in driving. You are right to point out that there can be many causes for such a failure - ADHD being just one of them but a well-known one at least. The failure to suppress attention to unimportant stimuli is one example of a failure of inhibitory processes. The failure of these processes may also have roles in other important aspects of driving and I think the real strength of interest in the ADHD area is that this population gives us an opportunity to investigate these processes more closely than might be possible in the wider population. I suspect, however, that this research is possible regardless of the researcher's personal medical history (which I do not believe is relevant in this context).

Re: Comments by Dan Keegan and Warren Harrison on 12 March, and Michael Cale on 13 March. Dan and Warren, I will respond to your comments shortly. Michael, there is nothing I can add to the discussion that has ensued between Warren and Bea about ADHD. You ask, however, about the availability of the CD ROM and, in a related email (thanks),about whether it will be evaluated. The CD ROM is not yet available to the public. However, if you would like to try and obtain a copy of it, the person to contact is Mr David Healy who is the Manager Road Safety at the Transport Accident Commission (TAC)here in Melbourne. The TAC funded the research and development of the CD ROM. His email address is david_healy@tac.vic.gov.au Re evaluation of the product. Our Centre has also been contracted by the TAC to evaluate the CD ROM product. We will commence this activity next week. We will be collecting outcome performance measures in the mid-range driving simulator here at the Centre and we will be collecting process measures using the Provus-Discrepancy model for curriculum evaluation. The simulator evaluation will involve 50 treatment and 50 control subjects. The treatment subjects will be trained using the CD ROM and the controls will be "trained" on selected modules from "Flight Simulator 98". After the training,which will occur weekly over a 6 week period,subjects will complete six transfer of training drives in the simulator - six immediately after training and another 4 weeks later. These drives will measure the degree to which skills acquired during CD ROM training transfer to facilitate safe driving performance in the simulator. We will measure both near and far skill transfer. We have also worked out a way of measuring the extent to which the training product makes subjects more or less confident in their driving ability. The evaluation will be completed and documented in September this year. We propose to present the preliminary findings of this work to the Traffic Safety on Two Continents conference in Sweden in September. After that, we will look for funding to carry out an on-road evaluation of the product using crash-based and other outcome measures. Michael.

Re: Mike Regan post on March 10, 1999 at 06:39 pm. Perhaps I can shed some light or fog on the issue that was raised. So far as I know the author is referring to the results of a single experiment that he conducted towards the end of the research period. He rightly states that a series of experiments were conducted during the project. The earlier experiments involved a different proportional task than he used. In that series of experiments, we had drivers split attention between a lane tracking and a search and recognition task. The technical reason for using these tasks was that I felt that the simulator was not an appropriate place for a headway task due to the fact that the cues for distance are necessarily based almost exclusively on relative size since the scene itself is placed approximately eight feet in front of the driver. With regards the search and recognition task I felt that such a task was preferable to a numerical or short term memory task such as Regan later used because it more closely represents the attentional activities that drivers must perform on the road.

Of greater importance than the differences between our experiments of course was the fact that we were both apparently able to train novice drivers attentional control skills. In our experiments, which involved training times far shorter than Regan used, we did find that truly novice drivers do not yet have sufficient basic skills to benefit from the training. Indeed such training seemed to interfere with the subsequent performance of such drivers. I therefore have some concern that the target population of 17-18 year olds (learner drivers in Victoria) that Regan intends to use may have a problem with the training (Wheeler, et al, 1997a).

We conducted two experiments to investigate the potential impact of ACT alone on driving performance, particularly as applied to risk perception. Results of the first again demonstrated that only the most experienced of the novice drivers may have benefited from such training and this was not conclusive (Wheeler, et al, 1997b). The second experiment attempted to determine if ACT would improve a driver's ability to deal with high workload situations. It should be stated that workload in this case dealt with driving situations and tasks rather than the kind of artificial tasks commonly used in research situations. Performance in dealing with hazardous situations was also measured in this experiment. This experiment also used probationary drivers who because of their increased driving experience appeared to be better able to absorb the training. The results seem to indicate that ACT does help experienced novice drivers deal with high workload situations. It was less obvious that the training did much to improve risk perception, though there may have been a slight trend in that direction (Wheeler, et al, 1997c).

In his ACT experiment Regan indicates that he gave the subjects 26 training trials, twice the number of trials that we used and closer to the recommended training trials proposed by Gopher, (i.e., approximately 100, if memory serves). I therefore assume that his subjects represent a better example of what might be done with ACT. It seems to me however that the interpretation that he has made of the results may be in line with our earlier finding that ACT does improve a driver's ability to deal with high workload, no mean feat in this rather squishy business. The inclusion of the secondary task of letter/number pair recall makes me wonder how he was able to establish that his drivers were better able to handle hazardous situations and not simply better able to handle increased workload.

Both Mike and I have tested the possibility that ACT improves driving performance by freeing up resources so that the driver could focus on appropriate areas of the driving scene. I believe that we have demonstrated through testing and replication that such is the case. I do not feel however that our work, to date, suggest that ACT in any way reduces the need for a driver to developed an appreciation of what must be attended to within that scene.

I am including the references for the work I've indicated here because, to my knowledge, they have not been included elsewhere.

Wheeler, W.A., Derry, H, Vinogradov, L., Godlye, S., Harrison, W., McCormick, L., Tierney, P., Tomasevic, N., and Wilson, N. (1997a). Attentional Control Training Experiment Series: Investigation into the Use of Embedded Part Task Training in Attentional Control. Monash University Accident Research Centre Technical Research Report, Clayton, Victoria, August 1997.

Wheeler, W.A., Derry, H., McCormick, L. Godley, S., Harrison, W., Tierney, P., Tomasevic, N., and Wilson, N. (1997b). Attentional Control Training Experiment Series: Investigation into the Effectiveness of Embedded Part Task Training in Attentional Control in Driving. Monash University Accident Research Centre Technical Research Report, Clayton, Victoria, August 1997.

Wheeler, W.A., Derry, H., McCormick, L., Godley, S., Harrison, W., Tierney, P., Tomasevic, N., and Wilson, N. (1997c). Time-sharing Experiment Series: Investigation into the Effectiveness of Proportional Emphasis Training on Dealing with Driving Workload. Monash University Accident Research Centre Technical Research Report, Clayton, Victoria, August 1997.

Re: Regan's post to Dan Keegan on March 10, 1999 - 11:13 pm. The experiment that Regan refers to (Wheeler, et al. 1997d) is the first of the experiments in ATC. Its purpose was to see if novice and experienced subjects could actually do the tasks we had devised to for such training. Interestingly, we found that our novice drivers were actually able to perform these tasks better than were our experienced drivers. One unusual aspect of this experiment that may help account for this is that our 19 experienced drivers were an older group than we were normally able to recruit (ages between 25 and 45 years). Also our measure of performance was based on baseline performance established for each individual, thus we were actually measuring how much performance decreased as the task became more demanding. As Mike says we did have indications that the two groups used different strategies to preserve overall performance as the task became more difficult. I believe that this would be consistent with observations that have been made by others on the differences in behavior between novice and experienced drivers.

Wheeler, W.A., Derry, H., Vinogradov, L., Godley, S., Harrison, W., McCormick, L., Tierney, P., Tomasevic, N., and Wilson, N. (1997). Attentional Control Experiment Series: Investigation into the Identification and Control of Cognitive Resource Limitations During Simulator Based Driving. Monash University Accident Research Centre Technical Research Report, Clayton, Victoria, April 1997.

To the authors: I suspect you will not be surprised if I disagree with your conclusion, as stated in the paper, that a behavioral approach that incorporates a "near miss" experience does not result in enhanced risk perception for similar events. The experiments we conducted during the first eighteen months of the project provided consistent indications that:

1) the "near miss" experience did result in a change in the novice drivers perception of similar situations outside the simulator (i.e., video tapes of typical driving situations)
2) that subsequent driving behavior in the simulator was often modified following a "near miss" experience, and
3) that how and when in the drivers experience a particular "near miss" experience is presented is crucial to being able to measure an effect.

Both the concept of learning from a "near miss" experience and the observation that there are significant constraints as to when such learning will occur are consistent with theory and a logical interpretation of how such events are experienced in the real world.

I am not suggesting that "mediated instruction" may not be a superior method for teaching novice drivers what, in the environment, may represent a hazard. I am suggesting however that the "near-miss" paradigm, coupled with an understanding of the various constraints in its use, does provide a reasonable tool for better understanding how it is that young drivers learn, under normal and natural conditions, to deal with the hazards that they meet.

Re: Warren Harrison to Dan Keegan on Friday, March 12, 1999 - 11:18. I tend to agree with Warren's skepticism concerning the substantial usefulness of a training program such as this in reducing the road toll among novice drivers. The CD-ROM product that the authors have developed is impressive because of the innovations that they have been able to incorporate in the technique of driver education by using the technology available to them. It must be acknowledged that in developing their training product the authors were under considerable constraints in time, money and the deliver mechanism that they could employ. As Warren is aware but others may not be, a considerable amount of time was spent during the first eighteen months of this project in trying to understand how and what novice drivers learn through experience. This proved to be a challenge well beyond the collective skills of the investigators and patience of the sponsoring agency. I believe that we made strides in that direction, particularly in identifying how inexperienced drivers may acquire an understanding of potentially hazardous situations that occur infrequently (Wheeler, et al. 1997e), but recognize that there are still too many unanswered questions. Unfortunately, time caught up with us and we had to get on down the road to other activities, including delivery of a training product.

The effort expended during the research phase of this project was phenomenal. The output in terms of completed experiments and subjects processed exceeded that of any other simulator based research team that I am aware of. And yet, we were unable to produce a theory based training product. One has to wonder why.

If I learned anything during the eighteen months that I was with the project it was that there is no magic bullet that will solve the novice driver problem. Non-the-less we must surely be able to do better than our present approach to preparing novice drivers, which relies largely on self-paced experimentation on the highways. I sense that when it comes to understanding how novice drivers learn to become safe drivers many of us are wandering around the conceptual swamp trying to find a logical and workable way out. We have no map and we have no compass. Every once in a while we might spy the light of another wanderer, be he situational awareness, risk homeostasis, avoidance learning, or attentional control. Invariably these lights soon fade. Thus we each acquire enough paper to start our own fires and so it goes.

It seems to me that if we are really interested in solving the novice driver problem we need to map the swamp, find and remember what we know about it and investigate all of the logical areas that we know nothing about. Forums such as this seem to help us see where some of the lights are, which is both interesting and informative. It seems to me however that if we are ever going to solve such a complex problem we need to identify what it is we know about the problem, what we think we know about it and what we really need to find out. And then we need to get on with the job of doing so. As a community, whether we represent funding agencies, researchers or those who must pick up the bodies after a crash, it seems we should all have an interest in developing a strategy that would do this. I am not aware of any such attempt having been made on the scale that the problem deserves. Perhaps it's time we made one.


Wheeler, W.A., Deery, H.A., Wilson, N.J., and Triggs, T.J. (1997) Progress Towards a Useful Model of Risk Perception in Driving. Proceedings - 5th SEAES Conference ASEAN Ergonomics Conference (ASEAN'97), Kuala Lumpur, Malaysia, November 6-8, 1997.

As redundant as it might sound the William's point was well made that no one magic bullet will solve all the problems. Trying to get order into the chaos of the problems related to novice drivers couldn't we conceptualise three different areas:

1. Problems related to lacking skills in attention, perception, cognition and reactions due to lack of experience, knowledge or tools such as this CD-ROM evidently try to deal with.

2. Problems due to the fundemental or temoprary inavailability of such processes due partially to ADD but also to other problems as Bea pointed out. Please consider that some 5% (dependant on country and study ) or more of the youngsters at school are diagnosed as suffering from ADD and are treated with Ritalin or similar medication. I have interviewed a large number of young drivers who were involved in accidents and often heard from those who had been classified as ADD that they didn't know "where the other car came from" or "what hit them" whilst the police/parents etc just classify them as risky drivers.
3. Problems due to motivational issues e.g. using motor vehicles to show off, role play and the like.

I do think the CD ROM approach might contribute substantially to problems arising from 1 and 3 (partially) and suggest we use the interest created by this congress to initiate effectivity studies on an international level.

The issue that needs to be understood here is that the model of attention used in developing the CD ROM and similar projects makes assumptions about attentional processes, and leaves out a number of factors involved in perception and attention. By definition perception is the meaning attached to what is seen - or attended to.

Having a long professional career of involvement
with Hyperactivity as it used to be called and
ADD/ADHD, I have developed many insights into
attentional and cognitive processes. As an ADD
Coach I deal now mainly with adults who have only
recently discovered that there is an explantaion for their inattention, distractibility, chronically being late and impulsivity to mention only a few of the symptoms that impact on driving. Annecdotal evidence suggests that adequately treated with neurofeedback ADDers no longer have these difficulties. It is assumed that connections are established and balances are restored.

However, Ritalin only works for 3 to 4 hours per dose and the person reverts to being highly symptomatic - due to the imbalance of neurotransmitters with obvious consequences for driving. ADDers are over represented in driving incidents.

The point that Michael Cale made about people with
ADD not noticing the other car etc is an illustration of what I meant by saying that ADDers do not choose to suppress anything. If visual stimuli are not processed - there is nothing to suppress. On the other hand high levels of distractibility means that ADDers
attend to a number of competing stimuli that often have nothing to do with the desired behaviours - in this case driving. Attentional Control Training in a simulator or in front of a computer screen will undoubtredly improve the performance of ADDers and others on a task of concentrating on target behaviours.

As data is collected about the effectiveness of the CD-ROM for Novice Drivers hopefully people will also collect information where possible on whether or not the subjects have ADD/ADHD.

My own approach in the Drive Alert For Life tape and book kits, and the reason for the success of
ADD Coaching, is that rehearsal and constant monitoring with a log is required. I suggest every 4 to 6 weeks for normals and weekly for ADDers. Perhaps ADDers will benefit from frequent retraining sessions with the CD ROM?
Hopefully together we can make an impact on the
ADD/ADHD driving statistics.

I apologize to Michael Cale and the others on this forum for getting carried away. I'm afraid that it's not the first time I've pulled a table thumping display on this issue and doubt it will be the last. Rather than further detract from the discussion of Mikes paper I'll move this question over to the General Discussion area where I hopefully we'll be able to get some ideas going.

Re: comments by Dan Keegan, Warren Harrison and Bill Wheeler. Tom Triggs and I have spent some time thinking about the important issues you have raised. Tommorrow I will post replies to all of the issues raised. Our apologies for the delay in responding.

Concerning the above message thread....there's quite a few messages here now and downloading them may be getting a bit slow on some computers/modems. At some point I may need to start archiving the earlier ones. Participants...please give me some feedback on this.

Moderator

Folks, I am one of many driver trainers, and one from Queensland. Much of what you describe in your notes to each other is above me at this time, but, given background and time I would get a better grasp of it. I understand the terms used, but have not the time to digest the information at the moment.
My wife Christine and I have a two person driving school in Brisbane and are active/executive members of the Australian Driver Trainers Association-Queensland. As such, if we or our association can work with anyone to improve driver education (our word, rather than training) we would welcome the opportunity.
Please continue with your work, we have a long way to go to improve driver education at all levels, and there will be many side benefits from your research.
Bea, can you give me some specific concepts about your Drive Alert for Life programs? I am not aware of them and we are interested in any areas of driver education that can improve our work.
Tony

Dan, It won't bother me if you archive earlier comments. It is easy to print a hard copy of previous comments for future reference.

Re: Dan Keegan on Friday March 12, 1999 at 03:17am.

These are good questions Dan. Our CD ROM product is intended to bias novice drivers' anticipatory avoidance behaviours towards those situations that place them at most risk (near transfer). However, the product is also intended to facilitate the transfer of skills acquired during training to new and novel situations that may also place them at risk (far transfer). If they already have pre-conceived biases based on personal experience, our product will help them if the bias is towards one of the situations we know places them at most risk (eg, right turn against crashes). If the bias is towards a less critical traffic situation, the product should also help them because it should at least give the novice driver safety-critical skills than can also be exercised in that situation. On a related matter, I should say that, based on some of our earlier simulator work (eg Wheeler et al., 1997), novice drivers do not always perform in risky situations in a manner consistent with their subjective estimates of risk in those situations. Just because they have a pre-conceived idea that a situation is risky doesn't mean they will necessarily drive accordingly in that situation.

You also ask if personal experiences after licensing quickly overcome any trained response to situations novice drivers have been told are risky. In our CD ROM product we don't simply tell people particular situations are risky. We expose them to potentially risky scenarios and, using a range of instructional techniques, facilitate the acquisition of perceptual and cognitive skills for anticipating and deciding on appropriate responses to those potential risks. Given that the product, as noted above, is also intended to facilitate the transfer of skills acquired during training to new and novel situations that may also place them at risk (far transfer), personal experiences post training are likely to have two effects. Firstly, these experiences will provide more and more contexts in which to apply and consolidate skills acquired during training. When this training effect would wear off we don't know. How personal experiences could have the opposite effect I simply don't know. The only possibility I can think of is that a driving instructor or parent trains the novice to respond in a manner that is entirely inconsistent with the training, which is probably unlikely. Alternatively, the person may not experience many or any of the risky situations experienced during training. However, this is unlikely to extinguish the mental model of risk perception and decision making acquired through training given that there will be many other situations encountered by the novice to exercise the model.

Re: Warren Harrison on Friday March 12 at 11:18pm

We do not believe that Warren's comments are relevant to the CD ROM product described in our paper.

Historically, a problem with many traditional driver training programs has been that they lack an appropriate instructional strategy. The approach of Incremental Transfer Learning (ITL; Wallace and Regan, 1998) is the general instructional strategy underlying our CD ROM and has been used successfully to train higher order skills in Australian military pilots. This approach views skill learning as occurring through task performance in a progression of contrived environments, such that each is more complex and demanding than the previous. In this sense, learners are considered to be incrementally transferring their past learning to contexts more and more similar to those which will eventually be encountered on-the-road. Moreover, ITL places considerable importance on the need to plan for both near-transfer and far-transfer of skills. Near-transfer refers to real-life skill applications where the on-road context is similar (or near) to those experienced in training. Far transfer refers to real life skill applications where the on-road context is novel and has not been addressed in training. Clearly, near and far transfer require different instructional methods and these are available through the ITL approach. Thus, the instructional strategy we have utilised recognises and utilises to advantage the powerful influence of experience: over time, on-road experience provides the novice driver with more and more driving contexts in which the skills (and mental model) acquired through training can be applied and refined.

Warren assumes that training has only a short term influence on driving behaviour. However we would comment that if, during this short time, even an elementary mental model of the traffic environment is developed which, with the benefit of subsequent experience, accelerates in the novice driver the ability to better anticipate and respond to emerging traffic hazards, the training has been effective.

Warren says that there "is a need to show that any driver training programs influence BOTH the skills of drivers (as an intermediate measure) and, more importantly, the crash risk of novice drivers. Given the general failure of driver-training in the empirical literature to this point, the onus for this demonstration clearly falls on the protagonists." In response to this comment we wish to say that we do not think of ourselves as "protagonists" in this matter. We also wish to re-iterate that we are about to embark on a 6 month evaluation of the product to collect both outcome measures (in a mid-range simulator) and process measures. After that we hope to find some funds to evaluate the product on the road. The proof will be in the pudding.

Re: Bill Wheeler on Wednesday March 17 at 04:08.

First, a general comment. Bill has made it clear in his comments, and rightly so, that he was responsible for designing and conducting a number of simulator experiments during the early phase of the program of research that preceded development of the CD ROM product. His important contribution to this early phase the project is acknowledged by the present authors. Tom Triggs and I have published in conference proceedings the findings from several simulator experiments we conducted after Bill Wheeler left the project. We have referred to Bill's earlier work in some of these publications and his work is referred to in the many internal reports that have been prepared for the sponsoring agency, the TAC. Bill's early work paved the way for some of the later experiments undertaken by the present authors.


Issue 1. Bill draws attention to differences between his ACT paradigm and that used by us in later experiments. The attentional control training paradigm employed in the Regan, Triggs and Deery, 1998c paper (referred to in the present conference paper) was different from that employed by Wheeler et al - for an important reason. Previous experiments by Wheeler et al. (as yet unpublished in the open literature) provide only partial evidence that subjects given ACT training using his technique achieved higher performance levels than those in control groups: ACT training led to statistically superior performance on only one of the training sub-tasks tasks (the shape recognition task) in only one of his experiments (Wheeler, 1997a), but this effect occurred only for older drivers; treatment subjects did not demonstrate significantly greater levels of performance on training sub-tasks as training progressed; and there was no significant generalisation of attentional control skill from the ACT training task to any of the "real world" transfer of training drives (i.e. the capture, conflict or emergency drives).

There are several possible reasons why Bill's ACT training may not have been as effective as it could theoretically have been given the spectacularly successful findings by Gopher (1992) and others:

(a) there may not have been sufficient training trials to enable subjects to acquire attentional control skill. In the studies by Gopher and his colleagues (1992), for example, subjects were given 50, 3-minute training trials. Wheeler et al. gave subjects no more than 12, 3-minute training trials in any one experiment;

(b) the subjects in Gopher's experiments were given on-line feedback during VP training on both their desired and actual level of performance on each task. Wheeler et al. did not provide feedback to subjects. Hence, it is possible that, in the absence of feedback, those who underwent ACT training in his experiments did not learn to properly prioritise attention to the sub-tasks in his training trials;

(c) the tasks used in Gopher's (1992) experiments were "resource-limited" (Norman and Bobrow, 1975). That is, if subjects allocated proportionately more attention to that task, performance would improve. In contrast, one or more of the tasks in the Wheeler et al. paradigm may have been "data-limited". An example of a data-limited task is trying to comprehend a telephone conversation when the reception is poor. Regardless of how much attention one devotes to the task, it is impossible to improve performance on that task. Indeed, subjects who undertook the ACT training in Wheeler's Experiment AC (Wheeler et al., 1997b) were able to give variable priority to the shape recognition and pursuit tracking tasks, but not to the compensatory tracking task, suggesting that at least the latter task may have been data limited;

(d) the lack of significant transfer of attentional control skills from the training task to the driving tasks (i.e. the capture, conflict and emergency drives) in the Wheeler et al Experiment TB (Wheeler et al., 1997c) may have occurred because the skill of attentional control was simply not learned during training, the attentional control skills required for the driving tasks were not sufficiently similar to the training task to allow generalisation to occur, or both. In other words, the driving tasks may not have required much attentional control. It should be noted that in developing the Space Fortress game, considerable effort was made to ensure that the demand elements (e.g. manual control, visual and spatial requirements) were similar to those in actual flight training.

The Regan, Triggs and Deery 1998c experiment was designed to address these and other issues in order to maximise the possibility of obtaining an attentional control training effect similar to that observed by Gopher and others. The critical distinguishing features of the experiment were as follows:

(a) the number of training trials was increased from 12 to 26;

(b) subjects were given feedback on their performance at the end of each training trial;

(c) two rather than three tasks were used in ACT training trials (a numerical calculation task and a headway maintenance task), consistent with Gopher's earlier studies which employed two tasks. The tasks were "resource limited". The headway task has high face validity and construct validity; it is intended to simulate the cognitive demands of driving in dense traffic. It does not matter what cues subjects use to judge headway distance; of importance is that they maintain a constant distance behind the car in front using whatever cues are available. The number task has high construct validity but less face validity. It is intended to draw attention continuously away from the driving task while at the same time requiring subjects to scan the driving environment for numbers, roughly approximating the perceptual and cognitive demands of scanning the road for potential hazards. The training tasks employed by Wheeler et al. were more abstract than these; and

(d) the three "real world" transfer of training sub- tasks (i.e. capture, conflict and emergency drives) used previously by Wheeler et al. were replaced by a single transfer of training (TOTD) drive which contained two sub-tasks: a speed maintenance task and a numeric calculation task.

Despite these differences, Bill Wheeler nevertheless paved the way for a series of subsequent experiments that led to a more refined ATC paradigm.

Issue 2: Bill raises a concern that the target group for our CD ROM product, advanced learner drivers, may not benefit from ACT training, based on the findings of his earlier experiments. This is not an issue, as the design of the CD ROM has been driven primarily by our recent experiments rather than the earlier, but important, work undertaken by Bill. Specifically, the ACT training technique employed in the CD ROM product is based on the training technique refined in the experiment described in the paragraph above, and in a published experiment before that by Regan et al, 1998 (In the Proceedings of the Road safety Research, Policing and Enforcement Conference, Wellington NZ, 16-17 November). In the experiment described above, we found that both learner drivers (average 41 hours experience) and more experienced drivers (average 360 hours experience) benefited from ACT training, but that learners derived even greater benefit in one of the transfer of training drives in the simulator. This was after only 26 training trials. We expect much greater transfer following the CD ROM training, which will involve well over a hundred training trials.

Issue 3: I have already addressed this issue in my response to Dan Keegan dated Wednesday March 10 1999 at 06:39pm.

Issue 4: We have not suggested anywhere that "ACT in any way reduces the need for a driver to develop an appreciation of what must be attended to within that scene."

Re: Bill Wheeler on Wedneday March 17 at 04:12am.

I think we have been misquoted here. We stated in the conference paper (p. 2), on the basis of your earlier experiments that" mere exposure to a near-miss event, at least in the simulator, is not on its own sufficient to guarantee significantly enhanced risk perception in young novice drivers, although the data supported a trend in this direction." We acknowledge that there were trends supporting the effectiveness of the technique that emerged from your earlier work, although these were not always statistically significant. We are not concluding in this statement that the near-miss technique is ineffective, although in a recent experiment directly comparing the relative effectiveness of this technique with that of mediated instruction (see Regan, Deery and Triggs, 1998, in the Proceedings of the 1998 Road Safety Research, Education and Enforcement Conference, Wellington NZ) we did find the near miss technique to be relatively ineffective.

We agree with your comment that the "near-miss" paradigm, coupled with an understanding of the various constraints in its use, does provide a reasonable tool for better understanding how it is that young drivers learn, under normal and natural conditions, to deal with the hazards that they meet. Indeed, on theoretical grounds, this technique should be useful in demonstrating to novice drivers the consequences of failing to exercise appropriate anticipatory avoidance behaviours. For this and other reasons we have utilised the technique in the CD ROM product.

Re: Bill Wheeler on Wedneday March 17 at 04:17am.

You state in the second paragraph of your comment" The effort expended during the research phase of this project was phenomenal. The output in terms of completed experiments and subjects processed exceeded that of any other simulator based research team that I am aware of. And yet, we were unable to produce a theory based
training product. One has to wonder why." We take exception to the comment that "we" (?) were unable to produce a theory based product. As you know, the design specification for the CD ROM product was developed after you left the project and is based on current theories of learning and instructional methods which have been proven, either by us or by our colleagues around the world, to be instructionally effective.

RE: Mike Regan's response to my comments

Mike's response to my comments gives a useful insight into the application of the ITL approach in their package. It does not, however, address my concerns (perhaps poorly expressed) about the relationship between experience, training/instruction, and the development of an appropriate mental model.

During any period of training/instruction, and afterwards, the learner driver will continue to accrue driving experience in the real world. The development of a mental model or representation of the driving environment is more likely to be influenced by this substantial real-world experience than by a (relatively) short period in training/instruction. While Mike suggests that even the development of an elementary mental model should assist the learner driver, this is only so if the elementary mental model is one that is consistent with the development of the more-complex mental model that will ultimately result from driving experience. If the elementary mental model developed through the training process is inconsistent with the experience-based model, there is a potential risk that the skill-development of the learner may actually be hindered in the longer term rather than assisted.

Assuming that the best mental model is one that represents the real-world most accurately, the mental model developed through experience is going to be complex and probabilistic in nature rather than deterministic - the driving environment is both complex and somewhat fuzzy! Some of the problems of novice drivers are probably the result of the development and application of a deterministic or rule-based mental model leading to a failure to perceive or respond to unexpected events.

In the context of the development of CD-ROM-based packages (in general, not just Mike's) I can see a risk that the package will promote either a rule-based model of the driving environment by teaching (even through increasingly complex experience) that particular cues are associated with particular events, or that the package will encourage the development of a mental model of the simulated environment of the package and will therefore have no real impact on the model of the driving environment.

Mike is right, of course, about the pudding! These are armchair arguments that need some good data!

Re: Bea Pullar (Bea) on Wednesday, March 17, 1999 - 04:56 pm.

Bea,Warren and Michael, it has been very interesting to follow your discussions on ADD/ADHD. It is interesting that our CD ROM product has 6 training modules, of which only one deals with attentional control - and yet virtually the whole discussion so far has been about attentional control. Your discussion (Bea)with Warren Harrison about the inability of ADD/ADHD sufferers to supress irrelevant stimuli is an interesting one. To me, it appears to relate back to the seminal work of Donald Broadbent and Alan Welford on the single channel hypothesis. They postulated that there is in the brain a "central executive", if you want to call it that, that limts the amount of information that enters consciousness. It acts as a selective information filter and, in an evolutionary sense, allows us to focus on the things in life that keep us alive. I devoted a PhD to this issue and came to the conclusion that, when we perform multiple tasks, there are different levels of processing that occur in the brain. Conscious activities, such as engaging in conversation, are controlled primarily by the central executive while more automated processes such as walking a delegated to subordinate pprocessors in the brain. At an even lower level are processors in the brain that control basic functions like breathing. However, the central executive has control over lower levels in some circumstances. If you injure your leg, for example, walking becomes temporarily a conscious activity under conscious control - under control of the central executive. Also, the central executive can perform a conscious task while automatic tasks are performed by lower level processes. We can walk and talk at the same time. It seems to me that in ADD/ADHD what is happening is that the central executive is not properly filtering unwanted information - it is allowing automatated process to fire off indiscriminately without intermittently intervening. A great deal of work has been done to understand how people without ADD filter information in mutliple task environments, and paradims such as dichotic listening (eg Cherry, 1953; Broadbent, 1958) have generated vast amounts of data. It seems to me that these, and more recent techniques of this kind, could be used to better understand at a micro level the behavioural manifestations of ADD/ADHD and to develop predictive models of how ADD/ADHD will effect driving and other behaviours under different circumstances.

Re Warren Harrison (Wh1960) on Thursday, March 18, 1999 - 11:57 pm. Warren, the content of this message appears to be identical to that of your earlier message. I think the points you make are valid and important - but there is nothing more that I can add to my reply to your earlier message on this issue. There is obviously a need for further research into the specific issues you raise, which would seem to fall under the general heading of transfer of training.

Re: Michael Regan on March 18, 1999 at 11:15 pm

Issue 1. Mike has provided a very good critique of the ACT we did and its shortcomings. I would only add that the shortcomings were largely a result of either program limitations (maximum length of any experimental session was 50 minutes and novice drivers had to be 16-17 year old students) or the technology (simulator capabilities). Our major challenge and goal was to see if we could make ACT work with an unsophisticated group (novice drivers not pilots) and if we could do so in a driving context. I'm gratified that we did this well enough for Mike to carry the ACT paradigm to a more certain application.

Issue 2. I believe that the reference intended was Triggs and Regan 1998 as per the paper published here rather than Regan et. al. 1998.

Re: Michael Regan on March 18, 1999 at 11:21 pm.

No misquote was intended. We both seem to acknowledge however that simply providing a "near miss" experience would not be sufficient so it's not surprising that it didn't work in the comparison that was done. Indeed once we had an indication that the paradigm might be appropriate most of the remaining experiments were attempts to establish what conditions had to be met to make a "near miss" in the simulator work to change the drivers behavior relative to similar situations in the future. The essential conditions appear to be:

1. The specific type of situation can not already have been encountered in "real world" driving otherwise it is not possible to measure an effect.
2. The "near miss" can not be presented in a way that allows the driver to believe he or she actually changed the outcome.
3. The outcome of the event must not have closure, a "fender bender", "traffic ticket" or even a major crash would all provide closure in ways that would allow the emotional focus of the situation to shift away from the circumstances that surrounded it.

Re: Michael Regan March 18, 1999 at 11:24 pm.

Mike, I recognized that the CD-ROM product that you've developed is based on sound instructional and educational theory. It appears to me that the same could be said of most, if not all, driver education programs.

What I was hoping might be accomplished in the project was that we could learn enough about how drivers acquire safety related behaviors so that we could use such knowledge and the basic learning theory that would underpin that knowledge as the basis for the training. I was obviously not able to accomplish more than develop reasonable concepts and some tentative indications during my time with the project. It's not obvious to me how the theories you've used to develop the CD-ROM reflect that goal either.

In fairness to you and what you've done however I'm sure that if it had been my job to finish up the project we would have been forced to do pretty much the same thing.

To Michael Regan: The discussion above has been enormously helpful to my understanding, both of the driving task and the development of instructional resources such as this CD-ROM. The responses to my previous questions have been most enlightening. I want to add something now to try to establish my understanding more firmly.

In the case of the student I mentioned earlier who tended to over-react to one kind of event - a car emerging from the right - because of a bad experience with this situation in the past .... this is how I interpret the results of your work as it would benefit this student:

1. The CD-ROM would alert her to a variety of other kinds of risks. It would also help her to achieve a balance between the attention she might give to various risky situations in traffic. In other words, the excessive emotional response to a car emerging from the right would become more appropriate. This is a near transfer situation.

2. The student would also get to experience situations which are similar to the risky situations but not so threatening (far transfer).

3. The student would become better at dividing her attention between different tasks and allocating attention efficiently to the most important tasks.

4. There is evidence to support the attention-dividing/allocation improvement, but no evidence yet that crash risk will be reduced.

It seems to me that based on the above, if my interpretations are accurate, the CD-ROM will be a very valuable tool for instructors. However, in terms of crash risk reduction I have questions. For example:

In the very early stages, beginner drivers tend to overreact to cars emerging at side streets, sometimes even slowing to a stop, even though the emerging car has a stop sign and the beginner has right of way. Later, with increasing confidence in their control, their right of way, and that emerging cars don't just charge out, their response becomes more balanced.

However, I can well imagine that, after a year or two of driving and never experiencing a problem with an emerging car, the typical novice driver's alertness to that car is almost extinguished. I would anticipate that most novices would hardly even be aware of it. This seems to me to be normal psychology. I see this as a huge problem for driver training that is focused almost exclusively on the risk of crashing.

I instinctively feel that training can effectively provide a suitable reaction to a car suddenly emerging into the driver's path (braking, steering, etc.). I also believe that the attentional training can help the driver see the emerging vehicle sooner, but somehow, no matter how powerful the risk perception training, I think we'll have a gradual extinguishing of alertness to risks that do not, over a prolonged period, materialize into an actual danger.

Does this seem like a fair assessment?

Since today is the last workday of the conference for our friends "down under" I thought it might be appropriate to express to them how interesting and informative their participation in the forum has been. The undogged determination of Mike Regan to answer all of our questions and of Warren Harrison to ask provocative ones were particularly helpful in making it a successful conference. I'm sure we're all looking forward to learning the results of the simulator test of the CD-ROM product and eventually the field test.

Hi Dan:
Congratulations on putting together this first successful conference. Sorry I didn't have the time to submit a paper or to participate in the discussions. I have read all the papers and have extremely lengthly responses to all of them for which I don't have adequate time to write. I have been out of state most of the month of March, and I just got back from the Lifesavers Conference in Seattle. A few quick comments about the use of this CD-ROM product. It seems like it can be of value to attentional training. In response to your comments of March 22nd: Drivers who have been trained in the Zone Control System are able to make adjustments to the vehicle's speed or lane positioning based upon restrictions (zone changes) that occur to their Line of Sight or to changes that affect their Path of Travel. The principles of the Zone Control System dictate what is acceptable speed selection, lane positioning and communication performance based upon the conditions that do exist, not on what may happen. Because there are only a few principles a driver must learn and they can be applied to zillions of situations, it is easy to develop into habit a style of driving that has low risk and more unconscious attention recognition of situations that are threatening to one's movement within the intended path of travel. For example: a driver using the principles of Zone Control would see a red traffic light as a closed front zone. Seeing a closed front zone would require a braking action to attempt arrival into an open zone, as well as
a check of the rearview mirror. A driver seeing a pedestrian crossing the street would see that as closed front zone also and respond in the same manner as seeing the red traffic light. And, a driver on a limited access highway seeing a lot of brake lights ahead caused by the sudden stopping of traffic would also see that in the same pattern of behavior as the two previous examples. So, if a driver learns to make appropriate responses that are the same for countless situations, the behavior can become appropriate risk prevention actions that become one's style of driving. They would be taking the action for the condition that is present, not for what could happen. Therefore, when the driver is confronted with a high risk situation such as the sudden stop of highway traffic, and with the habit of seeing it 20-30 seconds ahead, and with the habit of beginning to reduce speed in an early manner, and with the habit of checking the rear zone when a braking action is needed, then the driver may have more options for coping with the 18-wheeler to the rear being driven by a driver under the influence of sleep deprivation. Because the Zone Control System places emphasis on positive, precise risk-prevention, space management behavior, with use it doesn't become extinct­­it becomes more ingrained as part of one's pattern of driving.