Publications
Partners
CAPE TOWN - The attitude of South African drivers is the main cause of fatalities on the country’s roads, says roads and transport department deputy director general James Mlawu.
In reponse to the article published on Wheels24 in October 2012, reader Tyron Louw emailed us to share his thoughts on drivers in SA:
My name is Tyron Louw, I am just about to complete my Master of Science degree in Ergonomics at Rhodes University in the field of psychophysiology of driver fatigue - in particular, my MSc thesis sets out to better understand the two candidate mechanisms that try to explain the development of non-sleep related driver fatigue, as this is forms a large part of accident causation - in 2013 I begin a PhD in Physiology at the same university.
The Driving Question
Driving is a paradox. It is a skill that is relatively easy to master yet it demands the synthesis of many complex abilities. It is a relatively "new" skill, not one that evolution has crafted for us, yet it is so familiar to us that we rarely think about it at all. It is also one of the most over learned of all of our skills when one considers the regularity with which we perform it.
Finally, driving failure can be deadly. There is no need to regale you with the figures for death and injury, but the fact that in South Africa and the world over it remains the major cause of accidental death attests to it importance in terms of societal cost. Of course, the problem in South Africa is compounded by a myriad of technical, behavioural and economic factors, but in terms of the basic psychophysiological factors that influence performance, the South African population is no different to any other.
Given these observations, it is pertinent to ask why that in 2012, we still don’t know how drivers interact with their environment? We understand why factors such as sleep deprivation, monotony and distraction lead to accidents, but we still don’t fully understand the mechanisms underlying how this happens. Below we explore some reasons why this is the case.
The Driver Model
Let’s backtrack a sentence or two. To argue that we don’t understand driving is actually a little disingenuous, because we don’t have a satisfactory model of human performance in general. The simple reason for this is that we are not yet sophisticated enough to understand ourselves. Because human behaviour is context dependent, any models of behaviour have to involve context. Unfortunately, the dominant models we use to try understand information-processing fail to do this. Thus, when we study driving behaviour in any specific context we have to refer vaguely to constructs such as distraction, fatigue, attention, and the like, which are generally relevant but specifically impotent.
Satisficing versus optimising
Drivers drive well enough. That is, they drive sufficiently well to achieve the goal of the task – this is referred to as satisficing. An example might be lanekeeping. Drivers probably don’t care specifically where in the lane they are - it is not an explicit goal of performance - they just want to stay within it. However, when researchers come to measure performance, we act under the compulsion of optimisation.
That is, we study how much the individual deviates from our investigator–specified optimal goal. What other specific measures are available? Unfortunately, the conclusions we reach may totally be divorced from actual on-road behaviour because the goals and measures derive from satisficing and optimising, respectively.
Of course, there will be situations where excursions are not always “bad” behaviours but are crucially dependant upon the context (e.g., overtaking vs. driver sleep onset.) Thus, in order to take a holistic approach to understanding driver performance, we have to consider the driver and the vehicle in the specific context of concern.
The Predictability of Crashes
There is, of course, the problem of using crashes as a criterion, as they are highly unusual, nonlinear events that result from a sequence of linked precursors. In particular, crashes are not mere outliers drawn from distributions of normal driving. Further, some (if not many) crashes that can be traced back to human-error are unavoidable because of the restricted space and time available for driver response.
This has two implications: we cannot observe the phenomenon in real-time, which therefore means that we cannot accurately replicate the situation for further examination. So when we look for patterns in performance and behaviour to predict crashes, we cannot put them against a backdrop of normal driving, which sheds some light on the complexity involved in the analysis of the behaviour crashes.
In summary, the reasons why we cannot accurately account for crash-related driver performances revolve around three points:
1 Our inability to model and understand our own behaviour,
2 Our techniques for assessing driver performance are incongruent with the driving context, and
3 Crashes are almost impossible to predict with any real accuracy.
The central theme to these reasons is simple. The only way we will get a real grasp of driver behaviour and how it influences performance is when we gain a deeper and more detailed understanding of the driver context and when we understand more about contextual interactions.
Share your views in our Readers' Comments below or email us and we'll publish your article on Wheels24.
Reader: Reduce the speed limit
Reader: Try to obey the rules
Reader: Driver attitude to blame
Reader: Driver's disregard the rules
In reponse to the article published on Wheels24 in October 2012, reader Tyron Louw emailed us to share his thoughts on drivers in SA:
My name is Tyron Louw, I am just about to complete my Master of Science degree in Ergonomics at Rhodes University in the field of psychophysiology of driver fatigue - in particular, my MSc thesis sets out to better understand the two candidate mechanisms that try to explain the development of non-sleep related driver fatigue, as this is forms a large part of accident causation - in 2013 I begin a PhD in Physiology at the same university.
The Driving Question
Driving is a paradox. It is a skill that is relatively easy to master yet it demands the synthesis of many complex abilities. It is a relatively "new" skill, not one that evolution has crafted for us, yet it is so familiar to us that we rarely think about it at all. It is also one of the most over learned of all of our skills when one considers the regularity with which we perform it.
Finally, driving failure can be deadly. There is no need to regale you with the figures for death and injury, but the fact that in South Africa and the world over it remains the major cause of accidental death attests to it importance in terms of societal cost. Of course, the problem in South Africa is compounded by a myriad of technical, behavioural and economic factors, but in terms of the basic psychophysiological factors that influence performance, the South African population is no different to any other.
Given these observations, it is pertinent to ask why that in 2012, we still don’t know how drivers interact with their environment? We understand why factors such as sleep deprivation, monotony and distraction lead to accidents, but we still don’t fully understand the mechanisms underlying how this happens. Below we explore some reasons why this is the case.
The Driver Model
Let’s backtrack a sentence or two. To argue that we don’t understand driving is actually a little disingenuous, because we don’t have a satisfactory model of human performance in general. The simple reason for this is that we are not yet sophisticated enough to understand ourselves. Because human behaviour is context dependent, any models of behaviour have to involve context. Unfortunately, the dominant models we use to try understand information-processing fail to do this. Thus, when we study driving behaviour in any specific context we have to refer vaguely to constructs such as distraction, fatigue, attention, and the like, which are generally relevant but specifically impotent.
Satisficing versus optimising
Drivers drive well enough. That is, they drive sufficiently well to achieve the goal of the task – this is referred to as satisficing. An example might be lanekeeping. Drivers probably don’t care specifically where in the lane they are - it is not an explicit goal of performance - they just want to stay within it. However, when researchers come to measure performance, we act under the compulsion of optimisation.
That is, we study how much the individual deviates from our investigator–specified optimal goal. What other specific measures are available? Unfortunately, the conclusions we reach may totally be divorced from actual on-road behaviour because the goals and measures derive from satisficing and optimising, respectively.
Of course, there will be situations where excursions are not always “bad” behaviours but are crucially dependant upon the context (e.g., overtaking vs. driver sleep onset.) Thus, in order to take a holistic approach to understanding driver performance, we have to consider the driver and the vehicle in the specific context of concern.
The Predictability of Crashes
There is, of course, the problem of using crashes as a criterion, as they are highly unusual, nonlinear events that result from a sequence of linked precursors. In particular, crashes are not mere outliers drawn from distributions of normal driving. Further, some (if not many) crashes that can be traced back to human-error are unavoidable because of the restricted space and time available for driver response.
This has two implications: we cannot observe the phenomenon in real-time, which therefore means that we cannot accurately replicate the situation for further examination. So when we look for patterns in performance and behaviour to predict crashes, we cannot put them against a backdrop of normal driving, which sheds some light on the complexity involved in the analysis of the behaviour crashes.
In summary, the reasons why we cannot accurately account for crash-related driver performances revolve around three points:
1 Our inability to model and understand our own behaviour,
2 Our techniques for assessing driver performance are incongruent with the driving context, and
3 Crashes are almost impossible to predict with any real accuracy.
The central theme to these reasons is simple. The only way we will get a real grasp of driver behaviour and how it influences performance is when we gain a deeper and more detailed understanding of the driver context and when we understand more about contextual interactions.
Share your views in our Readers' Comments below or email us and we'll publish your article on Wheels24.
Reader: Reduce the speed limit
Reader: Try to obey the rules
Reader: Driver attitude to blame
Reader: Driver's disregard the rules