subscribe: Posts | Comments | Email

introduction: thesis traffic psychology


This is the introduction chapter of the thesis “From adaptive control to adaptive traffic behaviour” about traffic psychology and behavioural adaptation of drivers, by Wim van Winsum. The thesis is from 1996. It describes a number of behavioural experiments into car driving that were performed in a research driving simulator.

Other chapters of this thesis can be found here:



During the last decade the field of driver behaviour modeling has suffered from a lack of progress. This has been attributed to a number of different causes, the most important one being the preoccupation in traffic psychology with accidents and accident causation (Ran­ney, 1994). As a result it has never been clear whether driver theories should explain accidents or everyday driving. Several authors have stressed that the actual traffic situations instead of accident analysis should be the main focus and that driver theories should explain everyday driving instead of accidents (for example Rumar, 1988).

The second factor behind the lack of progress in driver modeling is the fact that the motivati­onal models which are dominant today have failed to generate testable hypotheses (Ranney, 1994) mainly because of the confusion between individual and aggregate levels of analysis (Michon, 1989). Also, the continuing debate concerning the validity of the risk homeostasis theory has stalled progress.

Michon (1985) has attributed the lack of progress in driver behaviour modeling to the failure to incorporate the results from the ‘cognitive revolution’ in psychology. He divided the task of car driving into three levels of skills and control: strategic (planning), tactical (maneu­vering) and operational (control). On the strategic level, trip planning and the selection of trip goals and route occur. On the tactical level, sometimes referred to as the maneuvering level, the driver negotiates prevailing circumstances. It includes maneuvers such as obstacle avoidance, gap acceptance, overta­king, choice of headway during car-following and speed choice. The operational level relates to direct lateral and longitudinal vehicle control. Michon postulated that a comprehensive model of driving should take these levels into account, and specify the relations between them. However, all existing models have focused almost exclusively on one level. Ranney (1994) regarded the hierarchical control structure between these levels as one of the most significant developments in the field of driver modeling. It forms a basis for the development of modern driver behaviour theories.

Huguenin (1988) saw the abundance of determinants and factors that operate simultaneously as the cause for the lack of a general theoretical basis or a comprehensive model of driver behaviour which has resulted in several theories that apply only to a limited problem domain.

The causes for the limited progress in driver modeling referred to by Ranney, Michon and Huguenin converge in the problem of ‘human behaviour feedback’ which has puzzled many traffic safety researchers during the last decade and triggered fierce discussions about the effects of safety measures. This issue of the apparently unpredictable human behaviour effects following road safety measures was explored by Evans (1985). He compared the expected safety effects with the actual safety changes in 26 studies and found evidence of changes greater than expected, as expected, smaller than expected, no safety change and perverse effects (safety change opposite in sign to expected). Evans concluded that no behavioural model was available to predict effects of changes in the road-vehicle-driver system. The same issue was referred to as ‘behavioural adaptation’ instead of ‘human behaviour feedback’ in a report of the OECD (1990). Behavioural adaptation was defined as “those behaviours which may occur following the introduction of changes to the road-vehicle-user system and which are not consistent with the initial purpose of the change …”. Because behavioural adaptation may strongly affect the success or failure of road safety measures, collection of driver behaviour data is at least as important as accident data. Accident and fatality data do not contribute as much to an understanding of the process that produced them as driver behavior data because they are only a summary or a final result of a complicated process.

The effects of ‘human behaviour feedback’ as defined by Evans and of ‘behavioural adaptation’ as defined in the OECD report are limited to changes related to the road and the vehicle. For example, in the OECD report a number of studies are referred to that present evidence of increases in speed if lane width or shoulder width is increased while accidents are reduced. Also, the presence of edge lines has been associated with speed increases and accident reductions. However, in chapter 2 evidence will be presented that adaptation is a much more general phenomenon that can be observed in the fields of individual differences, transient states, effects of age and so on. At present no theory is able to predict or explain the changes in behaviour after the introduction of a road safety measure, although several theories have claimed to explain some of the effects. The theory that has been associated most often with the ‘human feedback effects’ is Wilde’s risk homeostasis theory (Wilde, 1982). The most important reason for the fierce discussions evoked by this theory centers around the explanation it provides for the ‘human feedback effects’. The emotional discussions over the reasons for this phenomenon and surrounding any attempt at theory development have incapacitated the progress in driver modeling.

In chapter 2 a model of driving behaviour is developed in which the process of adaptation plays a dominant role. Although the meaning of the term adaptation in this thesis differs from its meaning in the OECD report, the underlying process is thought to be the same. In both cases behaviour on the tactical level is changed as a function of some factor. ‘Behavioural adaptation’ as defined in the OECD report results in a smaller safety benefit than expected after the introduction of a safety measure, because behaviour becomes more ‘risky’. This is sometimes referred to as ‘negative adaptation’: drivers choose a higher speed or follow at a smaller headway. It is often assumed that the number of traffic accidents would have been reduced more if behaviour had remained constant. In contrast, adaptation as described in this thesis sometimes may result in a process in which system safety increases because behaviour becomes less ‘risky’, in the sense that drivers sometimes choose a lower speed or follow at a larger headway in response to some factors. It is then assumed that both effects on behaviour are two sides of the same coin.

Most driver models can be characterized by an emphasis on either indivi­dual differences or on situational factors and are limited in scope to either the tactical level or the operational level. The emphasis on either of these is in part determined by histori­cal reasons. A comprehensive driver model should be able to handle both indivi­dual differences and situational factors as well as the operational and the tactical level of car driving behaviour. In this, several factors related to the driver, the vehicle and the road-environment have to be incorporated into a comprehensive framework.

The purpose of the present study is to construct and validate a model of driving behaviour in which these requirements are met, starting from a discussion of existing theories and models of driving behaviour in chapter 2. The essence of the model that is derived in the course of the next chapter is that human drivers operate at different levels simultaneously. Several factors affect the quality of operational performance. These factors may be related to individual differences in perceptual-motor ability affected by age or to temporary state-related effects induced by marijuana or alcohol. But situational factors such as sight distance may also affect operational performance. The main point discussed in chapter 2 is that both individual and situational factors that affect performance on the operational level will ultimately result in an adaptation of behaviour on the tactical level and in some cases also on the strategic level. Therefore the model is referred to as the adaptation model of car driving behaviour. In chapter 2 evidence is presented that adaptation of behaviour on the tactical level to changes in performance on the operational level may be crucial from a safety point of view. Accident involvement appears to be highest in cases where adaptation fails. This makes the study of the process of adaptation not only important from a scientific point of view but also from a traffic safety point of view.


Outline of the study. Chapter 2 serves as a theoretical section and discusses a number of driver models from the perspective of adaptation. A general model of driver behaviour is presented that emphasizes the interactions between the operational and the tactical level of the car driving task. The skill models, motivational models and adaptive control models are analyzed in terms of their emphasis on either the operational or tactical level and situational factors or individual differences. During this discussion a number of central problems in traffic psychology is examined in detail. The problem of the elderly driver is analyzed in terms of individual differences in perceptual-motor abilities. The effects of alcohol and drugs (marijuana) are analyzed in terms of state-related factors that result in a transient degradation of operational performance. The problem of the young male driver is associated with motivational factors and discussed in terms of the motivational models. Paragraph 2.5 connects the operational and the tactical levels of behaviour with the concept of safety margins, and makes the adaptation model more explicit for the lateral and the longitudinal driving control tasks. Paragraph 2.6 serves as a link between the theoretical model and six experiments that were performed to test a number of elements from the adaptation model. In these experiments one important aspect of the adaptation model is examined in detail: the extent to which individual differences in behaviour on the tactical level are related to individual differences in operational performance and perceptual-motor skills.

The experiments were performed in the TRC driving simulator. Chapter 3 discus­ses this research instrument and the contribu­tion of the author to its development in more detail. Two car driving tasks, negotiating curves and car-following, are studied in the chapters that follow.

Experiment 1 analyzes the driver task of curve negotiation and this is discussed in chapter 4. It focuses on the relation between steering performance and speed choice in curves with different radii. The car-following task and its relation with braking performance is examined in the experiments 2 to 6. These are discussed in the chapters 5, 6, 7, 8 and 9 respectively.

The general results from the six experiments and the relevance for the adaptation model are discussed in chapter 10, together with a number of general conclusions and some next steps.



  1. research driving simulators – Car driving simulator blog - […] you are interested in good examples of behavioural studies into car driving that have been performed in a driving…