Understanding attention selection in driving: From limited capacity to adaptive behaviour
Doctoral thesis, 2011

Accident analysis studies have consistently identified attention-related failures as key factors behind road crashes. However, less is known about how such failures lead to accidents. Traditionally, one reason for this knowledge gap has been a lack of sufficiently detailed data from the pre-crash phase, although this situation is currently changing with the advent of naturalistic driving studies. However, a remaining issue is the lack of an adequate conceptual model of attention selection applicable in natural driving situations. Existing attention models applied in the driving domain are generally based on the notion of attention as a resource with limited capacity, subject to overload in demanding conditions. Such models have mainly focused on dual task interference in experimental situations and but have put less emphasis on aspects central to attention selection in everyday driving such as expectancy and anticipatory attention allocation. The general objective of the present thesis was to obtain a better understanding of the relation between attention, performance and crash risk in real driving situations. To this end, a general conceptual framework for understanding attention selection in natural driving was developed, based on the view of attention selection as a form of adaptive behaviour rather than a consequence of limited information processing capacity. This also involved the development of a specific model of attention selection mechanisms and a series of empirical studies to support the model development. The main objective of these studies was to better understand the effects of working memory (or cognitive-) load on driving performance and the key mechanisms behind expectancy and proactive attention scheduling in driving. A key finding was that working memory load appears to selectively affect aspects of driving performance that can be characterised as controlled, while leaving reflexive and habitual, automatic, behaviours largely unaffected, an idea that resolves several inconsistent findings in the existing literature. Based on the proposed model, precise definitions of attention, expectancy, driver inattention and driver distraction were proposed. The thesis also suggests a general conceptualisation of the relation between attention selection and crashes. Finally, practical applications of the present findings in the areas of accident and incident analysis, countermeasure development and evaluation methods are discussed.

driver distraction


working memory load

adaptive driver behaviour

accident analysis

driving performance


Hörsal Delta, Hus Svea, Forskningsgången 4, Lindholmen
Opponent: Professor John Lee, University of Wisconsin


Johan A Skifs Engström

Chalmers, Applied Mechanics, Vehicle Safety

Areas of Advance


Subject Categories

Psychology (excluding Applied Psychology)



Doktorsavhandlingar vid Chalmers tekniska högskola. Ny serie

Hörsal Delta, Hus Svea, Forskningsgången 4, Lindholmen

Opponent: Professor John Lee, University of Wisconsin

More information