How to model different strategies in dynamic task environments

In a study about interaction with in-vehicle information systems, participants were put in a dual-tasking scenario. They had to operate a basic research test while driving in a driving simulator. Individual differences were found both in single tasking as well as in multitasking condition. It turned out that during multitasking participants seem to change their interaction strategies. In an immediate follow up single task setting, the participants retain these newly learned strategies. This paper introduces the experimental background of the study, presents corresponding ACT-R models that reflect the participants` behavior before and after the dual tasking treatment. Individual differences Modeling of individual differences in the context of usercentered design is of particular interest. From a cognitive point of view, psychology provides two perspectives on how to explain individual differences: a first approach focuses on different aptitudes or abilities of people. Differential psychology – one core discipline on individual differences – is interested in concepts such as intelligence or creativity. Ackerman (2005), for instance, is a prominent supporter of this approach. In his view, individual differences are based on ability determinants. He proposes to consider individual differences in respect of skill acquisition. The clarifying example of driving illustrates this position: at the beginning, we are novices altogether. Practice makes us more or less experts in driving. This first direction is denoted as parameter-based approach. In the context of cognitive modeling, prominent parameters that have already been considered are working memory (capacity) or processing speed (Schunn & Reder, 2001). For instance, Daily, Lovett & Reder (2001) propose an ACT-R model on individual differences that focuses on working memory performance. Lovett (2001) gives a clear distinction between architectural differences, such as processing speed and working memory capacity, and knowledge-based differences (Lovett, 2005). A second direction accounts for individual differences on the basis of individual, different strategies. Different strategies, for example, can be attributed to different groups of people (e.g., younger vs. older children). Facts about the world, experience and strategies belong to this approach. Schunn & Reder (1998) propose to combine the parameterbased with the strategy-based approach and to pay more attention to the aspect of strategies selection. The strategy that is most appropriate for a concrete situation is selected among a set of possible strategies. One major advantage of this perspective on individual differences is its explanatory power in many domains. A popular domain in the field of human-machine interaction is driving. The focus of this paper is on individual strategies in a multitasking scenario in a driving simulator. We argue that in our study, not all participants use the same strategy. Strategies, moreover, seem to be caused by adaptive behavior in the context of multitasking.