Cognitive Architectures for Modeling Driver Behavior

The Problem. The driving task involves a host of relevant cognitive, perceptual, and motor abilities. Cognitive architectures—frameworks for modeling human cognition and behavior—have evolved to integrate known theories of human cognition into unified theories that account for a wide range of cognitive phenomena. Role of Driving Simulators. Models of driver behavior based on cognitive architectures can be linked into driving simulators to act as virtual drivers, using simulated perceptual and motor processes to drive in the simulator environment. Typically, the output of these virtual drivers is analogous to that for human drivers, thus facilitating comparison between model and human data. Key Results of Driving Simulator Studies. This chapter highlights one model in particular developed in the ACT-R cognitive architecture. This driver model has been shown to account for several important aspects of driver behavior, particularly in the context of curve negotiation and lane changing. Scenarios and Dependent Variables. The ACT-R driver model can be placed in various environments with driving as the only task or with driving while performing another secondary task (such as dialing a cellular phone). The model’s simulation output includes predictions of both driver behavior as well as behavior on the secondary task (e.g., total dialing time). Platform Specificity and Equipment Limitations. The ACT-R driver model currently runs on a Macintosh platform and LISP environment. However, a recent Java re-implementation of the model within the Distract-R prototyping system has made the model more widely available on a range of platforms, in addition to facilitating its use by non-modeling designers and practitioners. 1. Keywords, Key points 1.1