Bridging Choice Modelling and Mathematical Psychology for Developing Dynamic Models of Driving Behaviour

Abstract

Modelling decisions in driving contexts, where the decision-maker has to make a single or a series of choices in a dynamic scenario is much more challenging than modelling decisions made in static scenarios. This is particularly the case when attributes associated with the alternatives and/or the choice set are changing quickly over time. These decisions have typically been modelled by splitting time into discrete intervals and assuming that the decision-maker evaluates the alternatives at each interval and makes a discrete set of choices based on utility maximisation principles.  On a parallel stream, in the field of mathematical psychology, `sequential sampling' (accumulator) models and dynamic quantum models have been developed to capture the accumulation of preference over time. However, the application of these `dynamic' models has been largely limited to scenarios with static attributes. In this paper, we develop dynamic models based on the modelling principles used in mathematical psychology for fast-evolving dynamic choice contexts. Specifically, we develop and operationalise two very different examples from mathematical psychology for dynamically changing attributes in the context of driving behaviour problems. We demonstrate that analytical and simulated versions of the models provide a better account of behaviour than models based on more traditional methods. These results show promising potential for improving the modelling of other truly dynamic contexts such as in gambling or emergency health interventions.

Keywords: Decision field theory, Quantum choice model, Driving behaviour, Dynamic choice contexts, Choice modelling