Beyond utility: incorporating eye-tracking, skin conductance and heart rate data into cognitive and econometric travel behaviour models

TL;DR

This paper explores integrating physiological data like eye-tracking, skin conductance, and heart rate into cognitive and econometric travel behaviour models, enhancing their ability to predict decision-making processes.

Contribution

It introduces a novel approach to incorporate physiological data into both econometric and decision field theory models for travel choices, addressing a significant research gap.

Findings

Physiological data improves static choice model accuracy.

Stress and eye-tracking data enhance dynamic decision models.

Decision field theory benefits more from physiological data than traditional models.

Abstract

Choice models for large-scale applications have historically relied on economic theories (e.g. utility maximisation) that establish relationships between the choices of individuals, their characteristics, and the attributes of the alternatives. In a parallel stream, choice models in cognitive psychology have focused on modelling the decision-making process, but typically in controlled scenarios. Recent research developments have attempted to bridge the modelling paradigms, with choice models that are based on psychological foundations, such as decision field theory (DFT), outperforming traditional econometric choice models for travel mode and route choice behaviour. The use of physiological data, which can provide indications about the choice-making process and mental states, opens up the opportunity to further advance the models. In particular, the use of such data to enrich 'process' parameters within a cognitive theory-driven choice model has not yet been explored. This research gap is addressed by incorporating physiological data into both econometric and DFT models for understanding decision-making in two different contexts: stated-preference responses (static) of accomodation choice and gap-acceptance decisions within a driving simulator experiment (dynamic). Results from models for the static scenarios demonstrate that both models can improve substantially through the incorporation of eye-tracking information. Results from models for the dynamic scenarios suggest that stress measurement and eye-tracking data can be linked with process parameters in DFT, resulting in larger improvements in comparison to simpler methods for incorporating this data in either DFT or econometric models. The findings provide insights into the value added by physiological data as well as the performance of different candidate modelling frameworks for integrating such data.