Explaining human interactions on the road requires large-scale integration of psychological theory

Interaction between road users is a societally important type of human interaction, which has been hypothesised to draw on a variety of cognitive mechanisms. These mechanisms are mostly studied and modelled in separate subfields of psychology, despite calls for increasing theoretical integration in the psychological sciences. Meanwhile, models of human road user interaction are lacking, required to enable deployment of automated vehicles in interactive traffic environments. Here, we demonstrate that to reproduce a set of well-established empirical phenomena in naturalistic driver-pedestrian interaction, we need to combine a large number of existing psychological models, integrating theories of sensory noise, Bayesian perception, evidence accumulation decision-making, long-term valuation of action affordances, behavioural game theory, and theory of mind. Our results demonstrate the feasibility and value of large-scale integration of psychological theory, and also underscore the high mechanistic complexity behind the interactions in which automated vehicles will need to be capable of taking part.