Using a mathematical representation of brain processes to explain decision making: adapting Friston's free energy principle for travel behaviour modelling

The Free Energy Principle (FEP; Friston 2009, 2010) has been hailed as a groundbreaking idea neuroscience, with the potential to account for a wide range of behavioural phenomena, including exploration and avoidance in uncertain or dynamic environments. Notwithstanding its inherently simple and elegant idea - agents minimise surprise, otherwise known as free energy, through a blief and action updating process referred to as active inference - real-world application has been limited due to computational complexity and the abstract nature of the mathematical framework. This paper introduces FEP to travel behaviour research, developing an active inference model of route choice under uncertainty and applying it to two experimental datasets. Results demonstrate that the FEP model outperforms alternatives based on reinforcement learning (RL), particularly for individuals displaying exploratory or belief- adaptive behaviour. While computationally more demanding, the FEP framework offers substantial advantages in capturing learning dynamics, habit formation, and uncertainty-driven adaptation in travel decision-making, positioning it as a promising neurocomputational foundation for future travel behaviour models.