Generalizing Drift Diffusion Models with a State-Dependent Framework

Abstract

Decision-making in natural environments often involves a trade-off between speed and accuracy, particularly in high-stakes scenarios like predator-prey interactions. Drift Diffusion Models (DDMs) have been used extensively to study speed-accuracy trade-offs, but such work has typically assumed that decisions are made independently of one another. In the wild, the consequences of one decision will often influence future choices. To address this gap, we introduce a new framework - State-Dependent Drift Diffusion Models (S3DMs) - which integrate state-dependent variables into the decision-making process. Using simulations of foraging scenarios, we demonstrate that S3DMs can predict markedly different behavioural outcomes compared to a standard DDM. By incorporating sequential and interdependent decisions, S3DMs offer a framework that is arguably both more realistic and more biologically relevant. Our research contributes to integrating functional and mechanistic explanations of behaviour and suggests a promising direction for future studies in behavioural ecology and neuroscience.