Neural Stochasticity Begets Drift Diffusion Begets Random Utility: A Foundation for the Distribution of Stochastic Choice

Abstract

The random utility model is the standard empirical framework for modelling stochastic choice behaviour in applied settings. Though the distribution of stochastic choice has important implications for both testing behavioural theories and predicting behaviour, the theoretical and empirical foundations of this distribution are not well understood. Moreover, the random utility framework has so far been agnostic about the dynamics of the decision process that are of considerable interest in psychology and neuroscience, captured by a class of bounded accumulation models which relate decision times to stochastic behaviour. This article demonstrates that a random utility model can be derived from a general class of bounded accumulation models, in which particular features of this dynamic process restrict the form of the relationship between observables and the distribution of stochastic choice.