Neural random utility: Relating cardinal neural observables to stochastic choice behavior.

Abstract

We assess whether a cardinal model can be used to relate neural observables to stochastic choice behaviour. We develop a general empirical framework for relating any neural observable to choice prediction, and propose a means of bench-marking their predictive power. In a previous study, measurements of neural activity were made while subjects considered consumer goods. Here, we find that neural activity predicts choice behaviour, with the degree of stochasticity in choice related to the cardinality of the measurement. However, we also find that current methods have a significant degree of measurement error, severely limiting their inferential and predictive performance.