Measuring multisensory integration in reaction time: the relative entropy approach

TL;DR

This paper proposes using relative entropy to quantify multisensory integration in reaction times by analyzing entire RT distributions, offering a more comprehensive measure than traditional mean-based methods.

Contribution

It introduces a novel approach using relative entropy to measure multisensory integration in reaction times, emphasizing distributional analysis over mean or median RTs.

Findings

Theoretical examples demonstrate the method's potential.

Highlights limitations of mean-based RT measures.

Suggests future empirical validation and statistical testing.

Abstract

A classic definition of multisensory integration (MI) has been proposed as ``the presence of a (statistically) significant change in the response to a cross-modal stimulus complex compared to unimodal stimuli''. However, this general definition did not result in a broad consensus on how to quantify the amount of MI in the context of reaction time (RT). In this brief note, we argue that numeric measures of reaction times that only involve mean or median RTs do not uncover the information required to fully assess the effect of multisensory integration. We suggest instead novel measures that include the entire RT distributions functions. The central role is played by relative entropy (aka Kullback-Leibler divergence), a statistical concept in information theory, statistics, and machine learning to measure the (non-symmetric) distance between probability distributions. We provide a number of theoretical examples, but empirical applications and statistical testing are postponed to later study.