Measuring Multisensory Integration in Reaction Time: Relative Entropy Approach

Area of Science:

• Cognitive Neuroscience
• Psychology
• Information Theory

Background:

• Multisensory integration (MI) is defined as a change in response to crossmodal stimuli compared to unimodal stimuli.
• Current methods for quantifying MI using reaction times (RTs) lack consensus, often relying on mean or median RTs.
• These traditional measures do not fully capture the complex effects of MI on response behavior.

Purpose of the Study:

• To address the lack of consensus in quantifying multisensory integration (MI) using reaction time (RT) data.
• To propose novel quantitative measures for MI that incorporate the entire RT distribution.
• To introduce relative entropy as a key statistical tool for assessing MI effects.

Main Methods:

• Critique of traditional RT measures (mean, median) for assessing multisensory integration (MI).
• Proposal of novel quantitative measures based on the entire distribution of reaction times (RTs).
• Application of relative entropy (Kullback-Leibler divergence) to measure differences between RT distributions.

Main Results:

• Numeric measures involving only mean or median RTs are insufficient for fully assessing multisensory integration (MI).
• Novel measures utilizing entire RT distribution functions provide a more comprehensive assessment of MI.
• Relative entropy is identified as a central statistical concept for quantifying the distance between RT distributions.

Conclusions:

• Traditional reaction time (RT) metrics are inadequate for a complete understanding of multisensory integration (MI).
• Analyzing the full RT distribution, particularly using relative entropy, offers a more robust approach to quantifying MI.
• This work lays the theoretical groundwork for advanced statistical analysis of multisensory processing.