Stimulus-Induced Pairwise Interaction Can Be Revealed by Information Geometric Approach

Abstract

Understanding the interaction of neural activities is one of the most important themes in neuroscience. To resolve this question, cross-covariogram analysis of two neurons’ activities is one of most extensively used techniques. This analysis is conducted mostly against the null hypothesis of independent firing. Here, we argue that an additional analysis is required to understand the role of correlation with respect to behaviourally relevant parameters such as visual stimulus. Specifically, we propose conducting this analysis against the null hypothesis of the activity in a control period. We show that information geometric approach can achieve this task. Furthermore, we demonstrate the validity of this method using data taken from the inferior temporal cortex. The results indicate the possible existence of a stimulus-modulated correlation.