A Tutorial on Representational Similarity Analysis for Research in Social Cognition

Representational similarity analysis (RSA) is a simple and widely used technique for comparing relationships between diverse types of measures using a common set of elements. For instance, RSA allows researchers to compare brain and behavioral measures of the same set of participants, or response patterns to the same set of stimuli across different conditions. This technique effectively quantifies relationships between complex, high-dimensional measures even when their measurement units are not directly comparable, contributing to its recent popularity. However, existing tutorials focus on neuroscience use cases, limiting their applicability to social cognitive research. In this tutorial, we cover an accessible introduction to RSA, discussing its strengths and limitations compared to other multivariate methods. We provide commented code and functions in R and Python in the context of two examples that showcase different applications of RSA, illustrating its potential to address questions of broad interest to the field of social cognition.