Representational Similarity Analyses

Representational similarity analysis (RSA) is a multivariate method that can be used to extract information about distributed patterns of representations across the brain. It is related to population vector analysis, a staple in the single-unit recording tradition. RSA is a flexible method that can be applied to many types of neuroimaging data, although the focus here is on its application to human functional magnetic resonance imaging. It is well suited to designs where items can be related to one another in various ways, especially if these relationships are continuous, rather than discrete. In short, RSA involves designing a task where trials can be isolated from one another, relating pairs of trials from different conditions to one another with a similarity metric (e.g., correlation), and, finally, making comparisons between the summary values of these condition-wise similarity metrics and drawing conclusions. RSA has been particularly valuable in advancing our understanding of memory by allowing researchers to ask questions about how information is represented either when information is learned or retrieved.