Localizing Brain Function Based on Full Multivariate Activity Patterns: The Case of Visual Perception and Emotion Decoding

Abstract

AbstractIt’s now common to approach questions about information representation in the brain using multivariate statistics and machine learning methods. What is less recognized is that, in the process, the ability to perform data-driven discovery and functional localization has diminished. This is because multivariate pattern analysis (MVPA) studies tend to restrict themselves to regions of interest and severely-filtered data, and sound parameter mapping inference is lacking. Here, reproducible evidence is presented that a high-dimensional, brain-wide multivariate linear method can better detect and characterize the occurrence of visual and socio-affective states in a task-oriented functional magnetic resonance imaging (fMRI) experiment; in comparison to the classical localizationist correlation analysis. Classification models for a group of human participants and existing rigorous cluster inference methods are used to construct group anatomical-statistical parametric maps, which correspond to the most likely neural correlates of each psychological state. This led to the discovery of a multidimensional pattern of brain activity which reliably encodes for the perception of happiness in the visual cortex, cerebellum and some limbic areas. We failed to find similar evidence for sadness and anger. Anatomical consistency of discriminating features across subjects and contrasts despite of the high number of dimensions, as well as agreement with the wider literature, suggest MVPA is a viable tool for full-brain functional neuroanatomical mapping and not just prediction of psychological states. The present work paves the way for future functional brain imaging studies to provide a complementary picture of brain functions (such as emotion), according to their macroscale dynamics.