Using human neuroimaging to examine top-down modulation of visual perception

Abstract

Both univariate and multivariate analysis methods largely have focused on characterizing how measurements from neural firing rates, EEG electrodes, or fMRI voxels change as a function of stimulus parameters or task demands-they focus on characterizing changes in neural signals. However, in cognitive neuroscience we areoften interested in how these changes in neural signals collectively modify representations of information. We compare methods whereby activation patterns across entire brain regions can be used to reconstruct representations of information to more traditional univariate and multivariate analysis approaches.We highlight findings using these methods, focusing on how a representation-based analysis approach yields novel insights into how information is encoded, maintained and manipulated undervarious task demands.