Two major challenges facing systems neuroscience today are (1) to relate computational brain theory with its notions of parallel computation and population-code representation to massively multivariate spatiotemporal brain-activity data as acquired with functional magnetic resonance imaging (fMRI) and cell recording and (2) to relate brain representations in animal models (e.g. nonhuman primates) to human brain representations. This project tackles these challenges with a focus on visual object recognition in human and nonhuman primates. Object recognition is a still poorly understood key problem of computational neuroscience with implications for cortical computation in general. We will test computational models and relate representational content of population codes between human and nonhuman primates by means of a novel multivariate technique called representational similarity analysis (RSA). The core idea of RSA is to characterize a given brain representation by a dissimilarity matrix of stimulus-evoked activity patterns and to visualize and statistically compare such dissimilarity matrices. In contrast to existing approaches, computational models here form an integral component of the analysis of brain-activity data. We will match up representationally homologous regions between human and nonhuman primate and determine which computational models best explain the empirical data (from human and nonhuman primate fMRI) for each brain region. Moreover this project will further develop the technique of RSA and provide an easy-to-use and freely available Matlab toolbox to the community. By richly relating brain theory to data and human to nonhuman primate studies, this project bridges major divides and will contribute to a more integrated systems neuroscience.