Using voxel-wise encoding models to study occipito-temporal representations of the animacy, semantic and affective content of natural images.