Proceedings of the Annual Meeting of the Cognitive Science Society

The translation of retinal images into recognizable objects and scenes is not yet well understood. Beyond edge-detection in primary visual cortex, higher stages of cortical representation are still uncertain. We use a multi-layer convolutionalneural network (Krizhevsky, 2012) to provide models for visual selectivities in the ventral visual pathway. We examine individ-ual neural units, or ”kernels”, in CNN layer 2, correlating kernel activity to single fMRI voxel activity for 1750 natural images(Kay, 2008). Building on G ̈uc ̧l ̈u (2015), we find most significant voxel-kernel correlations in V2, with additional matchesthroughout the ventral pathway. Notably, only 25% of kernels correlate with voxel responses — many voxels correlate with aconsistent small set of kernels. Inhibition of voxel response for kernel selectivities also was observed. Our results indicate alimited number of CNN kernels may be used to gain a finer understanding of voxel level representations in the mid-level ventralvisual pathway.