This document discusses closing the loop between biological and artificial vision by using deep convolutional neural networks (DCNNs) to model human vision. Specifically, it discusses: 1) Research showing DCNNs develop representations that become more similar to human brain regions involved in processing faces and places as the networks become more specialized for those categories. 2) Methods for using optimized images to visualize feature differences between DCNN layers and brain regions. 3) Opportunities for closing the loop by using DCNNs to efficiently test models of human vision, identify features detected in different brain regions, and find differences between brain regions.