Identifying Interpretable Visual Features in Artificial and Biological Neural Systems