We asked (a) if having a 1/n neural code make neural networks more robust, and (b) how does the neural code