Code to simulate linear-nonlinear-linear models. This project involves finding the optimal surround size for multilayered systems with respect to efficient coding.
Python
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
scripts
surround
.gitignore
README.md
setup.py

README.md

surround-size

One mystery in the retina is that there are multiple types of inhibitory interneurons that linearly contribute the same temporal feature but at different spatial scales. Why would a system ever do this? Why not just have one inhibitory interneuron at the average spatial scale?

This code base supports analyses done for explaning this problem using efficient coding, which supposes that the goal of the retina is to maximize information subject to a constraint on the retina's output variance.