/canonical_neural_operation

Computational models used to explain normalization done in the brain. Based off the paper: *Carandini, M. & Heeger, D.J. (2012). Normalization as a canonical neural operation. Nature Reviews Neuroscience, 13, 51-62.*

Primary LanguageJupyter Notebook

Neuro 1401: Building Models of the Brain

Team: George Moe, Seeam Noor

Group 2: Project 1

Harvard Spring 2021

Python Notebook answers the following questions based on the paper:

Presentation Slides can be found here

  1. Implement the normalization function (Equation 10 in Carandini & Heeger, 2012) and show how this function removes redundancy in an input (i.e., by decorrelating the pixels).

  2. Show how the normalization function induces winner-take-all competition in a population of neurons tuned to different orientations (see Figure 3e in Carandini & Heeger, 2012).

  3. Implement the adaptation version of normalization (Equation 12 in Carandini & Heeger, 2012) and show how this produces light adaptation in the retina.

  4. Discuss the empirical evidence for normalization in the visual system.

  5. Discuss the possible biological mechanisms that could give rise to normalization.

References: Carandini, M. & Heeger, D.J. (2012). Normalization as a canonical neural operation. Nature Reviews Neuroscience, 13, 51-62.