|
The Accuracy of Population Codes: from Orientation Selectivity Models to Nonlinear Decoders| old_uid | 1270 |
|---|
| title | The Accuracy of Population Codes: from Orientation Selectivity Models to Nonlinear Decoders |
|---|
| start_date | 2006/05/19 |
|---|
| schedule | 11h |
|---|
| online | no |
|---|
| details | invitée par Frédéric Chavane |
|---|
| summary | What is the accuracy of population codes? What are the factors that impact on this accuracy? How can we decode information from large pools of correlated neurons? We explore these questions in two studies: In the first study (Series, Latham & Pouget, Nat Neuro, 2004), we investigate how cortical connections influence the accuracy of a population code, using models of primary visual cortex. We compare two principal classes of models for orientation selectivity: one in which the tuning curves are sharpened through cortical lateral interactions, and one in which they are not. Detailed spiking models of these classes are constructed, and their activity is decoded. We report that sharpening through lateral interactions does not improve population codes but, on the contrary, leads to a severe loss of information. In addition, the sharpening models generate complicated codes that rely extensively on pairwise noise correlations. Our study generates several experimental predictions that can be used to distinguish between these two classes of models. In the second study, we provide a mathematical analysis of the properties of Fisher Information in large populations of correlated neurons, when noise is Gaussian and multiplicative (the variance of the spike count depends on the mean). In this situation, it is found that linear decoders are limited in the amount of information they can extract and nonlinear techniques become necessary. We analyze the performance of quadratic estimators (in terms of efficiency and amount of data needed) and show that these provide locally efficient estimators, which can be applied to real or synthetic neural data. |
|---|
| responsibles | Riehle |
|---|
| |
|