|
Efficient Coding in Visual Cortex| title | Efficient Coding in Visual Cortex |
|---|
| start_date | 2025/03/11 |
|---|
| schedule | 16h-18h |
|---|
| online | no |
|---|
| location_info | salle D2.2 & en ligne |
|---|
| summary | I will describe experiments in the primary visual cortex that test fundamental principles of sensory representation, focusing on how populations of neurons efficiently encode the retinal image. First, we examine the dynamics of a sparsification network, inspired by Olshausen & Field’s work on sparse coding. We show that stimulus onset initially triggers widespread cortical activation, reducing sparseness and increasing mutual information. As activity declines due to competitive interactions, sparseness increases while mutual information remains sustained. Notably, coding efficiency—quantified as the ratio of mutual information to metabolic cost—progressively increases, demonstrating an adaptive optimization of sensory representations over time.
Second, we describe a power law of adaptation, where the population response magnitude follows a power-law relationship with stimulus probability. We show how this power law emerges naturally from an efficient coding framework in which the neural population adapts to enhance both stimulus detection and discrimination while minimizing overall neural activity.
Together, these findings illustrate the explanatory power of efficient coding principles in understanding cortical function, revealing how neural populations dynamically optimize their responses to maximize information transmission while minimizing metabolic cost. |
|---|
| responsibles | Sarti, Citti, Petitot, Nadal, Ribot |
|---|
Workflow history| from state (1) | to state | comment | date |
|---|
| submitted | published | | 2025/03/10 12:47 UTC |
| |
|