Normative theory of functional interactions in primary visual cortex

Abstract

I will start by briefly reviewing empirical literature on two widespread observations in primary visual cortex (V1): contextual modulation (how the activity of a neuron in response to a target stimulus is modulated by contextual stimuli) and response variability (fluctuations in neural activity across repeated presentations of the same stimulus). I will argue that the descriptive model of divisive normalization suggests a precise relation between those phenomena, captured quantitatively through our recent extension to stochastic normalization applied to V1 data from macaque and mouse V1.
Motivated by the empirical observations and statistical modeling, I will then introduce a normative theory to explain the relation between normalization and variability. Specifically, we hypothesized that the computational goal of V1 is to approximate a probabilistic representation optimized to the statistics of natural visual inputs, and that the structure of V1 activity is best understood in the light of this goal. I will present a concrete computational framework that instantiates this hypothesis and reproduces a wide array of classical observations on contextual modulation and shared variability. Building on that foundation, I will argue that a complete understanding of this phenomenology must also account for the non-stationary statistics of natural inputs to which cortex is attuned. The theory makes detailed predictions about the sensitivity of V1 neurons to segmentation and grouping cues, including a surprising flexibility of functional interactions that we have confirmed recently with data recorded by our collaborators using multielectrode arrays in macaque V1.
I will conclude by presenting extensions of our computational models to deep probabilistic algorithms for natural image and video segmentation, and briefly outline ongoing work to generate and test new predictions on population-level coordination and dynamics.

About Ruben Coen-Cagli

I am Associate Professor in the Department of Systems and Computational Biology and Department of Neuroscience at Albert Einstein College of Medicine. I obtained my degree in theoretical physics from the University of Napoli Federico II in 2005 with a thesis on quantum computing. In my graduate work at the same university, I combined eye-tracking experiments, Bayesian modeling and a robotic arm simulator to study eye-hand coordination. In 2008 I started a postdoc with Odelia Schwartz, studying the link between statistical structure in natural images and neuronal responses in visual cortex, in collaboration with Peter Dayan and Adam Kohn. I joined the lab of Alexandre Pouget in 2012 with the SSN-IBRO postdoctoral fellowship, to study the origins of visual cortical and perceptual variability. I joined Albert Einstein College of Medicine in 2016.