Speakers > Ruben Coen-Cagli
Albert Einstein College of Medicine
Ruben Coen-Cagli is an Associate Professor at the Albert Einstein College of Medicine in New York. After a training in theoretical physics and obtaining a PhD from the University of Napoli studying eye-hand coordination with an interdisciplinary approach, Ruben Coen-Cagli joined the lab of Odelia Schwartz at the University of Miami to study the link between statistical structure in natural images and neuronal responses in the visual cortex. He then joined the lab of Alexandre Pouget to study the origins of cortical and perceptual variability. His current work addresses how probabilistic inferences are used in natural sensory processing, using in particular computer vision and machine learning approaches.
A central goal of vision science is to understand the principles underlying the perception and neural coding of the complex visual environment of our everyday experience. In the visual cortex, foundational work with artificial stimuli, and more recent work combining natural images and deep convolutional neural networks, have revealed much about the tuning of cortical neurons to specific image features. However, a major limitation of this existing work is its focus on single-neuron response strength to isolated images. First, during natural vision, the inputs to cortical neurons are not isolated but rather embedded in a rich spatial and temporal context which strongly modulates neural activity. Second, the full structure of population activity—including the substantial trial-to-trial variability that is shared among neurons—determines encoded information and, ultimately, perception. In this talk, I will first briefly and selectively review classical literature on how variability impacts neural coding, through the lens of population codes for low-dimensional stimuli. I will then present a normative theory of population encoding of natural images in primary visual cortex (V1). In this framework, V1 activity serves to approximate a probabilistic representation optimized to the statistics of natural visual inputs, and contextual modulation and variability are integral aspects of achieving this goal. I will present a concrete computational framework that instantiates this hypothesis, and new data recorded with neuropixel arrays in macaque V1 to test its predictions. Lastly, I will discuss statistical analysis methods we are developing, and their application to mouse V1 population data, to probe the underlying circuit mechanisms.
|
