Untangling population representations of objects. A closed loop approach to link neural activity to mouse behavior.

ongoing

A paramount component of intelligence is our ability to extract useful information in the world through our sensory observations. Object recognition is a fundamental problem in visual perception: every day we depend on our ability to identify objects in our visual environment, and our brain is capable of accomplishing it effortlessly and in a fraction of a second, in spite of immense variation in the sensory information that arrives in our retinas. Understanding the algorithm that the brain uses to do this complex task is a decisive conquest in neuroscience but in order to understand ethologically relevant visual processing, we need to understand how it drives behavior. Despite significant progress characterizing visual processing, we do not understand how the visual system solves visual inference problems in natural environments and we are still far from having a complete understanding of how the brain creates untangled transformation-invariant object representations in the perceptual/visual domain, that can subsequently be used to guide behavior. The proposed research effort aims to (i) create a state-of-the-art behavioral virtual navigation system for mice, (ii) combine it with recent advanced functional brain recording techniques and sophisticated neural data analysis to study how objects are represented in the activity of large populations of neurons across the visual hierarchy and beyond and (iii) causally relate these representations to the behavior of the animal. The outcomes of this project will provide significant insights into the computations used by the mouse visual cortex to extract relevant features from the environment, identify how distinct features are represented across the mouse visual areas and how in turn these representations guide the behavior of the animals.