BRAIN EAGER: A Proposed New Principle of Brain Organization

University Of California-San Diego, La Jolla CA

Investigators

Abstract

Most brain areas (like visual or motor cortex) are believed to have a map in which neuron location is related to the type of information that neuron processes. For example, in the visual cortex, neurons process information about a part of the visual world that is closely related to their map location in the brain. The idea explored in this project is that some important brain areas have an antimap, rather than a map, in which the type of information conveyed by a neuron is not at all related to its location in the antimap. Instead, information is spread out over the antimap in a way that makes it possible to get all of the available information from a small collection of any neurons in the antimap, as long as a critical number of neurons is selected. The conceptual basis for this antimap idea comes from a new field of mathematics and computer science called "compressed sensing", and compressed sensing places strict limits on the possible ways brain areas can communicate if an antimap is to be formed. A goal of the project, then, is to discover if the evolutionarily ancient brain areas noted above conform to these limits. The reason for the brain to use antimaps is that they provide information in a format that can be used to collect arbitrary pieces of information into a single "object" through learning. The goal of this EAGER project is to explore the new idea about how information is represented in four evolutionarily ancient brain areas: hippocampus, cerebellum, olfactory cortex, and basal ganglia, present in all vertebrates. To achieve this goal the PI will search the literature on anatomical and physiological characteristics of inputs to these four areas (and sub-parts of them), and will compile quantitative neuroanatomical data that will permit a comparison of what is observed with what the idea predicts. The outcome will determine the extent to which this new idea is tenable and lead to understanding its possible implications for computations in the various brain areas. This work is, by its nature, interdisciplinary as it relies on ideas from neurobiology, mathematics and computer science, and on the methods of theoretical physics. If the idea of antimaps is correct, all vertebrate (and perhaps invertebrate) brains use them to solve a wide variety of computational problems. Furthermore, knowing how information is represented in the brain is fundamental to understanding how the brain works, and if this new idea is correct it will play an essential role in achieving the goals of the BRAIN initiative.