Heller Lecture Series - January 21st 2020

Heller Lecture Series in Computational Neuroscience

 

Prof. György Buzsáki

NYU Neuroscience Institute and Center for Neural Science, New York University

Prof. Gyorgy Buzsaki

 

On the topic of

Ways to think about the brain

 

Neural computations are often compared to instrument-measured distance or duration, and such relationships are interpreted by a human observer. However, neural circuits do not depend on human-made instruments but perform computations relative to an internally defined rate-of-change. While neuronal correlations with external measures, such as distance or duration, can be observed in spike rates or other measures of neuronal activity, what matters for the brain is how such activity patterns are utilized by downstream neural observers. For example, episodic memory, supported by the hippocampus, is defined as “what happened to me where and when?” Accordingly, the recommended program in neuroscience is to uncover the neuronal mechanisms of the independent ‘what’, ‘where’ and ‘when’. According to this framework, numerous studies imply that the hippocampal system provides ’space coding’ and recent experiments suggest that the same neurons offer ’time coding’ as well. I will discuss the alternative option that hippocampal operations can be described by the sequential activity of neuronal assemblies and their internally defined rate of change without resorting to the concept of space or time. Brains make neither space or time nor do they sense them. Instead, one should try to understand how pre-existing evolving cell assemblies acquire ‘meaning’ (i.e., utility to the brain’s host) through action-based experience. I will also discuss how such pre-existing dynamic can support cognitive performance. Performance in sensory perception, time and space perception, decision-making, short-term memory and motor control obeys a log-scale law. What neuronal mechanisms can support such a wide dynamic range yet in a well-controlled manner? I will demonstrate that skewed (typically lognormal) distributions are fundamental to both structural and functional brain organization, including synaptic weights, firing rates, bursting, population cooperativity, microscopic and macroscopic connectivity, axon diameter and many derived measures such as place field number, size, information per spike, etc. In all brain states, a small minority of neurons and connections may be responsible for ‘good enough’ brain performance. This ‘backbone’, consist¬ing of the fast-firing minority of neurons in a postulated strongly connected network provides the brain’s ‘best guess’ for ‘good enough’ performance but deployment of the weakly active majority is needed for precision performance. The two ends of a continuous log-distribution of physiological parameters may also explain the perceptual contiguity between ‘similar’ and ‘different’. A minority of strongly interconnected neurons may generalize across situations and afford the brain the capacity to regard no situation as completely unknown. In contrast, the mobilization of the reservoir majority of weakly active neurons is needed to reliably distinguish one situation from another. These observations bridge anatomical structure with physiological function and the rules are used to establish how learning results in changing the synaptic matrix via plasticity.

Buzsaki, G. Rhythms of the Brain (Oxford University Press). 2006
Buzsaki G. The Brain from Inside Out (Oxford University Press). 2019

Heller21012020

 

When