What's going on in representational drift?

Abstract

Recent research has illuminated the phenomenon of “representational drift,” revealing that neural activity patterns associated with sensation, cognition, and action exhibit dynamic changes over extended periods, spanning days and weeks. This phenomenon has been observed in both rodents and humans, encompassing diverse cortical regions. We’ll highlight recent findings related to representational drift and related computational models that may shed light on its mechanisms. We posit that representational drift is an inevitable consequence of ongoing synaptic plasticity and plays a pivotal role in neural computation. It allows for continuous learning, aiding in the separation and association of memories that occur at distinct temporal intervals. Finally, we will present the outlook for future computational research to advance our understanding of representational drift and its associated circuit models. By delving deeper into this phenomenon, we aim to uncover its broader implications for neuroscience and cognitive science.

邹立诚

Master student of neuroscience

My research interest focus on the application of mathematics and theoretical physices to neuroscience, especially in the analysis of neuronal networks and the computational mechanisms of long-term memory and visual encoding.