Summary statistics of learning link changing neural representations to behavior
Open Access
- 29 August 2025
- journal article
- review article
- Published by Frontiers Media SA in Frontiers in Neural Circuits
- Vol. 19, 1618351
Abstract
How can we make sense of large-scale recordings of neural activity across learning? Theories of neural network learning with their origins in statistical physics offer a potential answer: for a given task, there are often a small set of summary statistics that are sufficient to predict performance as the network learns. Here, we review recent advances in how summary statistics can be used to build theoretical understanding of neural network learning. We then argue for how this perspective can inform the analysis of neural data, enabling better understanding of learning in biological and artificial neural networks.Keywords
All Related Versions
Funding Information
- National Institutes of Health
- National Science Foundation (DMS-2134157)
- Defense Advanced Research Projects Agency
This publication has 51 references indexed in Scilit:
- Drifting neuronal representations: Bug or feature?Biological Cybernetics, 2022
- The Geometry of Abstraction in the Hippocampus and Prefrontal CortexCell, 2020
- Separability and geometry of object manifolds in deep neural networksNature Communications, 2020
- Causes and consequences of representational driftCurrent Opinion in Neurobiology, 2019
- Spatial Embedding Imposes Constraints on Neuronal Network ArchitecturesTrends in Cognitive Sciences, 2018
- Neuroscience Needs Behavior: Correcting a Reductionist BiasNeuron, 2017
- Random synaptic feedback weights support error backpropagation for deep learningNature Communications, 2016
- Tonotopic reorganization of developing auditory brainstem circuitsNature Neuroscience, 2009
- Wiring Optimization in Cortical CircuitsNeuron, 2002
- Statistical Mechanics of Learning2001