Theoretical frameworks for credit assignment via behavioral timescale synaptic plasticity

Abstract Behavioral Timescale Synaptic Plasticity (BTSP) is a form of plasticity which drives place field formation in hippocampal pyramidal neurons. Unlike traditional learning rules, BTSP learns pre-post correlations on timescales of seconds and rapidly changes single-unit activity in only a few trials. To study how BTSP-like learning can be integrated into network models, we propose a generalized theoretical framework, gBTSP, which we apply to unsupervised and supervised tasks in feedforward and recurrent networks. We show that unsupervised gBTSP learns place field maps in the feedforward case and attractor memories in the recurrent case. For supervised learning, we derive how plateau events can reduce task error, enabling gBTSP to solve trajectory matching and delayed non-match-to-sample tasks. Finally, we find that the speed of credit assignment via gBTSP is limited by network depth or CA3-like recurrence, suggesting that additional mechanisms are necessary to account for BTSP-driven few-shot learning in naturalistic settings.

Authors

Institutions

Publication Details

Journal
Nature Communications
Published
2026-11-14
DOI
https://doi.org/10.1038/s41467-026-79000-y
Citations
3
Primary Topic
Memory and Neural Mechanisms
Type
article
Field-Weighted Citation Impact
4.46

Funders

Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Theoretical frameworks for credit assignment via behavioral timescale synaptic plasticity

Rui Ponte Costa, Claudia Clopath, Ian Cone
3 citations
Nature Communications
Memory and Neural Mechanisms
4.46
article

Theoretical frameworks for credit assignment via behavioral timescale synaptic plasticity

Rui Ponte Costa, Claudia Clopath, Ian Cone
article en
3 citations

Abstract

Abstract Behavioral Timescale Synaptic Plasticity (BTSP) is a form of plasticity which drives place field formation in hippocampal pyramidal neurons. Unlike traditional learning rules, BTSP learns pre-post correlations on timescales of seconds and rapidly changes single-unit activity in only a few trials. To study how BTSP-like learning can be integrated into network models, we propose a generalized theoretical framework, gBTSP, which we apply to unsupervised and supervised tasks in feedforward and recurrent networks. We show that unsupervised gBTSP learns place field maps in the feedforward case and attractor memories in the recurrent case. For supervised learning, we derive how plateau events can reduce task error, enabling gBTSP to solve trajectory matching and delayed non-match-to-sample tasks. Finally, we find that the speed of credit assignment via gBTSP is limited by network depth or CA3-like recurrence, suggesting that additional mechanisms are necessary to account for BTSP-driven few-shot learning in naturalistic settings.

Nature Communications
University of Oxford (GB), Imperial College London (GB)
Simons Foundation, Wellcome Trust, UK Research and Innovation, Engineering and Physical Sciences Research Council, Biotechnology and Biological Sciences Research Council
Openalex Percentile: Top 9%
Memory and Neural Mechanisms
4.46
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon