Online synaptic credit assignment in active dendrites
- 1Tsinghua University, China
Abstract
Credit assignment in neural networks is usually formulated as the computation of an abstract error gradient. Whether such a gradient can take a physical, causal form in biophysically detailed multi-compartment neuron models, and enable online, supervised learning, remains unclear. Here we show that the gradient of a detailed neuron's voltage with respect to a synaptic weight is itself a voltage. Differentiating the discrete backward-Euler update solved by standard simulators yields equations of the same form as the original voltage dynamics, driven by gradient currents. Replaying weight-specific gradient currents forward in time reproduces the exact gradient with high fidelity in an L5 pyramidal neuron across diverse input regimes (R2 > 0.998). Pairing the replayed gradient voltage with a local learning signal yields a causal, online learning rule. A single L5 pyramidal neuron with active dendrites learns to reproduce target voltage trajectories containing calcium plateaus and bursts, and recurrent networks of detailed neurons learn to generate target temporal patterns, far outperforming a readout-only control. These results demonstrate that synaptic credit assignment can be implemented online by the voltage dynamics of detailed neurons, potentially suggesting a physical substrate for gradient-based supervised learning in the brain.
Related
- Local translation couples synaptic activity to mitochondrial adaptation in dendrites
- ProtAgent: Post-Training Compact Protein Tool Agents with Interaction-Level Credit Assignment
- Local processors: Active dendrites and artificial intelligence
- Opposing Roles for the Spectraplakin Short Stop in Stable and Dynamic Dendrites Reveal Divergent DLK Signaling and a Role in Dendrite Regeneration
- Voltage dynamics of cortical dendrites in vivo
BibTeX
@misc{he2026online,
title = {{Online synaptic credit assignment in active dendrites}},
author = {G. He and S. Zhang and K. Du and T. Huang},
howpublished = {bioRxiv},
year = {2026},
month = sep,
doi = {10.64898/2026.09.14.751357},
url = {https://www.biorxiv.org/content/10.64898/2026.09.14.751357v1},
note = {CC BY-NC-ND},
}
RIS
TY - UNPB
TI - Online synaptic credit assignment in active dendrites
AU - G. He
AU - S. Zhang
AU - K. Du
AU - T. Huang
PY - 2026
DA - 2026/09/14
PB - bioRxiv
DO - 10.64898/2026.09.14.751357
UR - https://www.biorxiv.org/content/10.64898/2026.09.14.751357v1
AB - Credit assignment in neural networks is usually formulated as the computation of an abstract error gradient. Whether such a gradient can take a physical, causal form in biophysically detailed multi-compartment neuron models, and enable online, supervised learning, remains unclear. Here we show that the gradient of a detailed neuron's voltage with respect to a synaptic weight is itself a voltage. Differentiating the discrete backward-Euler update solved by standard simulators yields equations of the same form as the original voltage dynamics, driven by gradient currents. Replaying weight-specific gradient currents forward in time reproduces the exact gradient with high fidelity in an L5 pyramidal neuron across diverse input regimes (R2 > 0.998). Pairing the replayed gradient voltage with a local learning signal yields a causal, online learning rule. A single L5 pyramidal neuron with active dendrites learns to reproduce target voltage trajectories containing calcium plateaus and bursts, and recurrent networks of detailed neurons learn to generate target temporal patterns, far outperforming a readout-only control. These results demonstrate that synaptic credit assignment can be implemented online by the voltage dynamics of detailed neurons, potentially suggesting a physical substrate for gradient-based supervised learning in the brain.
ER -