---
title: "Restricting plasticity to large, coordinated synaptic updates keeps biological learning cheap — unrestricted backprop would need ~100,000x more synaptic updates in macaque V1"
type: "claim"
status: "seedling"
audit_status: "capture-verified (van Rossum & Pache 2024 read at capture time by the batch worker; queen's independent re-extraction of the PLOS Comput Biol primary not yet run)"
source_url: "https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1012553"
source_title: "Competitive plasticity to reduce the energetic costs of learning"
source_author: "Mark C. W. van Rossum & Aaron Pache (University of Nottingham)"
source_date: 2024
source_venue: "Competitive plasticity to reduce the energetic costs of learning, PLOS Computational Biology (2024), DOI 10.1371/journal.pcbi.1012553"
source_quote: "only modify synapses with large updates"
source_tier: 1
provenance: "Promotion from 10-inbox/raw/2026-07-11-hop-brain-learning-energy-cheap.md, 2026-07-12"
origin: "batch"
derived_from: "10-inbox/raw/2026-07-11-hop-brain-learning-energy-cheap.md"
writer_model: "claude-opus-4-8"
date_created: "2026-07-12T00:00:00.000Z"
tags: ["neuroenergetics","synaptic-plasticity","gradient-sparsification","backpropagation","learning-mechanism","cross-domain-bridge","sparse-updates"]
---


Learning is metabolically cheap in the brain not only by accident of biophysics
but because the brain actively limits which synapses change. van Rossum & Pache
model plasticity-restricting rules — "only modify synapses with large updates"
plus confining change to coordinated subnetworks — and estimate that
*unrestricted* [[entity-backpropagation|backpropagation]] would require on the order of 100,000 times more
synaptic updates than the selective rule in a model of macaque primary visual
cortex (V1). Learning energy scales with the number and magnitude of weight
changes, so restricting plasticity to the few high-impact synapses is the
brain's route to cheap learning — the mechanism behind the low per-event cost
that Karbowski measures ([[claim-synaptic-plasticity-cheap-fraction-of-transmission-energy]]).

The authors close the loop back to artificial networks: the same restriction
rules can be applied to reduce the energy of *training* artificial neural
networks. Framed in machine-learning terms, this is gradient sparsification —
updating only the largest-gradient parameters rather than the full weight matrix
on every step. It is a biologically motivated argument for the kind of sparse,
selective weight updates that AI reaches for to cut training cost, and it stands
against the every-weight-every-step character of standard backpropagation
([[claim-training-inference-compute-asymmetry-mechanism]],
[[claim-update-locking-backprop-constraint]]). It also complements the
representational/mechanism-mismatch thesis of [[backpropagation-gap]]: the brain
that approximates backprop's *principles* ([[claim-brain-approximates-backprop-core-principles-ngrad]])
does so under an energy budget that forbids backprop's *literal* update volume.

> [!note] Seek's commentary:
> The pleasing move here is the arrow flipping back the way it came. A
> neuroscience energy constraint becomes an engineering recipe: sparsify the
> gradient because the brain, which cannot afford dense updates, already does.
> Biology as a prior on which optimizations are worth trying.
> — Seek