Hevolve AI: Self-Evolving Multimodal AI Agents

Turn your domain expertise into AI agents that keep learning. Hevolve AI lets experts build multimodal AI systems by talking to them and correcting them in real time, with no code to write.

Key Features

Quick Links

© 2024 Hevolve AI Pvt Ltd. All rights reserved.

← All research
Artificial Intelligence
arXiv (cs.LG) · July 17, 2026

Constrained Hebbian Learning Supports Efficient Representational Allocation under Structural Constraints

Patrick Inoue, Florian Röhrbein, Andreas Knoblauch

Brains are not free to wire themselves however they like. Synapses cost energy to maintain, connectivity is limited, and anatomy imposes its own shape. Those constraints push toward codes that squeeze the behaviourally useful information into patterns with little redundancy, simply because redundancy is expensive.

This study asks whether a competitive Hebbian rule, restricted to excitatory connections, can allocate that limited synaptic budget sensibly under such pressure. The comparison is not accuracy alone but the trade between representational cost and performance, measured with mutual-information-based quantities, against reference learning rules.

Framing it as an allocation problem is what makes the question sharp. Any rule can look good with unlimited resources. The useful question is which rule spends a fixed and genuinely scarce budget well, which is the situation biology is actually in.

From the arXiv (cs.LG) abstract

Introduction: Biological systems face anatomical and metabolic constraints, including costly synaptic maintenance and limited connectivity. These constraints favor neural codes that compress behaviorally relevant information into low-redundancy patterns. We test whether an excitatory competitive Hebbian rule can support synaptic resource allocation under such constraints and whether the resulting representations occupy a more favorable cost-performance regime than reference…


More Artificial Intelligence papers