Neural networks maintain stable behavior during growth and change
Learning Time-Varying Connectomes for Stable Behavior Across Development
Neural and Evolutionary Computing
Summary
Neural circuits in living organisms change a lot as they grow, but somehow they keep working properly to control behavior. This paper shows how a single program can generate neural connections that change with age but still keep the behavior stable. The researchers used computer models of nervous systems and found that training controllers at a few key ages helps maintain good performance throughout development. Their method reveals that stable behavior doesn’t need fixed brain connections or activity patterns.
What this means in practice
- •For robotics engineers: Design controllers for robots that maintain stable behaviors as their hardware or sensors change over time.
- •For neural prosthetics developers: Create adaptive neural prosthetic devices that adjust connectivity patterns over time to maintain consistent function as user tissues change.
Authors
Christof Fehrman, Jacob J. Morra, Eva A. Naumann
Abstract
Neural circuits must maintain functionality to drive effective behavior while their size, topology, geometry, and intrinsic dynamics change across development. To model developing visuomotor circuits and behavior, we investigated whether a single compact generative encoding could produce functional recurrent controllers across such changes. Building on the HyperNEAT algorithm, we represented this program as a compositional pattern-producing network that received developmental time as an input to generate age-specific connectivity. The developmental substrate prescribed neuronal birth, position, sensory organization, and intrinsic time constants. We tested this approach using two geometrically distinct developmental substrates and two visuomotor tasks: optomotor responses and phototaxis. Controllers optimized at only the initial or mature substrate generally failed when evaluated at other ages. In contrast, controllers optimized across multiple developmental snapshots maintained substantially higher and more consistent performance across trained and unseen ages. Optimization at only three developmental snapshots was often sufficient to produce effective behavior across the tested developmental interval. The resulting connectomes changed substantially with age and did not converge on a single trajectory of sparsity or connection-weight statistics. Controllers also differed in whether they relied on neurons born later in development. These results demonstrate that a shared spatiotemporal time-varying connectivity program can produce recurrent neural networks that preserve behavior across substantial developmental changes. Together, these results provide a framework for studying how functional stability can emerge without requiring structural or dynamical stability.