Learning Structure – Neuroplastic AI

This public technical disclosure introduces the concept of Neuroplastic AI: topology-adaptive neural systems in which structural change can form an integral part of learning, adaptation, or deployment. Unlike conventional neural networks, where learning is limited to optimizing weights within a fixed architecture, this approach treats operations such as growth, pruning, rewiring, and modular reorganization as potential components of the learning process itself. The paper situates Neuroplastic AI in relation to neural architecture search, neuroevolution, dynamic sparse training, mixture-of-experts systems, pruning methods, continual learning, and biologically inspired structural plasticity. It outlines a broad technical design space encompassing different realization paths, trigger mechanisms, structural units, adaptation regimes, and control strategies, without presenting a single reference implementation. Originally published on Zenodo: https://doi.org/10.5281/zenodo.19735396 (April 24, 2026). This OSF record serves as a long-term mirror and preservation copy. Copyright © 2026 Thomas R. Glueck. All rights reserved. This work is made publicly available for documentation, citation, and public disclosure purposes. No copyright or patent license is granted. No reproduction, redistribution, modification, adaptation, sublicensing, commercial use, or creation of derivative works is permitted without prior written permission, except for short quotations and references as permitted by applicable law and standard scholarly citation, review, or commentary practice.