RESONA: From Symbolic Knowledge Models to Adaptive Spiking Neural Networks
This paper presents RESONA, a research proposal aimed at extending the existing research on symbolic knowledge representations and inference mechanisms developed within the SENSE project into the fields of neuro-symbolic artificial intelligence and neuromorphic computing systems. The research builds upon executable knowledge models, production rules, and semantic structures developed as part of the research activities at Xolution Research. The proposed research directly builds upon the work of Cotteret et al. (2024, 2025), who demonstrated the feasibility of implementing finite-state automata through distributed symbolic representations in attractor neural networks, as well as their implementation in recurrent spiking neural networks (RSNNs), including on neuromorphic hardware. RESONA seeks to extend these findings by investigating the neural realization of selected symbolic processes derived from the executable knowledge models developed within SENSE. Particular attention is devoted to the incremental adaptation of neural representations, the preservation of functional correctness under repeated modifications, and the identification of the limits of such adaptation. The research proposal envisages experimental validation of the proposed mechanisms through simulations, followed by implementation and evaluation on neuromorphic hardware. The aim of this paper is to formulate the research problem, hypothesis, and methodological foundations for investigating adaptive neuro-symbolic representations that could contribute to the development of more complex cognitive architectures in the future. Keywords: Neuro-Symbolic AI, Spiking Neural Networks (SNN), Recurrent Spiking Neural Networks (RSNN), Neuromorphic Computing, Distributed Symbolic Representations, Incremental Adaptation, Continual Learning, Executable Knowledge Models, Attractor Neural Networks, Symbolic Reasoning, Cognitive Architectures
Authors
- Libor Bešenyi (ORCID: https://orcid.org/0009-0007-2508-5863)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-15
- DOI
- https://doi.org/10.5281/zenodo.24236194
- Primary Topic
- Advanced Memory and Neural Computing
- Type
- article
- Field-Weighted Citation Impact
- 0.00