Abstract
TECHNICAL DISCLOSURE: ZenBrain — A Neuroscience-Inspired 7-Layer Memory Architecture for Autonomous AI Systems (v5)
This disclosure describes ZenBrain, a memory architecture for AI agents that integrates fifteen neuroscience models into seven distinct memory layers (working, short-term, episodic, semantic, procedural, core, cross-context).
Version 5 adds the Predictive Memory Architecture (PMA): six new biologically motivated components for neuromodulated memory lifecycle management. These include a four-channel NeuromodulatorEngine (dopamine/norepinephrine/serotonin/acetylcholine dynamics), a prediction-error-gated ReconsolidationEngine, TripleCopyMemory with divergent decay dynamics (fast/medium/deep copies with logarithmic deep-copy consolidation), a four-dimensional PriorityMap with amygdala fast-path, a StabilityProtector (NogoA/HDAC3 analog for memory protection), and a MetacognitiveMonitor for bias detection.
New experimental results include a PMA benchmark suite (6 algorithms), a full 15-algorithm ablation study, and a competitive comparison against static RAG and simple memory baselines. All experiments are fully reproducible with seeded PRNG and included as automated tests.
The architecture is validated by independent concurrent work: Anthropic's Claude Code Auto Dream (sleep consolidation in production), Tiwari & Fofadiya's multi-layer validation (arXiv:2603.29194), and the founding of the ICLR 2026 MemAgents Workshop.
Prior versions: v1-v4 (Zenodo DOI: 10.5281/zenodo.19353663). Open-source: github.com/zensation-ai/zenbrain
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Bering, Alexander, "ZenBrain: A Neuroscience-Inspired 7-Layer Memory Architecture for Autonomous AI Systems (v5)", Technical Disclosure Commons, ()
https://www.tdcommons.org/dpubs_series/9732