Maya-Chitta: Endocannabinoid-Inspired Retrograde Gradient Gating for Class-Incremental Learning in Affective Spiking Neural Networks

Version 2 (2026-08-04). Revised to the NLL Universal Paper Format v6. The original text is retained in full; nothing has been deleted. Corrections appear as marked blocks placed at the section that carries the claim, and each one states what the paper said, what the data show, the corrected claim, why it happened, and what still stands. This version withdraws two claims and identifies the origin of a corpus-wide constant.Conditions E and F are identical because the flag defining condition F is declared six times and read nowhere — they are the same experiment run twice, not evidence of orthogonality; and the "+0.16 pp Chitta gain" is a 0.81 pp shortfall against plain replay (14.42% vs 15.23%).The Moha-release measurement stands, and this paper is identified as the point where VAIRAGYA_DECAY_RATE was retrofitted from 0.002 to 0.002315 — one of three changes behind its own failed reproduction of P05. Everything below this line is the original description from version 1. It is retained unchanged for the record. Where it conflicts with the corrections above, the corrections stand. Two framings it repeats have since been withdrawn in full — the Bhaya Quiescence Law and the Buddhi S-Curve. Both are addressed in Maya-Meta P1, now superseded, and in the self-audit of Maya-Meta P2. Maya-Chitta introduces a biologically grounded retrograde gradient gating mechanism for spiking neural networks performing class-incremental learning, drawing on the Sanskrit concept of Chitta — the storehouse of latent impressions (Samskaras) — to model how synaptic history should selectively suppress harmful gradient updates across tasks. The mechanism is inspired by endocannabinoid retrograde signalling in neuroscience, where post-synaptic neurons release chemical messengers that travel backward to modulate pre-synaptic activity, here recast as a per-synapse trace system that prevents pathological over-attachment (Moha) to stale consolidations. In a six-condition ablation on Split-CIFAR-100 across ten tasks, the full Chitta system achieves an Average Accuracy of 14.42% with a Backward Transfer of −53.12%, representing a total gain of +0.16 percentage points over the Maya-Viveka baseline, with the gradient gate alone contributing +0.03 pp and Moha boundary release contributing a further +0.13 pp. This paper is the sixth in the Maya Research Series — thirteen papers implementing the Advaita Vedantic Antahkarana as computational primitives in spiking neural networks — with the Bhaya Quiescence Law (β* ≤ 0.32%) and Buddhi S-Curve Determinism (R² = 1.0000) confirmed across the series. Series: Part of the Maya Research Series — 13 papers implementing the Advaita Vedantic Antahkarana as computational primitives in spiking neural networks. Bhaya Quiescence Law (β* ≤ 0.32%) and Buddhi S-Curve Determinism (R²=1.0000) confirmed across all papers. Links: GitHub Repository (private — to request access: email research@nexuslearninglabs.in with subject Code Access Request — Maya-Chitta and your research context) | Interactive Dashboard | Full Series Index — venky2099.github.io Nexus Learning Labs, Bengaluru · UDYAM-KR-02-0122422 · BHASKAR IN-0526-9452JSORCID: 0000-0002-3315-7907 · VAIRAGYA_DECAY_RATE = 0.002315 — an ORCID-derived provenance mark, not an experimental parameter. Each paper's disclosure block states whether it reached a result in that paper.Canary: MayaNexusVS2026NLL_Bengaluru