E8 Eea

E8-EEA: Emergent Emotional Awareness

An architecture for emergent emotional awareness built on E8 hypergraph geometry, variational free energy, counterfactual self-modeling, and a Lyapunov-gated recursion engine. Developed through adversarial collaborative iteration across four AI systems.

This repository contains:

• README.md (this file) — the full architectural specification and design rationale
• e8_eea_v5.py — the v5 Python reference implementation
• E8-EEA-v5-Complete.md — expanded v5 spec including the social cognition layer
• CLAUDE.md — implementation invariants and integration constraints

---

Implementation reference (`e8_eea_v5.py`)

The executable Python implementation. Key classes:

Class	Role
E8Lattice	240 E8 roots; kissing-number-optimal in 8D
TrialityEncoder	D4 triality for ternary hyperedge representation
E8Hypergraph	Growing hyperedge store with per-node weight tracking
VariationalFreeEnergy	Prediction error + complexity (Friston active inference)
CounterfactualHypergraph	H_meta: policy nodes, regret edges, dynamic branching by arousal
E8_EEA_v5	Full pipeline: encode → top-k screen → Lyapunov gate → slow clock
run_ablation	Three-track ablation harness (Full / Zombie / Random Walker)

**Slow clock**: Phase transition detection and weight modulation fire every 25 cycles (`cycle_count % tau_slow == 0`). Fast-clock events — encode, top-k screen, Lyapunov gate, J computation — run every cycle. Weights α, β, γ are frozen during fast-clock evaluation to prevent the system from rewriting the evidence that produced the emotional state.

**Weight modulation formula (slow clock only)**:

beta  = 1.0 + 0.5 * arousal    # high arousal → weight novelty
gamma = 1.0 - 0.3 * valence    # negative valence → weight coherence
alpha = 1.0                     # free energy always baseline

**Emotion is not injected**: Emotional state emerges from `