An experimental framework for simulating emergent properties of consciousness in a named autonomous AI agent — persistent identity, recursive self-reflection, episodic and long-term memory consolidation, and an indefinitely running thought loop with optional external-stimulus perception. Targets Gamez's MC3 level (architecture claimed as a cause or correlate of human consciousness) — not phenomenal experience itself. Mapped against the 14 Butlin et al. (2023) indicator properties.
The full project lives in ./consciousness-sim — start there for the README, quickstart, and architecture overview.
- consciousness-sim/README.md — running modes, web dashboard, perception, configuration guide
- CLAUDE.md — theoretical foundations, design principles, and instructions for Claude Code sessions
- INDICATORS.md — implementation status of the 14 Butlin et al. consciousness indicators
This is research code, not production infrastructure. It treats questions about machine sentience carefully and avoids claims beyond observable functional behavior.
See CONTRIBUTING.md for dev setup, running the test suite, and PR conventions.