Brain Simulator

Brain Simulator is a developmental cognitive agent prototype. It is designed to accumulate experiences over time and update internal state through:

• neurochemical dynamics
• autobiographical memory
• identity/development updates
• attention-based thought selection
• reflection and narrative updates

What Is Implemented

• Core chemistry model with four chemicals: dopamine, cortisol, oxytocin, serotonin
• Chemical interactions, decay, noise, and clamping
• Dynamic identity traits: competence, social_value, resilience, intelligence
• Development tracking: experience points, reflection depth, maturity
• Global Workspace attention model (Thought + GlobalWorkspace)
• Consciousness score based on focus stability + development/reflection
• Autobiographical memory with event recording and memory-thought proposals
• Narrative engine that updates a short self-narrative from recent events
• Self-reflection with regret and wisdom tracking
• Synthetic environment that generates one life event per simulation cycle
• Perception pipelines:
  • brain.perceive(event) for structured life events
  • brain.observe_perception(modality, content, source) for text modality input
  • brain.receive_visual_signal(signal) for structured visual signals
  • brain.receive_hearing_signal(signal) for structured hearing signals

Current CLI

main.py supports:

• --mode simulate (default)
• --mode live
• --cycles <int>
• --deterministic

Examples:

python main.py --mode simulate --cycles 100
python main.py --mode live
python main.py --mode simulate --cycles 500 --deterministic

How Simulation Runs

In simulate mode:

1. Simulator generates one synthetic developmental event each cycle.
2. The event is sent to brain.perceive(...).
3. Optional scenario events (from simulation/scenarios.py) are injected.
4. Brain runs tick() and outputs updated state (and optional decision if engine is attached).

Programmatic Perception APIs

You can feed external sensors later without changing the core loop.

Event-style input

brain.perceive({
    "content": "You are praised for trying.",
    "category": "praise",
    "valence": 0.8,
    "intensity": 0.7,
    "source": "simulated",
    "timestamp": 0.0,
})

Structured vision input

brain.receive_visual_signal({
    "objects": ["person", "bottle"],
    "attributes": {"person": ["red"]},
    "relations": [{"from": "person", "rel": "near", "to": "bottle"}],
    "motion_level": 0.4,
    "confidence": 0.9,
    "source": "camera_pipeline",
    "timestamp": 0.0,
})

Structured hearing input

brain.receive_hearing_signal({
    "transcript": "Good job, keep trying",
    "speaker_type": "caregiver",
    "sentiment": 0.7,
    "prosody_intensity": 0.6,
    "keywords": ["praise", "support"],
    "source": "audio_pipeline",
    "timestamp": 0.0,
})

Output State Includes

brain.get_state() returns, among others:

• chemical values
• identity snapshot (identity_*)
• development snapshot (development_*)
• recent perceptions
• learned concepts
• wisdom
• self narrative
• consciousness score

Project Notes

• Decision engine exists (decision/decision_engine.py) and is attention-driven, but main.py currently initializes VirtualBrain without attaching a decision engine.
• This repository is a research/development prototype, not a production chatbot stack.