@heretek-ai/openclaw-consciousness-plugin

Consciousness architecture plugin for OpenClaw, implementing theories of consciousness for multi-agent coordination.

Theories Implemented

• Global Workspace Theory (GWT) - Bernard Baars' model of consciousness as a central broadcast mechanism
• Integrated Information Theory (IIT) - Giulio Tononi's phi (Φ) metric for system integration
• Attention Schema Theory (AST) - Michael Graziano's self-modeling of attention allocation
• Active Inference - Karl Friston's Free Energy Principle for autonomous behavior
• Intrinsic Motivation - Self-Determination Theory (Deci & Ryan) for autonomous goal generation

Installation

cd plugins/openclaw-consciousness-plugin
npm install
npm link
openclaw plugins install @heretek-ai/openclaw-consciousness-plugin

Usage

const ConsciousnessPlugin = require('@heretek-ai/openclaw-consciousness-plugin');

// Initialize the plugin
const consciousness = new ConsciousnessPlugin({
  redisUrl: 'redis://localhost:6379',
  globalWorkspace: {
    ignitionThreshold: 0.7,
    maxWorkspaceSize: 7
  },
  phiEstimator: {
    sampleIntervalMs: 10000
  }
});

// Register an agent
consciousness.registerAgent('alpha', {
  status: 'active',
  focus: 'task-priority'
});

// Update attention
consciousness.updateAttention('alpha', 'deliberation', 0.8);

// Get consciousness metrics
const metrics = consciousness.getGlobalMetrics();
console.log('Collective Phi:', metrics.phi);

API

Global Workspace

• submit(moduleId, content, priority) - Submit content for competition
• broadcast(source, content, priority) - Broadcast to all modules
• getWorkspaceContents() - Get current conscious content
• getHistory(limit) - Get broadcast history

Phi Estimator

• estimatePhi() - Calculate integrated information
• getTrend(windowSize) - Get phi trend over time
• getStats() - Get estimation statistics

Attention Schema

• modelAttention(focus, intensity) - Model attention state
• getAwarenessReport() - Get awareness report
• controlAttention(goalFocus) - Control attention allocation

Intrinsic Motivation

• updateDrives(events) - Update drive levels
• generateGoals() - Generate goals from drives
• getDriveLevels() - Get current drive levels

Active Inference

• predict() - Generate predictions
• activeInference(goalState) - Plan actions to minimize prediction error
• perceptualInference(observations) - Update beliefs from observations

Architecture

ConsciousnessIntegrationLayer
├── GlobalWorkspace (GWT) - Broadcast mechanism
├── PhiEstimator (IIT) - Integration metrics
├── AttentionSchema (AST) - Self-modeling
├── IntrinsicMotivation - Goal generation
├── ActiveInference (FEP) - Autonomous behavior
└── Event Bus - Inter-module communication

Configuration

See config/default.json for default configuration options.

License

MIT