Probing Machine Consciousness: Core Development Methods

How can machines develop core consciousness? This paper by Mathis Immertreu and colleagues investigates Antonio Damasio’s framework for consciousness to explore rudimentary self- and world-models in AI agents.

Probing for Consciousness in Machines, authored by Mathis Immertreu, Achim Schilling, Andreas Maier, and Patrick Krauss, demonstrates how reinforcement learning agents can form preliminary models of self and environment during task execution.

Key Highlights

• Damasio’s Framework: Applies the integration of self and world models as foundational to core consciousness.
• Reinforcement Learning: Trains agents in virtual environments to develop rudimentary models as a byproduct of task completion.
• Evaluation Probes: Uses classifiers to analyze neural activations for evidence of self- and world-model representations.

Connection to ACM

The Artificial Consciousness Module (ACM) aligns with this study by:

• Model Integration: Incorporating self- and world-models inspired by Damasio’s theory.
• Learning Frameworks: Leveraging reinforcement learning to simulate consciousness development.

For a detailed exploration of the methodologies and implications, access the full paper here.