Deinforcement Learning – Rethinking RL Assumptions

Summary

In our paper, we introduce the concept of Deinforcement Learning (DL), which challenges the fundamental assumptions behind reinforcement learning (RL), specifically proposing pain-based learning algorithms rooted in neuroscience and cognitive science.

Key Insights

• Deinforcement Learning questions reward-driven paradigms in AI.
• It proposes alternative learning frameworks inspired by cognitive neuroscience.
• Our study discusses implications for AI safety, ethics, and general intelligence.

Read the Full Paper

You can access the full paper in the Canadian Undergraduate Journal of Cognitive Science