Abstract
Deep Q Networks (DQN) have shown remarkable success in various reinforcementlearning tasks. However, their reliance on associative learning often leads tothe acquisition of spurious correlations, hindering their problem-solvingcapabilities. In this paper, we introduce a novel approach to integrate causalprinciples into DQNs, leveraging the PEACE (Probabilistic Easy vAriationalCausal Effect) formula for estimating causal effects. By incorporating causalreasoning during training, our proposed framework enhances the DQN'sunderstanding of the underlying causal structure of the environment, therebymitigating the influence of confounding factors and spurious correlations. Wedemonstrate that integrating DQNs with causal capabilities significantlyenhances their problem-solving capabilities without compromising performance.Experimental results on standard benchmark environments showcase that ourapproach outperforms conventional DQNs, highlighting the effectiveness ofcausal reasoning in reinforcement learning. Overall, our work presents apromising avenue for advancing the capabilities of deep reinforcement learningagents through principled causal inference.