CRAwDAD: Causal Reasoning Augmentation with Dual-Agent Debate

TL;DR

This paper introduces CRAwDAD, a dual-agent debate framework for causal reasoning in language models, significantly improving accuracy in causal inference tasks by enabling models to critique and refine each other's hypotheses.

Contribution

It proposes a novel multi-agent debate approach that explicitly models internal causal reasoning as an adversarial dialogue, enhancing the performance of reasoning language models.

Findings

Debate improves DeepSeek-R1 accuracy from 78.03% to 87.45%.

Debate enhances Qwen3 accuracy from 84.16% to 89.41%.

Counterfactual reasoning accuracy notably increases with debate.

Abstract

When people reason about cause and effect, they often consider many competing "what if" scenarios before deciding which explanation fits best. Analogously, advanced language models capable of causal inference can consider multiple interventions and counterfactuals to judge the validity of causal claims. Crucially, this type of reasoning is less like a single calculation and more like an internal dialogue between alternative hypotheses. In this paper, we make this dialogue explicit through a dual-agent debate framework where one model provides a structured causal inference, and the other critically examines this reasoning for logical flaws. When disagreements arise, the agents attempt to persuade each other, challenging each other's logic and revising their conclusions until they converge on a mutually agreed answer. To take advantage of this deliberative process, we specifically use reasoning language models, whose strengths in both causal inference and adversarial debate remain under-explored relative to standard large language models. We evaluate our approach on the CLadder dataset, a benchmark linking natural language questions to formally defined causal graphs across all three rungs of Pearl's ladder of causation. With Qwen3 and DeepSeek-R1 as debater agents, we demonstrate that multi-agent debate improves DeepSeek-R1's overall accuracy in causal inference from 78.03% to 87.45%, with the counterfactual category specifically improving from 67.94% to 80.04% accuracy. Similarly, Qwen3's overall accuracy improves from 84.16% to 89.41%, and counterfactual questions from 71.53% to 80.35%, showing that even strong models can still benefit greatly from debate with weaker agents. Our results highlight the potential of reasoning models as building blocks for multi-agent systems in causal inference, and demonstrate the importance of diverse perspectives in causal problem-solving.