Hypothesizing Missing Causal Variables with LLMs

2024-09-04

arXiv.org

AI Summary

This study addresses the challenge of causal inference in observational studies—where confounding and mediation bias impede valid causal conclusions and randomized experiments are infeasible. We formally introduce the “Missing Causal Variable Identification” task: given an incomplete causal graph, identify the causal role (cause, effect, or mediator) of missing variables. Methodologically, we propose a structured causal graph prompting template for large language models (LLMs), integrating domain knowledge injection, and construct the first controllable-difficulty benchmark for causal role reasoning. Experiments reveal that LLMs significantly outperform baselines in mediator identification compared to cause/effect identification, with several open-source models surpassing GPT-4. Contributions include: (1) a formal definition of the new task; (2) empirical characterization of LLMs’ capabilities and limitations in causal role reasoning; and (3) open-sourcing of the benchmark and evaluation protocol to advance trustworthy causal AI.

Technology Category

• Machine Learning: Causal Learning
• Reasoning under Uncertainty: Causality
• Knowledge Representation and Reasoning: Action, Change, and Causality

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

Scientific discovery is a catalyst for human intellectual advances, driven by the cycle of hypothesis generation, experimental design, data evaluation, and iterative assumption refinement. This process, while crucial, is expensive and heavily dependent on the domain knowledge of scientists to generate hypotheses and navigate the scientific cycle. Central to this is causality, the ability to establish the relationship between the cause and the effect. Motivated by the scientific discovery process, in this work, we formulate a novel task where the input is a partial causal graph with missing variables, and the output is a hypothesis about the missing variables to complete the partial graph. We design a benchmark with varying difficulty levels and knowledge assumptions about the causal graph. With the growing interest in using Large Language Models (LLMs) to assist in scientific discovery, we benchmark open-source and closed models on our testbed. We show the strong ability of LLMs to hypothesize the mediation variables between a cause and its effect. In contrast, they underperform in hypothesizing the cause and effect variables themselves. We also observe surprising results where some of the open-source models outperform the closed GPT-4 model.

Problem

• Identifying backdoor variables in causal graphs
• Completing partial causal graphs using contextual reasoning
• Testing LLMs' hypothesis generation for causal inference

Innovation

• Context-aware reasoning on parametric knowledge
• Completing partial causal graphs as benchmark
• Identifying backdoor variables through graph context