Causal Strengths and Leaky Beliefs: Interpreting LLM Reasoning via Noisy-OR Causal Bayes Nets

TL;DR

This paper investigates how large language models and humans reason causally by formalizing their judgments with noisy-OR causal Bayes nets, revealing differences and similarities in their reasoning signatures across multiple tasks.

Contribution

It introduces a novel framework using noisy-OR causal Bayes nets to compare causal reasoning in LLMs and humans across diverse tasks.

Findings

LLMs and humans show both aligned and distinct causal reasoning patterns.

The noisy-OR model effectively captures reasoning signatures of both LLMs and humans.

Model selection via AIC reveals differences in causal strength parameters between LLMs and humans.

Abstract

The nature of intelligence in both humans and machines is a longstanding question. While there is no universally accepted definition, the ability to reason causally is often regarded as a pivotal aspect of intelligence (Lake et al., 2017). Evaluating causal reasoning in LLMs and humans on the same tasks provides hence a more comprehensive understanding of their respective strengths and weaknesses. Our study asks: (Q1) Are LLMs aligned with humans given the \emph{same} reasoning tasks? (Q2) Do LLMs and humans reason consistently at the task level? (Q3) Do they have distinct reasoning signatures? We answer these by evaluating 20+ LLMs on eleven semantically meaningful causal tasks formalized by a collider graph ($C_1\!\to\!E\!\leftarrow\!C_2$ ) under \emph{Direct} (one-shot number as response = probability judgment of query node being one and \emph{Chain of Thought} (CoT; think first, then provide answer). Judgments are modeled with a leaky noisy-OR causal Bayes net (CBN) whose parameters $\theta=(b,m_1,m_2,p(C)) \in [0,1]$ include a shared prior $p(C)$; we select the winning model via AIC between a 3-parameter symmetric causal strength ($m_1{=}m_2$) and 4-parameter asymmetric ($m_1{\neq}m_2$) variant.