Reasoning as Pattern Matching: Shared Mechanisms in Human and LLM Everyday Reasoning
Challenges the assumption that human reasoning relies on abstract world models, with implications for cognitive science and AI alignment.
The paper shows that both humans and LLMs exhibit similar errors in everyday causal reasoning, and that LLM attention heads implement pattern-matching that can predict human errors, suggesting reasoning in both is pattern-matching rather than abstract modeling.
When large language models (LLMs) fail to generalize or make haphazard errors in reasoning, it is often taken as evidence that LLMs are not truly reasoning, but rather performing a kind of pattern matching. The implication is that people's behavior does not exhibit the same types of failures because human reasoning uses principled and abstract world models. We evaluate human participants and 25 LLMs on their ability to engage in common-sense reasoning about a variety of everyday situations and observe similar patterns of errors in both people and models. We then identify the set of attention heads driving LLM responses and find that these heads implement a form of pattern-matching. These attention heads allow us to predict seemingly inexplicable reasoning errors in people caused by ostensibly irrelevant prompt details. Taken together, our results suggest that everyday causal reasoning in people and LLMs is more consistent with a form of pattern-matching than with abstract world models.