Almanac
paper

Reasoning as Pattern Matching: Shared Mechanisms in Human and LLM Everyday Reasoning

paperactiveprovisionalreasoning-as-pattern-matching-shared-mechanisms-in-human-and-llm-everyday-reasoning-161ffc6c·1 events·first seen 5d ago

Aliases: Reasoning as Pattern Matching: Shared Mechanisms in Human and LLM Everyday Reasoning

More like this (12)

Recent events (1)

6arXiv · cs.AI·5d ago·source ↗

Study finds shared pattern-matching mechanisms underlie both human and LLM everyday reasoning errors

A new arXiv paper evaluates human participants and 25 LLMs on commonsense causal reasoning tasks, finding similar error patterns in both groups. The authors identify specific attention heads driving LLM responses that implement pattern-matching, and show these heads can predict human reasoning errors caused by superficially irrelevant prompt details. The findings challenge the common assumption that human reasoning relies on principled abstract world models while LLMs merely pattern-match, suggesting both may share a more unified cognitive mechanism.