LLMs model how humans induce logically structured rules

2025-07-04

AI Summary

This work addresses a core question in cognitive science—how humans represent logical concepts and perform inductive inference. It engages the longstanding debate over whether artificial neural networks can model abstract cognition. Method: We propose large language models (LLMs) as a novel computational framework for explaining logical concept acquisition, positing representational and inferential mechanisms distinct from classical Bayesian or symbolic models. Four rule-based logical concept induction experiments were conducted to systematically evaluate how well diverse LLMs fit human behavioral data. Results: LLMs achieve state-of-the-art model fit to human responses while exhibiting distinctive, interpretable prediction patterns. These findings offer a new computational account of the origin, structure, and development of logical concepts—bridging neural-scale computation with high-level symbolic reasoning without relying on explicit symbolic manipulation or probabilistic inference over structured hypothesis spaces.

Technology Category

• Cognitive Modeling & Cognitive Systems: Conceptual Inference and Reasoning
• Knowledge Representation and Reasoning: Computational Complexity of Reasoning
• Machine Learning: Statistical Relational/Logic Learning

Application Category

• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

A central goal of cognitive science is to provide a computationally explicit account of both the structure of the mind and its development: what are the primitive representational building blocks of cognition, what are the rules via which those primitives combine, and where do these primitives and rules come from in the first place? A long-standing debate concerns the adequacy of artificial neural networks as computational models that can answer these questions, in particular in domains related to abstract cognitive function, such as language and logic. This paper argues that recent advances in neural networks -- specifically, the advent of large language models (LLMs) -- represent an important shift in this debate. We test a variety of LLMs on an existing experimental paradigm used for studying the induction of rules formulated over logical concepts. Across four experiments, we find converging empirical evidence that LLMs provide at least as good a fit to human behavior as models that implement a Bayesian probablistic language of thought (pLoT), which have been the best computational models of human behavior on the same task. Moreover, we show that the LLMs make qualitatively different predictions about the nature of the rules that are inferred and deployed in order to complete the task, indicating that the LLM is unlikely to be a mere implementation of the pLoT solution. Based on these results, we argue that LLMs may instantiate a novel theoretical account of the primitive representations and computations necessary to explain human logical concepts, with which future work in cognitive science should engage.

Problem

• How LLMs model human induction of logically structured rules
• Comparing LLMs and Bayesian models in explaining human cognition
• Assessing LLMs as novel theoretical accounts of human logic

Innovation

• LLMs model human rule induction
• LLMs outperform Bayesian pLoT models
• LLMs suggest novel cognitive representations