Can Large Language Models Simulate Human Cognition Beyond Behavioral Imitation?

TL;DR

This paper introduces a novel benchmark based on longitudinal research trajectories to evaluate whether large language models can genuinely simulate human cognition beyond surface-level imitation.

Contribution

It presents a new cross-domain, temporal-shift benchmark and a multidimensional cognitive alignment metric to assess individual-level cognitive consistency in LLMs.

Findings

Current LLMs show limited ability to simulate authentic human cognition.

Enhancement techniques improve LLM performance but do not fully bridge the gap.

The benchmark reveals significant challenges in achieving genuine cognitive simulation.

Abstract

An essential problem in artificial intelligence is whether LLMs can simulate human cognition or merely imitate surface-level behaviors, while existing datasets suffer from either synthetic reasoning traces or population-level aggregation, failing to capture authentic individual cognitive patterns. We introduce a benchmark grounded in the longitudinal research trajectories of 217 researchers across diverse domains of artificial intelligence, where each author's scientific publications serve as an externalized representation of their cognitive processes. To distinguish whether LLMs transfer cognitive patterns or merely imitate behaviors, our benchmark deliberately employs a cross-domain, temporal-shift generalization setting. A multidimensional cognitive alignment metric is further proposed to assess individual-level cognitive consistency. Through systematic evaluation of state-of-the-art LLMs and various enhancement techniques, we provide a first-stage empirical study on the questions: (1) How well do current LLMs simulate human cognition? and (2) How far can existing techniques enhance these capabilities?