Revealing emergent human-like conceptual representations from language prediction

TL;DR

This paper demonstrates that large language models develop structured, human-like conceptual representations through language prediction, which align with human behavior and neural activity, despite lacking real-world grounding.

Contribution

It reveals that LLMs form shared, context-independent conceptual structures from language alone, advancing understanding of their internal representations and their relation to human cognition.

Findings

LLMs can derive concepts from linguistic descriptions based on context cues.

Representations in LLMs converge toward a shared, context-independent structure.

These representations predict model performance and align with human neural and behavioral data.

Abstract

People acquire concepts through rich physical and social experiences and use them to understand and navigate the world. In contrast, large language models (LLMs), trained solely through next-token prediction on text, exhibit strikingly human-like behaviors. Are these models developing concepts akin to those of humans? If so, how are such concepts represented, organized, and related to behavior? Here, we address these questions by investigating the representations formed by LLMs during an in-context concept inference task. We found that LLMs can flexibly derive concepts from linguistic descriptions in relation to contextual cues about other concepts. The derived representations converge toward a shared, context-independent structure, and alignment with this structure reliably predicts model performance across various understanding and reasoning tasks. Moreover, the convergent representations effectively capture human behavioral judgments and closely align with neural activity patterns in the human brain, providing evidence for biological plausibility. Together, these findings establish that structured, human-like conceptual representations can emerge purely from language prediction without real-world grounding, highlighting the role of conceptual structure in understanding intelligent behavior. More broadly, our work suggests that LLMs offer a tangible window into the nature of human concepts and lays the groundwork for advancing alignment between artificial and human intelligence.