How Humans and LLMs Organize Conceptual Knowledge: Exploring Subordinate Categories in Italian

TL;DR

This study compares human and AI-generated exemplars of subordinate categories in Italian, revealing low alignment but domain-dependent performance, and discusses implications for psychological and linguistic research.

Contribution

First to analyze subordinate category exemplars in Italian, evaluating LLMs' ability to produce human-like category organization across multiple tasks.

Findings

Low alignment between humans and LLMs in exemplar generation

Performance varies significantly across semantic domains

Highlights potential and limitations of AI in psychological research

Abstract

People can categorize the same entity at multiple taxonomic levels, such as basic (bear), superordinate (animal), and subordinate (grizzly bear). While prior research has focused on basic-level categories, this study is the first attempt to examine the organization of categories by analyzing exemplars produced at the subordinate level. We present a new Italian psycholinguistic dataset of human-generated exemplars for 187 concrete words. We then use these data to evaluate whether textual and vision LLMs produce meaningful exemplars that align with human category organization across three key tasks: exemplar generation, category induction, and typicality judgment. Our findings show a low alignment between humans and LLMs, consistent with previous studies. However, their performance varies notably across different semantic domains. Ultimately, this study highlights both the promises and the constraints of using AI-generated exemplars to support psychological and linguistic research.