Item

Abstract

Semantic features have been playing a central role in investi- gating the nature of our conceptual representations. Yet the enormous time and effort required to empirically sample and norm features from human raters has restricted their use to a limited set of manually curated concepts. Given recent promis- ing developments with transformer-based language models, here we asked whether it was possible to use such models to automatically generate meaningful lists of properties for ar- bitrary object concepts and whether these models would pro- duce features similar to those found in humans. To this end, we probed a GPT-3 model to generate semantic features for 1,854 objects and compared automatically-generated features to existing human feature norms. GPT-3 generated many more features than humans, yet showed a similar distribution in the types of generated features. Generated feature norms rivaled human norms in predicting similarity, relatedness, and cate- gory membership, while variance partitioning demonstrated that these predictions were driven by similar variance in hu- mans and GPT-3. Together, these results highlight the poten- tial of large language models to capture important facets of human knowledge and yield a new approach for automatically generating interpretable feature sets, thus drastically expand- ing the potential use of semantic features in psychological and linguistic studies.