Predicting Human Similarity Judgments Using Large Language Models

TL;DR

This paper introduces a new method using large language models to efficiently predict human similarity judgments from text descriptions, reducing data needs and outperforming visual-based models on naturalistic image datasets.

Contribution

The authors propose a domain-general, text-based approach leveraging language models to predict similarity judgments, significantly reducing data requirements compared to traditional methods.

Findings

Outperforms previous visual-based similarity models

Requires linearly fewer descriptions relative to stimuli

Effective across multiple naturalistic image datasets

Abstract

Similarity judgments provide a well-established method for accessing mental representations, with applications in psychology, neuroscience and machine learning. However, collecting similarity judgments can be prohibitively expensive for naturalistic datasets as the number of comparisons grows quadratically in the number of stimuli. One way to tackle this problem is to construct approximation procedures that rely on more accessible proxies for predicting similarity. Here we leverage recent advances in language models and online recruitment, proposing an efficient domain-general procedure for predicting human similarity judgments based on text descriptions. Intuitively, similar stimuli are likely to evoke similar descriptions, allowing us to use description similarity to predict pairwise similarity judgments. Crucially, the number of descriptions required grows only linearly with the number of stimuli, drastically reducing the amount of data required. We test this procedure on six datasets of naturalistic images and show that our models outperform previous approaches based on visual information.