Modeling, Measuring, and Mapping Individual Differences in Mental Object Representations

Human behavior shows consistent individual variation. We like different cuisines, watch different movies, and choose different careers. Personality psychologists ask people to describe their own or others' personalities, while cognitive psychometricians use behavioral tasks to measure individual differences in cognition. Our approach complements these methods by examining whether these differences extend to the most fundamental mental representations of everyday objects. To study this question, we use an odd-one-out task in which 779 U.S.-based participants are shown images of naturalistic objects. We observe that accounting for individual differences even in this basic task leads to better models of human choices. Furthermore, we demonstrate that these differences can be reliably measured and related to age and gender in 199 English-speaking participants. These results enhance our understanding of individual differences in mental representations and pave the way for personalized models of the world in social science and AI using realistic datasets.