hide
Free keywords:
-
Abstract:
To study how mental object representations are related to behavior, we estimatedsparse, non-negative representations of objects using human behavioral judgmentson images representative of 1,854 object categories. These representations pre-dicted a latent similarity structure between objects, which captured most of theexplainable variance in human behavioral judgments. Individual dimensions inthe low-dimensional embedding were found to be highly reproducible and in-terpretable as conveying degrees of taxonomic membership, functionality, andperceptual attributes. We further demonstrated the predictive power of the em-beddings for explaining other forms of human behavior, including categorization,typicality judgments, and feature ratings, suggesting that the dimensions reflecthuman conceptual representations of objects beyond the specific task.
