Function over form: The temporal evolution of affordance-based scene categorization.

Abstract

Humans can rapidly understand and categorize scenes, yet the specific features and mechanisms that enable categorization remain debated. Here, we investigated whether affordances-the possible actions a scene supports-facilitate scene categorization even when other similarly informative features are present. In Experiment 1, we generated triplets of images that were equally dissimilar on one feature dimension (affordances, materials, surfaces) but similar on the remaining two. Using an odd-one-out task, observers consistently chose the image that differed in its affordances as the outlier despite equally large differences in the other dimensions. In Experiment 2, we asked whether shared affordances also interfere with rapid categorization. When distractors shared affordances rather than surface features with a target category, observers committed significantly more false alarms, indicating that functional similarity creates stronger competition during scene categorization. Finally, in Experiment 3, we recorded ERPs to examine the time course of category representations. We used multivariate decoding to assess the quality of scene category representations. We found that both affordance- and surface-similar distractors yielded above-chance decoding starting around 60-70 ms after stimulus. However, the neural discriminability of target categories was reduced in the affordance-similar condition, starting around 150 ms. These findings suggest that affordances carry a privileged status in scene perception, shaping both behavioral category performance and neural processing.