Less is more: Aesthetic liking is inversely related to metabolic expense by the visual system

Energy efficiency drives evolution, and humans may have evolved pleasure-based signals to optimize actions. Could this extend to aesthetic pleasure? We tested this using both an in-silico model (VGG19) and human observers, finding strong evidence in both. First, we measured the metabolic cost incurred by VGG19—pretrained for object categorization or randomly initialized—processing 4,914 images, revealing an inverse relationship between aesthetic pleasure and metabolic cost, and only in the pretrained model. Next, we compared aesthetic ratings with metabolic activity in the human visual system during functional magnetic resonance imaging (fMRI). In both early (V1, V2, V4) and higher-level (FFA, OPA, PPA) regions, we observed the same inverse relationship between BOLD signals and aesthetic pleasure. These findings suggest aesthetic pleasure may at least partially arise from an affective heuristic favoring low-energy states, offering a unified framework linking visual discomfort with processing fluency, image complexity, and prototypicality.