"A Process-Informed Computational Framework for Affective Formation in Abstract Visual Art"

"All materials, data, and code used in this study. Understanding how affective meaning arises from visually structured but semantically sparse stimuli remains a challenge in affective computing. We propose a process-informed computational framework for affective formation in abstract visual art, decomposing the pathway from perceptual form to emotional response into interpretable intermediate stages. Using color field paintings as controlled low-semantic stimuli, we identify two partially dissociable routes to the experience of being moved: a structure-based route linking perceived visual abstractness to affect via bodily sensation, and an imagery-based route linking mental imagery to affect via the same bodily mediator. These routes are functionally distinct: the imagery-based route selectively predicts intersubjective agreement in experiences of being moved, whereas the structure-based route predicts more idiosyncratic experience. A double dissociation within the imagery construct further reveals that imagery vividness predicts the intensity of being moved, while imagery prevalence predicts the degree to which this response is shared across observers. Distinct classes of image statistics are selectively associated with specific stages of each route, with structure complexity features moderating the perception-to-body stage and orientation-related spectral features moderating the body-to-affect stage. To assess computational utility, we operationalize this architecture in a predictive framework and evaluate it using repeated cross-validation. Process-constrained models outperform dimension-matched PCA baselines and achieve performance comparable to a direct high-dimensional feature model in predicting being moved, while exhibiting reduced variability across train-test splits. Together, these findings offer a candidate principled architecture for understanding how affective meaning may arise from perceptual form alone, a capacity central to the experience of abstract art and related low-semantic aesthetic domains. "