The role of visual complexity and image authorship in shaping aesthetic experience: A comparison of original and artificial intelligence-modified art
Abstract
In this project, we investigate how visual complexity and the authorship of artworks, whether created by a human or modified by artificial intelligence, affect viewers’ aesthetic experience. Additionally, we examine how these factors influence participants’ evaluations of how beautiful and how intriguing they find the artworks. 71 participants viewed and evaluated 16 artworks drawn from an existing database, including original paintings and images modified by Neural Style Transfer. Using repeated-measures multivariate analysis of variance, we found significant main effects of complexity and authorship on aesthetic experience, as well as a significant interaction between these factors. Separate analysis of variances showed that higher complexity increased both beauty and intrigue ratings, while the main effect of image authorship on these dimensions was not significant. For beauty, this effect was further qualified by an interaction with authorship: among highly complex images, artificial intelligence (AI)-modified artworks were rated as more beautiful than original ones. Overall, these findings underline the central role of visual complexity in shaping aesthetic experience, as well as ratings of beauty and intrigue. Participants also reported higher ratings for original artworks compared to AI-modified images on selected dimensions of aesthetic experience including ease of perceptual organization, understanding, knowledge about the artwork and insight into the artist's intention. Notably, in AI-modified images, higher complexity enhanced perceptual clarity, understanding as well as perceived beauty, suggesting that aesthetic appreciation depends not only on perceptual fluency but also on meaningful structure and interpretability. Together, these findings highlight complementary effects of visual complexity and artwork authorship on subjective aesthetic experience.
// Source
Authors: Daria Makurat, Martyna Olszewska, Joanna Dreszer
Institutions: Nicolaus Copernicus University, University of Gdańsk