Dissertations, Theses, and Capstone Projects

Date of Degree

2-2025

Document Type

Thesis

Degree Name

M.S.

Program

Cognitive Neuroscience

Advisor

Edward Vessel

Subject Categories

Cognitive Neuroscience | Cognitive Psychology | Cognitive Science

Keywords

categorization, neuroaesthetics, visual perception

Abstract

How are categories of art internally represented, and how can we capture those representations systematically? Investigating category learning in the context of complex, multidimensional stimuli like visual artworks presents unique challenges due to their inherent variability and perceptual complexity. Artworks engage multiple cognitive processes and are shaped by both sensory-driven mechanisms and top-down influences like prior knowledge, making them uniquely suited for studying how structured categorical knowledge interacts with perceptual and aesthetic judgments. The present study addresses these questions by utilizing machine-learning tools to construct and implement a two-dimensional similarity space of 400 existing artworks painted by two artists from the Impressionist/Post-Impressionist art movements. Using a four-quadrant structure with equal number of stimuli (two quadrants per artist) as the basis for the category-learning paradigm, participants learned to categorize artworks with respect to the artist while only being exposed to one half of the similarity space. The structure of the space allowed us to measure participant responses to images that varied in both exposure and perceptual similarity to the trained images. The study examined how these variations influenced categorization performance as well as the relationship of categorization to metrics of confidence, familiarity, uniqueness, and aesthetic appeal. Results indicate decreased categorization accuracy for images closer to the category boundary, and also for those in unexposed areas of the space. The relationship of categorization performance to familiarity and confidence revealed a nuanced relationship whereas a significant correlation was found between categorization accuracy and aesthetic ratings across the test images, suggesting a positive effect of prior knowledge on aesthetic appeal. These findings highlight the role of structured knowledge in interpreting complex visual stimuli and demonstrate the utility of machine-learning tools in advancing our understanding of art perception and cognition.

Download

Share

COinS