An AI-generated art evaluation model that integrates computational aesthetics and cognitive psychology

This study addresses the critical gap between computational aesthetics and cognitive psychology by developing an evaluation model that explicitly operationalizes cognitive theories within AI frameworks for aesthetic assessment. The research employed a mixed-method approach integrating behavioral experiments, eye-tracking technology, and NASA-TLX cognitive load measurements to validate the correspondence between computational predictions and human cognitive processes across different levels of artistic expertise and visual complexity. The model architecture implements Reber’s processing fluency theory, Gestalt organizational principles, and dual-pathway processing mechanisms, demonstrating successful alignment between human aesthetic judgments and AI-generated assessments. The findings reveal that fundamental perceptual organization principles can be computationally instantiated while preserving cognitive validity, with the dual-pathway architecture showing dynamic contribution patterns that challenge traditional dichotomous models of aesthetic judgment. The cognitive interpretability analysis demonstrates that model variance substantially maps to established psychological constructs, while error patterns align with theoretically predicted sources including cultural, personal, and expertise-based variations. These results establish that AI-based aesthetic evaluation can achieve cognitive transparency through explicit mapping to psychological constructs, providing both theoretical validation of cognitive mechanisms and practical frameworks for developing human-aligned aesthetic assessment systems. The synthesis advances both cognitive science and computational aesthetics by demonstrating how AI models can serve as instruments for understanding human aesthetic cognition while cognitive theories provide essential constraints for developing interpretable artificial systems.