Qi Li (李琦), Chaoyuan Zhang, Qi Li (李琦)
Aesthetic models have expanded from accounts centered on formal order, prototypicality, complexity, and processing fluency to multilevel explanations involving predictive dynamics, affective valuation, self-related meaning, expertise, cultural context, and computational representation. This structured integrative review synthesizes foundational and recent evidence from empirical aesthetics, cognitive psychology, neuroaesthetics, and computational image assessment. The evidence shows that no single tradition supersedes the others: formal variables constrain perception; recurrent cognitive operations transform perception into appraisal; distributed neural systems support sensation, valuation, emotion, memory, and meaning; and context and individual differences determine which criteria are applied. Computational models have improved large-scale prediction, distributional scoring, and personalization, but predictive performance alone does not identify psychological mechanisms. To connect these levels, the review proposes a Situated Multilevel Aesthetic Model comprising object representation, predictive processing, affective valuation, self-meaning integration, and social-institutional context. The model treats aesthetic outcomes and their uncertainty as time-dependent functions of stimulus, viewer, and situation. Methodological priorities include construct-specific outcomes, factorial context manipulation, time-resolved measurement, multilevel and cross-cultural sampling, preservation of disagreement in datasets, and intervention-based validation of explanations. The resulting framework positions aesthetic modeling as an evidence-integrated enterprise in which psychological theory defines constructs, neuroscience constrains process, computation tests generalization, and humanistic inquiry identifies the institutional conditions of value.