PulseExploreJournal ClubDebatesTrendingResearchersJournals
Instagram
HomeExploreJournal ClubTrending
Synapse
⌘+K
Synapse
March 27, 2026Applied Sciences2 citationsOpen Access

Beyond the Black Box: An Interpretable Saliency Framework for Abstract Art via Theory-Driven Heuristics

View Full Paper
EVEvaldas VaičekauskasVAVytautas Abromavičius

Key Points

  • The central aim is to develop an interpretable saliency model for predicting attention in abstract art.
  • Constructed predictions using a combination of 35 heuristics based on perceptual psychology and art theory.
  • Optimized heuristic weights with a genetic algorithm and adjusted using a context-aware modulation mechanism.
  • Evaluated on eye-tracking data from 40 abstract paintings to assess predictive performance.
  • Achieved stable predictions comparable to the SalGAN baseline with a KL-divergence score of 1.11 ± 0.55.
  • Identified strong contributions from contrast, texture, and grouping heuristics.
  • Noted dynamic shifts in heuristic weights based on image-level features like edge density.

Abstract

Visual saliency modeling has achieved high predictive performance in natural image domains, yet its generalization to abstract art remains limited by the lack of explicit semantic structure and the scarcity of eye-tracking data. In such semantically ambiguous contexts, understanding the underlying drivers of attention is as critical as predictive accuracy. This paper presents an interpretable, ’white-box’ saliency framework tailored to abstract art, which constructs predictions through a weighted combination of 35 modular heuristics grounded in perceptual psychology and art theory, including contrast, grouping, isolation and symmetry. Heuristic weights are optimized via a genetic algorithm and refined by a context-aware modulation mechanism that adapts to image-level visual features. Evaluation against eye-tracking data from 40 abstract paintings demonstrates that the model with the expanded activation variant produces stable, meaningful predictions while achieving a competitive KL-divergence score (1.11 ± 0.55), which is comparable to the SalGAN baseline (1.11 ± 0.53). Analysis of the optimized weights reveals strong contributions from contrast, texture, and grouping mechanisms, while nearly half of the heuristics, including most horizontal symmetry heuristics are systematically pruned by the model. Moreover, context-aware modulation reveals that these weights are not static but shift dynamically based on image-level features such as edge density and intensity variation. By prioritizing transparency over raw predictive performance, this study demonstrates that explainable saliency models can function as robust investigative tools for decoding the principles of human visual perception in data-scarce domains.

Ask AI
Helpful
Bookmark
Share
View Full Paper

Cite This Study

Vaičekauskas et al. (2026) studied this question.

synapsesocial.com/papers/69c620ab15a0a509bde19320https://doi.org/10.3390/app16073145
Ask AI
Helpful
Bookmark
Share
View Full Paper