Exploring the Emotional Resonance of Color in Abstract Art Through the FeelingBlue Corpus

Understanding how colors evoke emotions is a rich area of study in art, psychology, and computational linguistics. The recent research embodied in the FeelingBlue corpus offers a pioneering perspective that situates color within its contextual framework, especially in abstract art, to unravel its emotional connotations. This multimodal dataset, grounded in both visual and linguistic data, advances our understanding of how color, combined with form, texture, and language, shapes emotional perception.

The FeelingBlue corpus comprises nearly 20,000 tuples of abstract artworks annotated by human evaluators for the emotions the images evoke, paired with textual rationales explaining these emotional responses. Unlike prior datasets relying on Likert scales, FeelingBlue employs a more consistent and replicable Best-Worst Scaling method, enhancing annotation reliability. The dataset’s focus on abstract art eliminates confounding factors like facial expressions or recognizable objects, isolating the influence of color and compositional elements such as line, stroke, and texture on emotional experience.

A core innovation presented with FeelingBlue is the task of Justified Affect Transformation: recoloring an image to intensify a target emotion while providing textual justification for the emotional shift. The researchers implemented an ensemble of deep neural networks that generate a new color palette conditioned on the original image, apply it, and then linguistically justify the transformation. The system’s outputs highlight patterns between altered color contexts and changes in emotional intensity, confirming that the subjective experience of color-emotion can be influenced and explained through computational means.

Importantly, the study emphasizes color in context — not just isolated color labels but how surrounding elements, like line and texture, modulate emotional perception. For example, blue hues in abstract paintings might evoke calmness or melancholy depending on their interaction with form and surrounding visual cues. This aligns with findings showing how colors and lines in abstract art systematically depict emotions such as sadness being associated with blue and particular line orientations (Anger is red, sadness is blue: Emotion depictions in abstract visual art, PMC).

Such insights bear practical value for artists, designers, and technologists. For instance:

• Artists seeking to evoke specific moods might use such research to better understand blue’s emotional versatility in abstract compositions. You can explore a curated collection of Blue Abstract Painting to see real-world artistic expressions of blue’s emotional impact.

• In digital design and marketing, knowing how color-context combinations influence emotion could inform user experience and branding strategies. For more on color psychology in design, resources like Pantone’s Color Institute and The Interaction Design Foundation explore how color choices affect human emotion and behavior.

• Computational models refined by FeelingBlue’s dataset could enhance content generation systems, such as AI-based image recoloring tools or emotionally aware generative art. Research on the perception of emotion in abstract artworks further supports the reliability of such AI-driven affective computing approaches (Perception of emotion in abstract artworks: A multidisciplinary review, ScienceDirect).

This body of work aligns with and extends prior research into color-emotion associations, moving beyond categorical links to explore nuanced, context-dependent emotional responses. The comprehensive multimodal approach has broad implications for affective computing, cognitive science, and creative industries.

Frequently Asked Questions (FAQ)

Q1: What makes FeelingBlue different from previous color-emotion studies?
FeelingBlue incorporates both visual context (line, texture, shape) and linguistic rationales, applied specifically to abstract art to control for confounding factors. It also uses Best-Worst Scaling for more reliable emotion annotations and introduces the novel task of Justified Affect Transformation, combining image recoloring and textual explanation.

Q2: Why focus on abstract art?
Abstract art emphasizes color and form without distracting representational elements such as faces or objects, allowing clearer study of how color and context affect emotional perception. Studies show that abstract art evokes a unique psychological distance, which shifts mindset towards broader emotional ideas (Abstract art evokes a sense of psychological distance, Reddit).

Q3: How can FeelingBlue’s findings be applied?
Applications include improving emotional expression in digital art, guiding artistic color choices (e.g., in blue abstract painting), enhancing emotion-aware AI tools, and informing marketing and design fields that use color psychology.

Q4: Where can I see examples of emotional use of blue in art?
Collections like the one at Blue Abstract Painting showcase varied uses of blue and its emotional richness.

Q5: Can AI already recolor images to change emotions effectively?
FeelingBlue presents a baseline with an ensemble neural network that can recolor images and provide rationales for emotional changes, showing promising but preliminary results that encourage further research.

For more on color-emotion research and applications, see authoritative resources like Pantone’s Color Institute, Interaction Design Foundation on color psychology, and academic surveys on affective computing.