Decoding Emotions in Abstract Art: Cognitive Plausibility of CLIP in Recognizing Color-Emotion Associations
This work addresses the gap in understanding how AI models interpret emotions in abstract art, which is incremental for cognitive science and AI alignment.
The study evaluated CLIP's ability to recognize emotions in abstract art, finding low but above-baseline accuracy, indicating poor alignment with human cognition, and identified stronger color-emotion associations in CLIP than humans.
This study investigates the cognitive plausibility of a pretrained multimodal model, CLIP, in recognizing emotions evoked by abstract visual art. We employ a dataset comprising images with associated emotion labels and textual rationales of these labels provided by human annotators. We perform linguistic analyses of rationales, zero-shot emotion classification of images and rationales, apply similarity-based prediction of emotion, and investigate color-emotion associations. The relatively low, yet above baseline, accuracy in recognizing emotion for abstract images and rationales suggests that CLIP decodes emotional complexities in a manner not well aligned with human cognitive processes. Furthermore, we explore color-emotion interactions in images and rationales. Expected color-emotion associations, such as red relating to anger, are identified in images and texts annotated with emotion labels by both humans and CLIP, with the latter showing even stronger interactions. Our results highlight the disparity between human processing and machine processing when connecting image features and emotions.