Beyond Traditional Stimuli: Establishing AI-Generated Images as Valid Tools for Emotion and Affect Research

Studies of emotion frequently require standardized emotional stimuli to elicit emotional responses in participants. While existing affective databases offer images with normative ratings of valence and arousal, selecting adequate stimuli becomes challenging when additional experimental criteria, such as specific combinations of theme and content, are necessary. The present study explored the feasibility of using generative AI, specifically text-to-image generators, to create tailored affective stimuli. Across two studies, participants rated the valence and arousal of 160 and 200 AI-generated images (negative and neutral). Our findings revealed that AI-generated images exhibit the typical valence-arousal patterns observed in standardized affective databases, demonstrating moderate to strong associations between these two emotional dimensions. These results highlight the potential of generative AI as a valuable methodological tool for creating customized affective stimuli aligned with distinct research objectives and experimental designs.