The Role of Cognitive Styles for Explainable AI

The increasing complexity of artificial intelligence (AI) poses challenges for users when assessing its predictions. Research has developed explainable AI (XAI) methods to help users understand AI model behavior. So far, however, XAI-approaches are mainly technology-focused – neglecting the influence of individual cognitive abilities and differences in information processing. Therefore, this study adopts a human-centered perspective, drawing on cognitive psychology’s concept of cognitive styles. Using a between-subject experiment (n=210), we investigate how users’ characteristic modes of processing information (rational and intuitive) affect their understanding of AI models for different types of explanations. Our results reveal the need to purposefully tailor explanations for individual users based on their cognitive styles. We, thus, contribute to the growing field of XAI by providing more insights into human factors in XAI design.