Visual categorization of complex naturalistic materials

Project C1 combines behavioral experiments, machine learning and brain imaging to investigate how we categorize real-world materials like plastics, textiles and metals. Our main goal is to map the entire hierarchy of material categories and their visual properties. Due to the enormous variety in their appearance, materials provide an ideal testbed for investigating the roles of perceptual (visually-driven) and conceptual (semantically-driven) features in categorization, as well as individual differences in categorization. Taking a ‘big data’ approach, we will train machine learning models of categorization and map the neural representations of essentially all material categories and their visual properties.