Abstract
A major challenge in the study of human cognition is understanding the mechanisms of semantic integration (SI). SI denotes the ability to judge semantic relations and form new meanings (e.g., for associative relations, "mirrors tend to co-occur with hairbrushes"). Though much research has been devoted to the way concepts are encoded and organized in the brain, little is known about the way the relations between such concepts are processed, especially when these relations do not rest on associative knowledge (e.g., "mirrors and peacocks both convey the concept of vanity"). Here we asked whether such abstract SI rests on the same neural mechanisms as associative SI, by assessing whether these processes differ in a quantitative or a qualitative manner. We contrasted ERPs induced by abstractly related, associatively related and unrelated pairs of concrete, real-life objects, presented sequentially. Subjects' task was to determine whether the objects were related and rate the strength of these relations. As expected, performance was less accurate and slower for abstract than associative or unrelated pairs, and ratings of relations strength decreased from associatively related, through abstractly related to unrelated pairs. Electrophysiological measures supported both a quantitative and a qualitative difference between abstract and associative pairs: the amplitude of the N400 component, held to index SI, was quantitatively modulated by semantic relations type. Notably, abstract pairs also elicited greater right over left hemisphere acitvations than associatively related pairs, suggesting a greater role for right areas in abstract relations processing.
Meeting abstract presented at VSS 2017