Electrophysiological Correlates of Event Segmentation: How does the human mind process on-going activity?

Research using videos of everyday activities and fMRI has identified a network of brain regions that are sensitive to the perceptually salient boundaries between events (Zacks et al., 2001). Here we present an electrophysiological investigation into the neural correlates of event segmentation using scalp recorded Event-Related Potentials (ERPs). Three studies investigate the influence of top-down knowledge (e.g. activity goal) and bottom-up influences (e.g. object involvement), on event segmentation. Task knowledge levels were manipulated by showing participants videos of activities that were familiar (e.g. making the bed), unfamiliar (e.g. setting up a microscope), or recently learned (building a table). In each case participants were required to view the videos passively, and then to view whilst making coarse and fine grained segmentations. Segmentation points were extracted and overlaid onto passive viewing data to allow ERPs to be formed time-locked to the segmentation boundaries. Analysis of coarse grain segmentation revealed a modulation over right-frontal electrodes, relatively positive going during the processing of unfamiliar activities and negative going during the processing of familiar and recently learned activities. In addition, fine grained segmentation was associated with a modulation over left parietal electrodes, negative going during familiar and unfamiliar objects relative to recently learned. The ERP differences demonstrate the activation of topographically dissociable neural components associated with fine and course grained event segmentation. Moreover, these ERP components are differentially sensitive to the amount of knowledge and experience that participants have about the activities being viewed.