PhD Defense by Joel M Mumma
Name: Joel M. Mumma School of Psychology – Ph.D. Dissertation Defense Presentation Date: Monday, July 15, 2019 Time: 2:00 PM Location: J.S. Coon 148 Advisor: Francis T. Durso, PhD (Georgia Tech) Dissertation Committee Members: Rick Thomas, PhD (Georgia Tech) Jamie Gorman, PhD (Georgia Tech) Howard Weiss, PhD (Georgia Tech) Joseph Magliano, PhD (Georgia State) Title: Understanding the Perceptual Segmentation of Situations via Event Segmentation Theory Abstract: Parsing the daily stream of activity into situations is essential for adaptive functioning in everyday life. Central to this imperative is the question of how people form and modify their mental representation of a situation. Across several literatures (i.e., social psychology, engineering psychology, and narrative comprehension), we identified a number of points of agreement about the properties of our mental representations of situations. We argue that Event Segmentation Theory (EST) provides a framework for understanding how these properties coalesce to give rise to our representations of situations. The goal of the present studies was to understand how EST might account for one property in particular - the hierarchical structure of our representations of situations. According to EST, people maintain a hierarchy of “event models” of ongoing activity in working memory, which represent events unfolding simultaneously on different timescales. Event models continually try to predict the near future and are updated in response to prediction error. Updating an event model gives rise to our perception of a “boundary” between events and is what people report during event segmentation tasks. EST posits that the hierarchy of event models in working memory arises from the differential predictive accuracies of coarse-event models (e.g., of situations) and fine-event models (e.g., of shorter events occurring within situations). We tested this hypothesis by orienting participants to their event models of the situations or of the fine events in a narrative film, either by having them indicate each time a new situation or a new fine event began. Throughout the film, we also assessed their confidence and predictive accuracy at moments when both variables should depend on the event model being interrogated. Across two studies, we obtained novel support for the general mechanisms of EST but converging evidence that participants only maintained fine-event models of activity, even though we found that their segmentation of the film depended on their orientation. We propose that the fine-grained segmentation of activity may reflect the updating of fine-event models whereas coarser-grained segmentation may instead reflect how people group fine events online, rather than the updating of coarse-event models (e.g., of situations) per se.
- Workflow Status: Published
- Created By: Tatianna Richardson
- Created: 07/01/2019
- Modified By: Tatianna Richardson
- Modified: 07/01/2019