Skip navigation
Please use this identifier to cite or link to this item:
Title: Predictability and Conceptual Similarity Structure in a Real-Life Narrative
Authors: Wang, Biyang
Advisors: Hasson, Uri
Contributors: Woolfolk, Robert
Department: Psychology
Class Year: 2014
Abstract: The brain constantly generates predictions as a means of survival: to ensure that an organism is able to perform accurate predictions to adapt to and prepare for events in the environment. But what influences predictability? In the present study, we used a real-life stimulus, separated into story segments, in in order to test the hypothesis that predictability of a story segment is related to the conceptual similarly of that segment with a prior segment. In, Experiment 1, a Predictability Survey was created that asked subjects to predict what immediately happens after listening to a story segment. The Intact condition appears to improve in accuracy over time compared to the Scrambled condition, although the difference was small. Experiment 2 asked subjects to rate the extent to which the events of Segment 1 affect their understanding of the events of Segment 2, in order to generate a recurrence plot of the conceptual similarity values. Computer- and human- generated plots show a significant correlation between their similarity judgments. By correlating the predictability time-course from Experiment 1, with the corresponding conceptual similarity values over the same timeline, results show that in humans, the ability to predict what happens immediately following a segment is positively correlated with the conceptual similarity between that predicted segment with the adjacent prior segment. Overall, the study shows that predictability is correlated with the degree of conceptual similarity between segments of a real-life story.
Extent: 98 pages
Type of Material: Princeton University Senior Theses
Language: en_US
Appears in Collections:Psychology, 1930-2017

Files in This Item:
File SizeFormat 
Wang_Biyang.pdf1.5 MBAdobe PDF    Request a copy

Items in Dataspace are protected by copyright, with all rights reserved, unless otherwise indicated.