Lindsey K. Wilson¹ (), Prasanth Chandran¹, Maverick E. Smith², Karissa B. Payne¹, Lester C. Loschky¹; ¹Kansas State University, ²Washington University in St. Louis

How do people perceive discrete events, with boundaries between them, from continuous experience? To measure that, we use picture stories as a highly simplified version of real-world events, and ask participants to segment them into discrete events. To investigate story event segmentation, we need a comparison condition in which segmentation is difficult. Here, we scramble the order of images. But are scrambled-order picture stories completely un-segmentable? Randomization should disrupt viewers' segmentation, but it might be above-chance. Our research question is, what is the nature of such above-chance segmentation of scrambled picture stories? We hypothesized viewers may perceive meaning in scrambled stories based on key images (images preceding or following event boundaries in the unscrambled stories), irrespective of their order. To test that hypothesis, we scrambled the order of six picture stories, had participants view each, then segment the boundaries between events, and then summarize each story. If segmentation agreement is greatest for key images, it would indicate that event segmentation is most influenced by them. Based on a power analysis, we plan to run 20 participants each, for 24 scrambled versions of each story (N = 480 total). Previously, when we presented these picture stories to participants in a coherent order (n=48), there were 38 images across all six stories with >50% boundary agreement. Our incomplete initial results (N = 11) for the randomized condition show a total of nine images with >50% agreement. This reduction from 38 to 9 boundaries (>50% agreement) is consistent with disrupted segmentation in the randomized condition. Three of these nine total event boundaries aligned with the same images from the coherent condition. Those images could be considered key images, irrespective of their order. Nevertheless, we need many more participants to determine the actually agreed-upon boundaries, and whether key images are indeed important for that.

Acknowledgements: This work was supported by the Kansas State College of Arts and Sciences Undergraduate Research & Discovery Scholarship