The local interactions between neighbors serve as the fundamental building blocks that give rise to human crowd behavior, but little is known about the actual mechanisms governing these interactions. This dissertation contains four experiments that quantify them and make it possible to test hypotheses about their properties and parameterization. In Experiment 1, a participant walked with three human ‘confederates’ while the heading direction or speed of a subset of neighbors was manipulated. Experiment 2 introduced and validated the use of virtual pedestrians by comparing participants’ responses to matched real-world and virtual reality stimuli. Experiment 3 consisted of three sub-experiments, which tested how participants responded to systematic manipulations in the heading direction or speed of a subset of 12 virtual pedestrians. Experiment 3A varied the number in the subset and their distance, Experiment 3B varied the number in the subset and the density of the crowd, and Experiment 3C varied the number in the subset as well as their eccentricity. Finally, in Experiment 4, real crowds of 16 or 20 walked together in a swarming scenario. Overall, the results indicate that pedestrians are coupled to their neighbors and that this local coupling is additive, decays strongly with distance, and decays weakly with eccentricity. As part of a broader endeavor to develop an empirically-grounded, cognitively-plausible understanding of human crowd dynamics, this research provides important constraints on models of pedestrian interaction, and sheds light on the fundamental mechanisms of self-organization that give rise to collective behavior.
Rio, Kevin W.,
"Mapping the Visual Coupling Between Neighbors in Real and Virtual Crowds"
(2014).
Cognitive Sciences Theses and Dissertations.
Brown Digital Repository. Brown University Library.
https://doi.org/10.7301/Z0057DBW
