Human crowds provide paradigmatic examples of collective behavior emerging through self-organization. We suggest that collective patterning in human crowds is promoted by anticipatory path-seeking behavior resulting in a scale-free movement pattern, called the Levy walk.
Human crowds provide paradigmatic examples of collective behavior emerging through self-organization. Although the underlying interaction has been considered to obey the distance-dependent law, resembling physical particle systems, recent findings emphasized that pedestrian motions are fundamentally influenced by the anticipated future positions of their neighbors rather than their current positions. Therefore, anticipatory interaction may play a crucial role in collective patterning. However, whether and how individual anticipation functionally benefits the group is not well-understood. We suggest that collective patterning in human crowds is promoted by anticipatory path-seeking behavior resulting in a scale-free movement pattern, called the Levy walk. In our experiments of lane formation, a striking example of self-organized patterning in human crowds where people moving in opposite directions spontaneously segregate into several unidirectional lanes, we manipulated some pedestrians ability to anticipate by having them type on a mobile phone while walking. The manipulation slowed overall walking speeds and delayed the onset of global patterning, and the distracted pedestrians sometimes failed to achieve their usual walking strategy. Moreover, we observed that the delay of global patterning depends on decisions made by pedestrians who were moving toward the distracted ones and had no choice but to take sudden large steps, presumably because of difficulty in anticipating the motions of their counterparts. These results imply that mutual anticipation between pedestrians facilitates efficient transition to emergent patterning in situations where nobody within a crowd is distracted. Our findings may contribute to efficient crowd management and inform future models of self-organizing systems.