September 2019
Volume 19, Issue 10
Open Access
Vision Sciences Society Annual Meeting Abstract  |   September 2019
Collective Decision Making in Human Crowds: Majority Rule Emerges From Local Averaging
Author Affiliations & Notes
  • Trenton D Wirth
    Cognitive, Linguistic, and Psychological Sciences, Brown University
  • William H Warren
    Cognitive, Linguistic, and Psychological Sciences, Brown University
Journal of Vision September 2019, Vol.19, 52a. doi:
  • Views
  • Share
  • Tools
    • Alerts
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Trenton D Wirth, William H Warren; Collective Decision Making in Human Crowds: Majority Rule Emerges From Local Averaging. Journal of Vision 2019;19(10):52a. doi:

      Download citation file:

      © ARVO (1962-2015); The Authors (2016-present)

  • Supplements

Collective motion in human crowds emerges when each pedestrian averages the motion of their neighbors (Warren, CDPS, 2018). Previously, we found that a participant is influenced by a weighted average of visible neighbors, with a weight that decays exponentially with distance (Rio, Dachner & Warren, PRSB, 2018). Yet crowds also make collective decisions, such as whether to follow a subgroup, that seem to imply a nonlinear decision rule (Pillot, et al., 2011; Leonard, et al., 2012). Faced with such a decision, does a pedestrian follow the group average, a quorum, the majority, or the minority?Our neighborhood model seems to suggest they would average the two groups. To test the question, we asked participants to “walk with” a virtual crowd (N=8 or 16) that split into two subgroups. Participants walked in a 12×14m tracking area while wearing a head-mounted display, and head position was recorded. On each trial, the virtual crowd walked forward (1–2 s), then a subgroup turned left (50%, 63%, 75%, 88% of the crowd) and the rest turned right by the same amount (angle between them α = 10°, 15°, 20°, 25°, 30°). Interestingly, both the participants and the neighborhood model tend to follow the majority. A regression analysis shows that the neighborhood model accounts for 42% of the variance in the human final heading. Thus, an apparently nonlinear decision does not require a nonlinear decision rule. Rather, the nonlinearity stems from the spatial separation of the two groups over time: The initial response is biased by the majority of near neighbors, who pull the participant away from the minority and increasingly dominate the neighborhood. To test this explanation, Exp. 2 eliminates the spatial separation by using two continuous streams of virtual neighbors. Conclusion: nonlinear decision-making emerges from spatial averaging over a small neighborhood.

Acknowledgement: NSF BCS-1431406 

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.