PNAS: Information transfer in human crowds

Posted by on Apr 23 2012 in Headlines4 comments

Pedestrian crowds can form the substrate of important socially contagious behaviors including propagation of visual attention, violence, opinions and emotional state. However, relating individual to collective behavior is often difficult, and quantitative studies have largely employed laboratory experimentation. In the only previous study of gaze-following in human crowds, Stanley Milgram et al. (1969) instructed stimulus groups to stop and stare up into a building window on a crowded street in New York City, measuring the probability of passersby adopting this behavior. Although informative to the effects of social influence in public settings, these findings raise several interesting questions. What is the strength of this contagion response? What are the functional reasons pedestrians follow the gaze direction of others? Can cues provided by the visual attention of others provide valuable information regarding the location and identification of pertinent, but weak or ambiguous, stimuli? To what degree this is response context dependent?

OPEN ACCESS – See the full study by clicking HERE.

We performed two studies in which we used semi-automated analysis to track the motion and head direction of 3,325 pedestrians in natural crowds (see picture on the right and video below), quantifying the extent, influence and context-dependence of socially transmitted visual attention. In our first study we instructed stimulus groups of confederates within a crowd to gaze upwards towards a single point atop of a building. Analysis of passersby demonstrates that the probability of pedestrians adopting this behavior increases as a function of stimulus group size, before saturating for larger groups. This response has important consequences for how information about environmental stimuli is acquired socially between pedestrians. Many group-living animals use quorums (i.e., k>1) in response to the presence or behavior of others, allowing them to make consensus decisions in which all (or the majority of) individuals adopt the same choice. Conversely, the proportional-saturating response established here did not generate strong consensus. We developed a model that predicts this gaze response will lead to the transfer of visual attention between crowd members, but is not sufficiently strong to produce a “tipping-point” or “critical mass” of gaze-following that has previously been predicted for crowd dynamics.

This first experiment also showed that visual attention spreads unevenly in space. Individuals with trajectories leading them to walk behind the visual orientation of the stimulus group exhibited a higher propensity to follow the gaze of the stimulus group. This suggests that gaze-following under these conditions is not due to social pressure or some form of obedience. Instead, individuals followed gaze direction more often when the person they were copying could not observe their behavior. A similar rearwards transfer of visual attention has also recently been observed in bi-directional pedestrian traffic (Gallup et al., in press), suggesting that within natural settings gaze-following is strongly mediated by social interaction and facilitates acquisition of environmentally relevant information.

A second experiment, in which passersby were presented with two male confederates performing suspicious/irregular activity, supports the predictions of our model. Although pedestrians appeared to follow each other’s gazes towards the stimulus, and this was particularly the case for the suspicious-acting condition, there was no tipping point at which large numbers of individuals simultaneously gazed in that direction. Overall, visual interactions between pedestrians occurred primarily within a 2-meter range. This localized response, and the propagation of copying mainly by those not under observation of others, further reduces its the efficiency for transferring information about the location of weak signals or subtle features of the environment.

To explore the context dependence in this response, these experiments were performed in two urban settings – a busy shopping street and a bustling city train station. In the thoroughfare pedestrian gaze was drawn towards the suspicious actors at close distances, while in the train station gaze was drawn towards nearby control activity but averted from those exhibiting suspicious activity at very close positions. We hypothesize that aversion in a crowded commuter station consisting of largely stationary individuals could be because close proximity and directed gaze communicate threat or dominance. Further analyses indicated that male pedestrians accounted for this close range gaze-avoidance.

While the above aspects of gaze-following response are reproduced robustly between experimental setups, the overall tendency to respond to a stimulus is dependent on spatial features, social context and the sex of the passerby. In summary, this approach has produced quantitative and qualitative insights about a candidate causal link in the relation between individual and crowd behavior: the propagation of attention.


  1. What are you looking at? People follow each other’s gazes, but without a tipping point (Discover Magazine)


  1. Gallup, A.C., Hale, J.J., Garnier, S., Sumpter, D.J.T., Kacelnik, A., Krebs, J. & Couzin, I.D. (2012) Visual attention and the acquisition of information in human crowds. PNAS, published online April 23rd, open access.
  2. Milgram, S., Bickman, L. & Berkowitz, L. (1969) Note on the drawing power of crowds of different size. Journal of Personality and Social Psychology 13, 79–82.
  3. Gallup, A.C., Chong, A. & Couzin, I.D. (2012) The directional flow of visual information transfer between pedestrians. Biology Letters, published online March 28th.


» Comments (RSS)
  1. [...] molto complessa che ha coinvolto oltre tremila persone, durata più di due anni (qui il video).Due gli esperimenti svolti, per tentare di rispondere anche a questi interrogativi: perché [...]

  2. [...] Experiment Make sure you check out the video of the analysis technique on Iain Couzin’s page here (it’s the one where everyone in the crowd looks like they’ve got a yellow arrow [...]

  3. [...] the researchers are interested in. Some of Ian Couzin’s work at Princeton looking at information transfer and decision-making in crowds is a great example of this kind of social science. Unfortunately the technology, participants, and [...]

  4. [...] the researchers are interested in. Some of Ian Couzin’s work at Princeton looking at information transfer and decision-making in crowds is a great example of this kind of social science. Unfortunately the technology, participants, and [...]

Add a comment