Abstract
Previous research shows that individuals can accurately determine the average characteristics of crowds (e.g., average emotion, gender, or identity) even when viewing crowds for a split second (Haberman et al., 2007, 2009;, DeFockert et al., 2009). While prior research has investigated whether participants can perceive average emotional characteristics from static displays of crowds, it remains uncertain how participants track emotional fluctuations of crowds in real-time. In this experiment, we display a crowd of continuously morphing emotional faces, and ask participants to track the average emotion of the crowd using an online method of adjustment approach. By tracking participants’ perception in real-time, we can determine whether participants are able to precisely report rapidlydynamically-changing crowd emotion. In our experiment, participants viewed crowds comprised of 1, 4 or 6 faces. While each individual face in the crowd changed its emotion every 50 ms, the overall average emotion of the crowd smoothly morphed between happy, angry, and sad following a semi-biased random walk trajectory (with 147 possible variations in between). Participants were able to precisely track the average emotion of the crowd as a whole, despite individual face percepts being disjointed. Additionally, our paradigm allows us to determine whether participants’ perception lags behind the average crowd emotion, and if so, by how much. A cross correlation analysis suggests that participants’ responses are tuned to the emotion of the crowd presented a few seconds earlier, potentially suggesting a perceptual lag or smoothing. Our results show that participants’ overall perception of a unified crowd emotion remains intact, despite high individual face variation. Our results extend previous experiments by showing that participants are able to precisely perceive the overall emotional tenor of the crowd—even in dynamic, rapidly fluctuating scenes.
Meeting abstract presented at VSS 2015