Promotors:
Dr. C. Gerritsen
Prof. Dr. K. Hindriks
G Roups and crowds are often seen as an accidental or planned congregation of individuals. However, in the way these individual people are connected, arranged and interact in groups, unexpected patterns can surface that transcend the mere sum of individuals. In other words, how we feel and act also depends on the situation and the others around us [116]. We talk about group chemistry, the will of the crowd or acting in the moment. Why does one crowd respond calmly to an incident, while another breaks into panic? Why does one group stand up to help a victim of public violence, while another stays passive or hinders first responders? This dissertation studies one contributing phenomenon called emotion contagion. In the presence of others, a strong emotion is often not experienced alone. Emotions can spread, inducing or altering emotions in others [524]. Emotion contagion occurs via a number of mechanisms and modalities, including mimicry (e.g. of facial, vocal and bodily expressions), category activation (e.g. from text and music), and social appraisals (e.g. via evaluations and attention) [165, 423]. This infectious quality of emotional expressions is often named as a contributing factor to large crowd incidents [197]. Stampedes and crowd crushes are tragic examples where fear spreads rapidly hand-in-hand with collective behaviour that is dangerous to its members [321]. Some examples that received large news coverage include that during Halloween in the streets of Seoul [328], the reoccurring incidents during the Hadj in Mekka [171], and the live-broadcast incident during the Astroworld festival in Houston [497]. Also collective anger is regularly linked to serious crowd incidents with human harm, as well as substantial economic damages. While most protests start out peaceful, some turn to collective violence and vandalism. Examples are riots following protests against the Covid19 measures in various countries world wide [553], against the election results in the United States [472], and against racism in the United Kingdom [470]. But even during events intended to be enjoyable and entertaining, negative emotions can flair, leading to confrontations among subgroups or with security personnel, such as around football matches [327], at festivals [207] or amusement parks [226]. 1 2 1 INTRODUCTION What if the vulnerability for emotion-driven incidents could be screened before public events take place [13]? Or predicted in real-time to warn and allow intervention before the situation escalates [85, 576]? What if security professionals and citizens alike could practice how to act in case of emotion-driven emergencies in the safety of a virtual crowd [593]? To these ends I aim to progress the development of computer models and applications that can simulate how people exchange emotions in crowds. In addition, I aim to further our understanding of emotion contagion in groups. Despite being linked to the spread of negative crowd behaviour, affective empathy is foremost an important mechanism in healthy human communication and groups dynamics [106, 526]. Therefore, thought should also be spent on how its desired effects can be protected in a changing world. Our society has digitalised rapidly over the past decades, where an increasing share of social interactions takes place online [165]. With social media, video conferencing and in the future social virtual reality, online groups have become a new front in managing the safety and social health of citizens. To this end, I also explore the exchange of affect in the context of communicating with computer-controlled as well as human-controlled virtual representations. Toward these aims, I combine techniques from the fields of affective computing and agent-based modelling.