Simulating Panic Amplification in Crowds via Density-Emotion Interaction

A considerable number of agent-based models have been introduced to study the spread of emotions in crowds. Since these studies often aim to simulate collective behaviour driven by emotional escalation, like during stampedes and riots, amplification of emotion is a key aspect in these models. However, the biological processes underlaying emotion amplification in these models often remain unclear, preventing validation and accurate parameter setting. The aim of the present study is to explore whether densityemotion interactions can theoretically explain events of panic amplification in dense crowds. Specifically, we extend the model DECADE, of which the process that drives emotional convergence is rooted in psychological and neurological literature, while support is lacking for the amplification process. Therefore, in this study we propose an alternative amplification mechanism that operates on the desire to maintain a personal space. A minimum distance is kept from others under normal conditions, but under stress other goals are prioritised, like escape. This results in personal space violations, where serious violations induce emotional stress in others. Additionally, pushing behaviour is considered when a stressed agent is prevented from escaping. The proposed model is validated with video of an evacuation incident, that was previously used to validate emotion contagion models. We conclude that the proposed amplification mechanism offers a plausible alternative that is biologically falsifiable, as it resembles emotion amplification in a dense crowd, while incidents in less dense crowds do not escalate. Further empirical study is necessary to establish whether such a mechanism indeed contributes to real-world stampedes.