Urban Impulses: Evoked Responses From Local Event Stimuli

In modeling human behavior, we expect people to make noticeable reactions to the events they witness. For people at a scheduled event like a concert or sports game, we can measure reactions by looking at geotagged social media posts. We work from a database of known events from a commercial ticket broker and a database of geotagged tweets to show how we can derive impulse response functions of tweet counts as event responses. Tweet counts typically rise in anticipation of the event and gradually fall after the event starts. We draw an analogy between evoked responses in functional magnetic resonance imaging (fMRI) from mental stimuli and social media responses from local event stimuli. Our analysis of event and tweet data shows that our derived impulse responses are statistically significant and that we can use the functions to accurately predict reactions to some event types. We give examples of impulse response functions derived from repeated events at different venues.