CASOS Working PAPER

"Generating Panic Within Populations" (PDF file)
Author: Carter T. Butts

Abstract
Previous research on the problem of belief panics - episodes in which numerous actors develop highly devergent beliefs for a brief period in the absence of direct evidence - has demonstrated the plausibility of belief feedback mechanisms as an explanation for panic. Building on this work, a model is here developed which allows for the emergence and dissolution of panic phenomena within structured populations of individual actors. The behavior of this model is then analyzed using a virtual experiment in order to identify the primary determinants of the rate of panic occurrence. Assumptions regarding saliency and communication are shown to be crucial aspects of the panic model as well as predictors of panic rates, along with network density and the rate at which external signals are introduced. Network clustering, while examined, is not found to be related to the panic rate.