Miller, I. D. & Cupchik, G. C. (2016). Agent Based Modeling of Emotional Selection in Urban Legends. In 28th Convention of the Association for Psychological Science. Chicago, IL, USA. doi:10.6084/m9.figshare.3427301



poster: Agent Based Modeling of Emotional Selection in Urban Legends


Urban legends: Previous work has shown that urban legends are subject to emotional selection during social transmission (Heath et al. 2001; Eriksson and Coultas 2014). Disgust, a high arousal negative valence emotion, was shown to predict sharing likelihood. Surprise and humor are also related.

Why no viral outcomes? In our first study, we perform a simulation in which agents share urban legends as part of a serial transmission task. We derived a behavioral model of sharing from Eriksson and Coultas (2014), then applied it to n=40 transmission chains with 4 agents each. In contrast to typical viral time series, this paradigm reliably “fizzles” without furious social contagion.

100x more participants: The likelihood of actually sharing a story is vanishingly low, so in study 2, we increased the number of participants to n=16,000. The results were not different; after 3 time steps, over 98% of agents never saw any urban legends.

Network topology: If more agents cannot exhibit virality when organized into serial chains, would the same transmission paradigm work with a network of higher average degree? In study 3, we succeeded at obtaining contagion using the same behavioral model as the previous 2 studies with a preferential attachment network. Under some conditions, the entire network is saturated.