Location: Room 357 Levin Building

Douglas Guilbeault
Network Dynamics Group
University of Pennsylvania

The Network Dynamics of Category Formation

How do societies develop categories for continuous sets of novel phenomena, as in the domains of art, science, and technology? Seminal work in cognitive science argues that given the same stimuli, people can independently arrive at the same categories as a result of universal and innate cognitive processes. Cognitive universals are said to account for “social coherence”, where distinct social groups and even distinct cultures have been shown to arrive at highly similar category systems. Social coherence is widely seen as incompatible with constructivist theories of category formation, which hold that categories are defined contextually through communication, leading to highly divergent, path-dependent category systems. Using a novel formal model, we predict that increasing the size of a social network can increase the similarity of the category systems that people construct. We test this hypothesis using a web platform to experimentally control the size of people’s social networks as they collaboratively labeled a novel continuum of arbitrary shapes. We find that in dyads (N=2), communication led to highly divergent category systems; but in large networks (N=50), it led independent groups to converge on remarkably similar category systems. We show that large social networks amplify weak population biases by enabling critical mass dynamics, where a subset of slightly more frequent words become exponentially more likely to spread as network size increases. These results suggest that the emergence of social coherence may be attributable to coordination dynamics and diffusion processes in social networks.