Collective Behavior in a Connected World

Abstract

Understanding how individual actions and interactions give rise to group-level properties is at the core of collective behavior research. In animal groups, measuring and modeling these processes has yielded predictive insight into how they avoid predators, navigate, cooperate, and make decisions. Often, they achieve these feats in the absence of leaders and without individuals being aware of the collective behavior itself. Remarkably, even in groups composed of unrelated, selfish members, natural selection has produced emergent and functional group behavior.

Collective functionality however, can be sensitive to perturbation and specific to environmental contexts. For example, the simple rules that govern how army ants follow pheromone trails allow them to navigate and make decisions in dense jungle forests with poor vision. In some contexts, however, these rules can go awry resulting in a death spiral whereby the ants march in a circle until they starve or rejoin the group. The context-dependent functionality of collective behavior raises serious questions for human groups. The rules that govern our collective behavior arose in the context of hunter-gatherer groups, yet we now face global challenges while communicating in increasingly digital ways.

This thesis begins by drawing on evidence across disciplines to argue that human collective behavior is unlikely to be sustainable in the absence of intervention. To this end, chapter one charts a course for a crisis-minded study of collective behavior, while highlighting some of the key challenges and critical next steps. In chapter two, this thesis reveal how network structures unique to the digital era impact the ability of groups to make accurate decisions in the face of uncertainty. Finally, chapter three examines a key hurdle to managing and understanding human behavior by providing evidence that social interaction and network topology pose challenges to conducting statistical inference.