Best-of-n decision making by human groups

Abstract

Collective decision making is a fundamental aspect of group behavior in both animals and humans, and often involves reaching a consensus on the best of n options, using empirical evidence. Although many parallels have been drawn between human and animal collective decisions, collective human behavior is rarely studied in the type of embodied scenarios that animals are often faced with. In this study, we placed human groups in a virtual setup similar to nest site selection in social animals, in which they explored a shared environment and reached a consensus based on their observations of empirical features. In groups of up to 10, participants had to reach consensus on the empirically largest of four candidate sites without verbal communication, instead using movement-based interactions in a custom-developed 3D virtual environment for online multi-participant experiments. The results showed that the speed and accuracy of consensus was importantly modulated by perceptual difficulty and information availability, but that no speed-accuracy trade-off was present. Participants attempted to reach consensus on the empirically largest site by flexibly adapting their use of social information to perceptual difficulty, their spatial position, and the time already spent supporting some option. When a minority of informed individuals were present, these individuals exercised greater independence and influenced the group to faster and more accurate consensus. These results extend previous findings on social decision making strategies in humans to nonverbal scenarios akin to those of social insects.