Decision Making in Multi-Agent Groups

Date

2022-07-30

Journal Title

Journal ISSN

Volume Title

Publisher

Abstract

We explore the ways temporally structured private and social information shape collective decisions. In our first model, we consider a network of rational agents who independently accumulate private evidence that triggers a decision upon reaching a threshold. When seen by the whole network, the first agent’s choice initiates a wave of new decisions but later decisions have less impact. In homogeneous networks, the overall probability of a randomly selected agent in such groups making a correct decision is bounded from above because of the impact of the first decider’s choice. In heterogeneous networks, the first decisions are made quickly by impulsive individuals who needed little evidence to make a choice. However, these early decisions, even when wrong, reveal the correct options to nearly everyone else. We conclude that groups comprised of diverse individuals can make more efficient decisions than homogeneous ones. However, when making decisions, we often rely on a mix of information that we have acquired individually and information that is commonly available. In our second model, we neglect the effect of social information exchange to consider whether the simple fact of information having an individual or a common source affects the quality of decisions. Multiple non-interacting agents make observations, some common and some private, and decide between two options when they have gathered sufficient information to reach one of two symmetric thresholds. In the presence of a mix of common and individual observations, the first agent to reach threshold is less likely to make the correct choice than the first agent reaching threshold when all observations are private or all observations are common. We explain this counterintuitive observation, and conclude that access to common information decreases accuracy for those whose early private information coincides with the common information.

Description

Keywords

Decision-making, Stochastic Differential Equations, Correlated Information, Group Decision Making

Citation

Portions of this document appear in: Karamched, Bhargav, Megan Stickler, William Ott, Benjamin Lindner, Zachary P. Kilpatrick, and Krešimir Josić. "Heterogeneity improves speed and accuracy in social networks." Physical review letters 125, no. 21 (2020): 218302.