Model of human collective decision-making in complex environments

TL;DR

This paper models human group decision-making in complex environments using a Markov process, revealing how social interaction strength and individual knowledge influence collective performance and the emergence of group intelligence.

Contribution

It introduces a continuous-time Markov model to analyze the effects of social interactions and knowledge levels on group decision-making in complex tasks.

Findings

Optimal social interaction strength balances consensus and search efficiency.

Moderate individual knowledge suffices for high group performance.

Group intelligence emerges at a critical social interaction threshold.

Abstract

A continuous-time Markov process is proposed to analyze how a group of humans solves a complex task, consisting in the search of the optimal set of decisions on a fitness landscape. Individuals change their opinions driven by two different forces: (i) the self-interest, which pushes them to increase their own fitness values, and (ii) the social interactions, which push individuals to reduce the diversity of their opinions in order to reach consensus. Results show that the performance of the group is strongly affected by the strength of social interactions and by the level of knowledge of the individuals. Increasing the strength of social interactions improves the performance of the team. However, too strong social interactions slow down the search of the optimal solution and worsen the performance of the group. In particular, we find that the threshold value of the social interaction strength, which leads to the emergence of a superior intelligence of the group, is just the critical threshold at which the consensus among the members sets in. We also prove that a moderate level of knowledge is already enough to guarantee high performance of the group in making decisions.