Criticality and Collective intelligence in Groups

We investigate the dynamical behavior of small group of agents in making decisions on complex fitness landscapes. In particular we identify specific conditions which lead the system to undergo “critical” phase transitions towards a state of high consensus characterized by high values of efficacy of the group measured in terms of the fitness. We identify this transition with the emergence of the collective intelligence of the groups. This transition is characterized by a significant amount of information flow from the fitness landscape to the group. This increased information flow is caused by a larger exploration of the landscape, which makes the agents be attracted by those regions of the landscape with higher fitness and simultaneously converge towards a high consensus state.

We show that the set of conditions leading to the collective intelligence state is strongly affected by the size of the group, their expertise, by the presence of leaders, by the structure of the group social network, by the presence of distrust, by the uncertainty associated with the environment, etc.

Our results identify the specific conditions under which the consensus seeking within the group results in ineffective decisions. We also show that, such a phenomenon, known as groupthink, can be strongly mitigated by the presence of distrust relationships that move the transition front so as to make the system closer to criticality and eventually induce the critical transition and trigger emergence of collective intelligence.