Quantifying collective intelligence in human groups

Abstract

Collective intelligence (CI) is critical to solving many scientific, business, and other problems, but groups often fail to achieve it. Here, we analyze data on group performance from 22 studies, including 5,279 individuals in 1,356 groups. Our results support the conclusion that a robust CI factor characterizes a group’s ability to work together across a diverse set of tasks. We further show that CI is predicted by the proportion of women in the group, mediated by average social perceptiveness of group members, and that it predicts performance on various out-of-sample criterion tasks. We also find that, overall, group collaboration process is more important in predicting CI than the skill of individual members.

An Active Inference Model of Collective Intelligence

“…Existing research of human collective intelligence is limited precisely by a lack of alignment between these two scales of analysis. On the one hand, accounts of local-scale interactions from behavioral science and psychology tend to construe individual humans as goal-directed individuals endowed with discrete cognitive mechanisms (specifically social perceptiveness or Theory of Mind and shared intentionality; see [15,16]) that allow individuals to establish and maintain adaptive connections with other individuals in service of shared goals [3][4][5][17][18][19]] (Riedl and colleagues [19] report a recent analysis of 1356 groups that found social perceptiveness and group interaction processes to be strong predictors of collective intelligence measured by a psychometric test.). Researchers conjecture that these mechanisms allow collectives to derive and utilize more performancerelevant information from the environment than could be derived by an aggregation of the same individuals acting without such connections (for example, by facilitating an adaptive, system-wide balance between cognitive efficiency and diversity; see [4]).…”

Beyond IQ: The Importance of Metacognition for the Promotion of Global Wellbeing

“…This kind of collective intelligence is a property of the group itself, not just the individuals in it. Although the individual intelligence factors of group members may play a role in collective intelligence (Bates and Gupta 2017), a meta-analysis that examined 22 studies, with 5279 individuals comprising 1356 groups, showed that c is predicted by the proportion of women in the group, mediated by average social perceptiveness of group members, and that it is validated in the sense of predicting performance on various out-of-sample criterion tasks (Riedl et al 2021). Yet, c is not nearly as well-validated as g (Coyle 2021).…”

Translating Member Ability Into Group Brainstorming Performance: The Role of Collective Intelligence

“…Recently, researchers have argued that groups can be characterized by their level of collective intelligence, or a latent factor representing the group's ability to work together which can be measured based on observations of performance across a variety of tasks (Woolley et al, 2010). This factor has been shown to be a stronger predictor of future group performance than various measures of individual member skill or ability (Engel et al, 2014;Riedl et al, 2021;Woolley & Aggarwal, 2020). Furthermore, ongoing work suggests that CI arises as a result of multiple interacting, yet distinct, collective socio-cognitive systems.…”

“…Social perceptiveness in groups increases behaviors that facilitate CI, such as attention to and synchronization of nonverbal cues (Chikersal et al, 2017) and a high level and even distribution of communication among members (Engel et al, 2014;Woolley et al, 2010), a pattern associated with more effective information sharing (Stewart & Stasser, 1995). Consequently, teams with high CI are able to engage in more effective task coordination in dynamic environments (Mayo & Woolley, 2021;Riedl et al, 2021), another process established in extant research to be critical to team performance (Hackman, 1987).…”

“…This is a time interval of work not dissimilar to that observed in knowledge work environments, where, for example, González and Mark (2005) document workers focusing on projects for an average of 6.5 minutes at a time, further broken up into tasks and collaborative interactions averaging 1 minute, 29 seconds each. 1 This format using multiple short tasks is also becoming increasingly common in experiments in computational social science (e.g., Bernstein et al, 2018;Lazer et al, 2009;Mason & Watts, 2011;Weidmann & Deming, 2020) to enable the capture and analysis of multiple episodes of performance and behavioral metrics, such as those needed to measure collective intelligence (Riedl et al, 2021). Therefore, testing the effects of brainstorming task structure on a shorter task will provide a perspective on its application to the increasingly fragmented environments characterizing more and more organizations, as well as enable the measurement and analysis of the influence of CI.…”