In research on individual intelligence, psychological researchers have repeatedly demonstrated that a single statistical factor—often called “general intelligence” or “g”— emerges from the correlations among individuals’ performance on a wide variety of cognitive tasks. The emergence of a general performance factor appears to be a common property of complex nervous systems, whether animal or human; performance is positively correlated across cognitive tests. This raises the possibility that, far from being an artifact of human society or culture, general intelligence factors may be a characteristic of complex systems, whether made up of brain modules, groups of humans, or even humans with computers. My first line of research explores the existence of a general collective intelligence factor in human groups. In our first paper on this topic (Woolley, Chabris, Pentland, Hashmi, & Malone, 2010), we used the same approach used by researchers in the construction of IQ tests, and administered a range of different kinds of tasks to groups. Our selection of tasks was guided by existing taxonomies categorizing tasks on the basis of the various group processes needed to perform successfully. Our goal was to select a wide variety of tasks, such that a group could not rely on a well-learned routine, or depend too heavily on the same member(s) to accomplish them all successfully. Thus we included tasks that required creativity or brainstorming, along with tasks that had singular correct solutions, as well as tasks that required the careful coordination of inputs from all members for successful completion. Our analysis of groups’ scores on the tasks demonstrated a collective intelligence or c factor for groups, and its ability to predict group performance on more complex tasks. We also explored variables that might be predictive of groups’ ability to develop c. One interesting finding was that the proportion of females in a group is significantly correlated with c. Moreover, member social perceptiveness appears to play a mediating role. Social perceptiveness is significantly positively correlated with c, and consistent with existing research, men in our sample had significantly lower scores than women on our social perceptiveness measure.
In further developing our understanding of c, in two additional studies we have explored its connection to group learning. Conceptually, we predicted that the same coordination and collaboration abilities that enable a group to adapt and perform a wide variety of tasks (i.e., develop high levels of c) would also enable them to assimilate new information and improve at a faster rate with experience on the same task. In addition, we also explored the connection between c and group synergy, or the ability of a group to perform better than its best member. While prior work documents the limited conditions under which groups can achieve synergy, no research has looked at what enables certain groups to do so more consistently than others. In a study of MBA student teams, we measured c at the beginning of the academic term, and found that it significantly predicted the rate of group improvement across a series of group-based exams. We also found that c was a strong predictor of a group’s ability to consistently outperform their smartest member (Woolley, Aggarwal, Chabris, & Malone, under review). In a laboratory study, we further find that c significantly predicts the rate at which a group learns to coordinate their individual choices in a tacit coordination game across a series of 10 rounds, above and beyond the variance explained by individual intelligence (Aggarwal, Woolley, Chabris, & Malone, in prep).
We have also continued to develop more efficient methods of measuring c and have explored their generalizability in two additional papers. In a laboratory study (Engel, Woolley, Jing, Chabris, & Malone, under review), we randomly assigned teams to collaborate face-to-face versus online, and used our newly developed, web-based tool to administer a short battery of tasks. We find that our new shorter battery of tasks performs similarly to the longer set of tasks administered in the lab in our original study, and is moderately predictive of groups’ future performance, both in the groups working face-to-face and those working online. Surprisingly, we also found that our “Reading the Mind in the Eyes” measure of social perceptiveness is as predictive of collective intelligence in online groups communicating only via text-based chat as it is in face-to-face groups. In a second paper, we build on these findings by combining them with additional studies conducted in online lab groups in Japan and field-based project teams in Germany working online and face-to-face (Engel, Woolley, Aggarwal, Chabris, & Malone, under review). We find that our shortened, web-based measure operates similarly in all of these settings.
Taken together, these preliminary results offer encouraging support for our basic approach. Continuing to develop our understanding of c is important for a number of reasons. First, on a conceptual level, understanding c as a measurable and stable attribute of a team, demonstrably separate from individual ability, will change the importance placed on team composition versus other team design and process elements that can be more effectively managed. Until now, group capability has been thought about primarily as the aggregate of individual member abilities. Second, on an empirical level, one major factor hampering the advancement of knowledge in teams research is the use of different criterion variables in different team studies, making the accumulation of and comparison among findings difficult to impossible. A reliable and generalizable measure of c would enable researchers to more readily evaluate the effects of team design or process variables in a manner we could expect to generalize to other settings. My collaborators and I have received funding for future work on this topic from the Army Research Office and NSF as well as the PNC Center for Financial Services Innovation. With the funding we have received we have further adapted the original collective intelligence test battery into an online tool that enables us to administer a collective intelligence test in a standardized way to all groups, whether working face-to-face or online. We enhanced the battery to span a wide and comprehensive range of group processes, building on existing task taxonomies, and have incorporated tasks that tap into both verbal and nonverbal domains. We are using the battery to conduct additional studies in the lab, as well as with even larger groups online and in organizational settings. In the near future we also hope to make it available to other researchers. In our ongoing work, we intend to further explore a number of questions, including: (1) Does status hierarchy affect the development of collective intelligence in teams? (2) How does collective intelligence relate to group size? At what point do we see declines in collective intelligence as group size increases? (3) What tools or practices can be used to support group processes to enable groups to be more collectively intelligent?