---

It is increasingly common for individuals to work in collaborative groups rather than alone. One finds this trend salient in a growing number of domains, from business and management to government and media to science and academia. (Even in the humanities, an increasing proportion of research papers are being co-authored.)

But what do we know about the nature of such collaborative work? What is the relation between the capacities of individuals and the capacities of the group? And how might cognitive enhancement technologies amplify the abilities of groups to solve the problems they're confronted with?

In a fascinating paper published in Science entitled "Evidence for a Collective Intelligence Factor in the Performance of Human Groups" (2010), Dr. Anita Williams Woolley and her colleagues (yes, the research was collaborative!) make two important discoveries:

1. They find that there is such a thing as collective intelligence. What is collective intelligence? Well, it's the analogue of general intelligence, or IQ, except it exists at the level of the group rather than the individual. What's useful about IQ - our best measure of an individual's general intellectual ability - is that it can be used to predict how an individual will perform in a number of different cognitive domains. (That's why it's called "general" intelligence!) Using the same basic approach for quantifying individual general intelligence, Woolley et al find that groups exhibit a similar property that is both measurable and allows for accurate predictions of how a group will perform on a range of cognitive tasks.

2. The researchers find some quite intriguing - and counterintuitive - correlations between properties at the level of the individual and the level of the group. For example, one might "pre-theoretically" think that group intelligence is a function of the average intelligence of that group's members. And one might "pre-theoretically" think that a group with a single exceptional individual would have a higher group IQ than one with, say, three above average but non-exceptional members.

However, Woolley and her colleagues find only a statistically weak correlation between the intelligence of groups and these two member-level properties. In other words, it's not possible to accurately predict how well groups will perform on a range of cognitive tasks simply by averaging the IQs of its members, or by noting a single exceptional individual within the group. These features aren't linked - or at least not robustly - to group IQ, despite what intuition might suggest.

The Key to Smarter Groups

What, then, determines how smart a group of collaborating individuals is? The researchers find three individual-level features that correlate in a statistically significant way to collective intelligence.

First, the greater the social sensitivity of group members, the smarter the group. Second, the more turn-taking within the group, the better the group performs. And third, the more women in the group, the higher the group IQ. For any reader who works on projects in groups, this is good information to know!

(By the way, what might gender have to do with group IQ? Well, the researchers surmise that groups with more women are smarter because women tend to be more socially sensitive than men. Thus, the gender factor is real but indirect - that is, it's mediated by the property of social sensitivity.)

What I find most interesting about the paper, though, at least from the perspective of transhumanism, is the following remark at the end: "More importantly, it would seem to be much easier to raise the intelligence of a group than an individual. Could a group's collective intelligence be increased by, for example, better electronic collaboration tools?"

This is interesting because virtually all talk of "cognitive enhancement" in the literature today focuses exclusively on the enhancement of individual intelligence. But what about group intelligence? Maybe there are technologies or strategies for cognitive enhancement that are applicable on the group rather than individual level? The authors themselves suggest, in the quote above, that increasing the information-sharing abilities of group members using "electronic collaborative tools" might enhance the intelligence of the group itself (without necessarily increasing the intelligence of individual group members).

Artificial Enhancement of Groups

But there's also the more speculative possibility, not mentioned by Woolley et al, of enhancing the social sensitivity of group members. What would happen if group members took, for instance, a pharmaceutical of some sort that enabled them to be more socially sensitive towards each other? What if some sophisticated technology were available that augmented the individual's ability to better listen to the ideas of others - to let others have time to speak and to be intellectually open to opposing views?

This would involve enhancement at the individual level, of course, but it would be explicitly aimed at enhancing the intelligence of groups. Furthermore, such enhancement would not really be cognitive in nature, since it would target the psychological and emotional capacities of individuals, such as one's capacity for social sensitivity, rather than one's general intelligence.

Indeed, given Woolley's findings, it appears that increasing the raw intelligence of individual group members cannot guarantee a smarter group. A group of cognitively enhanced individuals with extremely high IQs (because of their enhancement) thus might fail to outperform a group of "normals" if those "normals" prove to be more socially sensitive than their enhanced rivals.

I take this to be a nontrivial claim because: (a) "radical" cognitive enhancements will almost certainly become widely available in the near future, and (b) as noted earlier, there is a pervasive trend towards increasingly collaborative work in nearly all domains of human activity. Maybe the best way to improve the enterprises of science, government, business, and so on, would thus be to focus on enhancing group intelligence - a goal that may or may not have much to do with enhancing individual intelligence.

(Incidentally, there seems to be a point at which the above generalizations no longer hold. If, for example, a group consisted of three "normals" and one "posthuman" whose cognitive capacities exceeded our own in the way our own capacities exceed those of, say, a mouse, then this group might indeed outperform other groups with greater social sensitivity. And this group would outperform others simply by virtue of the profound intellectual abilities of a single exceptional member, namely the ultraintelligent posthuman.)

Group Mind vs. Extended Mind

One final thought on this issue: I find it extremely interesting - as well as perplexing - to think about how Woolley's group intelligence research might relate to the "extended mind thesis." What is this thesis? Basically (to make a long story very short), the central component of the extended mind thesis is called the Parity Principle. It states that "if, as we confront some task, a part of the world functions as a process which, were it to go on in the head, we would have no hesitation in accepting as part of the cognitive process, then that part of the world is (for that time) part of the cognitive process."

Thus, according to the Parity Principle, inanimate objects like a pad of paper, a calculator, a computer, Wikipedia, an iPhone, and so on, can all, under just the right conditions, constitute a literal component of one's cognitive system - of one's mind. But, one might wonder, what's stopping one from considering other brains to be literal components of one's extended mind too?

There are, indeed, many people who come to rely on the knowledge of others - spouses, friends, etc. - in exactly the same way that they rely on Wikipedia, their iPhones, and their own internal brain structures like the hippocampus. It follows, therefore, that another mind can indeed become a feature of one's own cognitive system (on the condition that the Parity Principle is true - obviously, one might reject this principle).

But how does this phenomenon relate to the idea that groups themselves can possess a kind of general intelligence? Imagine, for example, a group consisting of three people. Over time, each member of the group comes to rely on the knowledge had by the other two members. The minds of each member are thus gradually extended beyond the arbitrary boundaries of "skin and skull" to include the other individuals in the group. What they end up with, then, are three extended minds, each of which subsumes the entire group.

In addition to these minds, though, there is the cognitive property that Woolley et al term "collective intelligence." Such intelligence "emerges" from the interaction of the groups members, and consequently it too subsumes all three members. Thus: What exactly is the relation between an extended mind that includes a whole group of individuals and the collective intelligence of the whole group itself?* And furthermore, if groups themselves can have cognitive properties like individuals, is there any possibility of extending the "mind" of a group? What would this entail and how might it work?

Obviously, these are research questions involving two cutting-edge research programs (one focusing on group intelligence and one on extended minds). Although I don't have anything insightful to say in response to these questions right now, articulating them is the first step in any new research project! (What do readers think?)

* Of course, it's also possible that either one of these ideas - the group IQ hypothesis or the extended mind thesis - are wrong, which would make the conundrum described here a mere "pseudo-problem." Some problems need only be dissolved, rather than solved.