Does the cultural affinity of a group's members contribute to the group's collective intelligence?

Likely the 1,000's of you who have already submitted entries into the pending "HFC! CYPHIMU?” contest, the winner of which will be awarded a beautiful  "I am a citizen of the Liberal Republic of Science/I ♥ Popper!”  t-shirt (Jon Baron currently sits atop the leader board, btw), are bored and wishing you had something else to do.  

Well how about this?

First, read this fascinating study of "c," a measure of intelligence that can be administered to a collective entity.

 The study was first published in Science (2 yrs ago; fortunately, one of the authors pulled me from the jaws of entropy and  brought the article to my attention only yesterday!).

The authors show that the "collective intelligence" of groups assigned to work on problem tasks admits of reliable measurement by indicators akin to the ones used to measure "individual intelligence." An influential measure of individual intelligence is called the "g factor," or simply g. Thus, the authors call their collective intelligence measure "c factor" or "c."

C is predicted in part by the average intelligence of the group's members and by the intelligence of its smartest (highest-scoring on g) member. That it would be is not so surprising, given existing work on the predictors of group decisionmaking proficiency.

The really cool thing (aside from the proof that it was possible to form a reliable and valid measure of c) was the authors' finding that other interesting individual group-member characteristics also make an important contribution to c. One of these was how many women are in the group (compare with the recent claim by female members of the Senate that part of the reason Congress is so dysfunctional is that aren't enough female members; maybe, maybe not).

Another was the average score of the groups' members on a "social sensitivity" scale. Social sensitivity here measures, in effect, how emotionally perceptive an individual is. The better group members were at "reading" other's intentions, the more cooperatively and productively they engaged one another, the researchers found. This disposition in turn raised the "collective intelligence" of the group -- that is, enabled it to solve more problems more efficiently.  

Not mind-blowingly surprising, either I suppose. But if you think that social science is mainly about establishing mind-blowingly counterintuitive things, you are wrong, and will believe lots of invalid studies. Social science is mainly about figuring out which competing plausible conjectures are true. 

The conjectures that informed and were supported by this cool study were merely amazingly interesting, amazingly thought provoking, and likely amazingly useful to boot.

Second, now tell me what you think the connection might be between c and cultural cognition.  

As every schoolboy and -girl today knows, "cultural cognition" refers to the tendency of individuals to conform their perceptions of risk and other policy-relevant facts to ones that predominate in their cultural group. CCP studies this phenomenon, using experiments and other empirical methods to identity the mechanisms it comprises.

It is often assumed -- indeed, sometimes I myself and other studying cultural cognition say -- that cultural cognition is a "bias."

In fact, I don't believe this.  I believe instead that cultural cognition is intrinsic, even essential, to human rationality.

The most remarkable feature of human rationality, I'd say, is that individuals are able to recognize what is collectively known.  

Particularly, when a society is lucky enough to recognize that science's way of knowing is the most reliable way to know things, collective knowledge can be immense.  What's known collectively will inevitably outstrip what any individual member of the society can ever comprehend on his or her own--even if that individual is a scientist!

Accordingly, as my colleague Frank Keil has emphasized, individuals can participate in collective knowledge -- something that itself is a condition of there being much of it -- only if they can figure out what's known without being able to understand it. In other words, they must become proficient at knowing who knows what.  The faculty of rational perception involved in being able to figure this out reliably is both essential and amazing.

Well, it turns out that people are simply better at exercising this rational faculty -- of being able to reliably determine who knows what about what-- when they are in groups of people with whom they share a cultural affinity.  Likely they are just better able to "read" such people -- to figure out who actually knows something & who is just bull shitting.

Likely, too, people are better at figuring who knows what about what in these sorts of affinity groups because they are less likely to fight with one another. Conflict will interfere with their ability to exchange knowledge with one another.

Actually, there's no reason to think people can exercise the faculty of perception involved in figuring out who knows what about what only within cultural affinity groups.

On the contrary, there is evidence that culturally diverse groups will actually do better than culturally homogeneous ones if they stay at it long enough to get through an initial rough patch and develop ways of interacting that are suited for discerning who knows what within their particular group.

But in the normal run of things, people probably won't, spontaneously, want to make the effort or simply won't (without a central coordination mechanism) be able to get through the initial friction, and so they will, in the main, tend to learn who knows what about what within affinity groups. That's where cultural cognition comes from.

Generally, too, it works --so long as the science communication environment is kept free of the sorts of contaminants that make culturally diverse groups come to see positions on particular facts -- like whether the earth is heating up or whether the HPV vaccine has health-destroying side effects -- as markers of group membership and loyalty. When that happens, the members of all cultural groups are destined to be collectively dumb as 12 shy of a dozen, and collectively very unwell off.

So now -- my question: do you suppose the cultural affinity of a groups' members is a predictor of c? That is, do you suppose c will be higher in groups whose members are more culturally homogeneous?

Or do you suppose that culturally diverse groups might do better -- even without a substantial period of interaction -- if their individual members "social sensitivity" scores are high enough to offset lack of cultural affinity?

Wouldn't these be interesting matters to investigate? Can you think of other interesting hypotheses?

What's that? You say you won't offer your views on this unless there is the possibility of winning a prize?.... Okay. Best answer will get this wonderful "Cultural Cognition Lab" t-shirt.

Comments

[Ed Forbes]

OK......a shaggy dog story where the punch line is "group think".

If all you deal with is people who think like you, there is no need to stretch your brain. Group intelligence, as defined as problem solving ability, would be lower for groups that all think the same than for mixed groups.

[Angie Johnston]

It seems to me that one other crucial factor to consider here is the type of problem a collective entity is trying to solve. For instance, in the Wooley et al. (2010) article, participants set out to solve relatively uncontroversial problems in a culturally neutral domain (e.g., architecture, word completion, etc.). In these sort of situations, I would imagine that any effect that cultural affinity had would be explained by differences in the average social sensitivity of the group. If I had to make a prediction here though, I would guess that culturally homogeneous groups would perform better that diverse ones, simply because certain members of a diverse group might be distracted by various implicit biases, stereotype threat, etc. However, I don't feel very strongly about this prediction, and I think that social sensitivity would remain the primary predictor.

Importantly, though, I do think that you would see differences based on cultural affinity for issues relevant to cultural differences (especially for controversial issues). In these cases, I think that culturally diverse groups would outperform culturally homogenous groups. Specifically, I would predict that culturally homogenous groups would come to the table already agreeing on these sort of controversial issues, and act more like an individual rather than a group. In a sense, their group performance might simply be predicted by their highest IQ, and in fact they might lose the benefit of the 'c' factor entirely.

Interestingly, in the Wooley article there was only one task that groups were asked to perform that seemed to relate to cultural differences--Moral Reasoning. What's even more interesting is that this task did not correlate with any of the other tasks (individual or group) except for Speaking Turn Variance (see Supplemental Online Materials, p. 10). Moreover, Moral Reasoning had relatively low factor loadings in Study 1 (0.36), and especially low factor loadings in Study 2 (0.10; see Supplementary Online Materials, p. 7). Given that there was only really one culturally relevant sort of task in the initial study on 'c,' it is very hard to predict how group performance on these questions would differ. In fact, there was not much variance at all in the type of problems the groups were asked to solve.

This makes me wonder how flexible 'c' really is as a predictor of group performance. It seems like it does very well accounting for performance in neutral problem-solving domains, but how well does it predict other sorts of problem-solving that may be more culturally weighted or controversial?

[John F. Pittman]

I would also question the answer. An example is an engineering problem, where there are several good answers, but one "best" answer. Without such a control, I would question that any "grade" given to C reflects cultural or moral similarities than it can be used as an indication of potential effecient problem solving.