[Link] Study on Group Intelligence

post by atucker · 2011-08-15T08:56:40.889Z · score: 11 (11 votes) · LW · GW · Legacy · 6 comments


  The Paper:
  Some thoughts:

Full disclosure: This has already been discussed here, but I see utility in bringing it up again. Mostly because I only heard about it offline.

The Paper:

Some researchers were interested if, in the same way that there's a general intelligence g that seems to predict competence in a wide variety of tasks, there is a group intelligence c that could do the same. You can read their paper here.

Their abstract:

Psychologists have repeatedly shown that a single statistical factor—often called “general intelligence”—emerges from the correlations among people’s performance on a wide variety of cognitive tasks. But no one has systematically examined whether a similar kind of “collective intelligence” exists for groups of people. In two studies with 699 people, working in groups of two to five, we find converging evidence of a general collective intelligence factor that explains a group’s performance on a wide variety of tasks. This “c factor” is not strongly correlated with the average or maximum individual intelligence of group members but is correlated with the average social sensitivity of group members, the equality in distribution of conversational turn-taking, and the proportion of females in the group.

Basically, groups with higher social sensitivity, equality in conversational turn-taking, and proportion of females are collectively more intelligent. On top of that, those effects trump out things like average IQ or even max IQ.

I theorize that proportion of females mostly works as a proxy for social sensitivity and turn-taking, and the authors speculate the same.

Some thoughts:

What does this mean for Less Wrong?

The most important part of the study, IMO, is that "social sensitivity" (measured by a test where you try and discern emotional states from someone's eyes) is such a stronger predictor of group intelligence. It probably helps people to gauge other people's comprehension, but based on the fact that people sharing talking time more equally also helps, I would speculate that another chunk of its usefulness comes from being able to tell if other people want to talk, or think that there's something relevant to be said.

One thing that I find interesting in the meatspace meetups is how in new groups, conversation tends to be dominated by the people who talk the loudest and most insistently. Often, those people are also fairly interesting. However, I prefer the current, older DC group to the newer one, and there's much more equal time speaking. Even though this means that I don't talk as much. Most other people seem to share similar sentiments, to the point that at one early meetup it was explicitly voted to be true that most people would rather talk more.


Anything we should try doing about this? I will hold off on proposing solutions for now, but this section will get filled in sometime.


Comments sorted by top scores.

comment by gwern · 2011-08-15T18:17:15.069Z · score: 6 (6 votes) · LW(p) · GW(p)

This comment brought to you by my #lesswrong IRC logs. http://hbr.org/2011/06/defend-your-research-what-makes-a-team-smarter-more-women/ar/3#comment-238026281

"...First, what is reflected in their paper need not mean what they say in the article. And it seems like 'broken telephone' has occurred.

The main point of the paper they produced was that group intelligence was more accurately predicted by a 'c' factor much like individual intelligence is predicted by a 'g' factor.

Wolley states in the article:

"Things like group satisfaction, group cohesion, group motivation—none were correlated with collective intelligence. And, of course, individual intelligence wasn’t highly correlated, either."

According to the paper the 'c' factor correlated stronger with max individual intelligence and avg individual intelligence than any other factor including %women, social sensitivity except for speaking turn variance. So this statement is highly misleading.

Wolley also states in the article:

"And in our study we saw pretty clearly that groups that had smart people dominating the conversation were not very intelligent groups."

From the paper, there is no such indication of this. The methodology as explained in the paper for speaking turn variance did not measure how much smart people dominated the conversation, it showed how much a minority dominated the conversation. That is a very big difference from what Wolley said in the article. It may be that people who dominated the conversation were not the brightest but simply the most charismatic, essentially interfering the signal from their most valuable members. Either way more study has to be done here, and Wolley should be careful as to attribute conversation dominance with intelligence.

Finally, another very important question to ask is whether groups performed better than their max intelligence member within their group in regards to the intelligence testing portions of the test. My hunch is that groups do poorer than their max intelligent member in the intelligence tasks. Just to illustrate the point - I don't think a really good group of low intelligent folk could come up with general relativity (an extreme case I know!)

However, if a team environment is necessary, the goal may be to allow for the most intelligent to get the most attention rather than the most charismatic. Unfortunately, how you would do this i have no clue, and so the best alternative may be to force speaking turn variance.

Also drawing a straight line through a bunch of points is weak science (and even the authors are reluctant to draw too much from this especially in the case %women factor - a correlation of 0.23 makes me ambivalent especially if the factor is subsumed by speaking turn variance and social sensitivity factors)."

comment by jsalvatier · 2011-08-15T14:31:08.107Z · score: 3 (5 votes) · LW(p) · GW(p)

Well, one obvious point is that increasing the amount of turn taking, social sensitivity and females in meetup groups would be a positive thing.

It's worth noting that people don't always know when they are rambling. In the Seattle meetup group we had at least one person explicitly ask to be told when they're being unproductive in a conversation.

There was a thread a while back about conversational norms and rules (link). Some people liked these a lot.

I've heard that some groups succeed because they have an excellent group moderator who keeps people to the topic at hand and prevents people from dominating the conversation. Not everyone has this skill. Are there easy ways to learn it?

comment by atucker · 2011-08-16T05:36:43.573Z · score: 1 (1 votes) · LW(p) · GW(p)

Are there easy ways to learn it?

Are there easy ways to measure it? If so, I would enjoy going to a meetup where we try and get better and moderating.

comment by ScottMessick · 2011-08-15T17:59:07.515Z · score: 2 (2 votes) · LW(p) · GW(p)

Robin Hanson had an old idea about this which I liked: http://hanson.gmu.edu/equatalk.html

It's not going to be a silver bullet, but I think it would work well in contexts where the group of people who are in the conversation and how long it should last are well defined. Situations where an ad hoc committee is expected to meet and produce a solution to a problem, but there is no clear leader, for example. (Or there is a clear leader, but lacking expertise herself, she chooses to make use of this mechanism.)

It'd be nice to see a study on whether "EquaTalk" can produce the high "c" value observed in this study. (Disclosure: I didn't read or even skim the linked paper.)

comment by handoflixue · 2011-08-15T23:54:38.921Z · score: 1 (3 votes) · LW(p) · GW(p)

This makes me curious how much it applies to online groups as well; the website seems fairly male dominated.

comment by atucker · 2011-08-16T05:35:19.355Z · score: 2 (2 votes) · LW(p) · GW(p)

I'd imagine that it applies less to online groups. For instance, social sensitivity is incredibly blunted online, and on a forum/blog its much more difficult to dominate a conversation simply by talking louder and longer than everyone else. You can post more, but that doesn't force people to read you more.