Several people have been writing about intelligence recently, apparently because The Bell Curve somehow became a topic of conversation recently. (I can’t remember how it came up, despite constantly reading the boards where it did come up. Emergent properties, or proprieties, or improprieties, I guess.)

The best thing to do if you want intelligent discussion about the topic is to read Kevin Drum, Matthew Yglesias, Kieran Healy, Nathan Newman, Brad DeLong, The Sixth Internationalist, or, especially, Atrios here, here, here, here or here. I just have one fairly geeky point to append to those discussions.

One of the points that has come up a bit, more on the message boards on those blogs than in the actual posts, is whether there really is a relation more intelligent than. There is a recurring worry, quite reasonable I think, that trying to identify such a relation will involve somehow drawing illicit comparisons between those who are better at one kind of intellectual activity, say pattern recognition, and those who are better at another, say conceptualisation. But really there’s no reason for the mere existence of a relation more intelligent than to imply that such comparisons need be made. For we can believe there is a real relation here, but that it is . For some pairs of people, it is not the case that x is more intelligent than y, nor y more intelligent than x, nor are they equally intelligent. English does not treat non-linear relations well, particularly not non-linear comparatives, so the best we can say about such a case is that x and y are differently intelligent. Picasso, Joyce and Einstein, for instance, might well be regarded as simply differently intelligent to each other, neither more nor less nor equally intelligent.

Saying that the more intelligent than relation is real and non-linear seems to me to draw a nice middle ground between those who want to insist, on the basis of some obvious cases, that some people are more intelligent than others, and those who worry that ‘measures’ like IQ are fundamentally flawed. For if more intelligent than is non-linear, then any linear representation of intelligence will be necessarily distortive. There’s just no way to collapse a non-linear ranking onto a linear ranking, which a numerical ranking must be, without artificially ranking some kinds of intelligence above others, just because of the way that we draw the ranking.

If more intelligent than is linear, then the flaws in IQ measurement are correctable in principle, if it is not, then they are not. If more intelligent than is non-linear, then we’d expect that attempts to map it onto a linear scale would produce some successes, but inevitably there would be serious bugs that seem to move around, rather than vanish, when one tinkers with the system. To my relatively untrained eye, that looks a lot like what has happened with the data.

UPDATE: On reading more closely, I see that the point I’m trying to make here isn’t that different from the point Matthew Yglesias is making. The only difference is that I care about a geeky distinction that has, as far as I can tell, no implications for the rest of the debate. One hypothesis is that the phrase more intelligent than is indeterminate between a number of linear relations. Another is that it is ambiguous between all those relations. These are reasonable responses to the data. John Broome takes the intuitions I think are support for non-linearity to be reasons for believing in massive indeterminacy. But they aren’t my response. I think more intelligent than could be semantically reasonably determinate, and even have a fairly public meaning, and still denote a non-linear relation. If what you care about are conceptual flaws in IQ measurement, then any of the three hypotheses here will be sufficient to show they are really badly flawed. Indeed, they will all show something of the same flaw. Just which of the three is correct is very much a “philosopher’s question”.