Science, Faith, and Not Ruling Out Possibilities

Science, Faith, and Not Ruling Out Possibilities

Much has been said in the last two days about a Harvard law student's e-mail to a couple of other students — an e-mail that was some months later apparently widely forwarded by one of the recipients (without the sender's permission). The e-mail led to much criticism of the sender, coverage in Above The Law, the Boston Globe, an AP wire story, and the Harvard Crimson, a condemnatory statement from the Dean of Harvard Law School, an apology from the sender, and more.

I thought I'd say a few words about this, both in this post and in some others to come, because this seems to me to go to the heart of what a university should be, of what we should want our society to be, and of a scientific approach to questions of scientific fact. 

I would have happily avoided this topic if I could have. But I feel an obligation — as a professor, as a tenured professor, and as someone who feels strongly about the need to treat scientific questions as scientific questions and not as articles of faith — to speak up about it. I am not naïve enough to be surprised that an e-mail such as this would lead to public condemnation and a public outcry. But that the reaction has been unsurprising doesn't mean that it has been proper.

Let me begin with the e-mail, which was apparently a follow-up to a conversation between the student and the recipients at a dinner shortly before:

… I just hate leaving things where I feel I misstated my position.

I absolutely do not rule out the possibility that African Americans are, on average, genetically predisposed to be less intelligent. I could also obviously be convinced that by controlling for the right variables, we would see that they are, in fact, as intelligent as white people under the same circumstances. The fact is, some things are genetic. African Americans tend to have darker skin. Irish people are more likely to have red hair. (Now on to the more controversial:) Women tend to perform less well in math due at least in part to prenatal levels of testosterone, which also account for variations in mathematics performance within genders. This suggests to me that some part of intelligence is genetic, just like identical twins raised apart tend to have very similar IQs and just like I think my babies will be geniuses and beautiful individuals whether I raise them or give them to an orphanage in Nigeria. I don't think it is that controversial of an opinion to say I think it is at least possible that African Americans are less intelligent on a genetic level, and I didn't mean to shy away from that opinion at dinner.

I also don't think that there are no cultural differences or that cultural differences are not likely the most important sources of disparate test scores (statistically, the measurable ones like income do account for some raw differences). I would just like some scientific data to disprove the genetic position, and it is often hard given difficult to quantify cultural aspects. One example (courtesy of Randall Kennedy) is that some people, based on crime statistics, might think African Americans are genetically more likely to be violent, since income and other statistics cannot close the racial gap. In the slavery era, however, the stereotype was of a docile, childlike, African American, and they were, in fact, responsible for very little violence (which was why the handful of rebellions seriously shook white people up). Obviously group wide rates of violence could not fluctuate so dramatically in ten generations if the cause was genetic, and so although there are no quantifiable data currently available to "explain" away the racial discrepancy in violent crimes, it must be some nongenetic cultural shift. Of course, there are pro-genetic counterarguments, but if we assume we can control for all variables in the given time periods, the form of the argument is compelling.

In conclusion, I think it is bad science to disagree with a conclusion in your heart, and then try (unsuccessfully, so far at least) to find data that will confirm what you want to be true. Everyone wants someone to take 100 white infants and 100 African American ones and raise them in Disney utopia and prove once and for all that we are all equal on every dimension, or at least the really important ones like intelligence. I am merely not 100% convinced that this is the case.

Please don't pull a Larry Summers on me,

CRIMSON DNA

Here's my thinking on the e-mail itself; I'll have a few more posts shortly about some of the reaction to the e-mail.

1. Whether there are genetic differences among racial and ethnic groups in intelligence is a question of scientific fact. Either there are, or there aren't (or, more precisely, either there are such differences under some plausible definitions of the relevant groups and of intelligence, or there aren't). The question is not the moral question about what we should do about those differences, if they exist. It's not a question about what we would like the facts to be. The facts are what they are, whether we like them or not.

Given this, it seems to me that the proper approach to this question is precisely the same as the proper approach to other questions of scientific fact. One absolutely should not rule out the possibility that African Americans are, on average, genetically predisposed to be less intelligent. Likewise, to give examples involving three groups I myself belong to, one absolutely should not rule out the possibility that Jews are (say), on average, genetically predisposed to be more acquisitive, or more loyal to their narrow ethnic group than to broader groups, or that whites are genetically predisposed to be more hostile to other racial groups, or that being nonreligious is genetically linked, and that people who have those genes are genetically predisposed to be more likely to commit crime or cheat on their spouses or what have you. One should also obviously be willing to be convinced by evidence that shows that, by controlling for the right variables, we would see that those groups are, in fact, identical to other groups under the same circumstances. 

One should not rule out possibilities in the absence of conclusive evidence, for the simple reason that one then has no factual basis to rule out those possibilities. And since on many things the evidence will rarely be conclusive, one shouldn't rule out those possibilities categorically at all. And one should also be open to the evidence that exists, and to being convinced by it in one or the other direction (to the degree of conviction that is warranted by the evidence).

Now some claims may be so contrary to our current understanding of the world that we might say something like this: We shouldn't rule out the possibility in principle, but in practice the probability is so vanishingly small that we should exclude it from our analysis. That, for instance, might be one's view about claims that werewolves exist. First, it's just hard to imagine, given current science, what possible mechanism there might be that would turn humans into wolves every full moon. Second, one would think that if werewolves existed, we'd have good evidence of them, since proving their existence would be pretty easy.

But we still know very little about which genes produce intelligence, how exactly those genes operate, and even how intelligence can be defined. We obviously have vastly more left to learn about this. And there is certainly reason to believe that intelligence is heritable in some measure among individuals (though there is hot debate about the degree to which this is so). Such heritability, coupled with the possibility of differing selection pressures in different environments, provides a potential mechanism through which there conceivably could be intelligence differences among racial or ethnic groups.

So at this point it seems to me that the only scientifically sensible conclusion about this question, which I stress again is a question of what the facts really are, is that we can't be sure that there are no such differences: Again, we cannot rule out either the possibility that there are racial differences in intelligence, or that there aren't.

Or at least we cannot rule them out as a scientific judgment. (Perhaps there's some expert somewhere out there who is so knowledgeable and brilliant that he feels he can accurately predict all that we will ever know about this field, and therefore can rule out one or the other possibility; I doubt it, but in any case I'm pretty sure that no-one is this discussion is that expert.) Obviously, each of us has the perfect right to rule any factual possibility out as a matter of faith, moral, religious, or whatever else. We can say "I don't care what the evidence might say, I rule out this possibility because of my moral beliefs." Or we can say "My moral beliefs are actually capable of indicating to me not just what I should do, but what the scientific facts about the world actually are, and therefore I am completely confident about what those facts are, based on my confidently held moral beliefs."

But surely there ought to be no obligation on other people to adopt this sort of faith-based view on scientific questions. That's why it seems to me that the author's statement that "I absolutely do not rule out the possibility that African Americans are, on average, genetically predisposed to be less intelligent" — or a similar statement, as I suggested, about Jews, or whites, or the irreligious — is perfectly proper, and in fact is the way that people should approach scientific questions of all sort.

2. Of course, I take it that some people were inferring from the e-mail that the author doesn't actually mean just that she doesn't rule out this possibility, but rather that she actually thinks the possibility is likely true. If so, then to critique the e-mail one would have to further discuss whether in fact the possibility is likely true under the current, highly limited state of scientific knowledge. 

But there is no need to do that here. This e-mail was a follow-up to an earlier conversation, which apparently was not recorded. It was intended to be a private e-mail to other students who were parts of that conversation. One can't tell whether the e-mail was (a) actually a means of implicitly asserting that there probably are intelligence differences, or (b) a rebuttal to an allegation that the author wasn't scientifically minded enough in the discussion over dinner and was wrongly foreclosing scientific possibilities, or (c) part of a discussion about the nature of scientific evidence, or anything else. Sometimes, one might legitimately draw inferences about a person's views based on a statement that was meant to be self-contained, to the point of justifying public criticism of the inferred views and not just the literally stated ones. But one can't infer from this snippet of the broader conversation that the author means anything other than what she says: that she does not rule out a certain possibility, a possibility that I think cannot scientifically be ruled out.

I considered whether some of the language of the e-mail, such as (emphasis added) "In conclusion, I think it is bad science to disagree with a conclusion in your heart, and then try (unsuccessfully, so far at least) to find data that will confirm what you want to be true" suggests that the student believes that there is no existing data strongly suggesting the absence of genetic differences. If that were the right interpretation, then we'd have to discuss whether there is indeed such data. 

But my reading of this, given both this sentence and the rest of the e-mail, is that the author is saying that there has been no success in (to go further down the paragraph) "prov[ing] once and for all that we are all equal" in intelligence, and in providing evidence that would make one "100% convinced that this is the case." That's a restatement of the first sentence in the e-mail, and again it strikes me as being quite scientifically accurate: There can't be, at this stage of our knowledge (and possibly at any stage), proof "once and for all" that there are no such racial differences in intelligence.

3. On then to just a brief response to what I imagine would be some likely reactions.

a. Some might argue that belief in racial differences in intelligence could cause all sorts of immoral and harmful social and legal reactions. That might be so. But it's different from the question that the student was writing about, which is what is actually true. Lots of other facts that are actually true can yield, and have yielded, harmful social and legal reactions. That doesn't make those facts any less true — nor does it make it somehow improper for people to even be open to the possibility that certain facts might, in fact, be true.

b. Some might point to the history of unsound claims about racial differences in intelligence. And the history of errors in a field should indeed teach people to avoid those particular errors. But there's no "three strikes and you're out" for scientific theories: That some people in the past have posited various unsound theories with some general thesis doesn't mean that all theories with a related thesis are guaranteed to be false. One still cannot rule out the possibility that some other theory in that genre will in fact be correct. Again, that's just the way facts are: If something is true, people's having thought a bunch of similar-sounding things that are nonetheless false doesn't affect that truth.

c. Some might point out that intelligence and race are "socially constructed," which is certainly true in the sense that different societies may draw racial lines in different places, and may define what constitutes intelligence — or how it should be tested — differently. But while we can't just assume that there are some obviously correct definitions of either term, science often operates with terms that don't have an inherently correct definition. What usually happens is that people come up with possible definitions, there's debate about those definitions, there are studies done using different definitions, some results emerge that are common over a wide range of definitions and others that are highly sensitive to the definitions, and so on. Yet the right approach throughout this process is, again, precisely to "not rule out the possibility" that under some set of plausible definitions some result might be true, and to be willing to "be convinced" that under some set of plausible definitions some other result might be true.

It's also possible that over time it will turn out that the definitional question is so difficult (or the required measurements are so difficult) that no real pattern emerges in the results. Say, for instance, that under some definitions of intelligence one sees one result and under others one sees the opposite result, and there seems to be no good basis to choose any particular definition over another. That might mean that we have to reformulate the question, and that the original question might be abandoned as not accurately answerable in its original form. We can't rule out that possibility, either. But neither can we just assume that this is sure to happen.

d. Finally some might just argue that even the openness to the possibility that there may be racial differences in intelligence will offend people, and that the author should have recognized that the e-mail she sent to a couple of people might be forwarded to others who might be offended.

But this presupposes that it's somehow wrong for people in a free country to discuss scientific questions because of the possibility that some people might learn about that and be offended. That can't be right. 

It especially can't be right for students at a research university. But I think that it can't be right for anyone anywhere. I realize that in the real world there might be bad consequences to speakers who offend others, however legitimate the speaker's position — which, I stress again, is a position of openness to scientific evidence — might be. But we should work against that phenomenon, and its tendency to suppress honest discussion about scientific questions. We should not just give in to it as inevitable and, worse still, somehow right.

Disclosure: The student who wrote the e-mail will be clerking for the same judge for whom I clerked, so I thought I'd note this in case some thought it relevant. But I don't know her personally (perhaps I have talked to her once by phone, but I'm not sure I've even done that), and this post has nothing to do with the indirect employment connection.