Rethinking Testosterone

[To read Andrew Sullivan’s response, scroll to bottom.]

If Andrew Sullivan’s paean to testosterone (“The He Hormone“) in last Sunday’s New York Times Magazine were nothing more than a Philip Rothian screed on the underestimated glories of masculinity and a testimonial to a drug that restored potency to an H.I.V.-positive man, it would have been a perfectly acceptable (if overheated) personal essay. But instead it’s the provocative packaging for a supposedly objective report on the current state of testosterone research. And that’s a big problem right there. By letting Sullivan mix up his subjective reactions with laboratory work, by allowing him to blur the edges between his own powerful longings for a cure-all and several scientific studies whose authors–by the way–aren’t necessarily in the mainstream of testosterone research (though Sullivan doesn’t tell you that), the Times has put out under its imprimatur an account of testosterone, its therapeutic possibilities, and its larger social implications that is dangerously misleading.

First the news: Culturebox has interviewed several of the scholars and science writers mentioned by name in Sullivan’s piece, and most of them think he goes way overboard in his confidence that varying levels of testosterone (“the big T”) can explain appreciable differences in human competitiveness, social dominance, mental alertness, energy, self-confidence, etc.–though the critics don’t all agree about how far is too far. (Three of them were sufficiently like-minded to sign their names to a letter to the Times that says: “In particular, there are scant results from well-controlled experiments showing that testosterone affects behavior of normal men in the ways asserted by Sullivan.”) Robert Sapolsky, an eminent Stanford University professor of biology and neurology and an expert on testosterone who was not quoted in the article, told Culturebox: “[Sullivan] is entitled to his fairly nonscientific opinion, but I’m astonished at the New York Times.” On the other hand, objections such as these must be taken with a grain of salt because scientists, who are steeped in caution, are notorious for objecting to representations of their work by journalists, who traffic in bold generalizations. So let us consider some of Sullivan’s larger points–and the criticisms of them–one by one.

Are Sullivan’s exhilarating experiences with injections of synthetic testosterone representative? Can we learn anything from them? This one’s easy: Of course not. They’re not even relevant. There’s no way of knowing what’s responsible for Sullivan’s reactions. Did he feel his surges of energy and combativeness because he went from uncommonly low levels of testosterone to uncommonly high ones? (People with normal levels of testosterone do not report such dramatic results when injected with the hormone.) Could injections of any other chemical compound–estrogen, say, or nicotine, or Saint Johnswort–have produced the same outcome? Without controlled testing, no one can say. Sullivan alludes to the possibility that he was experiencing the placebo effect: “No doubt my previous awareness of the mythology of testosterone had subtly primed me for these feelings of irritation and impatience.” But knowing that doesn’t stop him from using his experiences as evidence anyway: “It seems plausible enough to ascribe some of this increased edginess and self-confidence to that biweekly encounter with a syringe full of manhood.”

Do we really know if testosterone “leads to behavioral differences” in human beings, as, in Sullivan’s words, “the evidence suggests that it does”? If it does, do we know how? No and no. Here we get to the fundamental flaw of Sullivan’s piece–what a friend of Culturebox’s calls “the old correlation-vs.-causation problem.” A mistake often made by people writing about science is assuming that coincidences mean more than a strict reading of the data permits. Again and again, Sullivan presents research showing that testosterone happens to be high (by some statistically meaningful amount) in the presence of this or that psychological trait (aggression, social dominance, sexual drive). Sullivan understands that co-existence doesn’t guarantee causality, because when reporting on the studies he carefully uses words such as “correlates” or “is associated with.” But then, in a subsequent sentence or paragraph, Sullivan often lapses into implying that high testosterone levels do in fact lead to these behaviors.

But can’t we at least say that it’s likely that testosterone causes aggression, etc.? Some sociologists, psychologists, and evolutionary psychologists do believe that high testosterone levels cause “dominance behavior”–or at least that testosterone and dominance behavior have a reciprocal impact on one another. Alan Booth, a professor of sociology and human development at Pennsylvania State University and one of the signers of the letter to the Times, says that of the many traits Sullivan lists as being “associated with testosterone”–energy, tenacity, self-confidence, high self-esteem, psychological dominance, strength–only dominance and strength can be pegged to testosterone. Testosterone clearly boosts muscle mass, and Booth feels the research supports a link between testosterone and the effort to dominate “interpersonal relationships.” Of the other traits, he says, “One can raise questions with each one of those.” For instance, he says that high testosterone is also associated with depression, which would tend to diminish self-esteem, confidence, tenacity, energy, and conceivably sexual drive. Even in the absence of depression, he says, he knows of no research directly linking T and most of those qualities.

It’s important to note that, their caution notwithstanding, Booth and his colleagues don’t even occupy the center position in the field of testosterone research. They mark out one edge. On the other side are biologists and neurobiologists and endocrinologists who somewhat counterintuitively hold that the causality runs the other way. Quantitatively speaking, this side has more research to point to. Sapolsky, in a recent essay called “The Trouble With Testosterone: Will Boys Just Be Boys?” lays it out baldly:

OK, suppose you note a correlation between levels of aggression and levels of testosterone among these normal males. This could be because a) testosterone elevates aggression; b) aggression elevates testosterone secretion; c) neither causes the other. There’s a huge bias to assume option A, while B is the answer. Study after study has shown that when you examine testosterone levels when males are first placed together in the social group, testosterone levels predict nothing about who is going to be aggressive. The subsequent behavioral differences drive the hormonal changes, rather than the other way around.

Now, Sullivan acknowledges that “the difference between cause and effect is often extremely hard to disentangle,” but he doesn’t appear to realize that if behavior causes testosterone rather than the other way around, then socialization is still as strong and valid an explanation for behavior as it has ever been, and none of the things he says about high-testosteroned men naturally being drawn to becoming trial lawyers or women not being designed for combat, etc., make any sense.

What do scientists know for certain? They know that testosterone is the cause of key developmental differences between male and female human fetuses. After that, what exactly it does in and to people is a mystery. Even Georgia State University psychology professor James M. Dabbs, probably the most supportive of Sullivan’s story and one of its primary sources, admits that “we don’t know too much about it. In those studies he cites, it’s 100 people here and 100 people there. It seems like you have to go beyond the data a little bit to make the statements he does.” Dabbs is alluding to the extreme thinness of testosterone research on human beings, compared with that done on animals. In a 1998 survey of the literature on testosterone and men written by Booth and Allan Mazur, a professor of public affairs at Syracuse University (another of the letter-writers), they explain in detail how hard it is to perform research on human subjects. The problem is that it’s difficult to find large groups of men to test over long periods of time unless they happen to be in unusual situations, such as reform schools, prisons, or the military. The scholars write: “The question remains: Is high T a cause of dominant and antisocial behavior? The question could be answered with a double-blind experiment comparing the behavior of normal men whose T levels had been altered with a control group.” But these studies haven’t been done, since there aren’t many normal men who’d mess with their testosterone levels for the sake of research.

But Sullivan has all those suggestive examples! On closer examination, Sullivan’s examples seem unduly selective, or much less telling than they sound. For instance, Mazur points out that in the study Sullivan cites that notes consistently higher T levels in actors than in ministers, those could easily be accounted for by disparities in age and body fat–actors being on the whole younger and skinnier than ministers. And at least one of Sullivan’s examples is a well-known scientific laughingstock. Sullivan alludes to a study of pregnant women injected in the 1950s with progesterone (which is similar to testosterone) whose daughters “later reported markedly tomboyish childhoods.” According to Sapolsky, that study is “one of the most famous and most famously discredited literatures in psychoendocrinology.” (“I just taught this one in my class last week!” he exclaimed to Culturebox.) First, contemporary scientists now dismiss the study’s suspect definition of “tomboyish” behavior–it included such things as expressing an interest in a career. Second, by the time the study was done, its subjects had all undergone extensive corrective surgery on their genitals, which had been masculinized as a result of exposure to the drug in the womb–a circumstance that gives you, at the very least, a powerful alternative explanation for why the girls would exhibit or identify with so-called male behavior patterns.

What about the studies that show that blacks have 3 percent to 19 percent higher testosterone levels than whites? Now here’s a touchy subject. None of the scholars Culturebox talked to had ever heard of the Journal of the National Cancer Institute study cited by Sullivan–which is not to say that it doesn’t exist, just that it is not widely known. The letter-writers told Culturebox that their research showed no significant differences in T levels between whites and blacks that couldn’t be accounted for by disparities in age and education and income levels. This gets complicated, of course, because other scientists would dispute whether education and income levels could play a part in this. The point is, Sullivan presents this inflammatory data as if it were matter of course, when it is at least controversial and possibly marginal.

What about the way he implicitly endorses testosterone gel for men and explicitly (though probably jokingly) endorses testosterone injections for women “to improve their sex drives, aggression and risk affinity”? It isn’t made clear in the article, but what he’s talking about here are steroids–the anabolic steroids that athletes take are essentially synthetic testosterone with the masculizining properties removed. In the case of men, everyone agrees that most men with minor variations in normal testosterone levels probably won’t feel much of anything, unless they take enormous doses, which would also entail putting the body at risk for prostate cancer and other side effects. As for women: “Here he’s speculating almost dangerously,” says J. Richard Udry, a professor of sociology and health at the University of North Carolina at Chapel Hill (and the third signer of the letter). “Then you start getting side effects, getting hair all over your face and body and causing your voice to get deeper, and it won’t go back up when you stop taking testosterone.”

So, to sum up, does testosterone explain the differences between men and women? It certainly explains why boys develop penises and secondary sexual characteristics in the womb and during puberty. Some people–not all–believe that it helps explain why men are more predisposed to seek dominance than women, and they are working to pin down exactly how this presumed predisposition might interact with the environment. But their findings are far from conclusive, because no one really understands how testosterone acts in the brain. Sullivan writes,

What our increasing knowledge of testosterone suggests is a core understanding of what it is to be a man, for better and worse. It is about the ability to risk for good and bad; to act, to strut, to dare, to seize. It is about a kind of energy we often rue but would surely miss. It is about the foolishness that can lead to courage or destruction, the beauty that can be strength or vanity. To imagine a world without it is to see more clearly how our world is inseparable from it and how our current political pieties are too easily threatened by its reality.

Given the dearth of evidence about human testosterone and the uncertainty about the little there is, the act of imagination required here is figuring out what Sullivan is talking about.

Full disclosure: Culturebox could be construed as having a conflict of interest in writing about Sullivan, since he recently gave a negative review to a book written by her husband. She does not believe this has affected her research into Sullivan’s essay, but it is obviously up to the reader to determine whether she has exhibited fairness and objectivity.