December 14, 2006

Flacks and hacks and Hitchens

In the January 2007 Vanity Fair, Christopher Hitchens has taken up the fashion for biologistic accounts of sex differences ("Why women aren't funny"):

The chief task in life that a man has to perform is that of impressing the opposite sex, and Mother Nature ... equips many fellows with very little armament for the struggle. An average man has just one, outside chance: he had better be able to make the lady laugh. [...]

Women have no corresponding need to appeal to men in this way. They already appeal to men, if you catch my drift. Indeed, we now have all the joy of a scientific study, which illuminates the difference. ... To annex for a moment the fall-about language of the report as it was summarized in Biotech Week:

[...] "Women appeared to have less expectation of a reward, which in this case was the punch line of the cartoon," said the report's author, Dr. Allan Reiss. "So when they got to the joke's punch line, they were more pleased about it." The report also found that "women were quicker at identifying material they considered unfunny."

Slower to get it, more pleased when they do, and swift to locate the unfunny—for this we need the Stanford University School of Medicine?

The Biotech Week article that Hitchens quotes is essentially a reprint of a Stanford University press release, "Gender differences are a laughing matter, Stanford brain study shows". And Hitchens' response to the article is classic: he sees the study as a wasteful confirmation of the obvious, telling us what everyone already knows about men and women.

The only trouble is, somewhere along the way from the researchers to the flacks to Hitchens, the message has gone wrong. Those things that Hitchens is so sure are obvious to everyone? Well, let's take a quick look.

The study in question is Eiman Azim, Dean Mobbs, Booil Jo, Vinod Menon, and Allan L. Reiss, "Sex differences in brain activation elicited by humor", PNAS 102 16496-16501, 2005.

Hitchens: "[women are] slower to get it, more pleased when they do, and swift to locate the unfunny".
Azim et al. 2005: "We found no between-sex differences in the number of stimuli found funny [t(17.531) = -0.029, P < 0.977], the subjective degree of funniness [t(17) = 0.895, P < 0.383], or the response time (RT) to funny [t(17.99) = 0.20, P < 0.944] or unfunny [t(16.22) = -0.769, P < 0.453] stimuli."

How in the world could Christopher Hitchens, who is a smart person, have made such a dumb mistake?

The short answer is that instead of reading the article, he trusted flacks and journalists.And then he added a bit of his own stereotype-driven misinterpretation of the flackery and hackery.

He's not the only one. The rest of the media response to this work -- which came out a year ago, it's not exactly a news flash at this point -- was the all-too-familiar steaming pile of falsehood and irrelevance. Some pieces, like Hitchens', asserted things about the study that are directly falsified by its conclusions. Others asserted things that the study didn't deal with at all: thus Anne Casselbaum wrote in Discover Magazine "Women Don't Understand (why Adam Sandler is funny)"

Stanford University humor researcher Allan Reiss has a reassuring insight for all the men whose girlfriends and wives roll their eyes at Adam Sandler movies: Women really do enjoy a good laugh as much as you do; they are just wired to focus on different aspects of humor.

In the study Reiss et al. did, the women found the same cartoons funny to the same degree as the men. But the study wasn't in any way designed to find the (no doubt genuine) differences among its subjects's various senses of humor, and in particular it shed no light whatsoever on the distribution of Adam Sandler's appeal by sex.

The Stanford study did find some sex differences in its sample, which I'll write about another time. These differences are interesting in themselves, though hard to interpret, and their connection to the PR (and to Hitchens' description of the PR) is fascinating. For now, though, I want to focus on two curious and completely characteristic facts about this study and its media uptake.

The first thing is that journalists, as a group, misdescribed the study's findings. Big surprise. It's pretty clear that none of those whose reports I've scanned actually read the paper with a critical eye, relying instead on the press release and on other journalists' stories. Stanford's PR department -- and Dr. Reiss himself -- appear to deserve a share of the blame for the resulting misrepresentations. But it's strange that a science-oriented publication like Discover would employ a writer who doesn't bother to look past the surface. And it's striking that Hitchens, who would never take a politician's press release at face value, is so completely uninterested in the facts of the science that he chooses to cite. This reinforces my conclusion that in today's public discourse, science is treated not as a search for the truth, but as source of edifying fables.

The second fact is more subtle and more pervasive. It's not just a fact about PR departments and journalists, it's embedded in the way that most psychologists and nearly all neuroscientists think. Read these two selections from the Azim et al. study, and think about them for a minute:

From the Materials and Methods section: "We scanned 20 healthy subjects (mean age, 22 years ± 1.9; 10 females)."

From the Conclusions section: "Males and females share an extensive humor-response strategy as indicated by recruitment of similar brain regions: both activate the temporal-occipital junction and temporal pole, structures implicated in semantic knowledge and juxtaposition, and the inferior frontal gyrus, likely to be involved in language processing. Females, however, activate the left prefrontal cortex more than males .... Females also exhibit greater activation of mesolimbic regions, including the nucleus accumbens ... "

So the subjects in this study were 10 males and 10 females, average age 22, recruited at Stanford Medical School. Presumably they were medical students, or grad students, or pre-med students. Anyhow, they were all between 20 and 24 years old, and they were all at Stanford.

But the paper's conclusions aren't about how Stanford med students' brains work. Instead, the conclusions are about what "males and females share" and what "females ... activate .. more than males" and so on.

This is the way that the Stanford researchers and the Stanford publicists talk about their findings, both in the research paper and in the press releases. And this spin is adopted, entirely uncritically, not only by Hitchens but also by all the news reports that I've seen on this work.

For example, Alok Jha, "Why females laugh longer at punchlines", The Guardian 11/8/2005, whose lede was:

Women find the punchlines of jokes more satisfying than men do, according to a study by scientists. They also use more of their brains to appreciate humour in the first place.

And Randolph E. Schmid, "Women May Enjoy Humor More, If It's Funny", The Associated Press, Monday, November 7, 2005, 6:39 PM (reprinted in the Washington Post), who began his article by writing

The difference between the sexes has long been a rich source of humor. Now it turns out, humor is one of the differences.

and who wrote throughout about "how the male and female brains react to humor", "how men and women process humor", and so on.

Would anyone accept a characterization of Americans' political or religious opinions, or their product preferences, based on a sample of 10 first-year Stanford medical students? Would a newspaper try to predict a national election from such a sample? Would a network executive rely solely on such a sample in estimating the response to a new comedy show? You'd have to unusually stupid or gullible to believe predictions about the American population at large - much less the human species at large -- that are based on ten 20-somethings enrolled at one first-rank American medical school at some point in 2003 or 2004.

So why are Dr. Reiss and his colleagues willing to treat such a sample as acccurately characterizing the nature of the brain responses to humor of human females and human males, taken as a whole? And why do science correspondents throughout the media, and a savvy political journalist like Hitchens, accept this extrapolation as truth, without a hint of skepticism?

There's an implicit assumption here that from the point of view of humor, a brain is a brain -- or rather, a male brain is a male brain, and a female brain is a female brain. Age, education, personality, cultural background, occupation -- none of that matters, and so none of that needs to be controlled for. We neuroscientists don't need no demographically balanced samples, we're measuring brains. Determining men and women's responses to humor is treated like determining the melting points of bismuth and antimony -- all you need to do is to measure a pure enough sample. There's some residual recognition that statistical variation needs to be averaged out, which is why N=10 rather than 1. But if you think of the enormous variation in either sex's sense of humor -- surely as richly varied as their attitudes towards politics or shampoo -- the assumption of sexual uniformity seems very strange.

The funny thing is, the very same authors, in the very same issue of PNAS, disprove this assumption -- based on a different experiment with the very same cartoons. The article is Dean Mobbs, Cindy C. Hagan, Eiman Azim, Vinod Menon, and Allan L. Reiss, "Personality predicts activity in reward and emotional regions associated with humor", PNAS 102 16502-16506, 2005. From the abstract:

Our analysis showed extroversion to positively correlate with humor-driven blood oxygenation level-dependent signal in discrete regions of the right orbital frontal cortex, ventrolateral prefrontal cortex, and bilateral temporal cortices. Introversion correlated with increased activation in several regions, most prominently the bilateral amygdala. Although neuroticism did not positively correlate with any whole-brain activation, emotional stability (i.e., the inverse of neuroticism) correlated with increased activation in the mesocortical-mesolimbic reward circuitry encompassing the right orbital frontal cortex, caudate, and nucleus accumbens. Our findings tie together existing neurobiological studies of humor appreciation and are compatible with the notion that personality style plays a fundamental role in the neurobiological systems subserving humor appreciation.

There's a striking amount of overlap between the brain areas whose responses to this particular set of cartoons varied with personality measures, and those whose responses varied with sex. That could be because sex and personality measures co-vary, either in general (as they surely do) or specifically in Stanford medical students who volunteer for fMRI studies (where the distribution of personality by sex might well be an atypical one). There may also be other factors, having nothing specifically to do with the interaction of humor, sex and personality, which affect the response to these cartoons so as to create what seem to be effects of sex and personality on the neurological processing of jokes.

Perhaps we'll see future studies showing characteristically different brain responses to cartoons as a function of socioeconomic status, educational level, age, political orientation, and ethnic background -- treated in each case as if some fundamental physical constant were being measured. If that happens, you can bet that whatever the measurements are, the news stories will pitch the results in terms of the relevant inventory of social stereotypes.

[I should add that in cases like this, neuroscientists are motivated to over-generalize by self-interest as well as by their discipline's cultural blindness to social factors. A story about the neurology of Stanford medical students' reaction to cartoons is unlikely to get any media play, but a story about the neurological basis of sex differences in humor is a different animal altogether. And science journalists share a similar sort of motivation, beyond their profession's tradition of assembly-line reworking of press releases: their editors will (I'm guessing) prominently display a story about the neuroscience of sex differences in humor appreciation, but bury or not run at all a story about marginal sex-linked differences in the brains of a few medical students watching cartoons.]

Posted by Mark Liberman at December 14, 2006 07:25 AM