The recent (self-)destruction of Harvard evolutionary biologist Marc Hauser is both hard to watch and impossible not to. When the university last month found Hauser guilty of scientific misconduct —ugly and serious words, those, meaning in this case either tweaking data or fabricating it outright—someone really, really big started a long fall in slow motion.
He won’t land for months yet, as everyone stays mum while federal funders and others investigate. In the mean time, what are we to make of his magnificent, downward arc? What was he trying so hard to prove, and why, to try to prove it, did he climb way out on a skinny limb, like some reckless rhesus?
It’s worth asking these questions, for such a scandal affects more than just the people at its center. Methods, theories, entire disciplines get needlessly sullied. Hauser will likely land hard. We might want to move some things out of the way before he hits.
Start with what he was up to. In the broadest sense, Hauser was trying to prove that we share with some of our primate kin certain basic, conceptual structures in our minds. First, he claimed (along with others) to have found in monkeys signs of the innate “universal grammar” that linguist Noam Chomsky had posited as being fundamental to human speech and thought. Then he made a more surprising assertion: Many of our primate kin, he argued, also share with us a universal moral grammar—or a “moral organ,” as he called it in his last book—that provides the necessary framework for ethical behavior. This was a Big Idea with Big Implications: TED-level stuff, and it shot him to stardom.
The question, of course, was Where’s the evidence? Hauser based both claims partly on clever, “spot the difference” experiments that test whether monkeys can make simple linguistic and moral distinctions. In one study, for instance, he tried to show that rhesus monkeys, cotton-topped tamarins, and chimpanzees could distinguish between intentional movements and those that seemed accidental. Each animal would be presented with two small containers, and then an experimenter would either a) grab one of the containers with a deliberate motion, suggesting the container might have food inside, or b) let his arm flop down in such a way that his hand would “accidentally” make contact with it, which presumably suggested nothing about its contents. Then the animal would be let at the containers. Which one would they check out first? On average, both monkeys and chimps were more likely to grab at a container that had been touched on purpose. Along with other experiments, this one suggested the monkeys shared some of the human ability to discern intentions; perhaps they possessed, too, the rudimentary “theory of mind” that’s considered a prerequisite for morality.
Hauser adapted many of his experiments from another lab in Harvard’s star-stuffed psychology building: that of infant-cognition researcher Elizabeth Spelke. Like him, Spelke (who’s about a decade older) made her bona fides searching for an innate perceptual module; she tested human babies to see whether they have an inherent sense of mathematical concepts, which she calls “numerosity.” To get at this question, she had to solve a tough problem: How can you tell what a baby knows or recognizes? Her experiments, drawing on work by Robert Frantz and Eleanor Gibson, used shifts in attention—where a baby was looking, essentially—as an indicator of whether he could distinguish one thing from another. Show a baby a regular stuffed rabbit six times, for instance, and he’ll stop paying attention to it. Then bring out a rabbit that’s the same in every respect, except it has four ears. If the baby stares for longer than he did at the two-eared rabbit, you’ll know he noticed the difference. Clever method.
These “preferential looking” studies produced wonderful results. Spelke used them, for instance, to show that while a 4-month-old knows a moving object will keep moving, it takes the mind of an 8-month-old to grasp the law of inertia and expect the object to hold a consistent path. She showed that 6-month-olds can tell the number 8 from the number 16, but not from the number 12. And so on. These early powers of discernment, she argues, reveal an innate sense of number. Similar studies have shown that babies know many other things—to use geometry to orient themselves, for example—far earlier than we’d suspected.
The Spelke studies—and the Hauser ones, too—pose one major problem, however: They’re laborious as hell. You must deal with many unpredictable test subjects in time-consuming trials while trying to ensure absolute consistency. Then you must watch the trials played back on a monitor and code the results—was that a quick look at the bunny there, or just a baby gazing about? And it’s best if the coding is done in a way that’s blind to exactly what the baby or monkey is seeing or hearing, so the coder can’t prejudge the response. All this takes many, many hours, of course, and creates additional room for error. And sometimes you spend weeks going through every step and find … nothing. Nothing new, nothing to report, nothing to publish.
This can prove horrifically frustrating for someone in a hurry. And Hauser, even by his own account, was a man in a hurry. This added to the thrill of watching him, of course: In just a few years, he appeared to show in animals the rough equivalents of what Spelke, Susan Carey, and Alison Gopnik had spent two or three decades establishing in humans.
Last month his fast success proved literally incredible. Yet the fault lies not in Hauser’s study paradigms nor even in his big ideas. (Others, like Spelke and primatologist Frans de Waal, have been wielding those to good effect.) I suspect we should blame instead his impatience—with the particular methods of his field, perhaps, but also with the slowness and uncertainty of science. In one instance of misconduct, he’s accused of bypassing protocols for watching and coding those dull films of one trial after another; as a result, he either saw monkey responses he desperately wanted to see or fabricated responses he didn’t see. In another, described in painful detail by Cognition editor Gerry Altmann, it appears Hauser simply made up the data altogether for one set of trials.
The ironies lie thick. One rap on Chomsky, for instance, holds that he didn’t much bother with experimental evidence; he simply said an innate grammar had to be there for kids to learn language so fast. It fell to others to poke around for those modules in the lab and produce some real data. People who study language, cognition, and evolution can and do argue over what those data mean. But at least they have something concrete to fight about. That’s what makes it science.
So give Hauser this: When it came to his theory of the moral grammar, at least the man wanted evidence. Problem was he wanted it bad.