Scientific Misdeeds

by by Nupur Shridhar

In August, after three years of internal investigation, Harvard University found Dr. Marc Hauser, a renowned evolutionary biologist (and, perhaps ironically, author of Moral Minds: How Nature Designed Our Universal Sense of Right and Wrong) “solely responsible” for eight counts of scientific misconduct. The resulting government investigation, which may take up to eight years to resolve, has revealed fundamental problems about the way scientists examine behavior—especially their own.

Hauser’s controversial work focuses on the linguistic capacities of cotton-top tamarin monkeys. In a paper published in Cognition in 2002, Hauser and his team concluded that his animal subjects could identify a novel sound from background chatter. Their experiment involved habituating infant monkeys to one of two consonant-vowel sequences—AAB (“wi wi di”) or ABB (“le we we”)—which served as controls. Using this model, Hauser concluded that the monkeys turned their head toward the speaker when they heard an unfamiliar test noise. This readily identifiable response to new stimuli proved, as Hauser sought to, that the ability to distinguish sounds is the fundamental evolutionary talent that allows for the primate’s intimate relationship with language.

But a student whistleblower working under Hauser revealed problematic holes in the experimental data: the lab tapes only show monkeys responding to novel sounds; there are no tapes of the controls, which are essential in proving any hypothesis. Without these baseline values, there’s no reason to believe Hauser’s incomplete conclusions. So did Hauser fabricate his controls (criminal) or simply forget to include them in his data (stupid)?

Without access to the investigation’s evidence, it’s impossible to do anything but hypothesize about what did or didn’t go wrong. Perhaps it was the double-blind nature of the experiment itself. The researchers, including Hauser, didn’t know what the monkeys in the soundproof booths were hearing because the noises were randomly generated by computers, so the controls could be missing or mislabeled. Or perhaps it’s the fact that Hauser and his students interpreted the monkeys’ movements differently. It takes many years of practice to understand and codify the subtleties of primate behavior—observers may understand movements differently. Of course, it doesn’t help that, just like every other animal, some tamarins are just smarter than others.

Consider Clever Hans, the horse who enthralled audiences in the early 1900s with his ability to tap out the answers to simple arithmetic problems. But in 1907, psychologist Oskar Pfungst proved that Hans wasn’t a math prodigy—he was just so good at reading his trainer’s body language that he could tell when he’d tapped his hoof enough times. Hans was brilliant, it turns out—just not in the ways his trainers had predicted. As Brown University evolutionary biologist Dr. David Rand notes, “It’s notoriously difficult to test these sorts of things. Behavioral study is fraught with complicating variables, and you can never really know what the animals are thinking.” Unsurprisingly, Hauser’s incomplete records are likely unable to account for all the possible sources of error.

But perhaps the problem derives more generally from science’s dependence on statistical manipulation. In order to be taken seriously (and to receive funding) every scientist needs to show that results are significant. Yet different statistical tests are more or less sensitive to irregularities in data. Some require high levels of correlation to prove significance; Hauser’s students have accused him of pressuring them to use tests that admit significance very easily.

What’s at stake, then, is the very validity of the theories we’re told to accept as fact. Research is essentially a numbers game, and the continual quest for significance reveals the range of possible errors worked into each ‘truth.’ Consider manufactured drugs, which have different efficacies in each user. Some medicines are for children, others work best in men, and others, like BiDil, a heart disease medication for “self-identifying African Americans,” may only work in individuals with a particular set of genes. For this reason, it’s possible to imagine how Hauserian manipulations of data could have lethal effects in pharmaceutical research—but it’s just as problematic to make assumptions about how and what we’ve evolved to think. As Emory University primate researcher Dr. Frans de Waal told USA Today, Hauser’s misconduct “leaves open whether we in the field of animal behavior should just worry about those three articles or about many more, and then there are also publications related to language and morality that include data that are now in question.”

It’s worth deconstructing this relationship between language and morality because, even if Hauser is an unethical scientist, his book’s conclusions on language and its importance to humankind seem to ring true. As Rand points out, what makes the scientific study of morality difficult is that natural selection has little to contribute to notions of good or evil. “Selfishness in animals, like a lion killing its prey or competing with other lions for mates, isn’t viewed as immoral, and people who are immoral or even just amoral aren’t genetic deviants. Just like with every heritable trait, it’s an issue of how much of the variation in moral behavior has to do with genetic variation as opposed to environmental variation—things like, did you have doting parents?” By this line of reasoning, evolutionary biologists have little authority to tell us who’s good and who’s bad.

Yet Hauser’s premise in Moral Minds is that language itself is a highly evolved organ that’s been selected for like our large brains and opposable thumbs. Only a handful of species have created vocabularies, and out of all of them, humans certainly seem to enjoy talking the most. Language, then, becomes the evolutionary trait that allows us to communicate our pain and pleasure and to negotiate these needs and emotions socially. What’s evil causes societal pain and what’s good increases societal pleasure. By extension, only by learning our cultural vocabularies—the words we charge with positive connotations (strong, nurturing) as well as the words some cultures leave out (homosexuality, interracial)—do we develop a sense of morality. Science certainly has the ability to study both the origins of language, through experiments like Hauser’s, and the kinds of neurological and hormonal circuits that are responsible for our experience of emotion.

As neuroscientist Sam Harris emphasizes in his TED talk, science can answer moral questions. “There are truths to be known about how human communities flourish, whether or not we understand these truths. And morality relates to these truths…I’m not saying science is guaranteed to map this space, or that we will have scientific answers to every conceivable moral question. I don’t think, for instance, that you will one day consult a supercomputer to learn whether you should have a second child, or whether we should bomb Iran’s nuclear facilities…but if questions affect human well-being then they do have answers, whether or not we can find them. And just admitting this—just admitting that there are right and wrong answers to the question of how humans flourish—will change the way we talk about morality, and will change the expectations of human cooperation in the future.”

It’s likely that in a few months, Hauser’s case will be long resolved and forgotten. Right now, however, it’s an opportunity to learn more about how difficult it is to confirm and replicate scientific results. Despite the rigorous peer review process and governmental standards for behavioral research, there is much more work to be done before scientists can truly understand animal behavior objectively, and what, even, it means to behave.

NUPUR SHRIDHAR B’11 looks forward to Hauser’s next book – Evilicious: explaining our evolved taste for being bad.