|Some Marc Hauser work was retracted.|
Misconduct case taints research
Colleagues fear questions surrounding Harvard scientist will hurt their work
In the Harvard Psychology Department, faculty have been meeting to discuss how to remove the cloud created by the scientific misconduct case of one of their most prominent colleagues, Marc Hauser.
Elsewhere, a scientist is considering repeating a key experiment Hauser conducted on the behavior of monkeys.
A month after Harvard said it found Hauser guilty of eight infractions involving three published papers and other unpublished work, scholars in and out of the university are struggling with how to respond, and particularly with how to establish the reliability of the rest of Hauser’s large and influential body of research.
The uncertainty is not just an academic concern. In popular books, news stories, and television programs, Hauser drew people into deep scientific questions that spark the imagination. Can nonhuman animals tell what others’ intentions are? What cognitive abilities make us uniquely human?
Now, many scientists fear that because Hauser contributed so much to the public perception of not only his own work, but of a field that looks for the evolutionary underpinnings of human cognitive abilities, the questions about him will also cast a broader shadow.
Susan Carey, who chairs the Harvard Psychology Department, said that discussions about what can or should be done are in their earliest stages, and no decisions have been made.
“We are members of the department where this happened, and we have students — whose work is most probably impeccable, who are being tainted by guilt by association,’’ Carey said in an interview. “We are also scientists, who are very much hurt when science is hurt — when there’s academic misconduct.’’
Julie Neiworth, a psychology professor at Carleton College in Minnesota, is considering trying to replicate the Hauser-led 2002 study that has been retracted from the journal Cognition.
“I just feel like it’s sort of a crisis at this moment,’’ said Neiworth, who has 16 cottontop tamarin monkeys in her laboratory, the type used in the study.
The retracted study was one example of a thread of Hauser’s pioneering work, in which he showed that small-brained cottontop tamarin monkeys, which seem far removed from humans, have some of the same basic cognitive abilities. The research tries to unravel what makes the human mind unique.
Hauser was a pioneer in extending experimental methods that had been used on human infants to monkeys. In the Cognition paper, he showed that monkeys could distinguish between patterns of syllables, an ability that had been shown in babies and could be a building block of human language.
“Since that piece [of research] is important, and is definitely being thrown out at this point . . . I thought it might clean things up a little bit’’ to repeat the study, Neiworth said.
Scientists routinely repeat colleagues’ experiments to confirm and build on the findings, but because there are relatively few primate laboratories, and there is little funding or professional reward for being the second person to do an experiment, results in primate research are not always replicated.
The broad idea that monkeys and other animals possess some humanlike cognitive skills is not in question. Even if findings haven’t all been replicated in primates of the same species, they can be supported by showing that an ability exists in other animals — ranging from other monkeys or apes to animals further removed from humans, such as birds or dogs.
Scientists emphasized that while animal behavior research is difficult to conduct, there are precise rules for experiments that, if followed, ensure the findings are valid.
“There’s been this sort of suggestion that this kind of work is imprecise and it’s really anybody’s call what the animal is doing and it’s not rigorous. That’s a big misperception,’’ said Joan Silk, a primatologist at the University of California, Los Angeles.
Animal behavior is measured by more than one observer, and precautions are taken to ensure that their results won’t be influenced by their expectations. For example, an observer may not know what experimental condition the animal is in, or may not know what the hypothesis of the experiment is. If multiple observers do not report seeing the same behaviors, the data can’t be used. Experiments are videotaped, and tapes are made available to other researchers who request them.
How these checks and balances work is illustrated by an experiment that Hauser reported on in a 1995 Proceedings of the National Academy of Sciences paper, which looked at the ability of cottontop tamarin monkeys to recognize themselves in a mirror. That paper was criticized by some scientists when it was published, but is not part of the current misconduct findings.
In the experiment, multiple observers coded the data, and one observer did not know the experimental condition. In a videotape of the experiment, which was provided to the Globe by Gordon G. Gallup Jr., a psychology professor at State University of New York at Albany who requested the raw data from Hauser, monkeys sometimes look in the mirror. When this happens, Hauser can be heard saying “stare.’’
Lengthy bouts of a monkey staring into the mirror calmly, instead of acting aggressively toward its reflected image, were one piece of evidence Hauser used to show the monkeys passed the mirror test. But Gallup said he saw no evidence on the tapes of mirror-guided behavior.
Hauser eventually repeated the experiment, but did not get the same result. He published the failure to replicate the experiment in 2001, but did not retract or correct the original 1995 paper. Failure to replicate a finding is not evidence of misconduct — experiments occasionally produce an erroneous finding by chance. If an experiment is not replicated, it could be that the initial finding is not valid. Or it could be that some factor about the second experiment was different.
Experiments can be repeated but scientific research depends critically on honesty, and this is why Hauser’s misconduct is so disruptive.
“We basically have to trust one another,’’ said Charles Snowdon, a psychology professor at the University of Wisconsin-Madison, who ran a tamarin lab for three decades. “We can’t possibly monitor everything in every laboratory.’’
Carolyn Y. Johnson can be reached at firstname.lastname@example.org.