Former Harvard professor Marc Hauser fabricated, manipulated data, US says
Marc Hauser, a prolific scientist and popular psychology professor who last summer resigned from Harvard University, had fabricated data, manipulated results in multiple experiments, and described how studies were conducted in factually incorrect ways, according to the findings of a federal research oversight agency posted online Wednesday.
The report provides the greatest insight yet into the problems that triggered a three-year internal university investigation that concluded in 2010 that Hauser, a star professor and public intellectual, had committed eight instances of scientific misconduct. The document, which will be published in the Federal Register Thursday, found six cases in which Hauser engaged in research misconduct in work supported by the National Institutes of Health. One paper was retracted and two were corrected, and other problems were found in unpublished work.
Although Hauser “neither admits nor denies committing research misconduct,” he does, the report states, accept that federal authorities “found evidence of research misconduct.”
In a statement, Hauser described the scrutiny of the past five years as a “long and painful period” and apologized to all who had been burdened by the investigation.
“Although I have fundamental differences with some of the findings,” Hauser wrote, “I acknowledge that I made mistakes. ... I let important details get away from my control, and as head of the lab, I take responsibility for all errors made within the lab, whether or not I was directly involved.”
Hauser agreed to a number of restrictions for three years, including excluding himself from serving as an advisor to the US Public Health Service, agreeing to have his work supervised if it is supported by federal funding, and having an institution vouch for the validity of any research supported by such grants.
According to his LinkedIn profile, Hauser now works at the Alternative Education Program at Cape Cod Collaborative, providing educational opportunities for at-risk youth.
“I am relieved that this investigation is now complete, allowing me to turn my full energy to the next chapter of my career,” Hauser wrote. “I look forward to making new contributions to human welfare, education, and the role of scientific knowledge in understanding human nature.”
The finding outlines a wide gamut of unethical scientific practices, which Harvard investigators detailed in a confidential document forwarded to the federal Office of Research Integrity, an arm of the Department of Health and Human Services. The agency also conducted its own additional analysis, according to the finding.
“What is the new information are the details of how many monkeys, or how many data points appear to have been manipulated or fabricated,” said Gerry Altmann, the editor of the journal Cognition and a psychology professor at the University of York said in an interview after reading the report Wednesday. “I don’t know what would happen if they were to explore more than the three studies that were published,” and the other unpublished papers mentioned in the report.
The report describes research misconduct in work supported by four National Institutes of Health grants, including scientific papers that were published, submitted for publication, and experiments that were never published. According to the federal findings:
-Hauser fabricated data in a 2002 Cognition paper that was later retracted, which examined monkeys’ ability to learn patterns of syllables. He never exposed monkeys to a particular sound pattern described in the experiment, despite reporting the results in a graph.
-In two experiments, researchers measured monkeys’ responses to patterns of consonants and vowels, a process called “coding” their behavior. Hauser falsified the coding, causing the results to pass a statistical test used to ensure that a particular finding was not just a chance result. Colleagues coding the same experiments came up with different results. Hauser “acknowledged to his collaborators that he miscoded some of the trials and that the study failed to provide support for the initial hypothesis,” the report said.
-A paper examining monkeys’ abilities to learn grammatical patterns included false descriptions of how the monkeys’ behavior was coded, “leading to a false proportion or number of animals showing a favorable response,” the findings stated. In an early version of the paper, he falsely reported that all 16 monkeys responded more strongly to an ungrammatical pattern than a grammatical one. Records reviewed by investigators found that one monkey responded in the opposite way and another responded equally. Hauser claimed that the behavior was coded by three scientists, when in fact he was the only one who measured their behavior. Then, when the manuscript was revised, he provided a false numerical description of the extent of agreement among multiple observers in coding behavior, despite being the only observer. All issues were corrected before publication.
-In a published experiment that examined monkeys’ responses to gestures, Hauser incorrectly reported results and also falsely said all trials were videotaped, when only 30 out of 40 were found for one trial. Hauser “was not responsible for the coding, analyses, or archiving but takes full responsibility for the falsifications reported in the published paper,” the report said. The experiments were later repeated, confirming the results.
-In another published experiment, the primates were identified with natural markings, tattoos, or ear notches to avoid re-testing the same animal, but only half the animals could be distinguished with such measures. The experiment was repeated and published, upholding the original finding.
-After initially coding rhesus monkeys’ responses to strings of sounds, Hauser and a research assistant discovered a problem with the coding procedures. Hauser recoded the behaviors, and the new coding differed in 36 cases from the original, nearly all of them in a way that would produce a result that was significant. Hauser “subsequently acknowledged to his collaborators that his coding was incorrect and that the study failed to provide support for the initial hypothesis,” the report said.
The problems came to light two years ago when the Globe reported that Hauser had sent letters to his colleagues informing them that a three-year investigation into his lab had found evidence of misconduct and that one paper would be retracted. Hauser took a leave of absence and, after colleagues voted to bar him from teaching in the psychology department, he resigned. But many scientists and colleagues have been waiting for the federal finding in the hopes it would elucidate more clearly what Hauser did wrong and perhaps help explain whether the problems cast a shadow over the rest of his body of work, which includes more than 200 scientific publications and collaborations with leading figures in diverse fields including evolutionary biology and linguistics.
Altmann said that Hauser had made positive contributions to his field, but that the shortcuts described in his experiments were unacceptable. Informally, he said, the field now recognizes some of his findings -- such as the one that was retracted from the journal Cognition in 2010 -- as unlikely to be successfully repeated, but no formal investigation is planned of his vast body of work. In that 2002 Cognition paper, Hauser and colleagues had found that cotton-top tamarin monkeys have the ability to learn patterns of syllables, a skill that had been seen in infants that was thought to play a role in the ability to learn language.
Hauser probed the evolutionary roots of human abilities such as language and studied whether morality was innate or learned -- questions that piqued the interest not only of scientists but of the general public. He wrote popular books and his work was frequently featured in the media. He was known for his interdisciplinary approach, running a laboratory with cotton-top tamarin monkeys but also collaborating with colleagues who studied infants and posing moral conundrums to people over the Internet.Carolyn Y. Johnson can be reached at firstname.lastname@example.org. Follow her on Twitter @carolynyjohnson.