What's lost in translation from research to headlines

posted Oct 13, 2009, 12:54 AM by Charles Henderson   [ updated Apr 22, 2012, 3:57 PM ]
 When science meets media
What's lost in translation from research to headlines
Published on August 10, 2009 by Alexandra Horowitz, Ph.D. in Minds of Animals

All of us working in the behavioral sciences face a similar paradox when our research gets featured in the traditional media (i.e., not an abstruse journal read by forty of our colleagues): while it is surprising and delightful to have attention paid to our work, we may feel some apprehension too, since misunderstandings about the work fly fast and furious.

When one's research subject is as familiar a creature as the domestic dog, the fast-and-furiousness is upped a notch. I was reminded of this last week, when I read an article in the New York Times about recent genetic research on African village dogs. "(T)heir samples, "the paper reported, "...have called into question a finding on the origin of dog domestication from wolves."

Wow! I was stunned. Dogs are not in fact ancestors of wolves (or a wolf-like canid)? The most accepted story of dog domestication is that a split from wolves occurred anywhere from 14,000 years ago (as far back as archeological evidence goes) to even 145,000 years ago (per mitochondrial DNA samples). Most researchers believe that dogs began loitering near early human settlements, scavenging, and being tolerated by the human population (and sometimes eaten by them). More tame than their forebears, those who were most agreeable to the local humans may have been allowed to live, and reproduce...and after some time, we began breeding them ourselves to suit us.

Reading on in the Times, I picked up a skeletal outline of the study. A Cornell researcher named Adam Boyko and his colleagues sampled the genetic diversity of African village dogs, and found that they were as diverse genetically as a previous sample of East Asian dogs. That high level of diversity was considered to be argument for a Eurasian origin of dogs.

I did a little scanning of other news reports. Discovery News added, "Modern humans originated in Africa, and now it looks like man's best friend first emerged there too."

Then I looked up the paper in the Proceedings of the National Academy of Sciences. For anyone not comfortable with mtDNA, microsatellite, and SNP marker genetic measures, it is not an easy read. But the authors' stances on the above points are clear:

On the matter of whether they have "called into question a finding on the origin of dog domestication from wolves", here's Boyko et al: "Dogs were probably domesticated from wolves". Hmm.

How about "now it looks like man's best friend" emerged from Africa? Boyko et al.: "we do not suggest that Africa is actually the site of dog domestication." In a quote in a much better National Geographic article, he elaborates: "...because there are no gray wolves there [in Africa]."

What has happened here? The same kind of thing that happened when a study of mine, on the prompt for the familiar "guilty look" in dogs (discussed more in this post), got picked up on the wires. Many reporters correctly interpreted the results (some by reading the actual journal piece!, bless them, and not just the press release), which were that "disobedience" didn't arouse any more guilty looks than "obedience" did. Instead, owner scolding was the prompt to the biggest "guilty look". That is, the guilty look may be a learned response to cues from the owner that the dog is about to be, or is being, punished.

Some news reports, however, would have you believe otherwise. " 'Guilty look' in dogs mostly owners' fantasy, study finds", one headline declared. Others claimed that the guilty look was either "a myth", "all in your head", or "the owner's fault". My favorite came from FOX News, who trumpeted: "Dogs don't feel guilt at all".

I got plenty of emails from indignant dog owners: how could I say that dogs don't have a guilty look? Here's a picture! Or informing me that their dogs clearly felt guilt, so I should reconsider my results...

Given that I was using the very existence of the guilty look as a premise for my study, the claims of its mythic status were simply wrong. And I also say nothing in the piece about whether dogs can or do feel guilt: I was just studying what prompted the look (which does give many owners the impression that dogs feel guilt...perhaps inappropriately).

In the case of the domestication study, most news articles I saw got it right. But so much research relies on nuances, well discussed in the journal articles but overlooked in summaries of the work, that small misrepresentations constantly occur. Often these snowball, as in the guilty-look study, into ridiculous claims. (I kept waiting for the climactic "Dogs have no conscience at all!" headline...)

It is too much to ask everyone to read every journal article before forming a judgment of what the research is about. And that is what the many very good science journalists do well: they distill the essence from the article and make it readable to a non-expert. But do take what you read with a grain of salt.

Even this post.