We have updates on the two mysterious Journal of Neuroscience retractions we reported on yesterday. One is that we have learned that there is a university investigation into the work of one of the teams that retracted one of the studies. More on that in a bit.
Two, the journal’s editor, John Maunsell, responded to our request for comment, and we’re quoting his entire email (with annotation) because we think it raises important issues:
Thank you for your question regarding our recent retractions. It is the policy of The Journal of Neuroscience to retract an article at the authors’ request at any time without requiring explanation. Our primary motivation is to minimize the barriers to retraction. We believe that authors are generally reluctant to retract articles, and we do not want to impose any requirements that could discourage authors from removing flawed articles from the literature.
We applaud any effort to minimize barriers to retraction. Where we seem to differ, as will become clear, is on how to lower those barriers while still serving science and transparency.
For the same reason, we do not publish descriptions of problems in a retracted article. Because retraction removes an article from the literature, eliminating it as a citable authority, describing which parts are or are not valid would be academic. There is no part of the retracted article that can be considered a valid, peer-reviewed observation. Scientists who have done research based on retracted observations would be better served by contacting the authors directly for complete details than relying on a brief published description of the most salient issues. By not identifying specific problems we also avoid the contentious and difficult practice of assigning blame for errors to particular laboratories or authors, which can also discourage retractions.
This has a certain logic, but we strongly disagree that scientists should be forced to take the extra step of contacting authors for complete details. Based on our experience, many of those authors are not forthcoming. And even if they were, isn’t it still better to make sure the scientific record is as complete and up-to-date as possible?
As far as the fact that assigning blame is “contentious and difficult,” we’re just not sympathetic. Most things worth doing are difficult. By shying away from reporting what actually happened, journals are encouraging misconduct, as we’ve argued before — and giving us far less reason to trust what’s in their pages.
We follow the guidance from the Committee on Publication Ethics (http://www.publicationethics.org/resources/guidelines) and distinguish misconduct from honest errors. Retractions made at the request of the authors are simply that. Retractions based on findings of inappropriate conduct are clearly indicated in the published retraction.
This we found particularly curious. It’s not our interpretation of the COPE guidelines, which clearly say that retraction notices should state
the reason(s) for the retraction (to distinguish misconduct from honest error).
It appears that the Journal of Neuroscience thinks this means that “this article was withdrawn at the request of the authors” is enough reason.
Again, we disagree. For one, authors can certainly retract because of misconduct. And as we’ve noted above, while we applaud anything that is an attempt to remove barriers from retraction, we fail to see how an uninformative notice fosters transparency or helps science. Perhaps COPE can weigh in here, by leaving a comment.
In one of these very cases, in fact, we have learned that there appears to be a university investigation into the work of one of the co-authors, Kenji Okajima, according to a report in Sankei Shimbun. The newspaper does not mention Okajima by name, but gives details about his age, area of research, and positions that make it clear it is him and a colleague whose work is being examined. Both of those authors are on the paper retracted this week.
That investigation seems to have been sparked by questions on this site, which does mention both authors’ names. (Both of those links are in Japanese, and we had help from a Retraction Watch reader in translating them.)
Okajima — who has also apparently studied the effects of wasabi and pepper on growing hair — has not responded to our requests for comment. We should also make clear that we have not learned anything at all about the reasons for the other retraction, as the lead author has not responded to us.
Back to Maunsell’s email:
In this regard I would like to point out that your web site, Retraction Watch, may be doing science a disservice by failing to emphasize that retractions fall into two very different categories: those initiated by authors to remove articles containing errors, and those initiated based on an investigation that finds fabrication, falsification, plagiarism or other unethical behavior. In searching your web site I could find no material that lays out this distinction for visitors. I worry that scientists visiting your site will be led to believe that all retractions arise from misconduct, which is far from the truth. By failing to make clear that retraction is an important mechanism for correcting honest mistakes, your site may imply to scientists that they must avoid retractions at all costs. You might greatly enhance and extend your site by providing your readers with general information about ethical behavior in scientific publication (there are many suitable links) and especially by emphasizing that there are different types of retractions.
This is a criticism that we have heard before, most recently last week from DrugMonkey, who posted a poll on what a retraction means in response to a Retraction Watch post asking what warrants a retraction. DrugMonkey’s feeling is that retractions should be reserved for fraud — an opinion with which we (and COPE, and many journals) respectfully disagree.
Still, as we wrote to Maunsell:
We appreciate all feedback and criticism, and we’ll take your suggestion to explore ethical issues in that vein. We should point out, however, that we routinely write about such issues. In our very first post, we note that fraud is the most rare reason for retraction.
We also praise journals and scientists who retract for honest mistakes, and tell the whole story. Just two examples.
With all due respect, we would say in return that it is opaque and frankly unhelpful retraction notices such as the two published this week in your journal that are doing science a disservice.
In other words, we believe the way to make science more transparent and self-correcting is to explain why a paper is being retracted. If lots of notices explain that there was no misconduct or fraud involved, retractions would be normalized, and we’d lower the barrier to them — which seems to be a common goal of ours and the Journal of Neuroscience. Others have suggested that perhaps corrections are warranted when fraud isn’t involved. We’d be happy to explore that, too.
But what we can’t understand is how uninformative retraction notices help anyone at all.
You write “…By shying away from reporting what actually happened, journals are encouraging misconduct…” But that assumes the medical journals are gatekeepers. They may think they are, and they should be… but they’re usually not. They’re basically fee-for-service bulletin boards for researchers who have a story to share — a story about what they’ve been doing in a lab or clinic. The publishers, editors and staffs of most medical journals don’t generally report anything. At best, they provide some suggestions (from their peer reviewers) and some proofreading services. They’re more like “journals” in the sense of “diary,” not in the sense of “journalism.” “Foster transparency” is unfortunately not in the medical journal job requirements.
I had to smile at John Maunsell’s assertion that “…describing which parts are or are not valid would be academic”. It is rare to come across the argument that academic journals are no place for academic arguments! But, seriously, as an editor, I put quite a bit of work into a paper, managing the reviewing process and helping to improve the quality of communication. I am not very keen on striking from the record a paper that has been published in my journal, unless there is very good reason. An author who requests a retraction from me will need to justify this with a careful explanation, and I shall publish that explanation as a formal retraction, one way or another. Not only should the decision to retract be taken with due care and consideration, it should be evident to all users of the journal that there are reasons, and that those reasons are indeed academic in the sense of upholding academic values, not merely in the sense of being practically irrelevant.
academic journals are no place for academic arguments!
Gentlemen. You can’t fight in here. This is the War Room!
Maunsell’s criticism of your site strikes me as disingenuous. Anyone with more than a passing interest in retractions knows full well that “retractions fall into two very different categories.” Their “opaque and frankly unhelpful retraction notices” give no information as to which category a retraction may fall into. It is those notices which leave us uninformed, not your site.
I’m not sure I agree that retractions should be easy, if that means they can be opaque.
If you want to publish a scientific paper, you have to spell out exactly what you did and why you arrived at your conclusions. I don’t think retractions should be any different. You can’t just say “Sorry – forget we ever said that. The published results may, or may not, have been wrong. We’re not saying either way.”
The scientific community, who read your paper (and your other papers) have a need to know what went wrong. The main reason being that we need to know what the real results were.
As an author of the COPE guidelines, I’d reiterate Ivan’s interpretation. I’m delighted that J Neurosci cites our guidelines (even though I don’t think it’s a COPE member) but disagree with their reading of them. I have heard of authors seeking to retract papers just BEFORE the results of an investigation were published to ensure that the retraction didn’t mention the misconduct. For this reason the COPE guidelines recommend that editors wait for the results of investigations (and maybe publish an Expression of Concern in the interim). And, no matter who retracts the paper, we recommend that the reason should be stated — largely to protect innocent authors who do the right thing and retract papers when they discover honest errors, who should be encouraged.
I think that it’s worthwhile to review the MEDLINE policy on retractions and partial retractions. http://www.nlm.nih.gov/pubs/factsheets/errata.html#retractions If the policy on partial retractions were followed then Maunsell’s concern that “There is no part of the retracted article that can be considered a valid, peer-reviewed observation.” would no longer apply.
MEDLINE doesn’t set policy on retractions — it’s just a bibliographic database. They mark an article as retracted if the journals tell them it’s retracted.
I’d like to make one further point. Let’s say there are two types of retractions: A, for honest mistakes; and B, for fraud or misconduct. J Neurosci is saying that by making retractions opaque, they are “lowering the barrier”, and will make retractions more frequent. I doubt it. For the (A) crowd, opaque retractions will make them less likely, because if no reason is given, then the public will generally wonder and always have doubts, as to whether there was misconduct. For the (B) group, I just have a gut feeling that most of them would resist retracting their papers regardless.
For me, a retraction without explanation pushes me to believe it is due to fraud/manipulation. If people are not forthcoming of the reason, they must have something to hide. Since we all make honest mistakes, embarrassing mistakes at times, there’s no reason to hide what the mistake was.
Everyone I know assumes that a short retraction without explanation is due to fraud.
In response to a 2011 Journal of Neuroscience article by Eric Leuthardt’s group, a letter was written to the editor by 4 individuals from different labs around the country, detailing a multitude of errors related to: faulty signal processing, lack of correction for multiple comparison, and masking images with semi-transparent white boxes(!). These fundamentally undermined the conclusions of the article. Although Leuthardt’s group attempted a response, the fact remains that the conclusions aren’t supported by their data, their statistics are flawed, and they masked their images (which is against the policy of every major journal in the world). Yet J Neurosci does not appear to have made an effort to retract the article (it continues to be cited with ~50 citations thus far). It is should very clearly be retracted.
My opinion is that J Neurosci has become a glamour journal, favoring story over science, and Maunsell is the one that made it that way during his tenure as editor-in-chief.
The Leuthardt article is here:
http://www.jneurosci.org/content/31/6/2091.short
The letter is here:
http://www.jneurosci.org/content/31/6/2091.short/reply#jneuro_el_85967