Nobel prize-winning lab retracts paper from Nature
Noah Gray
Monday, 10 March 2008 22:16 UTC
We discussed this topic on the Nature Neuroscience blog, Action Potential (and further discussed here), but I wanted to bring the discussion into this forum as well.
Briefly, a paper published by Linda Buck in 2001 was recently retracted in Nature. Now, a retraction always turns a few heads, especially when it involves a high-profile publication, and ESPECIALLY when it involves the work of a Nobel prize-winning scientist. But what makes this particular retraction even more interesting is the fact that an “Author Contributions” abstract was included by the authors with the retraction. It makes it pretty clear that the other authors place most of the blame for the irreproducibility on the first author.
Now, obviously, The Buck lab is doing the right thing by retracting a paper containing results that they can no longer reproduce, but it seems a bit odd to isolate one author when there was actually a “co-first authorship” on the publication. So the other first author should get the glory, but be protected from the retraction wrath? What do you think? Discussion topics about retractions in general:
1. For a retraction, should dissenting authors be given the opportunity to respond to the allegations against them before the retraction is published?
2. Continuing from the previous point, in that case, should the results and “case” of each side be presented for peer review before deciding on the retraction?
3. If your name is on a fraudulent manuscript, are you as equally liable as the perpetrator in the retraction?
4. It is obviously not practical for a productive lab head to review every single data point for every single figure in every single publication…or is it? In other words, how much trust should a PI afford to his/her students and post-docs?
Updated 12 March 2008 18:20 UTC
-
Replies
Jump to resultsResults
-
I’d be surprised if the first author was not fully aware of the situation and I’d be doubly surprised if Buck and co. had not in fact done everything in their power to resolve the matter with the first author before publishing the retraction. I’d be even more surprised if the first author had no idea that the retraction was coming.
If any of these expectations are not the case then Nature is journalistically required (is it not?) to provide a platform for the first author to defend himself. I have the confident expectation that Nature would do this and that this is the expectation of all involved.
That matter aside, the question of further responsibility remains. Obviously, a PI must trust their students and colleagues or the entire process is in jeopardy. It’s tempting to castigate the PI and co-authors by arguing that they must be more selective or more attentive, but one wonders just how practical such an expectation and criticism is in the face of real fraud.
The best questions must relate to how they behaved when the discovery was made and why it took so long. I’d like to know what is being done to review the other literature from this author and any research that has relied upon it.
-
The problem is people have forgotten what the point of publishing is. The point of publishing is to share results, so that other people can point out flaws in the work, or if there are no apparent flaws, then at least others can build on the work to generate new better ideas. Communication is what allows science to move forward. In a way, every publication is a challenge to the scientific community: ‘Prove us wrong’. When someone does that, including the authors themselves, it should be celebrated, because it means the self-correcting scientific process is working. Unfortunately, papers have become career points, and papers in prestigious journals have become status symbols. So corrections & modifications have been twisted into badges of shame rather than signs of good hard work, and leading scientists become defensive rather than revolutionary. That’s stupid. It’s great that the Buck lab is setting the record straight. Kudos to them for sharing what they saw in 2001, and kudos to them for sharing what they are now seeing in 2008.
My opinions on the discussion topics:
1. Should dissenting authors be given the opportunity to respond to allegations before a retraction is published? Of course. That helps figure out what went ‘wrong’. What is the point of a retraction unless there is some explanation for why the results differ? If it’s a technical mistake, then readers need to know so they can avoid it themselves. If it’s fraud, then readers need to know what the true results are in case the conclusions of the paper should be modified. If there is simply some weird unexplained variability in the assay that makes it unreliable, then readers need to know so that people working in that field can avoid drawing conclusions from that assay.
2. Should the results and ‘case’ of each side be presented? This sounds like a criminal court. Perhaps there is no ‘guilty party’, only an honest mistake. Certainly there are cases of outright fraud, and they should be dealt with accordingly. But maybe an hourly tech simply screwed up a reagent recipe that a hapless author relied upon for all his experiments. Maybe only some results are questionable. In those cases, the whole paper need not be retracted. The authors could simply revise their conclusions taking into account the new data. Most fields have controversies, suggesting our understanding of things is incomplete. That’s fine; that’s why we’re doing the research! Again, the idea that any paper’s conclusions are carved in stone is stupid, and belies a fundamental misunderstanding of how science works. Everything is simply ‘to the best of our knowledge’, based on the data we have. The whole point of science is to change and add to what we think we know about how the world works.
3. Are co-authors liable for fraud they didn’t commit? No, of course not. Why should we hold innocent people accountable for something they didn’t do? The reviewers and editors are supposed to be the ‘gatekeepers’ of published science. Should they be held responsible too?
4. Should a PI trust his lab members? Of course. Science is built on trust. PIs trust their lab members to honestly report what they see, Reviewers and editors trust authors to honestly report their results. The scientific community trusts that published results have been reported accurately, and carefully and fairly scrutinized by reviewers and editors before publication. Here is the sort of mistake that PIs SHOULD be held accountable for making: PIs are guilty when they do not report mistakes when they know or suspect they have been made (e.g. not retracting a paper if it’s unreproducible). PIs are guilty when they encourage a dishonest competitive scientific environment (ever hear of someone telling lab members not to share data before meetings, or sabotaging competitive grants or papers as a reviewer, or stealing ideas? This is unforgivable PI behavior). It is up to every PI to set the highest standards for science quality and community spirit. But it’s also important to share results in a timely fashion. So sometimes mistakes will be made. That’s fine, as long as they get corrected. In that regard, Dr. Buck should be applauded. It can’t be easy to retract a Nature paper. But getting things correct as best we can and sharing is what science is all about. Right?
-
David is right about the process.
I note that the first author sent an email denial to The Scientist.
-
Note also comments by the challenged author “ZZ” here on Nature:
http://www.nature.com/news/2008/080305/full/452013a.html#comments
Since his entire program appears to rely on the retracted paper and other apparently confirming results he appears to be greatly challenged. Is this perhaps a case of a scientist assuming the nature of the result and “finding just what they were looking for?”
-
Here is that Nature news link again
As Steven says, there are some pretty strong comments there. -
Some great points, many of which I agree with, but would like to add a few things.
I guess I found it to be an interesting thought experiment as to whether a retraction should wait until the dissenting author(s) have had a chance to reproduce the data themselves. The comments on the Nature news feature by the first author himself (ZZ) are very telling. He did sign off on the retraction, but still believes the data. It sounds like this is a case where the lab of LB can no longer produce the data, and therefore moved forward with the retraction. While commendable, it would be nice to at least have some sense of what experimental attempts were made by ZZ or LB to re-create the data.
I completely agree with Dave regarding the way that science should work. Studies are proven wrong all the time, but that doesn’t mean that the previous paper should be retracted, just that the methodology at the time did not allow for an accurate measure to be taken. Retractions are usually reserved for fraud and accidental mistakes. These are considered badges of shame, and at least for the former, should be. As for the latter, it is definitely embarrassing to publish something that was based on a poorly-made buffer, and although not entirely the fault of the researcher, still suggests an air of carelessness.
The reason why my eyebrow raised for this retraction was because of the unusual inclusion of the author contributions section. I looked back through dozens of retractions and couldn’t find anything like it. These authors must have had a reason for doing so, and unfortunately, it seems that the purpose may be to point fingers. Based on the openness that ZZ is presenting, I doubt that anything nefarious went on here, but it is disappointing to see the other authors place an arm’s distance between him and themselves.
-
As I read it external researchers were unable to reproduce the results, the Buck Lab looked back at the results and found “inconsistencies between some of the figures and data published in the paper and the original data.”
This needs to be clarified because it does state that the original data was misrepresented, not that the interpretation of the data was false.
There is little justification, therefore, for ZZ to repeat the experiments or to claim that his error had just cause.
-
I agree that it needs to be established whether or not fraud took place. The answer to that question colors the entire debate.
-
Although I obviously applaud the willingness of the authors to retract the paper (as others have mentioned: if retractions were treated less as a badge of shame and more as a ‘correction to the literature’, more authors might do it when appropriate), I also agree with comments here implying that the retraction’s wording is a bit strange and not particularly helpful for moving the field forward. As Steven said, it would be helpful to know exactly which experiments could not be replicated, and what the new results imply.
Perhaps editors should encourage authors to detail exactly why they are retracting a paper. Here is an example of a more helpful retraction (Current Biology 17(19):1720, Oct 9 2007): http://www.current-biology.com/content/article/fulltext?uid=PIIS096098220702009X
Thus, a retraction might be more of an ‘update’. Maybe journals should even use that term: ‘update’, in order to remove some of the stigma of providing one.
-
An ‘update’ to my post above:
In reviewing my previous post, it has come to my attention that the phrase “As Steven said, it would be helpful to know exactly…” should also have included an attribution to Noah. Thus, the sentence should read: “As Steven and Noah have pointed out, it would be helpful to know exactly…”
It is not clear exactly how this error occurred, but I regret any inconvenience that it may have caused to readers of this forum.
Results
-