University of Lisbon investigation that spawned neuroscience retractions found no evidence of misconduct
Yesterday, we reported on two retractions in the Journal of Neuroscience whose notices referred to a University of Lisbon report that had determined there was “substantial data misrepresentation” in the original articles. The notice didn’t say anything about misconduct, but when we see “misrepresentation,” we tend to think — as do many others — that there had been funny business.
But we heard back this morning from the senior author of the study, Ana M. Sebastião, and there’s a lot more to this story. It turns out that the University of Lisbon committee that wrote the report concluded, unanimously, that
the misrepresentations of the figures that were detected by the Editor of the Journal of Neuroscience in the two articles in question did not result from misconduct, i.e. these faults cannot be attribute to intentional fabrication or falsification. On the contrary, these errors seem to have occurred in good faith and in no way did these affect the scientific content of the articles and their conclusions.
The committee — full report available here — recognized
that these misrepresentations resulted from a careless action of 2 PhD students [Natalia Assaife-Lopes and Sofia Critovao-Ferreira] who admitted that at the time of preparing the final manuscripts had to work under great time pressure. Further the Committee acknowledges that the senior author (AMS) should take the main responsibility for not having detected the mistakes that occurred in the figures in both articles. All authors assumed their share of responsibility.
Here’s how the April 2012 report describes what happened:
In the case of Figures 1a and 4a of the paper by Assaife-Lopes et al. (2010), where the same images of a Western blot are repeated for two different conditions, the 1st author admitted that this occurred due to faulty “cut and paste” of very similar pictures. In the case of Figure 4f of the paper by Cristovao-Ferreira et al. (2011) the Western blots of Tubulin that were repeated resulted from using the selected part of the blot twice in the final figure. The Committee members examined the original blots to better understand what went wrong. This fault occurred at the last moment after the manuscript had been revised and was being prepared to be resubmitted. Apparently this was done under great time pressure to re-submit the manuscript. Most co-authors received this last version at the same moment as it was being sent to the Journal, and thus did not see the very final version before resubmission.
We’ve made the original Cristovao-Ferreira et al figure 4 available here, along with the corrected version here, so readers can judge the errors for themselves.
The committee recommended that the journal publish errata for the two studies — a move Sebastião had actually proposed in November 2011, months before the journal’s publisher, the Society for Neuroscience (SfN), asked the university to investigate in February 2012.
They also recommended that SfN rescind a moratorium it had placed on the University of Lisbon authors that would prevent them from publishing in the Journal of Neuroscience and presenting abstracts at the group’s annual meeting. (The moratorium did not apply to Moses Chao of New York University, one of the authors on one of the papers, which might have been a bit awkward, considering that Chao is immediate past-president of a group called the Society for Neuroscience.)
The moratorium was lifted on all authors except Sebastião, who will face a ban until September 2014:
The argument was that as corresponding author it was my duty to carefully check figures in the papers, which is true and is indeed written in the Journal rules. I do not contest that, rules are rules, though I consider the time period a bit excessive for errors that are very hard to detect by eye. As supervisors usually do, I checked quantifications against blots during the course of the investigation, but since the errors have been made while setting up the final Figures, due to a lack of attention of the first authors, they escaped to my eye while doing a final revision of the MS.
Now, it would be one thing if the Journal of Neuroscience didn’t believe the University of Lisbon report. After all, there is a conflict of interest when an institution investigates one of its own. But in an email exchange late last year with Sebastião, in which she tried (obviously unsuccessfully) to convince him to run errata instead of retract the papers — even offering to fly to the US to go through the data with him — John Maunsell, the editor of the journal, wrote:
We accept without reservation the conclusion from the investigation of the University of Lisbon that the errors were honest and that there was no misconduct. The retractions are based on the number and nature of the errors that appeared multiple articles, which have left the editors unprepared to stand behind these reports.
So why wouldn’t the journal include the four-word phrase “there was no misconduct” in the notice? And why does a ban make any sense, if such bans are about intent? We’ve taken issue with how the Journal of Neuroscience handles retractions before, and this unfortunately doesn’t instill any confidence.