A paper about the potential influence of neurotransmitters on the development of sexual orientation and psychiatric disorders that caught flack on social media a year ago has now been retracted – so recently that the corresponding author said he didn’t know about the retraction until we asked him about it.
Late last year, Neuroscience & Biobehavioral Reviews, an Elsevier journal, published an expression of concern for the article “Sexual orientation, neuropsychiatric disorders and the neurotransmitters involved.” It was published online in September 2021 and has not been cited, according to Clarivate’s Web of Science.
The notice said only that “some readers have raised concerns” about the article, which the journal was discussing the the authors, a group led by Dick Swaab of the Netherlands Institute for Neuroscience in Amsterdam.
We pointed out at the time that the expression of concern followed criticism on social media, including by Steve Maren, a biobehaviorial neuroscientist at Texas A&M University and member of the journal’s editorial board, who tweeted that he had emailed the editor in chief requesting the paper’s retraction.
The day after our original story published, Maren tweeted a detailed thread about how the review article “completely misrepresented” one of the studies it referenced:
He concluded:
Recently – we don’t know exactly when, because of how Elsevier publishes retraction notices – the journal retracted the article. The notice stated:
This article has been retracted at the request of the Editor in Chief of Neuroscience and Biobehavioral Reviews after concerns were raised with respect to the phrasing of comparisons drawn between humans and animal models. These comparisons were deemed unsupportable, and thus in the best interests of publication standards the Editor has concluded it is necessary to retract the paper. The authors disagree with the reason for the retraction.
We emailed Giovanni Laviola, the editor in chief of the journal, for more information, but haven’t heard back.
When we reached out to Swaab for comment last week, he was apparently unaware that the paper had been retracted, asking us,
Is the paper retracted? I thought we were still discussing the matter.
We confirmed that the online article said it had been retracted, and he responded that he would “go after it next week.” He has not yet replied to our follow-up request for comment but we will update with anything we learn. [Update, 1500 UTC, 12/21/22: Swaab declined to comment further, saying, “This is not a forum for a scientific discussion.]
Maren also learned that the article had been retracted from our email asking for comment, he told us. He summarized his concerns with the paper–the data misrepresentation, that the authors had overstated the relevance of animal models to human gender and sexual orientation, and that the paper exhibited a “false narrative,” namely, “that homosexual behavior is non-normative and induced by prenatal insults caused by maternal medication or drugs.” We’ve made his full comments available here.
Maren said he was glad to hear from us that the paper had been pulled:
I presented some of the concerns I detail above to the editor-in-chief and other than receiving an acknowledgment that the concerns had been received I had not further correspondence with the editors or Elsevier. In fact, your email is the first I have heard that the paper had been retracted.
Update, 1600 UTC, 12/23/22: Laviola sent us the following comment:
We can confirm that the article from Neuroscience & Biobehavioral Reviews was retracted on November 21, 2022 following a reassessment due to a number of complaints.
Experts concluded that the literature and research on animal models raised a few critiques but crucially the subsequent interpretation was not supportable, specifically the statements about the potential implications for humans. The reviewers felt that the human situation is far more complex, and that it was inappropriate to directly transfer information on mechanisms and processes from basic research in experimental animal models to the clinical setting.
The authors were informed that the Editor had decided to retract the paper and a draft text of the retraction note was shared with the authors in October, however due to an oversight, the authors were not informed of the publication of the retraction, which officially took place in November.
Like Retraction Watch? You can make a tax-deductible contribution to support our work, follow us on Twitter, like us on Facebook, add us to your RSS reader, or subscribe to our daily digest. If you find a retraction that’s not in our database, you can let us know here. For comments or feedback, email us at [email protected].
There is also something not quite good in regards to the literature search the authors performed.
The search strategy as given in the article, retrieves 4421 references d.d. 20-12-2022:
(((“sexual orientation”[MeSH Terms] OR (“sexual”[All Fields] AND “orientation”[All Fields]) OR “sexual orientation”[All Fields] OR “sex preference”[MeSH Terms] OR (“sex”[All Fields] AND “preference”[All Fields]) OR “sex preference”[All Fields]) AND (“neurotransmitters”[MeSH Terms] OR “neurotransmitters”[All Fields] OR “serotonin”[MeSH Terms] OR “serotonin”[All Fields] OR “dopamine”[MeSH Terms] OR “dopamine”[All Fields] OR “GABA”[MeSH Terms] OR “GABA”[All Fields] OR “oxytocin”[MeSH Terms] OR “oxytocin”[All Fields])) OR ((“neuropsychiatric disorders”[MeSH Terms] OR (“neuropsychiatric”[All Fields] AND “disorders”[All Fields]) OR “neuropsychiatric disorders”[All Fields] OR “neuropsychiatric diseases”[MeSH Terms] OR (“neuropsychiatric”[All Fields] AND “diseases”[All Fields]) OR “neuropsychiatric diseases”[All Fields]) AND (“neurotransmitters”[MeSH Terms] OR “neurotransmitters”[All Fields] OR “serotonin”[MeSH Terms] OR “serotonin”[All Fields] OR “dopamine”[MeSH Terms] OR “dopamine”[All Fields] OR “GABA”[MeSH Terms] OR “GABA”[All Fields] OR “oxytocin”[MeSH Terms] OR “oxytocin”[All Fields])))
The search strategy contains several severee flaws: some technical (redundant brackets, missing quotes) and some subject-related -the authors use three non-existent mesh-headings:
• “sexual orientation” [MeSH Terms]
• “neurotransmitters” [MeSH Terms]
• “neuropsychiatric disorders” [MeSH Terms]
Both types of errors yield an underrepresentation of the number of results.
When correcting these errors, and using correct mesh headings, the number of results increase from 4421 to 10,401:
(((“sexual behavior”[MeSH Terms] OR (“sexual”[tw] AND “orientation”[tw]) OR “sexual orientation”[tw] OR “sex preference”[MeSH Terms] OR (“sex”[tw] AND “preference”[tw]) OR “sex preference”[tw]) AND (“Neurotransmitter Agents”[Mesh] OR “Neurotransmitter Agents”[Pharmacological Action] OR “neurotransmitters”[tw] OR “serotonin”[MeSH Terms] OR “serotonin”[tw] OR “dopamine”[MeSH Terms] OR “dopamine”[tw] OR “GABA”[MeSH Terms] OR “GABA”[tw] OR “oxytocin”[MeSH Terms] OR “oxytocin”[tw])) OR ((“neuropsychiatr*”[tw] OR (“neuropsychiatric”[tw] AND “disorders”[tw]) OR “neuropsychiatric disorders”[tw] OR “neuropsychiatric diseases”[MeSH Terms] OR (“neuropsychiatric”[tw] AND “diseases”[tw]) OR “neuropsychiatric diseases”[tw]) AND (“Neurotransmitter Agents”[Mesh] OR “Neurotransmitter Agents”[Pharmacological Action] OR “neurotransmitters”[tw] OR “serotonin”[MeSH Terms] OR “serotonin”[tw] OR “dopamine”[MeSH Terms] OR “dopamine”[tw] OR “GABA”[MeSH Terms] OR “GABA”[tw] OR “oxytocin”[MeSH Terms] OR “oxytocin”[tw])))
Jan Schoones
So, not good then?
Yes, in my regards not good at all – the authors should have consulted a biomedical information specialist.
Research librarians are wonderful people, and experts in working through search terms and databases, but too often aren’t consulted. They are critical in any sort of systematic review
It seems concerning to me that scientific research is being critiqued by the political and cultural narrative of our time. If there is credibility to what they have said, we should reevaluate our beliefs based on the facts. We use animal studies all the time from which we extrapolate the data to make judgments concerning human health.
I totally agree with you, Justin.
Your conditional statement is uncontroversial. If there is credibility we should re-evaluate our beliefs. But, in this case, because of the deficiencies of the search query, the search only found 40% of the literature. For that reason alone, it should not much increase our prior confidence in the “credibility of what they have said”. So our re-evaluation is either a strengthening of prior beliefs, or a change of mind based on extrapolation from new findings that are not justified by the survey methods; in each case not worthwhile, and not safe. And, in the latter case, if our collective mind is changed, the damage is obvious – error may propagate.
Did you miss the parts where
a) a key conclusion rested on a graph that mislabeled the categories
b) the data search missed a significant portion of the relevant literature, in a way conducive to preferentially seeing a subset of outcomes that conflate or combine orientation and a wide range of disorders?
In addition, sexual behavior is vastly different in humans and bonobos, compared to effectively all other mammals (in which it is almost entirely estrous-driven). That’s a flashing warning sign, declaring that extrapolation in this issue is specifically unwarranted.
There are plenty of papers published that consider gender and/or orientation from the standpoint of biology. That’s not the problem. It’s very normal for bad science to be called out most loudly by those who have some interest in the topic. (There are also wonderful image nerds and graph nerds who do it for the love of finding bad papers; but they are a rare breed.) When astronomers call out bad astronomy data, there’s no backlash. When industrial chemists call out a bad catalysis paper, ditto. But for an orientation paper, the call out gets this hand-wringing backlash, over how it must be social engineering.
Interesting why THIS specific article was dissected so fully. Actually it is almost impossible to not find errors or misjudgement in a scientific article but we have peer review and then we can discuss a PUBLISHED paper. Some fellows made sure we have nothing to discuss – intelligently.
Some of Mr. Maren’s objections are unscientific. Honestly, these are frightening times.
Somehow, the expression « POTENTIAL implications for humans » has become a deviant inaccurate statement, even though ‘potential’ speaks for itself.
Does retraction imply improper research procedures or fraud on the part of the author’s. If so can involuntary retraction be the basis of a lawsuit for defamation?
The answers to these two questions must be, respective, no and yes. Genuine mistakes could lead to retraction, but would not be improper. But if the facts strongly suggest impropriety, or the editors represent it that they do, the retraction might be prima facie defamatory. Whether a defamation suit has good prospects must turn on the facts, including meta-facts like the good faith of those making the retraction. Also, it depends on the way defamation law works in the chosen venue. The case of Irving v Lipstadt failed because the jury did not believe Irving (but that is all we know). The arguments hinged on the underlying facts, and a professional critique of his methodologies, and on evidence about good faith.
And the neocons come crawling out of the woodwork.
‘the paper exhibited a “false narrative,” namely, “that homosexual behavior is non-normative and induced by prenatal insults caused by maternal medication or drugs.”’
So is this the crux of the issue? It must be nice to have the power to decide which narratives are true and which false. I think they’re called ‘fact-checkers.’
I checked out the retracted paper, and the comment by Marren.
There are two issues in my view.
1) Marren says the bar graphs were taken from:
Olvera-Hernández, S., Chavira, R., & Fernández-Guasti, A. (2015). Prenatal letrozole produces a subpopulation of male rats with same-sex preference and arousal as well as female sexual behavior. Physiology & Behavior, 139, 403-411.
However, the bar graphs seem to have been taken from a different paper:
Hernández, A., & Fernández-Guasti, A. (2018). Male rats with same-sex preference show higher immobility in the forced swim test, but similar effects of fluoxetine and desipramine than males that prefer females. Pharmacology Biochemistry and Behavior, 171, 39-45.
That paper reports: “Forty percent of the subjects treated prenatally with letrozole had same-sex preference compared with 5% shown by males treated prenatally with vehicle (chi square, X2 = 72.93, df 3; p = 0.001).”
(I still agree with Maren that the way the group was split into two subgroups is unclear/dubious)
2) The authors of the retracted article do not use the paper as part of the review itself but only as part of the introduction. The only thing they report about figure 1 is:
“Nowadays, useful animal models have helped to frame questions and to propose hypotheses relevant to human sexual orientation, as
described in Fig. 1. These tests, although imperfect, have been used to model certain aspects of human sexual orientation (Roselli, 2018).”
Is this a sufficient basis for retraction?