In 2005, a nasal spray of the hormone oxytocin ↗ made people hand more money to a stranger. The result, reported in Nature by Michael Kosfeld, Ernst Fehr and colleagues, became one of neuroscience's most repeated claims: the trust hormone. It then spent much of the next two decades failing to replicate. Fehr, an economist at the University of Zurich, has now published a large new study that rescues the effect by shrinking it.
Writing on August 6 in the Proceedings of the National Academy of Sciences ↗, Fehr, with Bodo Vogt and Paul Bengart at Otto von Guericke University Magdeburg and Carolyn Declerck at the University of Antwerp, report that oxytocin raised trusting behavior by about 15 percent, but only in men who started out low in the disposition to trust. Screen for that subgroup, they argue, and the effect holds. Mix everyone together, and it washes out. That, in their telling, is why the field spent years arguing.
The setup was a trust game, the same lab task Fehr used in 2005. One player, the investor, gets a pot of money and decides how much to send to an anonymous stranger. Whatever is sent gets tripled, and the stranger decides how much, if any, to return. How much the investor sends is the measure of trust, because sending more only pays off if you expect the other side to give some back. The researchers recruited 359 men who scored low on a standard trust questionnaire, gave each either a 24-unit intranasal dose of oxytocin (the same dose as the 2005 study, sold for medical use as Syntocinon) or a saline placebo, and had them play under complete anonymity.
The oxytocin group invested about 15 percent more. Pooling the new data with an earlier sample of 219 low-trusting people pushed the combined figure to 16.9 percent, and it stayed statistically significant. Within the low-trusting range there was no gradient and no interaction. The spray lifted the wary by the same amount whether they were slightly or deeply distrustful. The study was preregistered, meaning the hypotheses and the analysis were locked in before the data came in, which is the specific defense against the analyze-until-something-turns-up problem that has dogged this literature.
The honest way to read this is not that oxytocin builds trust. It is that a decade of null results may have been diluting a real but narrow effect. Cautions come with that. Everyone in the main study was a man, and everyone was preselected as low-trusting, so the result says nothing about women, and nothing about the average person the 2005 headlines implied. The outcome is money moved in a lab game, a stand-in for trust, not trust in a relationship or a negotiation. How much intranasal oxytocin even reaches the brain remains contested. And the framing is a retreat. The effect survives when you look only where there is room to move, which is a smaller claim than the one that made oxytocin famous.
None of this makes oxytocin a trust drug. The molecule is the same one hospitals give as Pitocin to start labor and the body releases during birth and bonding, and the point here is a narrow question about behavior, not a treatment. What the study does is turn a slogan back into a hypothesis with a boundary condition attached.
peptidemodel hosts oxytocin as a card against its receptor, OXTR ↗, and the section has tracked its social effects before, including a study where a stimulant's reward only formed in company and two peptides ran the switch ↗. The pattern across those pieces is consistent. Oxytocin's effects are real but conditional, showing up in specific people, contexts, and tasks rather than as a general dial on how social anyone is.