Oxytocin: The Two Studies Behind Conflicting Trust Headlines

A stranger gets to decide how much of your money comes back. Before you hand it over, researchers give you a nasal spray containing oxytocin or a placebo. The amount you send becomes their measure of trust.

This experiment helped make oxytocin famous. More than twenty years later, researchers are still testing what it can tell us. Two papers published in 2026 reached different conclusions, and both deserve a place in the story.

Charlotte Kroll and colleagues found no increase in trusting behavior. Bodo Vogt's team found roughly 15% higher transfers in a group selected for low baseline trust. If you have seen oxytocin described as a way to improve social connection, this disagreement matters: the claim depends on who was tested and what counted as improvement.

The useful question is how far a measured decision can take us toward understanding trust in daily life. Following the experiment gives us a much better answer than the nickname “trust hormone.”

The trust game gives an investor an initial amount of money. The investor can keep it or transfer some to another player, the trustee. The amount sent is multiplied before it reaches the trustee, who can return some money to the investor.

Sending money creates the chance for both players to gain. It also leaves the investor exposed: the trustee may send nothing back. Researchers use the size of the initial transfer as a measure of willingness to accept that social risk.

In Kroll's experiment, investors started with twelve experimental money units and could send zero, four, eight or twelve. The transfer tripled. Participants later played the trustee role with a different partner, but they made their investor decision before learning they would switch roles. Kroll et al., full research manuscript

That last detail matters. If you knew you would later play the other role, you might think about your own future choice while deciding what to send. The design tries to keep the first decision interpretable.

The task is useful because it produces a clear, repeatable number. Its meaning still has limits. A transfer can express confidence in the other person, a willingness to take a chance, or an interest in the potential reward. The game gives researchers something concrete to test while leaving much of a human relationship outside the room.

Michael Kosfeld and colleagues reported the influential result in Nature in 2005. Men received oxytocin or placebo before making decisions with real financial consequences. The key investor comparison included 29 men in each group.

Average transfers were 9.6 money units with oxytocin and 8.1 with placebo. Thirteen oxytocin recipients transferred the maximum amount on average, compared with six placebo recipients. In a separate task where a random process determined the return, both groups averaged 7.5 units.

That comparison suggested something more specific than a general rise in willingness to gamble. The difference appeared when another person controlled the return. The study did not find a corresponding increase in the trustee's willingness to repay. Kosfeld et al., 2005

This is a useful distinction to carry forward. Trusting someone and being trustworthy describe different actions. A peptide could change one measured decision without improving every part of a social exchange. The original experiment opened a research question that required more people, repeated tests and sharper definitions.

Kroll's team recruited young men and used a randomized, double-blind comparison. Neither participants nor the experimenters working with them knew which spray they received. The final analysis included 211 men: 99 received oxytocin and 112 received placebo.

The participants could spend ten minutes together in a common room before returning to private cubicles. Fifty minutes after spray administration, they played the trust game. This brief social contact was an intentional part of the experiment.

The treatment did not produce a statistically significant increase in transfers. The raw choices show why. About 39% of oxytocin recipients sent the full twelve units, compared with about 48% of placebo recipients. Those proportions alone do not establish that oxytocin reduces trust; the overall analysis did not support a treatment difference. Kroll et al., 2026

The researchers also combined their data with 321 participants from an earlier replication, producing a pooled sample of 532. That analysis again found no increase.

They then asked a more demanding statistical question: were the groups close enough to rule out effects larger than a range defined in advance? This is called an equivalence test. Their pooled result met that test within their chosen bounds. Their new sample on its own did not.

The distinction prevents two common reading errors. A nonsignificant result can leave substantial uncertainty. A successful equivalence test gives stronger information, but its conclusion depends on the size of effect the researchers decided to exclude. Here, the bounds concerned effects worth pursuing in typical laboratory studies. They did not establish that every possible social effect was exactly zero.

Vogt's team enrolled 359 men selected for low questionnaire scores on general trust. They played anonymously, without prior social contact. The study appeared in PNAS in August 2026.

Transfers increased by roughly 15% with oxytocin. The authors' full preprint reports averages of €8.29 versus €7.20 with placebo. A pooled analysis added an earlier low-trust sample and yielded 16.9%; it was not a separate replication. Published report; author preprint

Without a high-trust comparison group, this study cannot establish that the effect occurs only in low-trust people.

The recent studies varied in participant selection and prior social contact. Those differences offer possible explanations for their results. Neither experiment directly tested whether those differences caused the disagreement.

Kroll's team also examined whether baseline trust altered the treatment response within its sample. They found no clear evidence for that interaction. That result deserves to stay beside the positive study; selecting a particular group and testing variation across a broader sample are different research questions.

A stronger follow-up could deliberately vary social contact and recruit participants across a defined range of trust scores. It would need enough people to compare those conditions, with the analysis specified before the results arrived. This is a proposal for resolving the uncertainty, not a finding from either paper.

The route into the body adds another question. Oxytocin is a peptide made in the hypothalamus, a region of the brain. It has established roles in childbirth and milk release. A nasal spray aims to alter its activity experimentally, but a spray dose does not tell researchers exactly how much reaches each relevant brain region. Kroll's discussion identifies delivery and measurement as continuing sources of uncertainty.

Suppose a product promises better connection. Define the change you would want before accepting a laboratory percentage as the answer. Feeling comfortable during a conversation differs from trusting a stranger with money. Repairing a relationship adds shared history and another person's behavior.

For any treatment claim, look for a study that measured the outcome you care about over a relevant period. A one-visit experiment cannot tell you whether an effect lasts through months of ordinary life. The formulation and delivery method also need to match the evidence being cited.

The two 2026 reports make oxytocin more interesting to study. They give us a reason to ask which conditions produce a response, how reliably it repeats, and whether that response helps people outside the experiment. A useful clinical application would have to answer those questions with outcomes people can recognize in their own lives.

Which peptide claim should I trace back to its original experiment next? Reply with the claim or the paper.

For entertainment and education only. Not medical advice. Talk to your doctor before starting anything.