3 OCT 2026 — Most choices that affect other people come with uncertain consequences. A traveller flying to a conference cannot say exactly how much harm the flight does, and one taking the train cannot say exactly how much good. A study published on 1 October in Nature Human Behaviour finds that people fill that uncertainty in ways that flatter their own choice. The version favouring selfishness spreads more easily.
The researchers, Marius Vollberg and Jörg Gross of the University of Zurich and the University of Geneva, ran nine preregistered online experiments with 8,769 adults. Most participants were in the United States.
How the experiments worked
Participants chose between two options. Option A paid them 7 points and gave another person, the receiver, somewhere between 1 and 3 points. Option B paid them 5 points and gave the receiver somewhere between 3 and 5. Their own payoff was certain. The receiver's payoff was only a range, according to the paper.
After choosing, participants guessed what the receiver actually got under each option and were paid extra for accurate guesses. They were told only that the true number could be anywhere in the range. If they treated every value as equally likely, their guesses should have centred on the middle of each range regardless of what they had chosen.
What the guesses revealed
They did not centre on the middle. People who took the selfish option tended to inflate what the receiver got from their choice and play down what the receiver would have got from the generous one, so the two options looked more alike. People who took the generous option portrayed their choice as comparatively more beneficial and, less consistently, inflated how much they had given up.
The distortions appeared even when people had not yet chosen and when the choice was forced on them. To the authors, that suggests the bias reflects people's social preferences, not just an effort to justify a decision after the fact.
The contagion finding
In the final experiment, 850 participants saw another person's estimates before choosing. When those estimates were distorted in the selfish direction, participants' own estimates shifted and they became more selfish. Estimates distorted in the generous direction did not shift the average.
A small group with an outsized effect
The authors note that the selfish pattern came from a relatively small group, fewer than 10% of participants by their estimate, and that individuals varied widely in whether and how they distorted. They suggest this asymmetry helps explain how polarised views of the social impact of everyday choices form.
Limits of the study
The experiments used small sums in online games with mostly American participants, so they may not carry over to real decisions such as travel, shopping or policy. The authors flag several limitations in the paper, including how they measure distortion against the midpoint of each range. Their materials and preregistrations are posted on the Open Science Framework.
View author profile → · Editorial policy
