Summary: When people face an ethical dilemma, do they favor the greatest overall good or the fairest distribution? New neuroimaging evidence indicates that fairness frequently wins out. A laboratory study using an “icy water” harm-allocation task found that participants preferred to spread discomfort across multiple people rather than concentrate it on a single individual, even when that meant increasing total harm. Functional MRI data linked this preference to brain networks involved in valuing outcomes and mentally simulating others’ experiences.
Researchers contrasted two moral frameworks: utilitarianism, which seeks to minimize aggregate harm, and a Rawlsian approach, which prioritizes protecting the worst-off. University students inside fMRI scanners repeatedly chose between subjecting one person to a longer time with their hand in ice water or distributing shorter exposures across a group of three or four people. Critically, the combined duration for the group was deliberately larger in many trials, meaning the utilitarian option would have produced less total discomfort.
Key Facts
- Study trade-off: Participants weighed reducing total harm against avoiding disproportionately harming a single person.
- The “Icy Water” task: Choices contrasted one individual enduring a long exposure to cold water with a group of 3–4 people each receiving shorter exposures that summed to a greater total.
- Average cost of fairness: On average, participants accepted an extra 68 seconds of total discomfort across the group to prevent one person from suffering disproportionately.
- Active intervention: Even when a “do nothing” default was available so participants could avoid making an active choice, most intervened to reallocate harm fairly rather than leaving a single person targeted.
- Neural mechanisms: fMRI results implicate mentalizing regions (which model others’ subjective experiences) together with valuation networks in driving Rawlsian choices.
Study overview
The experiment, led by Woo-Young Ahn and colleagues, involved 52 paid volunteers from a South Korean university. While inside fMRI scanners, participants made repeated, incentivized decisions about how to allocate physical discomfort—placing a hand in icy water—between one person and a small group. Some trials displayed a default option that would have allowed participants to avoid pressing any buttons and thus avoid being the direct cause of harm. Despite that, most participants actively chose to spread harm across the group rather than concentrate it on the single individual.
Behavioral results showed a clear preference for fairness: participants were willing to increase total harm when it prevented singling out one person. The presence of a default option did not produce a major “do-nothing” bias, suggesting people were not primarily motivated by avoiding personal responsibility but rather by a desire to preserve equitable treatment.
Neural findings
Combining computational modeling with univariate and multivariate fMRI analyses, the researchers identified two separable components contributing to Rawlsian preferences: (1) a drive to minimize the maximum harm any single person receives, and (2) an individual-specific threshold for how much harm to one person is acceptable. These components mapped onto different brain systems: mentalizing networks that simulate others’ internal states and valuation systems that assess trade-offs between fairness and aggregate welfare.
Key questions answered
Q: Why would people accept more total pain?
A: For many individuals, equity feels morally weightier than efficiency. A single person being singled out for intense suffering is perceived as especially wrong, so people tolerate spreading discomfort across several individuals to avoid that outcome. The decision reflects an aversion to “singling out” rather than a calculation that only totals matter.
Q: Were participants reluctant to press buttons that caused harm?
A: Surprisingly, no. When a neutral default option was available, most participants did not simply avoid choosing; they actively intervened to reallocate harm more evenly. This behavior indicates that the motivation for fairness outweighed concerns about being the active agent of harm.
Q: What broader implications does this have for public policy or social systems?
A: The findings offer a neural explanation for why people often resist policies that are efficient on aggregate but appear to disadvantage the worst-off. Valuation systems in the brain can place greater subjective weight on protecting those who would be uniquely burdened, creating a tension between fairness and overall efficiency.
Editorial notes
- Article edited by a Neuroscience News editor.
- Journal paper reviewed in full by staff.
- Additional explanatory context provided by editorial staff.
About this ethics and neuroscience research news
Author: Woo-Young Ahn
Source: PNAS Nexus
Contact: Woo-Young Ahn – PNAS Nexus
Image: Image credited to Neuroscience News
Original research (open access): Decomposing the neurocomputational mechanisms of deontological moral preferences. Authors: Yoonseo Zoh, Soyeon Kim, Hackjin Kim, M. J. Crockett, and Woo-Young Ahn. PNAS Nexus. DOI: 10.1093/pnasnexus/pgag074
Abstract
Decomposing the neurocomputational mechanisms of deontological moral preferences
Research contrasting utilitarian choices, which maximize aggregate welfare, with deontological judgments grounded in rules has advanced understanding of moral decision-making, but less is known about the internal components of deontological preferences. This study separates two forms of deontological inclination—Rawlsian and Kantian—and contrasts them with utilitarian strategies using an incentivized moral dilemma task. Participants repeatedly decided how to divide harm between a single individual and a group of three to four people. Behavioral data showed a consistent preference for Rawlsian strategies: participants often chose to impose greater overall harm rather than let one person suffer disproportionately. Computational modeling identified two dissociable sources of individual variation in Rawlsian choices: a tendency to minimize the maximum harm to any one person and a subjective threshold governing acceptable harm to an individual. Univariate and multivariate fMRI analyses linked these dimensions to distinct neural systems—mentalizing regions that represent others’ experiences and valuation networks that compute trade-offs—clarifying neurocomputational mechanisms that shape decisions about fairness versus total welfare.