A 10 month proximity sensor study at Antarctica's Concordia Station found that the more time a 12 person crew spent near each other, the more tension they reported, and that they reorganized into shifting subgroups rather than a single group.
The usual story about small isolated groups is that shared hardship builds solidarity. A 10-month overwintering study at Antarctica's Concordia Station, the most remote permanently staffed research post on the continent, sitting on the polar plateau at temperatures near minus 80 degrees Celsius, suggests the picture is more complicated. Researchers found that the more time crew members spent in close physical proximity, the more interpersonal tension they reported, the opposite of what "hardship breeds closeness" would predict. The richer finding sits underneath: rather than merging into a single unit, the 12-person crew drifted into distinct subgroups, and the configuration of those clusters evolved throughout the mission.
The work was led by an international team anchored at the University of Zurich and the University of Bern, with senior author Jan B. Schmutz (Zurich) and co-lead Andrea Cantisani (Bern), published in the Proceedings of the National Academy of Sciences (PNAS) on May 26, 2026 (DOI: 10.1073/pnas.2533420123), according to a SciTechDaily report on the University of Zurich press materials published 12 June 2026. The full author list includes Andrea Cantisani, Jan B. Schmutz, Pedro Marques-Quinteiro, Lorenzo Dall'Amico, Ciro Cattuto, Mirko Antino, Walter J. Eppich, Katharina Stegmayer, and Sebastian Walther. The researchers designed the study to do something the usual interview-and-survey approach cannot: track who is actually near whom, and for how long, on a continuous basis across an entire Antarctic winter.
The empirical core is the methodology. The crew wore proximity sensors throughout the 10-month overwintering period, generating a stream of data on who shared rooms, corridors, and social spaces, and at what times. Layered on top of that, researchers ran periodic self-report surveys asking crew members to rate interpersonal tension. The combination allowed the team to test whether objective closeness, measured in meters and minutes, predicted subjective friction.
It did. More time spent physically near other crew members correlated with higher reported tension, not lower. The correlation is the inverse of what a "hardship builds bonds" reading would predict. The researchers are careful to note that the behavioral signal is proximity, not hostility: the sensors do not measure arguments or sentiment directly, only who is where. But the pattern is consistent and, on the researchers' reading, counterintuitive.
The second finding may matter more for the people designing future crewed missions to the Moon and Mars. Over the course of the 10 months, the 12-person crew did not converge into a unified group. They fragmented, naturally and repeatedly, into smaller clusters. The composition of those subgroups shifted over time rather than stabilizing, suggesting that isolated teams are not static social units but ongoing reorganizations, and that the natural unit of social support under isolation may be a sub-team of three or four, not the full complement.
For Mars mission planners, the practical implications are concrete. Crew selection typically focuses on interpersonal compatibility across the whole team. If subgroups are the natural operating unit, the design question changes: how do you build a crew whose small clusters each function well, and whose clusters cooperate? Habitat architecture matters too. If proximity drives tension, then shared quarters, communal galleys, and overlapping circulation paths become design problems, not amenities.
The study has obvious limits. The cohort is 12 people at a single station, with no control group, and the behavioral data is proximity rather than sentiment. Findings are correlational, not causal. The Concordia winter is also not Mars: resupply is impossible, evacuation is effectively impossible, and the crew knew both of those facts going in. The study cannot tell mission planners whether the same patterns hold at larger crew sizes, in mixed-gender or mixed-nationality teams, or over the two-to-three-year durations a real Mars mission would require.
What it can do is replace intuition with measurement. Future analog missions, the ground-based isolation studies run by space agencies and research consortia in the years ahead, can be designed around proximity sensing from the start, so that whoever walks onto a simulated Mars habitat will already be inside an instrumented version of the same question: who is near whom, and what does that do to the group?
The Concordia data does not predict the failure or success of any specific mission. It does suggest that the people building long-duration crew habitats should stop designing for the "crew" as a monolith and start designing for the subgroups that isolation actually produces.