Rehearsing Mars on Earth: The Logic of Analogue Research

AThe premise beneath analogue research seems almost too obvious to examine: if a place on Earth resembles Mars, rehearsing there will reveal what a Martian expedition would encounter. Funding documents and popular accounts alike rank sites by how convincingly they look and behave like the planet, and the word fidelity does the work of an argument. The premise is not so much wrong as incomplete. What a field campaign can transfer to a real mission is never the resemblance itself but the behaviour of one variable that the site happens to isolate cleanly, and a location reproducing several conditions at once may make the contribution of each impossible to disentangle. Researchers have borrowed a pair of terms from psychology to mark the difference: face validity, meaning that a study looks like the thing it stands for, and predictive validity, meaning that its results actually forecast what happens later. The two do not always travel together.

BYohanna Duarte, who directs the Coimbra Institute of Field Simulation, has assembled the only sizeable body of evidence bearing directly on the question. Between 2014 and 2020 her group tracked two sets of engineering procedures through to formal flight qualification: 37 that had been rehearsed inside a pressurised habitat built on volcanic terrain, and 41 that had been worked out in an unremarkable warehouse fitted with the same instruments and nothing else. Qualification testing later rejected 11 of the first set and 13 of the second, a difference far too small to survive the modest sample sizes involved. Duarte maintains that the expensive site bought realism rather than knowledge. She is careful about the scope of the claim. Her comparison covered hardware procedures alone, and she notes that a habitat costs some forty times what the warehouse costs to run for a single season.

CThe human side of analogue work carries a difficulty that no choice of location can repair. Between 2016 and 2021 six confinement studies, each lasting eight months with crews of six, recorded the familiar third-quarter dip in mood and the recovery that follows it, and the pattern has been treated as a rehearsal of the psychological arc of a real voyage. Yet every participant knew the date on which the door would open, and knew that a serious illness would be met by a vehicle rather than by a radio conversation lasting twenty minutes each way. Stress measured under those conditions is stress with an exit. The effect is not small: it was only among the volunteers who had been assured that evacuation could be arranged within a day that measured cortisol in the closing weeks fell back to roughly the level recorded before entry, a reassurance no real crew could be offered.

DPiotr Ambroziak, of the Boleslaw Laboratory of Human Performance, insists that this line of criticism misunderstands whom an analogue is meant to change. In his account the site does not simulate the planet for the benefit of the planet; it educates the engineers and physicians who will write the documents a crew must eventually follow, and an environment that feels real produces documents of a different quality. He compared nine design teams and found that those who had spent a season living inside a habitat later produced emergency checklists containing 24 per cent fewer omissions than the teams who had stayed at home. The finding is easy to state and hard to interpret. Ambroziak himself concedes that a checklist is not an outcome, that nine teams is a thin basis for any general rule, and that people who volunteer for a season in the field are unlikely to resemble the profession as a whole.

EA third difficulty belongs to the search for life rather than to engineering or psychology. Instruments intended to recognise the chemical traces of biology are calibrated at sites that teem with terrestrial organisms, so the very background a Martian instrument will not encounter is the background against which its thresholds are set. During a 2019 field trial in the Danakil basin an onboard analyser reported organic compounds at a concentration that would have been relayed home as a significant result; the compounds were later traced to a lubricant on the team's own sampling arm, and the misreading occurred only in the runs where that arm had been handled without gloves. Nobody involved was careless, and the fault lay in the setting rather than in the hands, which is precisely why moving the rehearsal to a cleaner desert would not remove the difficulty. Such episodes are usually described as embarrassments. They are better understood as data, because they establish the rate at which a procedure imports what it claims to find.

FNo single verdict covers all three strands, and pretending otherwise would be a disservice to both camps. On hardware this article follows Duarte: where the object of study is a valve or a seal, an expensive backdrop adds cost without adding information, and her comparison, though small, points in one direction only. On procedures and on the people who draft them Ambroziak has the better of the argument, since the mechanism he describes operates on Earth, where his evidence was gathered, and does not depend on any claim about Martian conditions. The two positions are less opposed than they appear. What would sharpen the whole enterprise is a discipline borrowed from clinical trials, namely a public statement, issued before a campaign begins, of which single variable the site is supposed to isolate and which result would count as a disappointment. Neither camp has any reason to resist a rule of that kind, since it costs almost nothing and settles in advance what a season in the field is actually for.