Fifty-four people sat in a lab in Melbourne wearing virtual reality goggles, drifting through a reconstruction of the International Space Station, and sniffed a cotton ball soaked in vanilla extract. Inside the simulated spacecraft, the vanilla registered as stronger than it had in a plain room. So did almond. Lemon did not move.
That is the result, and it runs the opposite way from the thing it was meant to explain. Astronauts complain that food in orbit tastes like nothing. A confined, cluttered, humming space capsule was supposed to dull the senses. In the simulation, it sharpened two of the three smells tested.
The work came out of RMIT University in July 2024, led by Dr Julia Low of the School of Science, and was published in the International Journal of Food Science and Technology. RMIT announced it as a first attempt to test food aroma perception in a space context without leaving the ground.
Inside the VX Lab
The spacecraft was built in Unity by students from RMIT’s School of Computing Technology and run at the university’s Virtual Experiences Lab. It renders the ISS in low Earth orbit with floating objects to suggest microgravity and equipment mounted on every surface to produce the sense of clutter and confinement that crews describe. Participants compared it against a neutral room.
The odours were deliberately mundane. Two millilitres of vanilla extract on a cotton ball. Two millilitres of almond extract. One millilitre of lemon essential oil, dosed lower so that all three arrived at roughly equal intensity. The published method lists the brands: supermarket baking extracts, not laboratory reference standards.
Samples were presented in randomised order with gaps between them, and participants were told to sniff the back of their own hand to clear the palate. The familiarity questionnaire included rosemary, wasabi and honey — odours that never appeared in the test — to keep people from guessing what was being measured. The cohort was 54 adults aged 18 to 39, screened for any history of motion sickness or vertigo.
Then the numbers. Vanilla was rated significantly more intense in the VR context than in the neutral one, at P = 0.009. Almond came in below P = 0.001. Lemon showed no significant difference across the two environments at all.
The benzaldehyde explanation
Vanilla and almond share a compound. Benzaldehyde is the molecule behind bitter almond, and it turns up in vanilla’s volatile profile too. Associate Professor Jayani Chandrapala, a food chemistry specialist at RMIT and co-author on the paper, told reporters the team believes “it’s this sweet aroma that gives that highly intensive aroma” inside the VR setting.
Lemon oil has no benzaldehyde. Its dominant volatile is limonene, which stimulates the trigeminal nerve alongside the olfactory bulb and carries a physical sharpness that sweet odours do not. The paper stops at volatile composition as the likely divider; it does not claim to have isolated a nerve pathway.
The other variable was the people. Clustering participants by how sensitive they were to the odours in general turned up a group that was less sensitive overall, and it was that group who rated almond as significantly stronger in VR, at P = 0.011. The context did not move everyone equally. It moved the people with the least acute noses the most.
What the second study found about emotion
The team ran the design again with different comparisons. That paper, first-authored by PhD scholar Grace Loke with Low and Dr Lisa Newman as co-leads, appeared in Food Research International in November 2024. Forty-four participants, eight odours, two simulated environments.
The odours widened out: vanilla, almond, lemon, lemon myrtle, eucalyptus, peppermint, vinegar and lemongrass. The environments were a reclined posture meant to mimic how an astronaut’s body sits in microgravity, and the same VR spacecraft. The VR context produced stronger intensity ratings than the posture for every odour except lemongrass.
This second study also measured how people felt. Emotional valence and arousal were both significantly higher inside the VR spacecraft, at p < 0.001. Self-reported stress stayed low in both conditions, with no significant difference between them.
And here the emotional story gets thinner than it looks. The abstract is explicit: emotional and stress responses did not generally affect how intense the odours seemed. Two narrow correlations survived. Valence tracked positively with almond and with vinegar. Stress tracked negatively with vinegar, and with nothing else.
So the environment moved the ratings. Mood, measured rather than manipulated, moved almost nothing. Low has suggested that a heightened sense of loneliness and isolation may contribute to how confined people smell and taste food, but she frames that as a possibility the data opens up, not a mechanism the data establishes.
The blandness problem this does not solve
Crews on long missions eat less than their nutritionists prescribe, and that shortfall is the reason RMIT started looking at aroma in the first place. On a six-month rotation it is a manageable problem. On a Mars transit it becomes a mission risk.
The standard account is plumbing. Microgravity drives bodily fluid toward the head, the sinuses congest, and smell dulls. Cabin air is filtered and recirculated. Food stays sealed until the moment of eating, so the volatiles that normally drift ahead of a meal never reach the nose. Low’s own framing of the puzzle is that the fluid-shift symptoms fade after a few weeks in orbit and the complaints do not, which suggests something else is running underneath.
The VR result does not obviously fit either account. In a simulated capsule, the sweet smells got louder. If confinement alone dulled perception, that is not what should have happened.
One reading is that the simulation captures the wrong half of the problem. VR delivers the visual and acoustic experience of a cluttered metal tube. It delivers none of the fluid shift, the radiation, the disrupted sleep or the eight-month accumulation of the same six faces. A ground analogue can isolate the psychological contribution precisely because it strips the physiology out — which also means it cannot tell you what happens when both run at once.
Personalised menus, and the people who are not astronauts
What both papers actually argue for is a shift from crew menus to individual ones. If the same VR capsule amplifies almond for low-sensitivity participants and leaves high-sensitivity participants unmoved, then designing one aromatic profile for six people is designing it wrong for most of them.
The hardware for that is unremarkable. Freeze-dried components, aroma capsules and controlled-release packaging already exist. What has been missing is evidence about which direction to push the aromatic load, and for whom.
RMIT has consistently pointed the application back toward Earth. Nursing home residents are the population the team names most often: people whose olfactory function has narrowed, whose surroundings are confined, and whose appetite falls away for reasons no one can fully separate. Hospital wards, Antarctic stations and submarines fall under the same description.
None of that is established. It is a hypothesis with two small datasets behind it and an obvious route to testing.
What the studies cannot show
Both experiments ran in a lab. The samples were 54 and 44 people respectively, all aged 18 to 39, none of them screened, trained or conditioned the way flown astronauts are. The mood measures were self-reported and the stress levels never rose far enough to test what real stress would do.
Three odorants, or eight, are not a meal. Real food is dozens of volatiles arriving together, plus texture, temperature, sound and the social business of eating with other people. Whether a shift in rated vanilla intensity has anything to say about whether an astronaut finishes dinner is not a question a cotton ball can answer.
The connection between smell and feeling is at least well attested from the other direction. Odours reach the amygdala by a shorter route than any other sense, and recent work at the University of Florida identified two genetically distinct amygdala cell populations that can attach either a positive or a negative charge to the same odour depending on where they project. The traffic also runs the other way: a study out of the University of São Paulo found that ambient smells change how quickly people read emotion in a face, and that facial expressions change how people rate the smell.
Low’s group is still working the problem. Her current project list includes multisensory extended-reality work on astronaut wellbeing in isolated environments, and research on appetite, reward and mood in spaceflight.
The cotton balls are still cheap, the extracts are still the ones you can buy at a supermarket, and the headset still shows a capsule that nobody in the room has ever been inside. Somewhere above all of it, at four hundred kilometres, a real crew is opening a real pouch of something, and the vanilla in it smells like whatever the moment makes it smell like.