Fooling Two Eyes and One Inner Ear

Part One · Two views

In 1968 a team led by Ivan Sutherland built a display that a person could wear on the head. It hung from the ceiling on a mechanical arm, partly because of its weight, and the arm was also one of its two head trackers. The rooms it showed were simple wireframes, boxes drawn in lines, and the Computer History Museum calls it probably the first head-mounted stereoscopic display. It was even see-through, and the second tracker was ultrasonic. Every headset since has had to fake the same three things. It must give each eye its own view, it must give the eyes' focusing something to work with, and it must keep up with a moving head, fast enough that the inner ear does not object. This essay takes the three in turn, then asks what anyone knows about children.

Take the first job. Stereopsis, the sense of depth, depends partly on binocular disparity, which is the small difference between the images in your two eyes. In adults they sit about 6 to 6.5 centimeters apart, so each sees from a slightly different spot. In 1838, Charles Wheatstone built the stereoscope. He wrote that the whole effect of relief comes from seeing two flat projections at once, one on each retina. Disparity is only one cue, since when you move, the motion itself gives depth, and a headset that tracks your head can supply that too. It also draws every frame twice. Designers care about interpupillary distance, the spacing of the pupils. On a Meta headset the lenses can move between settings such as 58, 63 and 68 millimeters.

Part Two · The focus problem

The second job is harder, and researchers have long treated it as a major problem. Your eyes do two things when you look at an object. They turn inward to aim at it, which is called vergence. Their focusing mechanism sharpens the image, which is called accommodation. In real life the two are coupled. In a headset, the lenses make light seem to come from one fixed spot, but the pictures can put an object nearer or farther, so the eyes aim at the object yet must focus on the screen. Hoffman, Girshick, Akeley and Banks reported in 2008 that this uncoupling often makes it harder to fuse the two pictures and causes discomfort and fatigue. They said a screen shows its pictures on one flat surface, so the focus cues describe the surface, not the scene. A 2016 review still called it a major problem.

How much does the mismatch matter? The 2008 team built a special screen with nearly correct focus cues. With those cues, viewers named 3-D shapes faster and did better on a timed test of depth. They saw fewer distortions, and they felt less fatigue and discomfort. That is good evidence that the mismatch costs something. But the cost shows up as feelings more clearly than as measurements. In a 2024 study, 18 young women played for 30 minutes at two levels of mismatch. Nausea, eye-movement symptoms and disorientation rose after the larger one. Yet their measured focusing and eye-aiming did not change significantly. The authors suggested that the symptoms came from a poor fit between the eyes and the inner ear. One small study of adults settles nothing, but it does show that the causes of the discomfort are knotted together.

Part Three · Clocks and canals

The third job is to follow the head. The motion chip inside a headset, the inertial measurement unit, holds an accelerometer and a gyroscope. It typically reports about a thousand times a second. Its weakness is drift, because tiny errors in what it reports pile up. Cameras report much less often, but they correct that buildup, and Meta's chief technology officer has said the two work together because the chip is faster but less accurate. Meta calls the whole approach inside-out tracking, because the device finds itself with its own cameras and sensors, not with boxes on the wall. Three degrees of freedom tell you which way you are looking, and six add where you are. There are limits. Cameras need light and things to see, and a chip cannot tell a head's speeding up from a car's, so a moving car confuses it.

Fast enough comes with a number, and the number deserves some suspicion. Michael Abrash of Valve told a 2013 conference that the whole pipeline from head motion to light should take somewhere between 1 and 20 milliseconds. One frame at 60 hertz already lasts 16.6 milliseconds, and in 2013 Abrash put a typical game's delay at 35 or more, so reaching 20 was hard on ordinary hardware. A company's support page says research puts a meaningful limit around 20 milliseconds, and below that, delays are generally not noticed. This is a rule of thumb, set by what people notice and not by physics, and it is a budget for the whole chain, not for the screen alone. Companies answer with faster screens: the 2016 Rift ran at 90 hertz, which makes each frame about 11 milliseconds, and the Quest 2 reaches 120. Too much delay, in Abrash's words, is a recipe for motion sickness.

Why does the inner ear care? Its semicircular canals tell the brain when the head rotates, and that system works with vision to keep your view steady as you move. A late picture can break that partnership. The leading idea about why VR makes people sick is sensory conflict. The senses that report motion, eyes, inner ear and body position, disagree. Commonly cited causes include a low frame rate, input lag and the focus conflict. Designers fight back with a fixed reference in the view, teleport-style jumps instead of gliding, and time, since many adapt with use. In a 2021 study of 11 adolescents, a game with a moving background raised body sway and sickness more than the same game with a fixed one. For motion sickness in general, sources differ on which childhood ages are most prone: one encyclopedia entry says 7 to 12, another 2 to 12.

Part Four · From Sensorama to Quest

The road to a headset people could buy was bumpy. In 1962 Morton Heilig introduced the Sensorama, a booth with a stereoscopic color picture, fans, smells and a moving chair. He made five short films for it, found no backer, and the work stopped. Jaron Lanier, who founded VPL Research in 1984, is credited with bringing the phrase virtual reality into popular use. In 1995 Nintendo's Virtual Boy, a tabletop console with a red display built from 224 LEDs and an oscillating mirror, drew complaints of headaches and eye strain, and it was gone in about a year. The turn came with Palmer Luckey's Oculus Rift. Its Kickstarter ran in 2012, Facebook announced a $2 billion purchase in 2014, and a consumer model shipped in 2016. The first standalone Quest, in 2019, removed the computer.

Few people need convincing that headsets are fun. The proof for other uses is thinner but real, and not always about headsets. In a randomized, blinded study of 16 surgical residents in 2002, those trained on a screen-based VR simulator until they met an expert standard dissected a gallbladder 29 percent faster. They made 1.19 errors a case against 7.38 for the others. Each group had only 8 people. A 2019 meta-analysis of 30 trials with 1,057 participants found VR exposure therapy for anxiety-related disorders about as effective as in-person exposure. NASA's Johnson Space Center has crews wear head-mounted displays to rehearse spacewalks. None of these is a study of ordinary children playing games.

Part Five · Children and the evidence

Now the question people search: how old do you have to be? As of November 2026, Meta's pages say a Quest account has a minimum age of 10, and children under 10 cannot use Quest 2 or 3. Ages 10 to 12 use parent-managed accounts, and Meta says ages may differ by country. Sony says PlayStation VR2 is not for children under 12. Apple says Vision Pro is for ages 13 and up, and that even 13-year-olds should be supervised. Meta calls headsets not recommended for younger or smaller-sized children. The American Academy of Pediatrics says most companies pick 12 or 13 because a headset may not fit well or may be heavy. The pages quoted here do not say what evidence the exact numbers rest on. The same pages cover the room. Meta warns that virtual boundaries do not physically stop you, and Apple says its headset will not detect all obstacles.

What do medical groups say? The American Academy of Pediatrics says brief exposure to VR has not been shown to cause harm in children, especially with parental supervision. It also says the safety data are limited. It suggests substantial breaks, about 15 minutes, between sessions to avoid cybersickness, which typically fades when use stops. It adds that children may struggle to describe when their eyes bother them, and may not take it off when something feels wrong. The same page reports the American Academy of Ophthalmology as seeing no reason to think headsets damage eye development, though long-term studies are lacking. The Canadian Association of Optometrists is more cautious. It notes that a child's visual system keeps developing, and says little is known. The two views fit together. No known harm is not the same as a long study showing none.

The studies behind those statements are short. In 2020, Tychsen and Foeller had 50 children aged 4 to 10 play a flying game on a PlayStation headset for two 30-minute sessions. Visual acuity, refractive error, eye alignment and stereoacuity did not change significantly. Balance fell about 9 percent, which missed statistical significance. Scores for eye discomfort, fatigue and motion sickness rose slightly but significantly. Three children quit in the first 10 minutes. The authors concluded that young children tolerate immersive VR without noteworthy effects on visuomotor function. In 2017, Turnbull and Phillips put young adults in a headset for 40 minutes. Binocular measures did not change, but the choroid, a layer of blood vessels behind the retina, thickened by about 10 microns. They suggested a headset may not be a myopia-causing stimulus. Both studies are about minutes, not years.

Here are the two positions, set side by side. One side says the best evidence so far shows no short-term damage to young eyes, so a flat age limit outruns the data. The other side says a 30-minute test cannot rule out effects of years of use on a visual system that is still developing, so the companies' caution is reasonable. Both can be right about what the studies show. What these studies have not measured is the long run. Do years of use change development? Does a five-year-old differ from a twelve-year-old? Will better optics remove today's problems? Researchers are still studying. For now the pediatricians' statement is a fair summary: brief use has not been shown to cause harm, and long-term studies are lacking.