Despite all this work, no one yet knows exactly what causes this visual fatigue [from watching 3-D video], or "asthenopia"; in any case, there's little reason to think it can ever be overcome.
One potential explanation for the discomfort lies with the unnatural eye movements stereoscopy elicits from viewers. Outside of the 3-D movie theater, our eyes move in two distinct ways when we see something move toward us: First, our eyeballs rotate inward towards the nose (the closer the target comes, the more cross-eyed we get); second, we squeeze the lenses in our eyes to change their shape and keep the target in focus (as you would with a camera). Those two eye movements—called "vergence" and "accommodation"—are automatic in everyday life, and they go hand-in-hand.
Something different happens when you're viewing three-dimensional motion projected onto a flat surface. When a helicopter flies off the screen in Monsters vs. Aliens, our eyeballs rotate inward to follow it, as they would in the real world. Reflexively, our eyes want to make a corresponding change in shape, to shift their plane of focus. If that happened, though, we'd be focusing our eyes somewhere in front of the screen, and the movie itself (which is, after all, projected on the screen) would go a little blurry. So we end up making one eye movement but not the other; the illusion forces our eyes to converge without accommodating. (In fact, our eye movements seem to oscillate between their natural inclination and the artificial state demanded by the film.) This inevitable decoupling, spread over 90 minutes in the theater, may well be the cause of 3-D eyestrain. There's nothing new about the idea—an article published in the Atlantic in 1953 refers to the breakdown of the accommodation-convergence ratio as a "difficulty [that] is inherent to the medium." And there's no reason to expect that newfangled RealD technology will solve this basic problem of biomechanics. ...
The eye-movement issue may even carry other, more serious risks. A long session of 3-D viewing tends to cause an adaptive response in the oculomotor system, temporarily changing the relationship between accommodation and convergence. That is to say, audience-members may experience very mild, short-term vision impairment after a movie ends. I won't pretend there's any hard evidence that these transient effects could develop into permanent problems. But if 3-D becomes as widespread as some in the industry claim—every movie in three dimensions, for example, and television programs, too—we'll no doubt have plenty of data: Small children, their vision systems still in development, could one day be digesting five or six hours of stereo entertainment per day. There's already been one published case study, from the late-1980s, of a 5-year-old child in Japan who became permanently cross-eyed after viewing an anaglyph 3-D movie at a theater.
--Daniel Engber, Slate, on why you should watch Avatar in 2-D