cognitive science
and more
Can you brain-train your way to perfect eyesight?

Over the past month I’ve seen a huge increase in the number of visitors to the Gabor-patch generator on this site. A Gabor patch is a type of stimulus that psychologists like to use for experiments. It’s a pretty weird stimulus, as you can see in the example below, not really useful for anything except experimentation. So why the sudden interest? Why are thousands of people suddenly generating Gabor patches?

A Gabor patch

The rush on Gabors appears to have been triggered by a paper that appeared last month in Current Biology. In this paper, Deveau and colleagues claim that you can dramatically improve vision through repeated training on a simple visual task that uses–you guessed it!–Gabor patches. Even more remarkably, the participants in the study, who were university baseball players, even showed a marked improvement in on-field baseball performance!

Whoah! Improving your eyesight simply by looking at some weird images! If you can’t wait to get started, you can buy the training program in the form of an iPad app called ULTIMEYES Pro ®. The app, priced at a mere $5.99, is developed by Carrot Neurotechnology, a company founded by the senior author of the paper.

But wait, what’s that smell? Oh yes … It’s something fishy.


Let’s start with a bit of background. The main claim of the paper is that real-life vision can be improved through a simple visual brain-training program. How plausible is this claim? Most visual impairments are due to defects of the eye’s lens. For example, as you get older, your lens loses some of its flexibility, and you are consequently less able to adjust focus, and to see sharply at certain distances. The lens also becomes cloudy, which results in blurred vision. Obviously, brain training will not do anything to improve the quality of your eye’s lens.

But even if you cannot brain-train your eye back into shape, you may be able to train your brain to compensate for imperfections of the eye’s lens. Just like you can use software to unblur a blurry photo (but not as well as Hollywood would have you believe!), your brain may learn to reconstruct poor visual input by processing visual input more effectively. This is the idea behind perceptual learning, the subfield of visual neuroscience that has inspired ULTIMEYES Pro ® and similar vision-training programs.

Perceptual learning works. If you perform a simple visual task for a long time, you become better at it. But unfortunately this type of learning doesn’t generalize very well. For example, if you train yourself to discriminate the orientation of a Gabor patch to your left, you hardly become better at discriminating a Gabor to your right. Some tasks generalize a bit better than this, but generally speaking perceptual learning is highly specific to the task that you trained on. Therefore, you wouldn’t expect visual training with Gabor patches to lead to a general vision improvement, and certainly not to better baseball skills. Deveau and colleagues recognize this problem, but claim to have overcome the ‘curse of specificity’ with their ‘novel integrative perceptual learning program’.

To sum up, it is not entirely impossible that perceptual learning could lead to real-life vision improvements. But so far there is little evidence to suggest that it does, and the claims made by Deveau and colleagues are therefore very bold. And bold claims need to be backed by solid evidence.

So what about the evidence? Let’s again take a step back and review some basic research methodology. If people follow a training program (or take a drug, follow a therapy, etc.), they are pretty much guaranteed to get better, even if the training doesn’t work. This is purely the power of suggestion. For example, you can make someone’s headache go away by giving them a yeast pill and calling it an aspirin. This is widely acknowledged, and therefore, if a new drug is tested, there is always a control group that takes a placebo (a fake pill, a fake treatment, a fake training, etc.). Importantly, no-one, including the researchers themselves, should know who is taking the real drug, and who is taking the placebo. That’s the only way to prevent suggestion from exerting its incredible power.

Unfortunately, there was no proper control group in the study by Deveau and colleagues. The control group consisted of baseball players who did not receive any training. Therefore, the trained players knew that they were being trained, and the control players knew that they were not. The improvements observed in the trained group might therefore simply result from the power of suggestion. Of course, the training may have been effective, but due to the study’s poor design we cannot tell. In a properly designed study, the control group would have performed a fake training program that would not have led to vision improvements, but did make the participants feel like they were being trained.

What I find quite astonishing is that in the scientific paper the authors cite anecdotal reports to illustrate how well their training program works. According to the trainees, they could “see the ball much better”, were “able to distinguish low-contrast things”, and could “see further”. Still not convinced? Just listen to what Tina has to say: “After using ULTIMEYES® for only a couple of weeks I began to see a difference. Less eye strain, and I could now see people clearly. Definitely noticed a difference and the eye chart test proved it!” Well … ok, if Tina says so. Such self-reports are extremely susceptible to suggestion, and are a very unreliable source of evidence.

So what are we left with? With the interesting hypothesis that perceptual learning leads to real-life improvements in vision. But this hypothesis is only very weakly supported by a methodologically weak study. And with a massive conflict of interest to boot, because the senior author is commercially exploiting the training program.

But what annoys me most is that this paper gives hope, quite possibly false, to the many people who suffer from vision impairments. And it tricks these people into paying for an app, which is quite possible worthless.

Verdict: bad science.


Deveau, J., Ozer, D. J., & Seitz, A. R. (2014). Improved vision and on-field performance in baseball through perceptual learning. Current Biology, 24(4), R146–R147. doi:10.1016/j.cub.2014.01.004

Fahle, M. (2005). Perceptual learning: specificity versus generalization. Current Opinion in Neurobiology, 15(2), 154–160. doi:10.1016/j.conb.2005.03.010