Perception training vs brain training: what is the difference, and does either work?

Two products, two claims

A brain-training app asks you to play games — remember a sequence of tiles, find the odd shape, do fast arithmetic — and tells you that doing so will make you better at something outside the app: your memory in meetings, your attention while driving, your mind in old age. The games are the means; the general ability is the claim.

A perception-training app asks you to make a sensory judgement — rebuild a colour, find a pitch again, hold a press for a remembered duration — and tells you how far off you were, in units. The claim is that the judgement gets more accurate with practice. Nothing is promised about your memory, your attention or your mind. The measured thing is the whole product.

Those two claims have been tested separately, and they have had very different outcomes.

Does brain training work? What the evidence says

In October 2014 the Stanford Center on Longevity and the Max Planck Institute for Human Development published a consensus statement signed by more than seventy cognitive scientists and neuroscientists. Its position was that the scientific evidence did not support claims that brain games reduce or reverse cognitive decline, and that the marketing of such products was outrunning the science. Two months later a second letter, signed by more than a hundred and thirty scientists and practitioners, argued the opposite: that a substantial body of work showed certain training regimens do improve cognitive function in ways that generalise.

To settle which letter was right, Daniel Simons and colleagues reviewed the published intervention studies the industry cited, in a paper that appeared in Psychological Science in the Public Interest in October 2016. They set out what a convincing study would need — proper control groups, enough participants, outcomes measured outside the trained tasks — and applied those standards to the literature. Their conclusion: brain training reliably improves performance on the trained tasks, shows smaller gains on closely related tasks, and shows no convincing improvement in everyday cognitive performance or distantly related tasks. They also observed that the two 2014 letters disagreed largely because their authors held different standards for what counts as evidence.

The regulator reached a similar place by a different route. In January 2016 the US Federal Trade Commission announced that Lumos Labs, the company behind Lumosity, would pay US$2 million to settle charges of deceptive advertising. The FTC's complaint said the company had made unsubstantiated claims that its games improved performance at school and work and protected against conditions including dementia, and had used customer testimonials without disclosing that they were collected through a contest. The settlement required the company to hold competent and reliable scientific evidence before making such claims in future. Lumosity remains a well-made product; its marketing language changed.

Does perception training work? What the evidence says

The research field behind perception training is perceptual learning, and its central result points in the opposite direction from brain training's marketing. Practising a sensory discrimination with feedback produces a lasting improvement, and the improvement is specific: it stays close to the trained stimulus, the trained location and the trained precision. Transfer to related tasks is partial; transfer to unrelated abilities is not something the field claims.

That specificity is exactly why the narrow claim survives scrutiny. A perception-training app that says "your hue discrimination will improve" is saying what the literature says. One that said "your hue practice will sharpen your memory" would be making a brain-training claim with a different coat of paint. The honest version of perception training is modest, measurable and, so far as the evidence goes, true.

Side by side

How to read an app's claims

You do not need to read the research to tell the two apart. The marketing copy and the result screen will tell you.

1. Look for the general nouns. IQ, memory, focus, attention, cognition, brain age, mental fitness. Each is a promise of transfer, and transfer is what the 2016 review could not find.
2. Look for units. A perception score comes with a distance: degrees of hue, cents of pitch, a percentage of a duration. A brain-training score is usually a point total or a percentile against other users, which measures the game, not you.
3. Look for the scoring rule. If the app will tell you how a score is computed, it is measuring something. If the formula is a secret, it is probably a motivational number.
4. Look for the word "prevent". Any claim about dementia, decline or ageing is a medical claim. No consumer app has the evidence for it.
5. Look at what happens when you stop. A product that lowers your score for missing days is managing your behaviour. One that lowers its confidence in a stale reading is managing its own honesty.

Where Percore stands

Percore is a perception-training app and says so without hedging. The first question in the FAQ on its landing page is whether it is brain training, and the answer is no: it makes no claim about intelligence, memory, focus or any benefit that carries over to work or life, and it does not diagnose, treat or assess anything medical. What it measures is how close your reproduction was to the target you were just given — a colour, a pitch, a duration, a shape — and it shows you that over time, against your own history only. The formulas are published on the senses page and in How Percore scores. If you want a better memory, this is not the app, and it will not pretend otherwise. If you want to know whether your eye for colour is steadier than it was in September, it is.