Why training a game trains the game

Near transfer is improvement on a task closely resembling the one you practised — you drilled a memory span task, and you are now better at a different memory span task. Far transfer is improvement on something structurally different: your work, your studying, an untrained ability, everyday functioning.

Almost every disagreement about brain training is a disagreement about which of these two a study showed. The sellers quote the first and imply the second. Once you hold them apart, most of the literature becomes readable in an afternoon.

Sala and colleagues' 2019 second-order meta-analysis — a meta-analysis of meta-analyses — found that working-memory training reliably produces near transfer, with a size that depends on the population, and estimated the true far-transfer effect against active control groups as close to zero.

Simons and colleagues' 2016 review, 83 pages of it, put the same conclusion in a sentence worth reading twice: there is extensive evidence that brain-training interventions improve performance on the trained tasks, less evidence that they improve performance on closely related tasks, and little evidence that training enhances performance on distantly related tasks or everyday cognitive performance.

Gobet and Sala returned to the field in 2023 and described it as in search of a phenomenon. That is not a hostile framing from outside; it is two researchers who have spent years pooling the trials describing what the pooling keeps producing.

A passive control group does nothing while the trained group plays a game for six weeks. At the end, the trained group scores better on a range of measures — and so would a group that had spent six weeks doing crosswords, or being visited weekly, or simply expecting to improve. Some of what looks like training is the trial itself.

An active control group does something equally engaging and equally structured, so the comparison isolates the training rather than the attention. When the field moved from the first design to the second, the far-transfer effects shrank toward zero. That is the single most important methodological fact in this literature, and it is invisible in any summary that reports only whether scores went up.

The same pattern holds where you would most want it not to: working-memory training does not enhance older adults' cognitive skills in the pooled evidence, and a 2024 review found no supporting evidence for far transfer of perceptual and cognitive training to sport.

It is good for the task. Getting genuinely better at a hard, adaptive task is a real skill acquisition, and the record of it is real: your own scores, on the same exercise version, on the same input mode, over time. That is not a proxy for anything. It is the thing itself.

It is also honest company for the rest of a plan. Two of the highest-value habits in the seven-day protocols the Snapshot produces — recalling before reviewing, and protecting a single block of attention — have better evidence behind them than any game does, and neither is a game. A product that sold you the game and hid the habit would have its priorities exactly backwards.

And a hard task is intrinsically worth playing for people who like hard tasks. That is a modest claim, and it is one we can actually keep.

No game here will ever be described as making you sharper or smarter. No game here promises anything about your work, your studying or your memory in the rest of your life. No score will be called a brain age, a level, a percentile or a rank. No game result feeds any conclusion about you elsewhere in the product without its own separately graded justification.

One line sits under every exercise saying what it trains, and it names the task. Where a member asks whether this makes them smarter, that line plus this article is the answer, and neither of them says yes.

There is a reason the industry needed a rule rather than a preference. The category's largest brand paid two million dollars to settle Federal Trade Commission charges over claims its games improve performance at work and at school and delay age-related decline — claims made, the complaint said, without the science to back them. Every one of those is a claim this product is structurally unable to make.

A grade belongs to one claim, in one population, for one preparation — never to a compound, a technique or an article. The weakest link is the part of the chain between the evidence and the thing you would actually do that is doing the least work.