The transfer problem
Practise anything and you improve at it. The question that matters for cognitive health is whether that improvement transfers — whether getting faster at a matching task makes you better at remembering a name, following a conversation, or managing a household.
The honest summary of the research is that near transfer happens reliably and far transfer mostly does not. People get better at the trained task and at tasks closely resembling it. The gains tend not to travel into everyday function.
What the large trials found
Commercial brain-training products have been through enough scrutiny, including regulatory scrutiny of their advertising, that the pattern is reasonably clear. The improvements are real and measurable within the games. The evidence that they produce the everyday benefits the marketing implied is weak.
Which is worth knowing before spending money, and is the same standard this site tries to apply to its own product.
What does seem to generalise
- Learning something genuinely new and difficult. An instrument, a language, a craft with a technique to acquire. The common thread is sustained effort at something you are initially bad at.
- Conversation. Another person is unpredictable, has to be modelled in real time, and will not wait while you work out your answer. A crossword does none of that.
- Movement. Still the item with the strongest evidence base in the whole field, which remains inconvenient for anybody selling anything.
So should you stop doing the crossword?
No. Do it because you enjoy it, which is a perfectly good reason and does not need a cognitive justification attached. The error is not doing puzzles; the error is doing puzzles instead of the things with better evidence, and believing the swap is covered.
An hour of crossword and no walk is a worse week than half an hour of each. That is the whole of the argument.