The five ages of the brain, and what they do to a brain-age score

Cognitive testing

Say "brain age" and you probably picture a slope. You start near the top, the line runs gently downhill, and a test tells you how far along it you have got. The picture is tidy. It is also the one this site sells you a number against, which strikes me as a reason to handle it carefully rather than a reason to keep quiet about it.

Two recent studies make that slope look wrong, and they get there by different routes.

Four turning points, not one slope

A Cambridge team went through diffusion MRI scans from 3,802 people between birth and ninety. They were after wiring rather than performance. Diffusion imaging follows how water moves through tissue, and from that you can infer where the connecting fibres run. Their analysis, published in Nature Communications in November 2025, found that the reorganisation of those connections does not run as one continuous process. It breaks at four points: around 9, around 32, around 66 and around 83.

Five stretches fall out of that, and the authors call them epochs. Childhood goes up to about 9. Adolescence, in wiring terms, runs from 9 to about 32, which is a good deal more of a life than the word usually claims. Adulthood holds from 32 to about 66. Early ageing carries on to about 83, and late ageing covers the rest.

The biggest break sits at 32. Alexa Mousley, who led the work, put it this way: "Around the age of 32, we see the most directional changes in wiring and the largest overall shift in trajectory." She is describing the biggest change of direction there, which is not the same thing as the steepest decline.

Two cautions belong with that, and the researchers name both themselves. The scans come from a lot of different people at different ages rather than from the same brains followed across decades, so the epochs describe a population and not a life. The far end of the range is thin, too: there are fewer scans of people over 83 than of people at 30, which leaves that last turning point resting on less evidence than the first three.

A different clock, in the tissue

The second study looks somewhere else. Researchers led by Bing Ren examined gene regulation and the three-dimensional folding of the genome, cell by cell, in tissue from the human hippocampus, the structure most associated with forming new memories. Their report in Science this year describes a marked shift somewhere between roughly 50 and 75. Many of the brain's resident immune cells fall in number and give way to cells with more inflammatory characteristics. The cells forming the blood-brain barrier weaken, and the organisation of the genome inside the nucleus degrades.

That window does not line up with any of the four wiring turning points, which is the part I keep coming back to. It is a different measure, taken on a different timetable in a different piece of the brain. The hippocampus is also only one structure, and the study says nothing about what any of it felt like to the people whose tissue was examined.

What this does to a number like ours

Put either finding next to a brain-age score and the mismatch is hard to miss. The Cambridge work measures wiring. The Science work measures molecular housekeeping. Our test measures neither. It counts how many items you get right and how quickly you answer them, then maps that onto an age scale, and we have said before that the mapping is a presentation choice rather than a measurement.

Nothing in either study lets a ten-minute test work out which epoch you are in. You cannot read a turning point in a population curve off one person's performance, any more than you can read a national birth rate off one family.

The studies do support the smaller claim sitting underneath our own disclaimer. If the underlying biology changes direction at least four times, and if a second and unrelated process reorganises the tissue somewhere in the fifties and sixties, then a single number sliding down a single line was never going to be the right shape for describing any of it. The number is a snapshot of performance on a particular morning. It is not a position on a curve.

The part you can actually use

You can take something practical from this. It is duller than the headline, but it survives contact with the evidence.

1. Compare yourself to yourself. Population turning points say nothing about your Tuesday. Run the same test twice a month apart, on the same device, at roughly the same time of day, and that will tell you more than any comparison with a stranger.
2. Watch the breakdown, not the headline. Our brain age test reports recall, working memory, processing speed and reasoning separately, and there is a reason for that. The four do not move together, and the ways they come apart tell you more than the number that averages them.
3. Expect speed and knowledge to part company. It is the pattern already visible in processing speed, where the raw measure drifts down while the things you know carry on accumulating.

The five epochs are a new picture of how the brain gets put together across a lifetime, and the hippocampus work is a new picture of how it keeps itself in repair. Neither one tells you your brain age. If anything, they are the clearest argument yet that the phrase was always doing more work than it could carry.