If you are 62 years old, is your brain necessarily 62? The question sounds like wordplay until you notice that it has an operational answer. Train a model on enough MRI scans from people of known ages and it learns what brains tend to look like at each age. Show it a new scan and it returns a number. Subtract the real age from that number and you have a quantity researchers call the brain age gap — and for most people it is not zero.

What Is Actually Being Measured

Nothing mystical, and nothing about how old someone feels. The model has learned the statistical regularities of brain structure across age — volumes, cortical thickness, the shape of things — and it reports the age at which a brain like this one is most typical.

A gap of +4 does not mean a brain has aged four extra years in any mechanical sense. It means the structural pattern resembles that of people about four years older, on average, according to a model trained on a particular population with a particular scanner mix.

That distinction is easy to lose and it changes what the number can be used for. Which is why the interesting question is not what any individual’s gap is, but whether the gap behaves systematically across groups.

The Case-Control Answer

A 2026 study assembled structural MRI from 45,900 healthy controls and 2,698 patients across nine disorders grouped into four families — developmental, addiction, dementia and other psychiatric — and compared each group’s predicted age difference with controls, accounting for age, age-squared, sex and scanning site (Liang et al., PLoS Medicine, 2026).

The gap was consistently greater across disorders, and the ordering is the informative part. The largest effects were in dementia: Alzheimer’s disease at a standardised effect of 0.97 (95% CI 0.82 to 1.13) and mild cognitive impairment at 0.45 (0.34 to 0.56). Addiction followed — combined alcohol and tobacco use disorder at 0.84, tobacco at 0.72, alcohol at 0.62. Then the psychiatric group: schizophrenia 0.53, bipolar disorder 0.46, major depressive disorder 0.28.

And then the result that does not fit a simple ‘illness ages the brain’ story: the developmental disorders were not different from expected. ADHD and autism, in this analysis, did not show the elevated gap the other groups did.

The Same Question, Asked a Different Way

A second 2026 analysis pooled 65 MRI-based studies rather than assembling raw scans, and reported its results in years rather than effect sizes (Mhanna et al., GeroScience, 2026).

The largest gaps were in multiple sclerosis, at a mean adjusted 7.81 years, and Alzheimer’s disease at 5.57 years. Schizophrenia came in at 4.40 years, mild cognitive impairment at 3.77, and Parkinson’s disease at 3.52 — smaller, but statistically clear.

Two independent teams, two different methods, converging on the same broad ordering is worth more than either result alone. It is also where the agreement stops being comfortable: the meta-analysis reports stroke and bipolar disorder as descriptive only, because two cohorts and one cohort respectively were too few for a formal analysis.

One Genuinely Interesting Detail

The meta-analysis found that the brain age gap was largely independent of chronological age in the neurodegenerative disorders — but that it increased with age in schizophrenia.

That is a different shape of finding. A gap that stays constant regardless of how old someone is behaves like a fixed offset. A gap that widens with age behaves like a divergence.

It is one result from one pooled analysis and should be held loosely. But it is the kind of detail that distinguishes a measure that is merely correlated with illness from one that might eventually say something about trajectory.

What This Cannot Do

It cannot tell you your brain’s age. The gap is a group-level research measure with substantial variation between individuals inside every group, and no clinical service estimates it for the general public. A number derived from a model trained on one population, on particular scanners, does not transfer cleanly to one person on a different scanner.

It cannot establish cause. Every finding above is a comparison at one point in time between people who have a diagnosis and people who do not. Whether an elevated gap precedes a condition, follows it, or shares a cause with it is not answered by any of this.

And a null is not proof of absence. The developmental disorders showing no elevated gap is a real and useful result, but it means this particular measure did not distinguish those groups — not that nothing differs.

The meta-analysis is also candid about its own machinery: it assessed publication bias with Egger’s test and applied trim-and-fill, which are the tools you reach for when you suspect the published record is not a complete record.

So Is Your Brain 62?

Probably not exactly, and there is no way for you to find out. What the research establishes is narrower and more interesting than a personal number: that structural brain aging can be estimated at all, that the estimate varies systematically between groups, and that the variation is ordered in a way two independent 2026 analyses broadly agree on.

The practical implication is about the framing rather than the measurement. Chronological age and biological aging are separable in principle, which is what makes prevention research coherent — the whole enterprise assumes trajectory is not fixed by the calendar.

That is the same reasoning behind what the prevention evidence actually supports, and it sits alongside the broader point that neural patterns respond to changed input rather than being fixed.