October 7, 2026 Independent · Est. 2026 Subscribe
Paper breakdown

Some people reach 75 with the memory of a 45-year-old. A brain scan is starting to read the difference.

Stanford researchers built one score from a resting brain scan that tells super agers from Alzheimer's patients about 85 percent of the time. In more than 1,300 healthy adults, it flagged who was drifting toward decline, and did it earlier than amyloid, tau or genes.

November 18, 2026 · 7 min read · Fact-checked
Share X Bluesky LinkedIn Facebook Email

Some people reach their late 70s and beyond with memories that work like a much younger person's. Researchers call them super agers, and the label is strict. In the work behind this piece, a super ager is someone 75 or older who scores on memory tests at the level typical of a 45-year-old, while keeping normal thinking and planning for their age. They are not simply aging well. They are holding onto something most people slowly lose.

Why some brains hold up while others fall to Alzheimer's is one of the biggest open questions in the field. A team at Stanford, led by Yu Zhang with lead author Kanhao Zhao, took a fresh angle on it. Instead of hunting for a single faulty protein, they looked at how the brain's regions talk to each other at rest, and asked whether that pattern alone could tell a resilient brain from a vulnerable one. The study was published November 4 in Science Advances.

Their answer is a single number they call the ARC score, short for Alzheimer's resilience-vulnerability connectome. A low score means a brain whose activity looks like a super ager's. A high score means one that looks more like a brain with Alzheimer's. The striking part is what that score did in people who were still perfectly healthy.

What the team measured

The scan at the center of the study is resting-state functional MRI. A person lies still and does nothing in particular, and the machine tracks small ups and downs in blood flow all over the brain. Regions whose activity rises and falls together are counted as connected. Map every one of those relationships and you get a functional connectome, a portrait of the brain as a working system rather than a picture of its anatomy.

To learn what resilience and vulnerability look like in that portrait, the team started with two extreme groups, both 75 or older. One was 180 super agers. The other was 110 people with Alzheimer's disease. The scans came from several large open research collections, including the Open Access Series of Imaging Studies, the Harvard Aging Brain Study, and the Alzheimer's Disease Neuroimaging Initiative. From the gap between those two groups, the team built the ARC score, a single readout of where any given brain sits between them.

On its home turf, the score worked. It could tell super agers from people with Alzheimer's disease about 85 percent of the time. Tested on a separate group of patients it had never seen before, that fell to about 71 percent. Both numbers are worth reading slowly. They describe how cleanly the score separates two groups picked to be as different as possible, not how well it diagnoses anyone. The team is explicit that the ARC score is not a diagnostic test.

Super-ager-like low ARC score Alzheimer's-like high ARC score most older adults sit somewhere in between
A sketch of the idea, not the study's data. The ARC score places a brain on a line between two extremes. Scans like a super ager's sit toward the low end, scans like an Alzheimer's brain toward the high end. The team built the score from 180 super agers and 110 people with Alzheimer's, then read it off the brains of more than 1,300 healthy adults.

The real test was in healthy brains

Separating people who already have Alzheimer's from people who clearly do not is a low bar. The harder question is whether the same pattern shows up early, in brains that still look fine. So the team applied the ARC score to more than 1,300 adults between 55 and 70 who had no sign of cognitive trouble, and then followed what happened to them. The datasets tracked people for up to 12 years, though most covered about five to six.

In that healthy group, a brain's ARC score carried real information about its future. It predicted who would go on to show cognitive decline, who would build up more of the amyloid-beta that clumps into plaques, and who would be diagnosed with mild cognitive impairment or Alzheimer's. The team weighed the score against the markers doctors already lean on, including age, genetics, and amyloid and tau scans. "Our model shows comparable or better prediction ability," Zhao said, "and it can do so at an earlier time point." Adding the ARC score on top of those existing markers sharpened the prediction further, which suggests it is carrying something they miss.

That earliness is the whole point. "What we really care about," Zhang said, "is defining early markers that can detect earlier signs of cognitive decline." A marker that moves before the usual ones is a marker that could buy time.

A score is a snapshot, not a sentence

Several cautions keep this in proportion. The first is that the two groups the score was built on, super agers and people with Alzheimer's, are the extremes. Comparing them makes for a clean signal, but it is not the same as sorting typical older adults, who live in the wide middle. The researchers picked the extremes on purpose, to cut down on the biological noise that muddies the average brain, and that choice is a strength and a limit at once.

The second is that a score is a snapshot. Brain activity shifts over time, and the share of people whose scans look super-ager-like tends to shrink with age, as injuries and other conditions add up. A low score today is not a guarantee, and a high one is not a verdict. In the healthy group, the clearest divergence turned up in the datasets that ran longest, and since the follow-up mostly ran only a few years, the long view is still thin.

The deepest caution is the honest one. The score predicts. It does not explain, and it does not treat. Nobody has shown that changing a person's brain pattern changes where they end up. As the team frames it, the finding is a first step toward defining a target, not a target anyone has hit.

Why this is hopeful

For twenty years, most of the money and effort in Alzheimer's has aimed at one thing, clearing the amyloid that forms plaques. That work has paid off, though only modestly and by a margin still argued over, as we covered in our look at whether the new drugs work. This study sits in a different lane. It treats resilience as something you can measure, and perhaps one day move, rather than something you either have or you do not. It is a close cousin of the resilience research we wrote about earlier this month, reached through brain wiring instead of spinal fluid.

The researchers are already looking for the handle. Zhang points to two halves of the problem, knowing which part of the brain to aim at and knowing how to change it. Tools like cognitive training and transcranial magnetic stimulation, a non-invasive way to nudge brain activity that is already used for depression, are candidates, though neither has been shown to slow cognitive decline. The team is preparing a trial that would measure people's ARC scores before and after cognitive training, the kind of test that could turn a marker into a method. "If we can precisely identify such an early indicator," Zhang said, "I'm optimistic that we can reverse or prevent this process through advanced intervention technologies." That is a hope, stated plainly as one, and the honest framing is what makes it worth taking seriously.

A score is a snapshot, not a sentence.

The bottom line

Stanford's team turned the gap between the sharpest old brains and the most impaired ones into a single number read from a quiet scan that takes a few minutes. It tells those extremes apart about 85 percent of the time, and in more than 1,300 healthy adults it flagged who was drifting toward decline earlier than the markers in use today. It is not a diagnosis, and it is not a treatment. What it offers is a measurable picture of resilience, and a reason to think the brains that hold up are doing something specific, something a future therapy might learn to copy. In Alzheimer's Awareness Month, a map of what protection looks like is a good thing to have.

Get the next post in your inbox.

A new deep dive by email. No fluff. Unsubscribe anytime.

We use your email only to send you the Review (via Buttondown), and we never share it. Privacy.