TL;DR
-
Muse analyzed 189 users (ages 31-83) across 1,228 Stroop attention-test sessions
-
Users who trained more consistently tended to show lower Stroop interference, reflecting better cognitive-control performance.
-
Consistent practitioners in their 60s-70s outperformed sporadic practitioners in their 30s-40s (192 ms vs. 219 ms interference)
-
Training consistency was linked to lower interference regardless of age (β = -54 ms, p = 0.048)
-
The cognitive performance gap between consistent and sporadic practitioners didn't widen with age. Both groups aged at a similar rate.
-
This is an observational finding, not a clinical or medical claim. Muse is not a medical device
Cognitive control, the split-second ability to override an automatic response, tends to slip a little more each decade. It's one of the more reliable patterns in aging research.
That’s why when Muse researchers looked across one of the world's largest at-home EEG datasets, for an answer to the question: can the way people train their brain affect how well their brain performs?
One pattern stood out: the people who showed up more consistently tended to perform better.
What is cognitive control, and why does it fade with age?
Cognitive control is the umbrella term for two everyday abilities:
-
impulse control (pausing before an automatic reaction) and
-
selective attention (filtering out what doesn't matter so you can focus on what does).
It's what lets you stop yourself mid-interruption, catch a mistake before you hit send, or stay locked onto a task in a noisy room.
This ability declines gradually with age, and the pattern shows up at several levels:
-
in age-related changes to dopamine signaling in the prefrontal cortex,
-
in reduced white-matter integrity linked to slower attention-task performance in older adults, and
-
in weaker electrical brain-response markers of control during attention tasks.
Can I measure my brain’s age?
Yes. Brain age can be estimated from EEG data using a foundation model trained on resting brain rhythms, and Muse is built to do exactly this. One way researchers study healthy cognitive aging is by comparing a person's "brain age," how old their brain looks based on resting EEG patterns, to their actual age. In Muse's analyses, brain-age estimates among Muse users differed from chronological age by about ±7 years. Peer-reviewed research using Muse recordings also reported a similar finding.
The methodology
Muse analyzed 189 users, ages 31 to 83, comparing those who trained most weeks of the year against those who trained only occasionally, using the Stroop task to measure the result. Full study details are below.
[long-term-perspective]What is the Stroop task?
The Stroop task shows users colour words printed in matching or mismatching ink. For example, the word RED may appear in blue ink, and the user must respond to the ink colour.
Stroop interference is the extra time taken to respond to mismatching trials compared with matching trials. Measured in milliseconds, lower scores indicate better cognitive-control performance.
Try it yourself - say the ink color out loud, not the printed word:
RED BLUE YELLOW GREEN
[/long-term-perspective]
What did the data show?
The takeaway, in plain terms: training regularly matters.
Consistency, not volume, carried the association. Training consistency was inversely associated with Stroop interference (β = −54 ms, p = 0.048), and the gap between consistent and sporadic practitioners held at every age band tested, a persistent 40 to 50 millisecond offset rather than a widening one.
Consistent practitioners in their 60s and 70s averaged 192 ms of interference. Sporadic practitioners in their 30s and 40s averaged 219 ms. On this test, the older, consistent group outperformed the younger, sporadic group.
<long-term-perspective>
"What's striking is how steadily this held across the whole age range. The people who practised regularly kept their edge on a demanding attention task, and the most consistent older adults matched younger ones. That's the kind of real-world pattern that becomes visible when thousands of at-home sessions are studied over time."
— Dr. Walter Greenleaf, Neuroscientist and Digital Health Expert, Stanford University
</long-term-perspective>
The study at a glance
Cohort: 189 Muse users, ages 31 to 83, who completed at least 3 Stroop sessions within six months of their first Stroop task.
Training data: roughly 53,000 completed biofeedback sessions across the group, queried over the year preceding each Stroop task.
Consistency: the percentage of weeks in the prior 52-week window with at least one Muse session. Consistent users trained in about 59% of weeks; sporadic users in about 6%.
The task: a 50-trial Stroop task in the Muse app. Half the trials show a color word in matching ink ("RED" in red); the other half clash ("RED" in blue). Users tap the ink color. The mismatch trials are what require pushing past the automatic reading response, and the extra time that costs is what got measured.
Result: consistency predicted lower interference, independent of age (β = −54 ms, p = 0.048).
Limits: observational and cross-sectional. Users chose their own training habits, so this points to an association, not proof that training sharpens attention. People who train consistently may also differ in other healthy habits this analysis can't account for.
Learn how resistance training supports brain health
Track your own brain age
Muse is bringing brain-age tracking into the app, so you can follow this measure over time rather than as a one-time snapshot. Premium members can also track Alpha Peak, a measure of cognitive performance drawn from resting EEG, alongside brain age for a fuller picture over time.
Try now on Muse S Athena with Muse Premium.
Findings are observational, drawn from a self-selected cohort of 189 Muse users, and correlational, not evidence of a causal, therapeutic, diagnostic, or efficacy effect. Brain-age estimates typically vary by about ±7 years. Muse is not a medical device.
FAQs
Q: How often should I do neurofeedback brain training to see benefits?
A: Our analysis found that training in most weeks predicted better attention performance than training a lot but sporadically. There's no established minimum dose, but the data points toward regularity over volume, roughly one session most weeks outperformed occasional longer stretches of practice.
Q: Does brain training actually work as you get older?
A: The evidence is still developing and our analysis is observational, not a controlled trial, but in this dataset, consistent Muse users in their 60s and 70s performed as well on an attention test as sporadic users decades younger. It doesn't prove training reverses cognitive aging, but it's a real-world pattern worth taking seriously.
Q: What age does cognitive decline typically start?
A: Cognitive decline in areas like processing speed and cognitive control can begin as early as someone's 30s, gradually accelerating with age. It's driven partly by changes in the prefrontal cortex, the brain region responsible for overriding automatic responses. Rate and timing vary a lot between individuals.
Q: What is neurofeedback and how does it work?
A: Neurofeedback is a technique where a person receives real-time feedback on their own brain activity, typically measured by EEG, and uses that feedback to help guide their brain state during practice. Muse headbands use this approach through biofeedback sessions that respond to a user's live brain signals.
Q: Is EEG-based brain age backed by peer-reviewed research?
A: Yes. Muse's brain-age estimates, drawn from resting EEG data and a foundation model, land within about ±7 years of chronological age in Muse's analyses, a finding consistent with peer-reviewed research using Muse recordings (Banville et al., 2024, Imaging Neuroscience).
Q: In the context of brain-training, can consistency in a habit matter more than the amount of time you put in?
A: In this analysis, yes, for attention performance specifically. Training regularity was a stronger predictor of Stroop interference than total practice time or age. Whether that generalizes to other cognitive habits is an open question this data doesn't answer.
