Do Brain-Training Games Actually Work? Lumosity, the ACTIVE Trial, and the Evidence

Do Brain-Training Games Actually Work? Lumosity, the ACTIVE Trial, and the Evidence

The honest answer to the question "do brain training games work" is yes, but almost never in the way the apps promise. Train on a game and you get better at that game. Whether that improvement carries over to your memory, your job, or your odds of avoiding dementia is a separate question, and for two decades the science said the answer was mostly no.

Then a 20-year follow-up to the largest cognitive training trial ever run complicated the story. One specific type of training, done in a lab, appears to lower dementia risk decades later.

So the picture is not "brain games are a scam" or "brain games rewire your brain." It is more interesting than either headline. Here is what the evidence actually shows.

Key Takeaways

• Most commercial brain games produce near transfer (you improve at the trained task) but little far transfer (improvement in untrained, real-world abilities).
• In 2016, the FTC fined Lumosity's parent company $2 million for advertising benefits it could not back with science.
• The ACTIVE trial, the largest cognitive training study ever run, found that only one training type, speed-of-processing, showed lasting protective effects.
• A 2026 follow-up linked that single training type to a roughly 25% lower dementia risk up to 20 years later.
• The lesson for brain games evidence: the type of training matters far more than the hours you log.

The Core Problem: Near Transfer vs. Far Transfer

Brain training lives or dies on one concept: transfer. Near transfer is when practice on a task improves performance on very similar tasks. Far transfer is when it improves different, untrained abilities, the thing every app actually sells you.

Spend a month on a number-matching game and you will get faster at number matching. That is near transfer, and it is real. The marketing claim is that this speed bleeds into your memory, your focus at work, your reaction time on the road. That is far transfer, and it is the part the research keeps failing to find.

This distinction explains almost every confusing headline you have read. A study showing people "improved" usually measured the trained skill. A study showing "no benefit" usually measured something the person never practiced. Both can be true at once.

The question of brain training transfer is not academic. It is the entire commercial promise. If the only thing you get is better at the game, you have bought an expensive hobby, not a cognitive upgrade.

The Lumosity Reckoning: When Marketing Outran the Data

In January 2016, the Federal Trade Commission settled with Lumos Labs, the company behind Lumosity, over deceptive advertising. The brand had built a business on the idea that a few minutes of play several times a week delivered serious real-world benefits.

According to the FTC, the ads suggested the games could delay age-related decline and protect against dementia and Alzheimer's, among other claims. The agency's verdict was blunt. "Lumosity preyed on consumers' fears about age-related cognitive decline, suggesting their games could stave off memory loss, dementia, and even Alzheimer's disease," said Jessica Rich, Director of the FTC's Bureau of Consumer Protection. "But Lumosity simply did not have the science to back up its ads."

The penalty was concrete. As part of the settlement, Lumos Labs, the company behind Lumosity, agreed to pay $2 million in redress and to notify subscribers of the FTC action and provide them with an easy way to cancel their auto-renewal.

This is the founding document of the modern skeptic's case. The FTC was not arguing the games did nothing. It was arguing the company sold far transfer it could not prove. That is the core of the lumosity science debate, and it reframed how regulators and researchers talk about the whole category.

The ACTIVE Trial: The Strongest Evidence Either Way

If you want to know whether structured cognitive training can change real outcomes, the ACTIVE trial is the study that matters most. It is the largest, longest, and most cited.

What the Study Did

The Advanced Cognitive Training for Independent and Vital Elderly trial was funded by the National Institutes of Health and run across six US cities. It enrolled 2,802 adults into the study in 1998 to 1999 to assess long-term benefits of participants randomized to three different types of cognitive training, memory, reasoning, and speed of processing, compared to a control group who received no training. In the three training groups, participants received up to 10 sessions of 60 to 75 minutes of cognitive training over five to six weeks.

About half of each training group later received short "booster" sessions. This was not an app you played on the bus. It was supervised, adaptive, and tightly controlled.

The 10-Year Results

The first eyebrow-raiser came at the decade mark. In a ten-year follow-up of the randomized trial with 2,802 persons at a mean baseline age of 73.6 years, participants in each intervention group reported less difficulty with instrumental activities of daily living. In plain terms, the trained groups felt more capable at everyday tasks like cooking and managing money than the untrained control group.

That is a hint of far transfer, the holy grail. But self-reported daily function is a soft measure, and the effects were modest. The trial's bigger surprise came much later.

The 20-Year Dementia Finding

In February 2026, a new analysis tracked the same participants two decades out, and the result was specific rather than sweeping. In a 20-year study of adults 65 and older, those who completed five to six weeks of adaptive speed-of-processing training, along with a few booster sessions, were markedly less likely to develop dementia, including Alzheimer's disease, even two decades later.

The magnitude is worth stating precisely. Participants who received the boosted speed training had a 25% lower dementia risk compared to those who received no training, making it the only intervention in the trial to show such a lasting protective effect.

Read that twice. Memory training did not do it. Reasoning training did not do it. Only one narrow form of training, processing visual information quickly under time pressure, moved the needle. Of three training types tested, only speed-of-processing training under time pressure with adaptive difficulty reduced dementia risk, cutting it by 25% (hazard ratio 0.75), while memory training and reasoning training had no marked effect.

So, Does Brain Training Improve Memory?

Here is the uncomfortable synthesis. The question "does brain training improve memory" got a clear answer in the ACTIVE trial, and it was no. The memory-training group, the one explicitly built to improve memory, showed no lasting dementia protection.

The one thing that worked was not a memory drill at all. It was speed of processing. This is the deepest lesson in far transfer cognitive training: benefits, when they appear, are weirdly specific and rarely match what the label says.

It also means most commercial brain games are aiming at the wrong target. Generic "train your memory" puzzles are the exact category the ACTIVE data suggests does the least.

How the Options Compare

The pattern is clear. The closer a "cognitive" intervention gets to your whole body, the better the far-transfer evidence looks. The pure mental gym is where the promises get thin.

What This Means for You

You do not need to delete your puzzle apps. Games are fine entertainment, and staying mentally engaged has its own value. Just calibrate the expectation. You are buying enjoyment and near transfer, not a memory upgrade.

If your goal is long-term brain health, the evidence points toward sleep, cardiovascular fitness, and social and cognitive engagement, the unglamorous basics. The one brain-game exception with hard outcome data, speed-of-processing training, is narrow and specific, not the typical app.

The broader skill here is reading brain games evidence the way the FTC and the ACTIVE researchers do. Ask whether a study measured the trained task or a real-world outcome. Ask whether the effect was specific or general. That habit protects you from the next Lumosity.

Frequently Asked Questions

Do brain training games work for everyday cognition?

For everyday cognition, the evidence is weak. Most commercial games reliably improve performance on the trained task (near transfer) but show little carryover to untrained, real-world abilities (far transfer). The one well-documented exception, speed-of-processing training in the ACTIVE trial, was a specific lab protocol, not a typical app. If you enjoy the games, play them, but treat real-world cognitive gains as unproven for most products.

Why did Lumosity get fined?

In 2016, the FTC settled with Lumosity's parent company, Lumos Labs, over deceptive advertising. The company had promoted its games as able to delay age-related decline and guard against conditions like dementia and Alzheimer's. The FTC found the company lacked adequate science for those claims and required it to pay $2 million in redress and make subscription cancellation easier. The case became a landmark for how cognitive-product marketing is regulated.

What is the difference between near transfer and far transfer?

Near transfer means improvement on tasks very similar to the one you trained. Far transfer means improvement on different, untrained abilities, like a number game boosting your work focus. Near transfer is common and easy to demonstrate. Far transfer is rare and hard to prove, and it is the benefit most brain-training products advertise. Almost every dispute about brain games comes down to confusing these two.

Does brain training improve memory?

Largely no, based on the strongest evidence. In the ACTIVE trial, the memory-training group showed no lasting protection against dementia. The only training type with durable real-world benefit was speed of processing, which targets how fast you take in visual information, not memory drills. So the activity most apps sell as "memory training" is the one with the least supporting outcome data.

What made the ACTIVE trial different from a phone app?

The ACTIVE trial was a supervised, adaptive, NIH-funded randomized controlled trial across six cities with 2,802 participants and 20 years of follow-up. The training was structured, difficulty scaled to the user, and outcomes were measured rigorously over decades. Commercial apps are self-directed, unsupervised, and rarely studied with that rigor. The format and accountability gap is one reason lab results do not automatically apply to the app you downloaded.

Is any brain training worth doing?

Speed-of-processing training has the best evidence, with a 20-year link to lower dementia risk in the ACTIVE trial. Beyond that narrow case, the highest-yield "cognitive" investments are aerobic exercise, quality sleep, and genuine mental and social engagement. These have broader and more consistent far-transfer evidence than puzzle apps. If you train cognition formally, choose protocols with published outcome data, not just marketing.

The Real Lesson: Specificity Beats Hype

Brain training is neither miracle nor fraud. The honest reading of the data is that practice makes you better at what you practice, and only rarely at anything else. The single bright spot, speed-of-processing training, earned its reputation through 20 years of follow-up, not a slick ad campaign.

That is the standard worth carrying into every cognitive claim you encounter. Did the study measure the trained skill or a real outcome? Was the effect specific or general? How long was the follow-up? The answers usually separate the signal from the sales pitch.

The category's history is a reminder that good intentions and engaging design are not evidence. Outcomes are.

Related from Roon

• The neuroscience of neuroplasticity and how the brain actually changes
• What makes a true nootropic, and what the science requires

How Roon Thinks About Cognitive Claims

This article exists because the gap between what a product promises and what it can prove is the most important thing to scrutinize, especially in cognition. That is the standard we hold ourselves to at Roon. We would rather tell you what the evidence supports than sell you far transfer we cannot back.

To be clear about what Roon is and isn't: a sublingual pouch is not brain training, and it will not lower your dementia risk or replace sleep and exercise. What it does is offer a measured, ingredient-transparent approach to short-term focus. Each pouch pairs 80 mg of caffeine with 60 mg of L-theanine, plus 25 mg of methylliberine (Dynamine) and 5 mg of theacrine (TeaCrine), built for a 5 to 10 minute onset and a 6 to 8 hour window with no jitters, no crash, and no tolerance buildup.

If you value evidence-first thinking over hype, that is the same lens we ask you to bring to us. Try Roon when you want honest, near-term focus, and keep the long game where the data says it belongs: in your sleep, your fitness, and your habits.

Written by Roon Team