The claim and why it is seductive
The brain training industry is built on a simple and appealing premise: just as you can improve your physical fitness by training specific muscles, you can improve your cognitive abilities by training specific mental "muscles." Train your working memory → get better at remembering things. Train attention → get better at focusing. Practice pattern recognition → get better at solving problems.
It is a compelling analogy. It drives a $4 billion global industry. And the research says it is mostly wrong — or at minimum, dramatically oversold.
The 2014 Stanford consensus
A letter signed by over 70 neuroscientists and cognitive psychologists (including Nobel laureates) stated: "We object to the claim that brain games offer consumers a scientifically grounded avenue to reduce or reverse cognitive decline when there is no compelling scientific evidence to date that they do."
The important caveat
This does not mean all cognitive training is worthless. It means that most commercial apps show near-zero transfer to real-world tasks beyond the specific games they train. Some narrow interventions — dual n-back, direct digit span training, speed-of-processing training — do produce genuine, transferable gains. The distinction matters enormously.
Near transfer vs. far transfer: why most apps fail
The critical concept in cognitive training research is the distinction between near transfer (improving on tasks similar to what you trained) and far transfer (improving on dissimilar real-world tasks).
Near transfer (what apps achieve)
Better score on the trained game
Improvement on highly similar tasks
Measurable in controlled lab settings
Far transfer (what people want)
Better memory in daily life
Better attention at work or school
Reduced cognitive decline with age
Meta-analyses covering 50,000+ participants consistently find effect sizes of d=0.1–0.2 for far transfer from commercial brain training programs — effects so small they are arguably indistinguishable from test familiarity effects. Users get better at taking tests, not at the underlying cognitive function.
The exceptions: what actually transfers
Speed-of-processing training (ACTIVE trial)
Strong evidenceThe most important real-world data comes from the ACTIVE trial (Advanced Cognitive Training for Independent and Vital Elderly), a 2,832-person, 10-year follow-up study. Speed-of-processing training — specifically the "Useful Field of View" exercises — reduced dementia risk by 29% at 10 years. This is the largest real-world effect size found for any cognitive intervention.
What makes it different: Speed-of-processing training uses adaptive visual tasks that push processing speed at the limit — not static difficulty games. The key is that the task genuinely requires faster real-world visual processing, not just faster button-pressing on an app.
Dual n-back working memory training
Moderate evidenceDual n-back training — where you simultaneously track the position and sound of stimuli n steps back in a sequence — shows the most consistent far transfer of any working memory intervention. Effects of 0.2–0.4 standard deviations on fluid intelligence and untrained working memory tasks have been replicated in multiple studies. The catch: it requires genuine effortful practice, not casual gaming. Most people hate doing it because it is genuinely hard.
Action video game training
Moderate evidenceUnlike puzzle or strategy games, action games (first-person shooters, real-time strategy) show genuine transfer to untrained attention, visual processing speed, and multitasking tasks. Causal studies find effects of 10–20% on reaction time and attention after 20–30 hours of training. The key mechanism is the genuine cognitive demands: real-time decision-making, rapid target detection, and sustained attention under pressure.
What actually works better than brain training apps
| Intervention | Effect size (vs. brain training) | Evidence quality |
|---|---|---|
| Aerobic exercise (150+ min/week) | 3–5× larger effect | Very high |
| Sleep optimization (7–9 hours) | 2–4× larger effect | Very high |
| Learning a new skill (music, language, craft) | Similar or larger | Moderate |
| Social engagement / complex conversation | Comparable | Moderate |
| Mindfulness meditation (consistent practice) | Comparable | Moderate |
| Commercial brain training apps | Baseline | Weak (far transfer) |
The cognitive reserve model
The strongest evidence for reducing long-term cognitive decline points to building "cognitive reserve" through lifelong engagement in genuinely novel and complex activities. This means learning truly new skills — not the mild novelty of app games. Cognitive reserve appears to delay the clinical onset of dementia even in the presence of significant neuropathology.
What Human Benchmark is (and isn't)
We want to be honest about what our tests do and don't provide. Human Benchmark tests are measurement tools, not training protocols. They measure genuine cognitive abilities — reaction time, working memory capacity, pattern recognition speed — using validated paradigms.
Using our tests repeatedly will improve your score on those specific tests through familiarity effects — and that improvement is real in the narrow near-transfer sense. But we make no claim that scoring 200ms on the reaction time test will make you a better driver or athlete unless accompanied by genuine practice in those activities.
What the tests can do: establish a genuine baseline, track the effects of real interventions (sleep, exercise, caffeine), motivate behavior change through accurate measurement, and identify relative cognitive strengths. These are valuable — as long as the measurement is not confused with training.
Measure. Don't just train.
Establish your genuine baseline today, then use real interventions — sleep, exercise, deliberate practice — and track the change objectively.
See all cognitive tests