Methodology · updated for version 1.2
How the Attention Index works
The short answer: the Attention Index is a 0–100 summary of how you performed on three short tasks in the Attention Age app. It is not an age, a percentile, a ranking or a diagnosis. Here is exactly how it’s built.
1. What you play
The five-minute baseline has three timed tasks:
- Signal Watch — tap white signals, never the red decoy (a go/no-go task).
- Interference — judge the direction of the middle arrow while surrounding arrows may point the other way (a flanker task).
- Tracking — follow a few highlighted shapes as they move among identical ones.
2. Five category scores
Raw measurements from those tasks become five category scores from 1 to 99:
| Category | Task | Weight |
|---|---|---|
Sustained Attention Share of white signals caught, and whether accuracy held from the first third of the task to the last. | Signal Watch | 15% |
Response Control How often you held back on the red decoy (no-go trials). Tapping a decoy lowers it. | Signal Watch | 15% |
Consistency Reaction-time variation (standard deviation ÷ mean). Speed alone is not rewarded — steadiness is. | Signal Watch + Interference | 30% |
Distractor Handling How much conflicting flanker arrows slowed you down or caused errors versus calm trials. | Interference | 25% |
Object Tracking How accurately you re-found highlighted shapes after they moved among identical ones. | Tracking | 15% |
Each raw measurement is placed on a 1–99 scale using fixed reference points chosen during design. Those reference points are not population norms and don’t come from a validated dataset, so a category score means “how this compares with the app’s fixed scale”, not “how you compare with other people”.
3. The Attention Index
The index is the weighted average of the five category scores using the weights above, rounded and kept between 1 and 99. Consistency carries the most weight because it draws on two tasks and tends to be the steadiest measurement within a short session.
4. Validity checks
Every session is checked for patterns that suggest it doesn’t reflect a genuine attempt. The result is still shown, with a caution explaining why, when:
- a task recorded no usable trials,
- more than 60% of signals were missed,
- nearly every decoy was tapped too (indiscriminate tapping),
- average reaction time was under 150 ms (faster than a deliberate response), or
- Interference accuracy was close to chance.
5. Comparing sessions honestly
Weekly re-tests use the same tasks, settings and scoring so they can be compared with your baseline. Changes smaller than 5 points are described as “about the same”. Every result — including lower ones — stays in your history.
6. Practice effects
Scores on short tasks usually rise over the first few attempts simply because the rules become familiar. That is improvement at the task, not evidence of a broader change in attention. The app says so on early re-tests.
Can an app calculate your cognitive age?
Not honestly without large, representative, age-referenced data collected under controlled conditions and properly validated. Attention Age doesn’t have that. Versions before 1.2 displayed results on an age-style scale; that presentation has been retired. Older results remain in your history, labelled as earlier versions. Because the category formulas didn’t change, their index values are directly comparable with new results.
7. What the index does not measure
- It is not a medical, clinical or diagnostic test and can’t indicate whether you have any condition (including ADHD or dementia).
- It doesn’t measure intelligence, and it isn’t compared with any population or age group.
- Improvement on these tasks doesn’t show improvement in everyday attention, work or study.
8. Why results vary
Sleep, stress, time of day, caffeine, the device, grip and chance all move short reaction-time tasks. Treat any single result as one data point and look at the trend across several weeks.
Try the browser versions: the free attention test, the go/no-go game and today’s Daily Signal. Browser timing differs from the app, so the numbers aren’t directly comparable.