Attention Tests
Four laboratory tasks that catch attention in the act of failing.
Attention is measured by watching it break down. Every task in this hub is designed so that the interesting event is an error or a slow trial, not a fast one.
The Stroop test sets two pieces of information against each other, then prices the clash in milliseconds. The sustained attention test makes pressing so habitual that holding back gets hard, then counts the times you failed to hold back.
The psychomotor vigilance task does almost nothing for three minutes and counts your lapses. It is the standard laboratory measure of sleep loss. The trail making test makes you swap between two sequences and prices the swapping.
None of these is a concentration score. Each gives you one specific number attached to one specific way of failing. That is far more useful than a single figure claiming to represent your focus.
Three of the four report a gap rather than a total. Subtracting one half of a test from the other removes everything the two halves share, including the hardware lag no browser can see. The vigilance task is the exception, and it is the one worth repeating, because practice barely moves it.
The 4 tests in this domain
Numbers and reading in this domain
The gap between two scores is the measurement
Three of the four tasks here report a gap rather than a total, and that is deliberate. Your Stroop score is your clashing time minus your matching time. Your trail making score is part B minus part A.
Subtracting one half from the other removes everything the two halves share. That is exactly the stuff we cannot control in a browser: your screen lag, your mouse, your eyesight, how sleepy you are.
It also removes the main reason totals mislead. A slow person with excellent control can post a worse raw Stroop time than a fast person with poor control, while showing half the interference. The raw time is mostly about speed. The gap is about control.
Why one of these tasks is boring on purpose
The psychomotor vigilance task is dull by design. It was built at the Walter Reed Army Institute of Research and refined by Dinges and Powell in 1985.
The aim was to beat the two things that ruin most cognitive tests, which are practice and natural ability. Almost nobody gets better at this task by repeating it, and being clever barely moves the score. What moves it is how much sleep you have had.
That makes it the most useful task here for measuring yourself over and over. A lapse count that has doubled over a fortnight is real information about your sleep. No memory game can tell you that.
What a bad score usually means
Five explanations, in rough order of likelihood. You were tired. You were interrupted. You were on a phone. You had never seen the task before. Or your hardware is slow.
Only after all five does anything about your attention come into it. Those five explain the overwhelming majority of poor scores.
These tasks are used clinically. That use involves a trained examiner, standard equipment and a quiet room. It also involves figures adjusted for your age and education, and someone who knows the rest of your history. A browser tab supplies none of that.
So if your results here worry you, print nothing and diagnose nothing. Take the concern to a doctor, who can order the assessment that actually answers the question.
Running these tasks properly
Sit at a desk. Use a mouse or a real keyboard rather than a trackpad if you have one, and then use the same device every time. A trackpad and a mouse differ by more than most of the effects you are looking for. Close anything that notifies. Put the browser full screen.
If you plan to compare across days, keep the order and the time of day fixed. Treat your first run of any new task as practice that does not count.
The sustained attention and vigilance tasks in particular react to whether you have just had coffee. Control for that if you want your numbers to mean anything.