Response restraint

Go / no-go test scored on commission errors

Press the spacebar for a disc, press nothing for a ring, and the run counts the rings you pressed on anyway — 150 shapes in about three and a half minutes, free, with the result on screen and no account anywhere. Four out of five shapes are discs so that pressing stops being a decision, which is the only condition under which withholding is worth measuring. The rate comes out beside the two numbers that set it, the go-to-no-go ratio and the interval between shapes, because it means nothing without them.

100% free
No signup
150 trials
4 discs per ring
Rate with a 95% interval

150 shapes arrive one every 1,400 ms — about three and a half minutes with two breaks in it. 120 of them are discs and you press for every one; 30 are rings and you press for none. The point of the four-to-one split is that pressing stops being a decision, and what gets measured is whether you can stop a hand that has already started.

The run, spelled out

Shapes: a filled disc or an open ring of the same diameter, in one pale gray, no hue
Your job: press Space for a disc and nothing at all for a ring
Mixture: 120 discs to 30 rings — 4 to 1, which is what makes pressing automatic
Beat: one shape every 1,400 ms, held by setting each wait from the length of the trial before it
Window: 800 ms from the shape appearing, then the trial closes
Void: pressed inside 150 ms, pressed before a shape was up, or interrupted by this tab going behind something
Order: runs of the same shape capped at 12; this one has a longest disc run of 10 and 5 places where two rings touch

Two rings in a row are left in on purpose. Forbidding them would mean every ring is followed by a disc, and a rule like that is worth more to the hand than to the design: after one ring you would know the next shape without looking.

Practice runs on a 1.5-second window instead of 0.8 and names the shape after every trial. Its 20 trials reach no figure on this page.

How to take the go/no-go test

One key, two shapes, and a rhythm that does most of the work.

Put one finger on the spacebar and leave it there
A disc means press; a ring means do nothing at all. The two shapes have the same diameter and the same pale gray and differ only in whether the middle is filled, so no part of this depends on telling colors apart. If you have no keyboard, the wide bar under the panel does the same job — say so to yourself first, because tapping changes what the run measures and the result will tell you by how much.
Run 20 practice trials that name every shape back to you
Practice gives you 1,500 ms instead of 800 and says after each trial what the shape was and what you did. None of it reaches any figure, and the pressing habit it builds is the thing being tested afterwards, so it is worth having. Skip it if the shapes are already obvious to you.
Hold the rhythm through 150 trials and read the rate
Shapes land every 1,400 ms whatever you do — a fast answer lengthens the wait that follows it so the next one still arrives on the beat. Two breaks come after trial 50 and trial 100. At the end you get the number of rings you pressed on out of the rings that reached you, a Wilson interval on that proportion, the discs you missed, a d-prime separating how well you saw the shapes from how ready you were to press, and the average speed of the three trials before each error.

Technical specifications

Trials	150 scored — 120 discs and 30 rings, a 4:1 mixture — plus 20 optional practice trials that reach no figure. Two breaks, after trial 50 and trial 100
Stimulus	A filled disc and an open ring of identical outer diameter, both in one pale gray on a dark panel. Shape and fill only: there is no hue in the task and therefore no second route around one
Response	The spacebar, read by physical position, held down by one resting finger. A full-width on-screen bar takes a tap instead, and the result counts those separately because a thumb that has to travel was never poised to be inhibited
Beat	One shape every 1,400 ms. The trial machine closes a trial the moment a key lands, so each wait is set from the length of the trial before it — otherwise a withheld ring would push the next shape back by the whole 800 ms window and the gap itself would give the last shape away
Response window	800 ms from the shape appearing. Nothing after that counts, and a disc that goes unanswered is recorded as a miss rather than dropped
What voids a trial	A press inside 150 ms of the shape painting, a press that arrives before it paints, and any trial this tab spent behind another window. A voided trial is requeued so the run still delivers 150
Run limits	No more than 12 of the same shape in a row. A tighter cap is possible at this mixture and is the wrong idea: at four to one it forces almost every ring to sit between fixed-length stretches of discs, and a countable ring is not an unexpected one
What leaves the page	Nothing until you press copy, which writes plain text to your own clipboard — the rate, the two design numbers, the outcome counts and the run seed, with no trial-by-trial times in it

Frequently asked questions

Why are four out of five trials discs? Would an even split not be fairer?

An even split would not be a go/no-go task at all. The whole paradigm rests on making the press automatic before asking you to hold it back: at four to one your hand starts the movement before the shape has been identified, and the ring then has to catch something already under way. At one to one you would simply be classifying shapes, with no habit to interrupt, and the errors you made would be errors of perception rather than of restraint. The published versions run between three and four go trials per no-go for this reason, and the rate they report is not comparable across those settings.

What exactly does the commission rate depend on besides me?

The mixture, the interval between shapes and how much speed pressure the task applies — and each of those moves the rate across most of its available range. Widen the gap between shapes and the same person makes far fewer errors, because there is time for the decision to catch the hand. Push the ratio to nine discs per ring and the errors climb. That is why this page prints its ratio and its measured interval next to the rate every time the rate appears, and why a commission rate you read somewhere else, quoted bare, is not a number you can compare yours to.

Two rings came one after the other. Is that a mistake in the sequence?

No, and forbidding it would be the mistake. If a ring were never allowed to follow a ring, then every ring would tell you the next shape is a disc — a free, reliable cue arriving exactly where the task needs you to have none. The shuffler here only caps how long a run of one shape may get, and lets the rings clump where the draw puts them. The run panel counts how many back-to-back pairs your sequence happened to contain, so you can see what you were given.

I pressed on several rings. Does that point to ADHD?

No. It does not point anywhere, and the arithmetic is the reason rather than caution: the rate is set as much by this page's ratio and rhythm as by the person taking it, so two runs of the same task with different settings produce different rates from the same hand. There is also no defensible population figure to compare yours against, which is why none appears here. Clinical versions of this task exist, they use fixed timings that are sold with their own norms, and they are one input among many in an assessment that a website cannot carry out.

Why does the page care what my speed was doing before each error?

Because a press on a ring can come from two opposite failures and the score alone cannot tell them apart. A hand outrunning the decision produces errors after a burst of unusually fast trials; attention that has drifted away from the screen produces them after unusually slow ones. The page takes the three go trials immediately before each ring, averages them, and compares the rings you pressed on with the rings you held — with an interval on the difference, because with a handful of errors per run the split is descriptive and should be read as one run rather than as a trait.

I missed a few discs. Is that the same kind of error?

It is the opposite one, and reading them together is more informative than either alone. A missed disc is an omission — the hand was not there when it was needed — while a press on a ring is a commission. Someone with no commissions and many misses was not inhibiting well; they were disengaged, and a rate of zero bought that way is not a good score. The d-prime and the bias figure make the same point more formally: d-prime says how far apart the two shapes sat for you, and the bias says which way your hand was leaning, and they move independently.

Why is my average reaction time not the headline number here?

Because this task is deliberately built to make that number uninteresting. The rhythm and the four-to-one mixture exist to turn the press into a reflex, so how fast that reflex fires is mostly a fact about the beat it was trained on rather than about you. The go speed is still reported, and the pre-error window uses it, but the measurement this page is for is what happens on the one trial in five where the reflex has to be stopped.

What a go/no-go actually measures, and what the rate cannot be compared to

The task is old, simple and routinely misdescribed. What it measures is restraint: a response that has become prepotent through repetition, and the ability to keep it from being emitted when the stimulus says not to. Everything in the design serves that one word. The lopsided mixture builds the prepotency; the fixed interval between stimuli keeps it built, because a rhythm you can settle into is what turns pressing into something the hand does rather than something you decide; the short response window stops you from waiting out each trial to be sure. Remove any of the three and the errors that remain are perceptual, not inhibitory. That is why a page that reports a commission rate without saying what its ratio and its interval were has reported a number with no referent — and why both appear beside the rate here every time it is printed.

The confusion worth clearing up is with the paradigm next door. The stop signal task looks like this one and measures something different: there the signal arrives after the go stimulus, so the response is already under way and has to be cancelled in flight rather than withheld. Cancellation has a latency that can be estimated; restraint does not, and this page never pretends otherwise — it counts errors, which is the honest quantity a task without a second signal can produce. Nearby but different again are the interference tasks, where nothing has to be stopped and something has to be ignored: the Stroop test puts the distractor inside the target and the flanker task puts it beside one. Filtering without any speeded response at all is the selective attention test, and a run across several of these at once is what the executive function assessment assembles. If what interests you is impulsivity where the pressure is reward rather than rhythm, the balloon analogue risk task asks the question from the other side.

The third thing this version does that most browser versions skip is look behind each error. A commission rate treats every failure as the same event, and they are not: some arrive on the back of a burst of unusually quick responses, which is a hand that has outrun its own decision, and some arrive after unusually slow ones, which is a person who had stopped watching several seconds earlier. The two produce identical scores and different explanations, so the run reports the average of the three go trials in front of every ring and compares the failed ones with the held ones directly. It also reports a d-prime and a bias term, which separate how distinguishable the shapes were from how ready your hand was to move — two things a single error count adds together and cannot take apart again.

Why no reference figure appears beside your rate

The commission rate is almost entirely determined by the design: the proportion of no-go trials, the inter-stimulus interval and any speed pressure move it across most of its possible range. A number quoted without those three parameters describes nothing, and published studies use too many different combinations to average.

So the page reports your own count and the width a run of 30 rings gives it, and offers no external comparison at all — which is a smaller claim than the one you will find elsewhere, and the only one that survives being checked.

A reaction time here is the interval between the frame that painted the stimulus and the timestamp the browser attached to your key, both read from the same monotonic clock. What neither can see is the display pipeline behind it, so on a 60 Hz screen roughly 16 ms of every figure below is the machine rather than you. That is the timing floor: two numbers closer together than that are the same number, and this page reports no precision it cannot support.

This is a measurement exercise, not a clinical assessment. It reports what you did on this page against a stated reference and nothing more — it cannot establish ADHD, an impulse-control disorder or an attention problem of any kind. Only a qualified professional, working with more than a browser, can make that judgment.

Where your 150 presses are counted

Every number on this page is worked out by JavaScript running in the tab you are reading it in. Your answers, your reaction times and your score are never uploaded, logged or kept — which is also why the test carries on working after you disconnect from the network, and why nothing here can be held back behind an email address.

The timestamps sit in one JavaScript array in this tab and are used for two subtractions — a shape onset from a keypress, and one cell mean from another. Nothing is written to storage and nothing survives a reload, which is why a finished run disappears if you refresh and why the copy button is the way to keep it.