Executive control

Executive function assessment with three separate scores

Three short tasks in one run — 24 conflict trials, 24 task-switching trials and 6 planning problems, about five minutes, free and with nothing held back. The result is a difference in milliseconds for inhibition, a second difference in milliseconds for shifting, and a count out of six for planning, printed side by side and never added together. Adding them is the one thing a battery must not do: the whole reason to run three tasks is that the same person can be ordinary on two and unusual on the third.

100% free
No signup
54 scored items
Three scores, no composite
~5 minutes

Three short sections, 54 scored items, about five minutes. Each section produces its own figure and the page never adds them, because the point of running three is that the same person can be unremarkable on one and unusual on another.

Inhibition — the arrow in the middle
24 trials. Five arrows appear; name the direction of the middle one and ignore the four beside it. The figure is how much slower you were when the neighbors disagreed with the target.
Shifting — two rules, one digit
24 trials. A digit appears with a word above it saying which question to answer about it: odd or even, or under or over five. The figure is how much the answer costs on the trials where the question just changed.
Planning — count the moves without moving
6 problems. Two bead arrangements appear side by side and you press how many moves separate them. Nothing is draggable: the whole point is to measure the working out with the doing taken away.

Sections 1 and 2 answer with two keys — F for left or odd or under five, J for right or even or over five — and section 3 with the number keys 2 to 5. Every one of them has on-screen buttons as well.

Practice adds 9 items with feedback after each one and no bearing on any figure. The response window only applies to the scored trials.

How to take the executive function assessment

Three sections, two keys for the first two and four for the third.

Section 1 — answer the middle arrow and ignore its neighbors
Five arrows appear in a row and you press F if the middle one points left, J if it points right. On half the trials the outer four point the other way, and the figure this section produces is how much slower those trials were. The response window is 1.5 seconds on purpose: this effect shrinks when there is time to be careful, so a generous window would measure care rather than conflict.
Section 2 — the same two keys, but the question keeps changing
A digit appears with a word above it. On ODD / EVEN, F means odd and J means even; on UNDER / OVER 5, F means under and J means over. The word changes without warning, and the cost of an unannounced change is the measurement. The word and the digit arrive together, so there is no time to get ready — which produces the largest switch cost a design can produce, and makes this figure incomparable to a study that gave a second of warning.
Section 3 — say how many moves, without making any
Two bead arrangements appear side by side on pegs holding three, two and one bead. Press 2, 3, 4 or 5 for how many single-bead moves take the left board to the right one. Nothing is draggable: a normal planning task measures thinking and doing at once, and removing the doing is the only way to get a planning number that has no motor component in it at all.

Technical specifications

Sections and items	24 flanker trials (12 congruent, 12 incongruent), 24 switching trials (12 switch, 12 repeat) plus one unscored lead-in, and 6 planning problems — 54 scored items. Practice adds 9 more, all excluded from every figure
Inhibition figure	Incongruent minus congruent mean response time, with a 95% interval on that one difference. Reference span in healthy adults: 30-100 ms — Eriksen & Eriksen (1974), Effects of noise letters upon the identification of a target letter in a nonsearch task, Perception & Psychophysics
Shifting figure	Switch minus repeat mean response time, with a 95% interval, plus the accuracy in each cell — a switch cost that shows up as errors instead of milliseconds is the same failure in different clothes. No reference span is quoted
Planning figure	Correct move counts out of 6 with a Wilson interval, the exact binomial probability of doing that well from four options by guessing, and the median time to answer. Guessing averages 1.5 of 6
The planning board	Three beads on pegs holding 3, 2 and 1, which gives 36 reachable arrangements. Every stated minimum is a shortest path found by searching that whole graph, so the answer key is a proof rather than a designer's intention
Switching design	Cue and digit appear simultaneously, so the preparation interval is zero. Digits 1, 3, 6 and 8 give the same answer under both rules and 2, 4, 7 and 9 give opposite ones, drawn in equal numbers so the two cells are not confounded with that conflict
What voids an item	A response inside 150 ms of the stimulus painting, one that arrives before it paints at all, and any item during which this tab went to the background. A voided item is added back at the end of its own section
No composite score	There is none, and none is coming. Summing a millisecond difference, a second millisecond difference and a count out of six needs a weighting nobody has published, and the sum would hide which of the three was out of line — which is the only thing three tasks can tell you that one cannot

Frequently asked questions

Why three scores instead of one executive function number?

Because the three abilities are correlated enough to share a name and separable enough that one score cannot stand for all of them. The measurement that established this extracted three latent factors from nine tasks and found them correlated in the range 0.42 to 0.63 — clearly related, and clearly not the same thing. A single number would be defensible only if that range sat near 1.0. What makes three numbers worth having is precisely the case where they disagree, and a sum deletes exactly that case.

The literature says inhibition, updating and shifting. Where is updating?

It is not measured here, and the reason is a boundary rather than an oversight. The canonical updating tasks — keep track of the latest member of several categories, monitor a stream of tones, recall the last four letters — are memory tasks, and memory in every form belongs to a different site in this portfolio rather than to this one. Rather than dress a memory task up as something else, this page substitutes planning, which the clinical tradition has treated as an executive function since Shallice's work in 1982 and which no other page in this section samples. The three sections are therefore inhibition, shifting and planning, and the About section below says what the substitution costs.

Why is section 3 not a proper tower puzzle?

Because a tower puzzle measures planning and execution together and this section is meant to isolate the first. When you move beads, the number you produce contains the route you worked out, the moves you made, the moves you took back and the time your hand spent traveling — which is why the dedicated Tower of London page splits its clock in two. Removing the beads entirely leaves the question and nothing else: look at two boards, work out the distance between them, press a number. It is a smaller measurement, and it is a cleaner one.

Is a large switch cost bad?

It is uninterpretable on its own, which is why the page prints the two cells' accuracy next to it. A switch cost can be spent in either currency: some people take an extra 300 ms on a switch trial and stay accurate, others answer in the same time and get the switch trials wrong, and those two runs describe the same difficulty while producing opposite millisecond figures. The design also inflates the cost deliberately — the cue and the digit appear together, so nobody gets to prepare — which means the number here is larger than one from a study that showed the cue first, and comparing the two would be comparing designs rather than people.

Can I use this before a hiring assessment?

Only as a demonstration of what conflict and switching feel like under a clock, because no employer's assessment is built from these three tasks. Cognitive-ability tests used in hiring are timed batteries of reasoning items with their own norms held by their publishers; what this page gives you is a feel for answering fast under an unannounced rule change, which is a real component of those experiences and not a preparation for any particular one. The reasoning-item practice lives elsewhere on this site.

Why do two of the sections use the same two keys?

To keep the motor part of the measurement constant across them, and to make one specific comparison honest. Both differences reported in milliseconds are subtractions within a section, so any cost carried by the keys themselves cancels — but if section 1 used two keys and section 2 used four, the two difference scores would sit on different baselines and a reader would inevitably compare their sizes anyway. Reusing F and J means the difference between the two figures is about the two operations rather than about the hands. Section 3 uses four number keys because its answer has four possible values, and it reports no millisecond difference at all.

Nine practice items seems like a lot. Can I skip them?

Yes, and there is a button for it on the start screen. They exist because two of the three sections change what the keys mean partway through the run, and a visitor discovering the second mapping during the first scored trial contributes a 900 ms trial to a cell whose whole purpose is a mean. The practice items have feedback after each one and no response window in the sections that are timed, so nothing about them is a measurement — and if you have taken tasks like this before, the scored blocks teach you nothing you already know.

What 'executive function' names, and what a five-minute run can measure of it

The term covers the things that make a plan happen rather than the plan itself: holding a goal while something else demands attention, changing rule when the rule stops working, and working out an order of operations before starting. The measurement that turned that list into something testable came in 2000, when nine tasks given to 137 undergraduates, nine tasks, three latent factors were reduced to three latent factors — inhibition, updating and shifting — that were correlated at 0.42-0.63 and still statistically distinguishable. Both halves of that result matter. The correlations are why the three belong under one name; the separability is why a single score for all of them cannot be defended, and why this page prints three figures with three units and no arithmetic joining them. Miyake, Friedman, Emerson, Witzki, Howerter & Wager (2000), The unity and diversity of executive functions and their contributions to complex 'frontal lobe' tasks: a latent variable analysis, Cognitive Psychology.

This run measures two of those factors and swaps the third. Inhibition is sampled with a five-arrow conflict task and shifting with an unannounced change of rule, both of which reduce to a difference between two of your own conditions — the only kind of reaction-time figure a browser can defend, since your monitor, your keyboard and how awake you are cancel out of a subtraction. Updating is absent because its canonical tasks are memory tasks, and memory is not measured on this domain at all; the honest options were to leave the factor out or to relabel a memory span as something it is not, and this page leaves it out. In its place is planning, which the neuropsychological tradition has treated as an executive function since 1982 and which the other pages in this section do not sample. The cost of the swap is that the three figures here are not the three factors of the 2000 paper, and the page should not be read as reproducing that analysis.

The reference figures behave differently in each section, and that is worth knowing before you read your own. The inhibition difference has a published span behind it — 30-100 ms, per Eriksen & Eriksen (1974), Effects of noise letters upon the identification of a target letter in a nonsearch task, Perception & Psychophysics — so your figure can be put beside it, with the caveat that spacing and response mapping move the effect around. The planning count has no published reference, because reading a move count off a static board is a reduction nobody has normed. And the switch cost has none for a more interesting reason: the two switching measures that do carry published numbers are unusable here, one because its norm tables are a clinical product and the other because the examiner correcting each error is part of the published score. If you want any one of these three measured properly rather than sampled, the pages are next door — the Stroop test and the flanker task for conflict, the stop-signal task for cancelling an action already under way, the Wisconsin card sorting test for abandoning a rule, the Tower of London test for planning with the beads put back, and the selective attention test for the filtering question asked without a clock on it.

The figure this page is built on

Correlation between the executive factors: 0.42-0.63 between the three latent factors.

Miyake, Friedman, Emerson, Witzki, Howerter & Wager (2000), The unity and diversity of executive functions and their contributions to complex 'frontal lobe' tasks: a latent variable analysis, Cognitive Psychology · 137 undergraduates, nine tasks, three latent factors

These correlations are between latent factors extracted from three tasks each, not between single task scores — a single flanker difference and a single switch cost from one browser session agree with each other far less well than this range suggests, because each carries its own measurement error on top. The figure is here to explain why the three numbers are printed apart rather than added, and it is not a prediction about any one person's three results.

Why no single score, and no percentile on any of the three

The behavior rating inventories used clinically are published instruments with age-normed T-scores, and their scores feed diagnostic decisions this site does not participate in.

Memory is deliberately not measured on this site, so there is no memory figure to compare yours against. That is a decision about what this site is for, not a gap in the published research: span and recall are a separate ability with their own literature, and a page here that reported one would be reporting something it never set out to measure.

The published norms are pen-and-paper, administered by an examiner who stops the participant on every error and has them correct it before continuing. That error-correction procedure is part of the score, and a browser version that simply rejects a wrong click is timing a different task; mouse or touch input against a hand moving a pen changes the motor component again. The stratified cell means and standard deviations are also not reproduced here because this file could not verify them to the digit.

One more comparison worth having in front of you before section 1: the conflict effect in the arrow task is smaller than the color-word one it is usually mentioned beside. The published spans are 30-100 ms against 50-200 ms for the Stroop task, and the reason is where the distraction sits — beside the target rather than inside it. A 40 ms figure here is not a worse result than a 120 ms figure there; it is a different subtraction.

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 executive function disorder or frontal-lobe damage. Only a qualified professional, working with more than a browser, can make that judgment.

Where the three figures are worked out

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.

All 54 response timestamps stay in a JavaScript array in this tab and are used once each, to form two subtractions and one count. Nothing is written to storage and nothing is sent anywhere, which is why a reload loses a finished run and why the copy button hands you plain text rather than a link to a saved result.