Coding task

Digit symbol substitution test — ninety seconds, one count

A key of nine marks paired with the digits 1 to 9 sits at the top of the screen, one mark appears below it at a time, and you type the digit it is paired with for ninety seconds. The score is the number you converted correctly — free, no account, and the number appears the moment the clock stops. The key is shuffled for every run, so coming back tomorrow measures substitution again rather than how well you memorized yesterday’s pairs.

100% free
No signup
9 pairs
90 seconds
Key reshuffled each run
No scaled score

Nine marks, nine digits, and ninety seconds to convert as many marks into digits as you can. The key is visible the entire time, so nothing here is a memory task — what is being measured is the rate at which a lookup can be made and acted on, over and over, without the rate falling apart.

The key — it stays on screen for the whole run

What the ninety seconds actually contain

Item: one mark, shown large, with the key above it
Answer: the digit paired with that mark — number row or keypad, or the buttons
Score: marks converted correctly before the whistle
Key: drawn afresh for this run, so coming back tomorrow is not a memory test
Errors: counted, never corrected, and the run does not stop for one
Void: whichever mark is up when the whistle goes, plus any mark the tab was hidden during

Every mark carries a spoken name — “two waves”, “barred circle” — and the key announces which digit each name pairs with, so the task can be worked entirely through a screen reader. Be aware of what that changes: hearing a name and recalling its digit is a verbal lookup, while seeing a mark and finding it in a row of nine is a visual search. Both are real measurements and they are not the same one, so a count taken that way belongs beside your own earlier counts taken the same way and nowhere else.

Take 9 untimed items first, one per pair, each telling you the right digit afterwards. The clock does not exist during them and none of them reaches the score.

How to take the coding task

One lookup, one keypress, repeated until the whistle.

Read the key once, then stop trying to learn it
Nine marks, each with a digit under it, stay visible for the whole run. Glancing back at the key is the task, not cheating at it: this measures how fast you can look something up and act on it, which is why nothing is ever hidden. People who try to memorize the nine pairs first lose the seconds they spend doing it and gain them back slowly, because the pairs learn themselves as you go.
Take the nine untimed items if the interface is new to you
One item per pair, no clock, and the right digit shown after each one. They run on the same key the timed block will use, so the practice is not wasted, and nothing in them reaches the count. Clear the checkbox to skip them; the clock costs you nothing until it starts.
Convert marks for ninety seconds and let the errors stand
Type its digit on the top-row numbers or the keypad, or tap one of the nine buttons. There is no way back to a mark you got wrong and no penalty subtracted for it — the score is simply how many you got right, so hesitating over a doubtful one costs more than being wrong about it. When the clock expires the mark on screen is discarded rather than counted, and the result shows the count, the rate per minute, the accuracy and how much faster the last thirty seconds were than the first.

Technical specifications

Key size	9 pairs — the count the Wechsler form uses. Each is a mark drawn for this site as an SVG path, not a reproduction of any published symbol set
Interval	90 seconds, counted from frames delivered to this tab. Time with the tab in the background is not charged to you, so a run interrupted twice takes longer than 90 seconds of wall clock and still contains 90 seconds of coding
Direction	Mark to digit, the reverse of the pencil form. A drawn answer would measure mouse control, so the answer is a keypress on 1 to 9 — the same reversal the Symbol Digit Modalities Test made in print in 1982
Score	Marks converted correctly before the whistle. Errors cost the item and nothing more: no penalty term is subtracted, and one item in nine (11.1%) would come out right from a blind guess
Response	Digits 1-9, taken from the top row or the numeric keypad and read by physical key position, plus nine on-screen buttons that work throughout. A tapped run is reported separately because aiming nine buttons costs time a resting hand does not
Practice	9 optional untimed items, one per pair, each showing the correct digit for 700 ms afterwards. They are excluded from every figure and they run on the key the timed block will use
Reported split	Three 30-second thirds, each with its own count, error count and median time per mark, plus a Welch interval on the change from the first third to the last
What leaves the page	No request is made to any server during a run. A count leaves this tab only when you ask for it, as plain text on your clipboard, and the per-mark intervals are not part of what it carries

Frequently asked questions

Why does this show a mark and ask for a digit, when the paper test is the other way round?

Because the paper test asks you to draw, and a browser cannot accept a drawing without measuring your mouse instead of your head. On the pencil form you copy a symbol into a box under each digit, and the speed of that copying is partly handwriting; move it to a screen and it becomes cursor control, which varies far more between people than handwriting does. Reversing the direction — mark on screen, digit on the keyboard — keeps the lookup and the substitution and removes the drawing. Smith's Symbol Digit Modalities Test made the same reversal in 1982, on paper, to stop a motor impairment reading as a cognitive one.

Does the key change if I take it again?

Yes, every run, and that is deliberate. A fixed key would be partly memorized after one attempt, and the second run would measure recall of nine pairs rather than the speed of looking them up — the score would climb and the ability would not have moved. Reshuffling means the only thing you carry into a repeat is familiarity with the interface, which is what the nine untimed items are there to give you on the first run anyway. The seed that produced your key is in the copied summary, so a particular run can be rebuilt exactly.

Why is there no scaled score, IQ or percentile?

Because the conversion tables are somebody else's licensed property and are not ours to hand out. There is a second reason that matters more for the number itself: those tables were built on a pencil, a printed form and an examiner with a stopwatch, and this run has a keyboard, a display refresh and no examiner. Even a table sitting in the public domain would be scoring you against a different task. What a count from here compares with is another count from here.

Do errors count against the score?

No — an error costs you the item and nothing else. Nothing is subtracted, so the score is the number of correct substitutions and not correct minus wrong, which means a doubtful mark is worth answering rather than dwelling on. Accuracy is reported separately, alongside a 95% interval, because a fast count with a poor accuracy figure has bought its speed with mistakes and the two numbers only mean something together. Most people finish this task above 95% correct; if yours is much lower, the count is measuring how you handled uncertainty rather than how quickly you work.

What happens to the mark I was in the middle of when the clock ran out?

It is discarded, not counted and not scored as wrong. There is no defensible way to credit a mark that was on screen for an unknown fraction of the ninety seconds, and the paper form has the same rule — the row you are halfway through when the examiner says stop does not count. The results panel names it in the account of what was thrown away, so the number of items started and the number that reached the count never silently disagree.

Why ninety seconds rather than two minutes?

Ninety seconds is the Symbol Digit Modalities Test's interval and two minutes is the Wechsler one, and this page runs the shorter of the two because it uses the shorter test's direction. Mixing them would be the worst of both: a count taken over the Wechsler duration with the Symbol Digit response, comparable to neither. The interval is stated everywhere the count is for the same reason a distance is stated with a running time — a count of marks means nothing without the window it was collected in, and the rate per minute is printed beside it so a run you stopped early is still readable.

Why did my last thirty seconds go faster than my first?

Because the key stops needing to be read. Every time you look up the same mark, the pairing gets a little easier to retrieve, and by the final third most people are checking the key only for the two or three marks they find least distinctive. That improvement is the normal shape of this task and it is why the results panel splits the run into thirds instead of reporting one average: a single mean hides a curve that was still rising. The interval printed with the change tells you whether your own improvement was larger than the noise, and one that straddles zero means the two thirds were the same speed.

What the coding task measures, and why it appears in every battery

Substitution tasks are older than the tests that made them famous — the pairing of arbitrary marks with digits was a laboratory exercise decades before Wechsler put a version of it into a clinical battery, and it survived every revision of that battery since. The reason is unglamorous: it is a lookup repeated under time pressure, so it recruits visual search, short-term pair retrieval, decision and motor output all at once, and a person who is even slightly slowed anywhere in that chain finishes measurably fewer items. Salthouse (1996), The processing-speed theory of adult age differences in cognition, Psychological Review, made this the load-bearing case for treating speed as the mechanism behind a great deal of age-related difference in cognition, and coding is the marker that argument leans on hardest. Its sensitivity is also its limitation: a task that picks up almost any disturbance is a good detector of change within one person and a poor description of who that person is.

Two features of the browser version are departures worth naming rather than apologizing for. The first is the direction, covered above: mark to digit rather than digit to drawn mark. The second is what this version leaves out. The clinical procedure has an optional second stage after the timed block, in which the participant is asked to reproduce the pairings from memory with the key taken away — the incidental learning check, which asks what you absorbed while you were busy doing something else. It is a genuinely interesting measurement and it is not here, because it is a memory measurement rather than a speed one, and this site draws its boundary at what a person brings to a task rather than what they retain from it. Leaving the key visible for the whole ninety seconds is the same decision from the other side: nothing on this page rewards memorizing, so nothing on it is measuring memory by accident.

The marks themselves are drawn for this site and are deliberately unlike anybody’s published key. That is partly a copyright matter and mostly a validity one: a leaked symbol set stops measuring anything the moment it is available to practice against, which is exactly why the real ones are controlled. If you want the same underlying ability estimated from more than one task rather than from ninety seconds of one, the processing speed test runs three short blocks and reports whether they agree with each other, and the brain age test puts a speeded block beside an ability that moves the other way with age. The neighboring paradigms measure different things through similar-looking work: the trail making test adds a switch between two sequences, the Wisconsin card sorting test changes the rule without telling you, the Posner cueing task moves attention before the target arrives, and the Tower of Hanoi removes the clock entirely and counts moves instead.

The figure this page will not print: Wechsler subtest norms — Digit Symbol/Coding, Block Design, Matrix Reasoning, Vocabulary

The norm tables are the product. They are copyrighted, sold with the test kit under a qualification requirement, and reproducing them would be both an infringement and an invitation to treat a browser imitation as a Wechsler score.

The whole thing costs nothing and asks for nothing: no account, no address, no wait, and nothing withheld until you identify yourself. Every figure the run produces — the count, the three thirty-second thirds, the accuracy interval and the seed the key was dealt from — is on screen the instant the clock stops, and the copy button puts all of it on your clipboard as plain text so the run survives the tab that made it.

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 a processing-speed deficit, an attention disorder or the early cognitive change people arrive at this task worried about. Only a qualified professional, working with more than a browser, can make that judgment.

Not affiliated with, endorsed by, or connected to the owner of Wechsler. Wechsler is a trademark of its owner and is used here only to name the assessment this page prepares for. The questions on this page are our own: no part of the published test is reproduced, and a score here is not comparable to a score from the real instrument.

Where the ninety seconds 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 key your run was built from, the order the marks were dealt in and every per-mark interval live in JavaScript variables inside this tab. They are read once, to produce the count and the three thirty-second summaries, and then cease to exist along with the tab. That is the trade a page with no server makes: a refresh takes a finished run down with it, so the clipboard is the only route a count has out of here.