AbilityBenchCognitive ability testing
Cognitive ability tests that show their working
98 free tests of reasoning, attention, hearing, vision, language and space, all of them running in this tab with no sign-up and no email wall. Start with the twelve-item sampler below — three items each of verbal, numerical, abstract and spatial reasoning, thirty seconds an item — and it tells you which of the four formats cost you time and which page measures it properly. Every figure this site prints beside a score names the study it came from, and where no defensible figure exists the page says so instead of inventing one.
- 100% free
- No signup
- 98 tests
- 12-item sampler, 6 minutes
- Every reference figure sourced
Twelve items, three each of verbal, numerical, abstract and spatial reasoning, at 30 seconds an item. It samples the four formats and tells you which of them cost you time; it is not an IQ test and it does not produce one, because twelve items of our own writing have no population to be ranked against.
What this run does, before you start it
- Items
- 12 scored, three each of verbal, numerical, abstract and spatial
- Time
- 30 seconds an item, running from the frame that paints it
- Scoring
- one point an item, four options, no penalty for a wrong answer
- Chance
- 3 of 12 — a run of pure guesses scores three on average
- Practice
- two untimed items with worked answers, offered first, never scored
- Void
- an item you were on when the tab went to the background is discarded and re-run, and so is one answered inside 150 ms of appearing
Six of the twelve items are drawings. They are black on white, so no part of this page depends on telling colors apart — but they do need sight of the figure, and a visitor using a screen reader should expect the abstract and spatial thirds to be unanswerable and read the verbal and numerical scores on their own.
All 98 tests
Ten sections, from reasoning with 16 pages to oddities with 9. Each card says what that page measures that its neighbors do not.
Reasoning
16 testsCognitive Ability Test
Ninety-eight browser tests that measure instead of guessing: reasoning, attention, hearing, vision, language and space. Each one shows its method, its scoring and where its norms come from — and none of them asks for your email to hand back your own result.
You are on this page
Verbal Reasoning Test
Timed true / false / cannot-say items over short passages, the format graduate employers actually use, with the trap explained afterwards: cannot-say means the passage is silent, not that the claim is false.
Practice Test, Watson-Glaser Format
Practice for the critical-thinking appraisal used in law and consulting hiring, built on the same five sub-tests. Not affiliated with, or endorsed by, the assessment's owner.
Matrix Reasoning, Raven's Format
Matrix items in the classic progressive format, with the item-construction rules named on the page and the reason a browser version cannot yield a clinical IQ score stated plainly.
Pattern Recognition Test
Sequences across number, shape and position, reporting which of the three you are fastest at — most people are markedly better at one.
Cognitive Reflection Test
The three questions with an intuitive answer that is wrong, plus the extended set, scoring how often you overrode the first thing that came to mind.
Numerical Reasoning Test
Tables and charts read against a pace marker rather than a cutoff, with a calculator allowed, because the tested skill is finding the right two numbers rather than dividing them.
Abstract Reasoning Test
Shape sequences where you infer the rule and pick what comes next, scored on accuracy and time separately because the two trade against each other.
Critical Thinking Test
Inference, assumption, deduction, interpretation and argument evaluation as five separately scored sub-scales rather than one blurred total.
Logical Reasoning Test
Deductive items where a valid conclusion must follow from the premises alone, including the ones that are true in the world but do not follow — the single most common way people lose points here.
Non-verbal Reasoning Test
Reasoning with no language at all, which is what makes it comparable across native tongues and useful when English is not yours.
Deductive Reasoning Test
General rule to specific case, with syllogisms and conditional items, and a worked explanation of affirming the consequent when you fall for it.
Mechanical Reasoning Test
Gears, pulleys, levers and fluid pressure as diagram items, the Bennett-style format used for maintenance and trade hiring.
Inductive Reasoning Test
Specific cases to general rule over figural series — the half of reasoning that predicts learning speed rather than schooling.
Diagrammatic Reasoning Test
Process flows where operators transform symbols and you deduce what each operator does — the format used for software and engineering roles.
Matrix Reasoning Test
A three-by-three grid with one cell missing, scored by rule complexity, so the report tells you which kind of rule you missed rather than only how many.
Hiring
13 testsPractice Test, Wonderlic Format
Fifty mixed items in twelve minutes, the pace that makes this test hard rather than the items themselves. Practice only, with original questions. Not affiliated with, or endorsed by, Wonderlic.
Practice for the Hogan Assessment
What the personality-based inventories measure and how the scales are reported, with practice items in the same shape. Not affiliated with, or endorsed by, Hogan Assessments.
Situational Judgement Test
Workplace scenarios ranked by effectiveness, with the scoring explained: a real key is one employer's expert consensus rather than a right answer, which is exactly why this page has none and says so.
Practice Test, CCAT Format
Fifty items in fifteen minutes across verbal, math and spatial — the pace that most candidates underestimate. Original items. Not affiliated with, or endorsed by, Criteria Corp.
Mechanical Aptitude Test
Practical physics as diagram items — gear direction, pulley advantage, lever balance, fluid flow — the battery used for maintenance, utility and armed-forces roles.
Practice for the Predictive Index
Practice for the cognitive assessment's format: fifty items, twelve minutes, verbal, numerical and abstract mixed together. Original items, not the real bank. Not affiliated with, or endorsed by, The Predictive Index.
Aptitude Test Practice
A mixed set that samples every format on this site's hiring pages, so you can find which one costs you time before an employer does it for you.
Practice for SHL Tests
Practice in the formats used across the publisher's verbal, numerical and inductive batteries, including the adaptive pacing candidates find unfamiliar. Not affiliated with, or endorsed by, SHL.
Practice for the Caliper Test
Practice for the profile's combined personality and abstract-reasoning sections, with the forced-choice format explained. Not affiliated with, or endorsed by, Caliper.
Practice for Korn Ferry Assessments
Practice covering the numerical, verbal and logical batteries used in the firm's talent assessments. Not affiliated with, or endorsed by, Korn Ferry.
Personality Assessment for Jobs
How forced-choice work inventories are scored, why consistency checks catch faking, and what answering honestly costs and gains.
Practice for the Criteria Cognitive Test
What the raw score out of fifty actually means, and why no single percentile exists for it: the publisher norms by role and by employer, not against one working population. Takes a result from the practice drill and reads it back. Not affiliated with, or endorsed by, Criteria Corp.
Pre-employment Assessment Guide
What each family of hiring test actually measures, how long it runs, and how much it predicts — with the meta-analytic figures side by side and no claim about what any employer accepts. Written for the candidate rather than the buyer.
General
8 testsPractice IQ Test, Mensa Format
A practice set in the format of supervised admission tests, with the honest part stated up front: no online test can qualify you for anything, and only a proctored session counts. Not affiliated with, or endorsed by, Mensa.
Attention Span Test
How long you hold performance before it degrades, measured across a sustained block rather than asserted from a questionnaire.
Numeracy Test
Everyday quantitative reasoning — percentages, ratios, rates, unit pricing — at the level adult numeracy frameworks actually define, each item set in the situation it comes from.
Brain Age Test
Speed, memory-free reasoning and attention reported as three separate scores with every norm sourced, and an argued refusal to convert any of them into an age.
Mental Math Test
Arithmetic with no scratch pad and no deadline, reporting accuracy and median solve time by operation so the slow operation is the finding.
Processing Speed Test
How fast you handle simple material accurately, measured with symbol matching over a fixed interval — the ability that declines earliest and most measurably with fatigue.
Math Aptitude Test
Whether you reason with quantities rather than whether you were taught a syllabus, using items that need no formula recall.
Quick Math Test
A sixty-second sprint over mixed operations, for when you want the number and not the analysis.
Attention
8 testsStroop Test
The color-word conflict in its canonical form: congruent, incongruent and neutral blocks, reporting your interference effect in milliseconds against the published 100-200 ms range.
Flanker Task
Eriksen's arrows, measuring the cost of ignoring the arrows beside it — congruent and incongruent trials with the conflict effect reported separately from raw speed.
Selective Attention Test
Finding the signal while distractors compete, with your accuracy cost per added distractor plotted rather than summarized.
Executive Function Assessment
A short battery across inhibition, switching and planning, reported as three separate scores because they dissociate in the same person.
Continuous Performance Test
A long, dull vigilance block that measures sustained attention and impulsivity together — and never returns a diagnosis, because that is not what a website can do.
Go / No-Go Test
Response inhibition scored on commission errors, which is the number that matters here — going fast is easy, stopping is not.
Concentration Test
A grid-cancellation task in the d2 tradition, scoring speed, accuracy and the fluctuation between them across the block.
Problem Solving Test
Novel problems with no learned procedure, scored on solution path as well as outcome.
Paradigms
10 testsWisconsin Card Sorting Test
Sort by a rule you are never told, which changes without warning; scored on perseverative errors, the measure the task was built to produce.
Digit Symbol Substitution Test
The coding task from the Wechsler tradition, counting correct substitutions in ninety seconds and showing how the rate moved across the run.
Trail Making Test
Parts A and B, with the B-minus-A difference reported — the derived score that isolates switching from raw speed.
Tower of Hanoi
The classic disc puzzle with move counting against the 2ⁿ−1 optimum, playable from three discs to eight.
Lexical Decision Task
Word or non-word, as fast as you can, with the frequency effect measured on your own trials and an interval that says whether ten words a band were enough to separate them.
Simon Task
Stimulus position conflicts with response side; the Simon effect is the difference, and the page reports yours in milliseconds.
Tower of London Test
Planning measured as moves-above-minimum and first-move latency, not just whether you finished.
Posner Cueing Task
Valid and invalid cues at 100 and 700 ms: what a wrong cue costs you, and why at the longer delay the cue starts to hurt.
Balloon Analogue Risk Task
Pump for reward, stop before it bursts; scored on adjusted average pumps, the standard index of risk taking under uncertainty.
Stop Signal Task
Estimates your stop-signal reaction time with a staircase procedure — the latency of the internal brake, which cannot be observed directly.
Vision
8 testsColor Perception Test
Hue discrimination by arranging a gradient in order, scored by displacement — a perception measure, not a color-blindness diagnosis.
Depth Perception Test
Four monocular depth cues — relative size, occlusion, texture scale and motion parallax — each scored with its own interval, starting with the pairs where two cues disagree.
Blind Spot Test
Finds your optic disc with the classic converging-dot method and shows how completely the brain fills the hole in.
Dominant Eye Test
The Miles and Porta methods run in sequence, because they disagree in a meaningful minority of people and one result alone would be misleading.
Peripheral Vision Test
Detection away from a fixation point you must actually hold, mapping roughly how far out you still catch a change.
Stereopsis Test
Random-dot depth that only appears if both eyes cooperate — no anaglyph glasses required, and the page says what it cannot see.
Contrast Sensitivity Test
Gratings that fade toward your threshold across spatial frequencies, with the caveat that a browser cannot calibrate your display luminance.
Optical Illusion Test
Illusions with a measurement attached: you set the matching value, and the page reports how far the illusion moved you.
Hearing
7 testsPerfect Pitch Test
Name notes with no reference tone across randomized octaves and timbres, with the strict criterion applied — near misses are not absolute pitch.
Tone Deaf Test
Pitch and melody discrimination in the Montreal battery tradition, reporting where you sit against the roughly 4% of people with genuine amusia.
Auditory Processing Test
Speech in noise, gap detection and temporal ordering — the measures that separate hearing a sound from making sense of it. Never a diagnosis.
Rhythm Test
Tapping to a beat and detecting when a pattern shifts, scored on timing variability rather than a pass mark.
Interval Ear Training
Ascending, descending and harmonic intervals with adaptive difficulty and a per-interval confusion table, so practice targets what you actually miss.
Relative Pitch Test
Identify notes against a given reference — the skill musicians actually use, and the one that can be trained.
Shepard Tone Illusion
The endlessly rising scale, generated live with the octave stack exposed so you can hear how the trick is built.
Language
13 testsSynonym Test
Adaptive multiple-choice synonyms that converge on your level in about forty items instead of marching through a fixed list.
Vocabulary Test
Estimates your vocabulary size with real and plausible fake words, so guessing is caught and corrected for rather than rewarded.
Reading Speed Test
Words per minute with a comprehension check afterwards, because a speed figure without comprehension is not a reading measure.
Reading Level Test
Places your comprehension against grade-level bands, with the readability formula behind each band named.
English Level Test
Places you on the CEFR scale from A1 to C2 across grammar, vocabulary and reading, with what each level means in practice.
Reading Comprehension Test
Passages with inference, main-idea and detail items scored separately — people who miss inference rarely miss detail.
Spelling Test
Adaptive dictation of the words people actually misspell, with British and American forms both accepted and flagged.
Verbal Fluency Test
One minute per category and per letter, scored on total, clustering and switching — the three numbers this task was designed to yield.
Reading Age Test
Measures the readability of text you paste — four published formulas side by side — and explains why a reading age is a score on one national school assessment rather than a number a browser can produce.
Verbal IQ Test
Vocabulary, analogies and comprehension combined into a verbal composite, with the norms sourced and the browser caveat stated.
Word Association Test
Response latency and self-consistency on repeated cues across a standard cue list, unscored by construction — there is no right answer, and no association dataset small enough to ship to a browser.
Sentence Completion Test
Grammar and register judgment in context, the format used in language placement rather than in projective psychology.
Language Processing Test
Sentence-picture verification, negation and embedded clauses timed to the millisecond, measuring how fast meaning assembles rather than how much you know.
Spatial
6 testsSpatial Reasoning Test
Rotation, folding and cross-section items reported as three sub-scores, because being good at one predicts little about the others.
Map Reading Test
Bearings, routes and scale under time pressure — the applied end of spatial ability, and the one that transfers to real navigation.
Mental Rotation Test
Shepard-Metzler cube figures with response time plotted against rotation angle — the linear slope is the actual finding, and you get to see yours.
Block Design Test
Reproduce a pattern from colored blocks against the clock, with time bonuses applied as the standard scoring does.
Paper Folding Test
Fold, punch, unfold: the classic visualisation task, scored with the guessing correction the original used.
Spatial Orientation Test
Point to a target from an imagined heading; scored on angular error, the measure that predicts real-world wayfinding.
Oddities
9 testsAphantasia Test
The VVIQ imagery questionnaire paired with a behavioral item, so the page measures something instead of only asking you to rate yourself.
Left or Right Brain Test
A set of lateralised tasks, and the page states the finding honestly: the personality version of this idea is a myth, and here is what actual hemispheric asymmetry looks like.
Misophonia Test
Trigger-sound sensitivity in the questionnaire format the research uses, with actual sound samples and an easy way out of any of them.
Face Blindness Test
Unfamiliar-face memory built on synthetic faces of our own, with percent correct and a confidence interval rather than a norm — the published instrument's faces and its norm table are not ours to reproduce. Covers the term prosopagnosia that people arrive searching for. Not a diagnosis.
Synesthesia Test
Consistency across repeated trials, which is what actually separates synesthesia from a vivid association — a genuine synesthete matches themselves months later.
Alexithymia Test
Original items grouped by the three factors the literature describes — identifying feelings, describing them, and attention turned outward — reported separately, with the difference between not feeling and not naming spelled out.
Learning Preferences, VARK Format
The visual-auditory-read-write-kinaesthetic inventory, reported alongside the evidence that matching teaching to a style does not improve outcomes. Not affiliated with, or endorsed by, the VARK trademark holder.
Sensory Processing Test
Sensitivity and seeking across the sensory modalities, reported per modality rather than as one score, since profiles rarely move together.
Time Blindness Test
Estimate elapsed intervals without a clock and see your production and reproduction error — a measurement, not a self-report.
How to take the cognitive ability sampler
Twelve items, four formats, one number per format and a route to the page that measures each one at full length.
Practice two items with the clock off
Take the two untimed practice items first if the formats are unfamiliar. Each one shows the worked answer and waits for you to click on rather than moving on by itself, and neither counts towards anything. Skip straight to the timed twelve if you already know what a figure series looks like.
Answer twelve items at thirty seconds each
Click an option or press 1 to 4. The clock beside the item counts from the frame that painted it, an item you leave unanswered scores zero rather than stopping the run, and an item you were on when the tab went to the background is discarded and put back at the end so you still answer twelve.
Read the split, then open the format that cost you
The result gives correct out of twelve against a chance floor of three, the exact probability of doing that well by guessing, and a score for each of the four formats. Three items cannot rank anybody, so treat a low format score as a reason to open that page rather than as a finding, and copy the plain-text summary if you want to keep it.
Technical specifications
| Tests on this domain | 98 across 10 sections — one homepage sampler and 97 dedicated pages: Reasoning, Hiring, General, Attention, Paradigms, Vision, Hearing, Language, Spatial, Oddities |
|---|---|
| Sampler length | 12 scored items at 30 seconds each — 6 minutes if you use every second, plus 2 untimed practice items if you take them |
| What runs on a server | Nothing. The pages are static files and every test is JavaScript in your own tab; there is no API to call and no account behind one |
| What is stored | Nothing. No answers, no response times, no score, no cookie set by this site for the tests themselves |
| Timing floor | About 16 ms on a 60 Hz display before you are involved, from display and input latency no browser API can see past. Times are measured from the paint of the stimulus to the event timestamp, never from a setTimeout |
| Where items come from | Written for this site. No published item bank is reproduced anywhere on the domain — not Raven's matrices, not the Wonderlic or CCAT sets, not the Wechsler subtests |
| Reference figures | 48 carried with an author, a year and a title, of which 2 are filed against this page; 43 more are recorded as gaps with the reason each one cannot be printed and what the page does instead |
| Percentiles printed | None. Not one of the reference figures is a mean-and-SD distribution collected under conditions a browser reproduces, so no page here converts your score into a rank |
Frequently asked questions
Is this an IQ test?
No. The sampler reports how many of twelve items you got right and how that compares to the chance floor of three, and it stops there. An IQ figure is a position in a standardization sample, which needs a published mean and standard deviation collected on people taking that exact test under supervision — twelve items written for a website have none of that. The practice sets in the format of supervised admission tests are on the Mensa-style IQ test page, and that page says the same thing at more length.
Is it free, and do I have to sign up?
Free, and there is nothing to sign up to. There is no account, no trial, no partial result and no page on this domain where a score appears after an email address does. That is the entire competitive position here: the rest of this corner of the internet runs forty questions and then asks for nineteen dollars or a mailing list subscription.
Do you store my answers?
No. Your answers, your response times and your score exist as JavaScript variables in the tab you are reading this in and are gone when you close it. Nothing is posted anywhere, which is also why every test keeps working with the network disconnected and why the copyable summary is the only way a result leaves the page at all.
Why is my score here different from another site's?
Because the item difficulty is different, and neither score is on a common scale. Two sites can both call something a verbal reasoning test and set items three years apart in difficulty, and without a shared standardization sample there is no arithmetic that makes the two numbers comparable. The one thing that does transfer is the split between formats: being consistently slower on figure series than on word problems is a fact about you rather than about whose items you answered.
Can I use a result from this site in a job application?
No, and no employer will accept one. Hiring assessments are administered by the employer or their vendor under conditions they control, partly so that the person answering is known to be the candidate. What these pages are useful for is knowing the format before it counts — how a situational judgement item is actually scored, what fifty items in twelve minutes feels like — which is what the hiring section is for.
How long does the whole site take?
The sampler is six minutes at its limit and most tests here run three to ten. The long ones are long on purpose: a continuous performance test that finishes in ninety seconds has not measured sustained attention, and every page that runs longer than a few minutes says at the top how long it takes before you start it.
I have an assessment on Thursday. Where do I start?
Take the sampler, then open the hiring page named after the test you were told to expect. The formats repeat across vendors far more than the branding suggests, so time on any well-matched practice set transfers; the thing that does not transfer is item content, which is why nothing here reproduces a real item bank.
What “measures rather than guesses” means on this site
The difference is whether the number came from something you did. A questionnaire that asks you to rate your own concentration from one to five and returns an attention score has measured your opinion of your concentration; the Stroop test makes you name the ink color of a word that spells a different color and reports the milliseconds that conflict costs you. Both produce a number. Only one of them would change if you were tired. Ninety-eight pages here are built on the second kind wherever the second kind exists — reading speed with a comprehension check on the reading speed test, hue discrimination by arrangement on the color perception test, note naming with no reference tone on the perfect pitch test. Where only self-report exists — imagery vividness on the aphantasia test is the honest example — the page pairs the questionnaire with a behavioral item and says which half produced which part of the result.
The second half of the rule is about the reference figure beside your score. This site carries 48 of them and every one names an author, a year and a title; it also carries 43 recorded gaps — figures a page will reach for and not find, each with the reason. Some of those are the most interesting paragraph on the page that hits them. The eight-second human attention span has no primary source and never did, so the attention span test reports how your own performance decayed across a block instead. Trail-making norms cannot transfer to a browser because the examiner correcting each error as it happens is part of the published score. A generic situational judgement test has no scoring key at all, because a real one is scored against the consensus of experienced people in one specific job at one specific employer — which is exactly what makes it a valid selection tool and exactly why a version on a website cannot mark you right or wrong. It teaches the format instead, and says so.
The reference figures this page is allowed to print
Adult silent reading rate, non-fiction: 238 wpm
Brysbaert (2019), How many words do we read per minute? A review and meta-analysis of reading rate, Journal of Memory and Language
This is a mean with no individual-level standard deviation behind it, so it supports a comparison and not a percentile: a page may say a result is above or below the average adult rate and must not say which percentile it falls in. The figure is for reading for comprehension, not skimming, and a rate measured without a comprehension check afterwards is a scrolling speed.
Mensa qualifying percentile: 98th percentile
Mensa International's published admission criterion — The criterion is a percentile, not a score. The score that reaches it differs by test.
The threshold is the 98th percentile of the general population and nothing else — the familiar numbers 130, 132 and 148 are that same percentile expressed on three scales with different standard deviations, which is why they disagree. Always name the scale alongside the number. No unsupervised test qualifies anybody: admission requires a supervised, standardized administration, and this site does not provide one.
The third rule is about the clock, and it is why several of these pages exist at all. A response time here is the interval between the frame that painted the stimulus and the timestamp the browser attached to your keypress, both read off the same monotonic clock; it is never the gap between a setTimeout firing and a handler running, which measures the timer queue. What that still cannot see is the display pipeline — the compositor, the cable and the panel — so on a 60 Hz screen roughly 16 ms of every reported figure is the machine rather than you. Every page that prints milliseconds says that on the page, reports no precision the platform cannot support, and throws away any trial during which the tab went to the background rather than averaging in a forty-second response to something nobody was looking at. If you want the practice sets rather than the paradigms, the two busiest are the Mensa-style IQ practice test and the Wonderlic practice test, and the language pages people reach for most are the synonym test and the vocabulary test. The work inventories — including Hogan assessment practice — are preparation for somebody else's instrument and carry an explicit non-affiliation line saying so.
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 any condition at all, on any page of this site. Only a qualified professional, working with more than a browser, can make that judgment.
What happens to the twelve answers you give
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 sampler holds your item order, your option order and your responses in the page for as long as the tab is open. The copy button is the only route out of it, it produces plain text on your own clipboard, and what it contains is the counts, the chance floor and the run seed — never the answers themselves.