Why so many different tests?
Because "cognitive ability" is not one thing. Working memory, inhibition, processing speed and spatial reasoning are separable — people are routinely strong in one and ordinary in another, and brain injuries can damage one while leaving the others intact.
So psychology built a task for each. Below are eight you will actually meet, what each one is really measuring, and where it came from. If you are facing a cognitive assessment in a hiring process, most of what you see will be a variant of something on this list.
1. Stroop task — inhibition
What you do: name the ink colour of a colour word that says something else.
What it measures: the ability to suppress an automatic response. Reading is automatic for literate adults, so the word's meaning arrives whether you want it or not and competes with the answer you need.
Origin: John Ridley Stroop, 1935. One of the most reproduced effects in psychology, still used clinically.
Watch for: people improve by avoiding the conflict — defocusing so the word is less readable. That raises the score without raising inhibition, which is worth knowing before you interpret one. Try it →
2. N-back — working memory updating
What you do: say whether the current item matches the one N steps back.
What it measures: maintaining and continuously replacing a rolling set of items. Unlike a span task, you cannot rehearse a fixed list, because the list moves every trial.
Watch for: this is the task at the centre of the brain-training dispute. Practice improves N-back reliably; transfer to fluid intelligence has largely failed to replicate. See does brain training actually work? Try it →
3. Corsi block-tapping — visuospatial span
What you do: watch cells light up in sequence, reproduce the order.
What it measures: short-term memory for spatial position. The spatial counterpart to digit span, and the two dissociate — patients can be impaired on one and not the other.
Origin: Philip Corsi, 1972. Typical adult span is 5–6, a little below digit span.
Watch for: chunking. Seeing a path instead of five positions turns five items into one, which is how high scorers actually do it. Try it →
4. Mental rotation — spatial reasoning
What you do: decide whether two shapes are rotations of each other or mirror images.
What it measures: the ability to transform a mental image. The classic finding is that response time scales linearly with rotation angle, which is strong evidence that something genuinely analogue is happening.
Origin: Shepard and Metzler, 1971.
Watch for: the average sex difference here is real in the literature and heavily overstated in popular accounts — distributions overlap substantially and the gap narrows with practice. Try it →
5. Balloon Analogue Risk Task (BART) — risk-taking
What you do: pump a balloon for points, bank before it bursts.
What it measures: behavioural risk preference, and more informatively consistency — whether you change strategy after a burst even though trials are independent.
Origin: Lejuez et al., 2002.
Watch for: the optimal strategy is a fixed moderate number of pumps, not "until it feels risky". Intuition does not track expected value well, which is the point. Try it →
6. Iowa Gambling Task (IGT) — learning from feedback
What you do: draw from four decks; two are net positive, two are net negative and hide it behind larger immediate rewards.
What it measures: integrating outcomes over time rather than reacting to the last one.
Origin: Bechara, Damasio et al., 1994 — built to explain patients with normal IQ who nonetheless made ruinous real-life decisions.
Watch for: the famous result is that healthy participants start avoiding the bad decks before they can explain why. The methodology has been criticised, but that core observation holds up. Try it →
7. Go/no-go — response inhibition
What you do: respond to targets, withhold from lookalikes, under time pressure.
What it measures: stopping a movement already underway. Note that missing a target and hitting a no-go signal are different failures — the first is attention, only the second is inhibition.
Watch for: any whack-a-mole-style game with a penalty object is a go/no-go task in disguise. Try it →
8. Task switching — cognitive flexibility
What you do: apply rule A or rule B depending on a cue that changes unpredictably.
What it measures: the cost of reconfiguring what you are doing. Trials right after a switch are reliably slower — often by hundreds of milliseconds — and a residual cost remains even when you are warned in advance.
Watch for: this is the direct evidence against multitasking. What people call multitasking is switching, and every switch pays this tax. Try it →
What these tests cannot do
Three limits worth holding onto, whichever test you are looking at.
- A browser is not a lab. Screen latency, input device, ambient noise and your own motivation all move the numbers. Comparing your result to a stranger's is close to meaningless; comparing your own result across days is informative.
- None of them is an IQ test. They measure narrow, specific capacities. A standardised intelligence battery takes an hour or more with a trained administrator and is normed on a large sample.
- None of them diagnoses anything. Clinical use requires norms, controlled administration and a clinician interpreting the pattern. A single score from a website is not a finding.
If you are facing an assessment
Practical, in order of value:
- Get familiar with the formats. The points you lose to unfamiliarity in the first few items of a task are the ones you can genuinely recover. This is near transfer and it works.
- Sleep the night before. Larger effect than any amount of practice.
- Check your equipment. On speed tasks, a laggy touchscreen is measured as you.
- Answer trait questions consistently. Personality sections almost always include consistency checks; performing an ideal candidate produces contradictions that are easy to detect.
To see several of these areas in one run, try the full cognitive test.