Cognitive Tests Explained: Types, Scores and What They Actually Measure

The three things people mean by “cognitive test”

Search for “cognitive test” and you will be handed a dementia screening questionnaire, an IQ test and a personality quiz, all on the same page of results. These measure different things, for different people, for different reasons. Sorting out which one you are looking at is the single most useful thing you can do before you take one.

Clinical cognitive screening tests

These are short, standardised tests administered by a clinician, usually in a primary-care office. Their purpose is triage: to decide whether a person's cognition warrants a fuller neuropsychological workup. They are quick by design, and the price of that speed is precision.

MoCA vs MMSE: why the newer test displaced the older one

The MMSE, published by Folstein and colleagues in 1975, was the standard for a generation. Its weakness is that it was built to detect dementia, and it is comparatively blind to the milder stage that precedes it. When Nasreddine and colleagues introduced the MoCA in 2005, they tested both instruments on the same participants. Against mild cognitive impairment, the MMSE identified 18% of cases; the MoCA identified 90%. Against mild Alzheimer's disease the MMSE identified 78% and the MoCA 100%.

That sensitivity comes at a cost. In the same study, specificity — the ability to correctly clear people who are cognitively healthy — was 100% for the MMSE and 87% for the MoCA. The MoCA catches far more real impairment, and it also flags more healthy people incorrectly. That trade-off is the central fact about cognitive screening, and it is almost never mentioned in articles that simply list the cutoff scores.

Access and licensing

Neither of the two best-known screens is freely usable any more. The MMSE was distributed informally for decades, but the copyright was acquired by Psychological Assessment Resources in 2001 and authorised forms must now be purchased. The MoCA, long downloadable at no charge, moved to a licensed model with an official training and certification programme. As of 2026, certification is required for clinicians who interpret individual sub-scores or domain profiles, and remains optional for those who use only the total score.

The cutoff score problem nobody mentions

Almost every page describing the MoCA repeats the same sentence: a score of 26 or above is normal. That number comes from the 2005 validation study, and it has a well-documented flaw.

Carson, Leach and Murphy re-examined the MoCA's cutoffs in the International Journal of Geriatric Psychiatry in 2018. They found that the 26/30 threshold produces an inflated rate of false positives, and that the effect is strongest in exactly the people most likely to be screened: those who are older, and those with fewer years of formal education. Their analysis found that an overall cutoff of 23 out of 30, rather than 26, maximised correct detection while minimising missed cases.

Later work has reinforced the point from another direction: optimal cutoffs differ by language, by country and by educational background, which is why the search for one universal threshold has largely been abandoned in the research literature even as a single number continues to circulate online.

A screening test is not a diagnosis

A cognitive screen produces a number. A diagnosis requires a clinical history, an examination, laboratory work to rule out reversible causes such as thyroid disease or B12 deficiency, often brain imaging, and in many cases formal neuropsychological testing that runs to several hours. No screen substitutes for any part of that.

The evidence on population screening is genuinely unsettled. In 2020 the US Preventive Services Task Force concluded that “the current evidence is insufficient to assess the balance of benefits and harms of screening for cognitive impairment in older adults,” and issued a Grade I statement — meaning neither for nor against.

This sits alongside a separate requirement that is often confused with it. Detection of cognitive impairment is a mandatory element of the Medicare Annual Wellness Visit in the United States. Notably, the Centers for Medicare & Medicaid Services deliberately declined to specify which instrument to use, on the grounds that no single screen satisfies every need. Detection there may rest on direct observation and on concerns raised by the patient or their family, not necessarily on a scored questionnaire.

Cognitive ability tests: measuring level rather than decline

The second family of cognitive tests asks a different question. Instead of “has something changed?”, they ask “where does this person stand relative to others?” These are the instruments most people mean when they say IQ test.

What separates these from a magazine quiz is not the questions. It is the norming. A score of 112 on the WAIS is meaningful because the test was administered to a standardisation sample chosen to represent the population, and your raw performance is expressed as a position within that distribution. Without a representative norm group, a score is just a count of correct answers.

Ability tests are also used clinically, but for a different purpose than screening. A neuropsychologist may administer the WAIS to establish a person's current cognitive profile in detail — comparing, for example, preserved vocabulary against slowed processing speed — which is a far more informative picture than a single screening number. Our guide to the WAIS intelligence test covers its indices and the recent debate over its factor structure, and types of IQ tests compares the main instruments.

What cognitive domains are being tested?

Both families of test sample from the same underlying set of mental functions. Clinical screens sample many domains shallowly; ability tests sample fewer domains deeply.

• Attention — sustaining focus and resisting distraction. Tested by digit span, serial subtraction, cancellation tasks.
• Working memory — holding and manipulating information over seconds. Tested by reversed digit span, running span, n-back tasks.
• Episodic memory — recalling events and word lists after a delay. Tested by three- or five-word recall, list learning.
• Language — naming, fluency, comprehension. Tested by confrontation naming, category fluency, sentence repetition.
• Visuospatial function — perceiving and constructing spatial relations. Tested by clock drawing, cube copying, block design.
• Executive function — planning, set-shifting, inhibition. Tested by trail-making, the Stroop task, and abstraction items. See executive function for why these measures disagree with each other.
• Processing speed — how quickly simple decisions are made. Tested by symbol search and coding tasks.

Are online cognitive tests accurate?

It depends what you are asking them to do. Computerised cognitive assessment is a legitimate and growing field: validated batteries such as the NIH Toolbox Cognition Battery and Cambridge's CANTAB are used routinely in research, and self-administered instruments designed for longitudinal tracking have been published in peer-reviewed journals.

The tests that appear at the top of a web search are usually not those. A free browser test has no proctor, no verified identity, no controlled conditions and, most importantly, no representative norm sample. It can give you a reasonable indication of relative reasoning ability. It cannot detect dementia, and it cannot substitute for an hour with a psychologist. The Mensa Norway IQ test is a good case study in both the appeal and the limits of the format.

Myths and facts about cognitive tests