What is abstract reasoning?

Short answer: Abstract reasoning is the ability to work out the rule, pattern or relationship in material you have not seen before, and then apply it to a new case. Tests measure it with nonverbal puzzles such as series, matrices, odd-one-out and analogy items. Researchers use such puzzles to measure fluid intelligence, which Chuderski (Journal of Intelligence, 2022) calls the broad ability most strongly tied to general intelligence, at least in homogeneous samples. It is closely related to IQ but not the same thing: a full clinical IQ test samples several other abilities too.

Common forms of abstract reasoning items (every example below is made up for this article, and none comes from a real test):

• Series: continue a sequence. A circle, then a circle with one line through it, then two lines, then three. The next figure is a circle with four lines.
• Matrices: fill the missing cell of a grid. In a 3 by 3 grid of stars, each row runs small, medium, large from left to right, and each column keeps one color. The empty bottom-right cell is a large star in that column's color.
• Classification (odd one out): find the figure that breaks the rule the others share. Take a square, a hexagon, a rectangle, an octagon and a triangle. Only the triangle has an odd number of sides.
• Analogies: apply the relationship in one pair to another. A small circle is to a large circle as a small square is to a large square.

Each of these types is named in research on fluid reasoning tests (Borter, Schlegel and Troche, Journal of Intelligence, 2023; Chuderski, 2022).

What is abstract reasoning?

Abstract reasoning means finding the general rule behind a few specific cases. The rule is what makes it abstract: it holds for cases you have not met yet, so you can use it to predict the next one.

The APA Dictionary of Psychology has no entry headed "abstract reasoning". Its nearest entries describe the same skill:

• Abstract thinking: "thinking characterized by the use of general ideas or concepts", in contrast with concrete thinking, which stays tied to specific objects and events.
• Abstraction: "the formation of general ideas or concepts by extracting similarities from particular instances."
• Abstract attitude, credited to Kurt Goldstein: "the ability to grasp essentials and common properties, to keep different aspects of a situation in mind and shift from one to another, to predict and plan ahead, and to think symbolically and draw conclusions."

Researchers describe the usual test in plainer terms: "Typical fluid intelligence tests involve identifying abstract rules governing relatively complex figural stimuli patterns and selecting the response option that best matches these rules" (Chuderski, Journal of Intelligence, 2022).

Abstract reasoning examples

You use abstract reasoning far outside a test room. These everyday examples are our own illustrations, and none comes from a source or a test:

• Your electricity bill jumps every January and every July. You work out that heating and cooling drive it, and you set money aside before next winter.
• A friend always picks the quiet restaurant with outdoor seating. The next time you choose for them, you apply that rule without asking.
• Someone tells you not to put all your eggs in one basket during a job search. You read it as advice about spreading risk.

Each one moves from specific cases to a general rule, then back to a new case. The APA Dictionary describes inductive reasoning as the kind that "underlies the process of developing hypotheses from particular facts and observations".

Knowledge still matters. A lifespan textbook by Lally and Valentine-French (2017) notes that "Abstract reasoning in a particular field requires a knowledge base that we might not have in all areas."

How abstract reasoning is tested

Many abstract reasoning tests use figures instead of words. The best known is Raven's Progressive Matrices, which the APA Dictionary of Psychology (read 14 October 2026) describes as "a nonverbal test of mental ability consisting of abstract designs, each of which is missing one part". Published in 1938, it "is often viewed as the prototypical measure of general intelligence", in the dictionary's words. Our guide to Raven's Progressive Matrices explains how it works.

Several item types can sit inside one test. Cattell's Culture Fair Test has subtests for series, classifications, matrices and topologies, as described in a 2023 study of 200 participants (Borter, Schlegel and Troche, Journal of Intelligence, 2023). The authors call it a "(not culturally independent) measure of general reasoning ability and fluid intelligence". Chuderski (2022) adds figural analogies to the list of typical fluid intelligence tests.

Clinical IQ tests include abstract reasoning tasks as one part of a longer test:

• WISC-V (children). Matrix Reasoning and Figure Weights feed the Fluid Reasoning Index. Pearson says that index requires inductive and quantitative reasoning and "abstract thinking", among other skills (Pearson, read 14 October 2026).
• WAIS-5 (adults). Pearson describes a new subtest, Set Relations, as "designed to measure logical deductive reasoning, fluid intelligence, and simultaneous processing." Matrix Reasoning has no strict time limit, but a 30-second guideline, because "the vast majority of examinees who will respond correctly do so within 30 seconds" (Pearson, read 14 October 2026). See our guide to the WAIS IQ test.

Abstract reasoning and fluid intelligence

Intelligence researchers group rule-finding tasks like these under fluid reasoning. A 2024 study of 3,927 youths across six intelligence tests classed Wechsler Matrix Reasoning as induction and Figure Weights as quantitative reasoning, both under fluid reasoning (Caemmerer and colleagues, Journal of Intelligence, 2024). The same study placed Wechsler Similarities under lexical knowledge, which is a crystallized ability rather than fluid reasoning.

The APA Dictionary sets fluid intelligence, the processes "used in dealing with relatively novel tasks", against crystallized intelligence, the knowledge measured by vocabulary and general information tests.

For how fluid ability changes with age and where it sits in the CHC model, see what fluid intelligence is.

Is abstract reasoning the same as IQ?

The case for a close link is strong. Chuderski (2022) writes that "compared to all other abilities, fluid intelligence most strongly loads general intelligence factor." Caemmerer and colleagues (2024) found that "The influence of g and fluid reasoning were difficult to separate statistically".

Those findings concern abilities measured across many tests. A single test is a different matter: in a bifactor model across several large samples, Gignac (Intelligence, 2015) found that Raven's shares only about 50% of its variance with g. A clinical IQ also blends verbal, working memory and processing speed subtests with fluid ones (Pearson, read 14 October 2026), as what an IQ test measures explains.

So a strong result on abstract puzzles tells you about one ability that sits close to general intelligence. It does not stand in for a full IQ, and it says nothing about vocabulary or knowledge.

Abstract vs logical reasoning

The two terms overlap. The APA Dictionary defines reasoning itself as "thinking in which logical processes of an inductive or deductive character are used to draw conclusions from facts or premises." Abstract reasoning uses those same processes to find general rules in specific cases, which is why most test items are inductive.

The more useful split is inductive versus deductive:

• Inductive reasoning draws "inferences and general principles" from "specific observations and cases" (APA Dictionary). If the premises are true, the conclusion is "only probably true" (Salmon, 1963, as summarized by the Internet Encyclopedia of Philosophy).
• Deductive reasoning is "the form of logical reasoning in which a conclusion is shown to follow necessarily from a sequence of premises" (APA Dictionary). In the same account by Salmon (1963), if all the premises are true, the conclusion must be true.

Most abstract reasoning items are inductive: you see examples and infer the rule. Wechsler Matrix Reasoning is classed as induction (Caemmerer and colleagues, 2024), and SHL calls its pattern test inductive. A few items are deductive, such as the WAIS-5 Set Relations subtest that Pearson says measures "logical deductive reasoning".

Piaget and abstract thinking

In Jean Piaget's theory of cognitive development, abstract thinking belongs to the fourth and last stage, the formal operational stage. The APA Dictionary of Psychology (read 14 October 2026) says it begins "around age 12". Psychology 2e, a free OpenStax textbook, says "about age 11". Together, that is about age 11 or 12.

The change is in what a young person can reason about. "Whereas children in the concrete operational stage are able to think logically only about concrete events, children in the formal operational stage can also deal with abstract ideas and hypothetical situations" (OpenStax, Psychology 2e).

Not everyone uses it everywhere. Piaget himself wrote that different children "vary in terms of the areas of functioning to which they apply formal operations, according to their aptitudes and their professional specializations" (Human Development, 1972). Lally and Valentine-French (2017) say the ability "does not apply to all situations or all adults", and that "Formal operational thought is influenced by experience and education."

The stage model has critics. OpenStax notes that "several contemporary studies support a model of development that is more continuous than" Piaget's separate stages.

Where abstract reasoning tests are used

Employers use them in hiring. SHL, a hiring-assessment publisher, says its inductive reasoning test tasks "include working out the pattern in a set of images and identifying the correct next image" (SHL product page, read 14 October 2026).

Admissions tests used them too. Before 2025, the UCAT, a UK medical and dental school admissions test, had five separately timed subtests (UCAT test format page, 2024). Its Abstract Reasoning section gave candidates 12 minutes for 50 questions. The official page listed question types such as deciding whether a shape belongs to Set A, Set B or neither, and choosing the next shape in a series.

Then the section was dropped. In an announcement dated 11 December 2024, the UCAT Consortium said Abstract Reasoning "will be withdrawn from the test". Its UCAT 2025 page gives two reasons. The first was prediction. "Our analysis has shown that Abstract Reasoning has lower predictive validity than the other sections", meaning the subtest was "not as effective as the other subtests in predicting candidate performance at university". The second was coaching: "The test also has high coachability." Today's UCAT "consists of four separately timed subtests", and abstract reasoning is not one of them (UCAT test format page, read 14 October 2026).

Published research had pointed the same way. A 2022 systematic review of 24 studies found that "For over 70% of univariate data points, the UCAT exerted no statistically significant predictive validity", with abstract reasoning showing the least evidence among the subtests (Bala and colleagues, Medical Teacher, 2022).

What makes an item harder

Mainly the kind of rule and the number of rules. A 2025 study summarizing Carpenter's work says "the difficulty of RAPM items depends on the type and number of rules involved in their resolution" (Lozano, Embretson and Revuelta, Journal of Intelligence, 2025). Finding each rule depends on its type. Keeping track of all of them depends on working memory and on how many there are. Carpenter, Just and Shell (1990) sorted matrix rules into five types, which our Raven's guide describes.

Summarizing Carpenter, Just and Shell (1990), von Gugelberg writes that "With increasing item difficulty, more rules and item features must be maintained simultaneously" (Journal of Intelligence, 2026). The 2026 study itself recorded the eye movements of 300 participants on a figural matrices test, and higher reasoning scores went with constructive matching: working out the missing piece first, instead of eliminating answer options one by one. It is a finding about how people differ, not a method that raises a score.

Can you improve abstract reasoning?

Scores on abstract reasoning items can rise. Whether the ability underneath rises is a separate question, and the research on these items speaks mostly to scores.

Scores rose on the UCAT section. The UCAT Consortium cited the subtest's "high coachability" and said candidate performance on it increased, and response times decreased, over the past 10 years (UCAT 2025 page, read 14 October 2026). Of the 14,332 UK school-leaver candidates aged 16 to 20 who sat the UCAT in 2017, 5,439 answered a questionnaire on how they had prepared (Kulkarni, Parry and Sitch, 2022). In that survey, "performance appeared to increase with greater preparation time categories for the abstract reasoning and quantitative reasoning subsections only". That is a survey result: it links preparation time to scores but does not show that one caused the other.

Some item types got easier within one test. In a study of 765 military recruits on 34 matrix items built on Carpenter's rule types, Lozano, Embretson and Revuelta (2025) found that "Decreases in difficulty were observed for both pairwise progression and figure addition or subtraction" as people worked through the test. They suggest these practice effects are "conceptually tied to intelligence."

Training and schooling. A 2016 review found "no convincing evidence of any reliable improvements" in nonverbal ability when working memory training was compared with a treated control condition (Melby-Lervåg, Redick and Hulme), and a year of education is linked to about 1 to 5 IQ points (Ritchie and Tucker-Drob, 2018). Across cognitive tests in general, retaking a test gives an adjusted effect size of .26 (Hausknecht and colleagues, 2007). What fluid intelligence is covers these studies in full.

Practice makes the format familiar and lifts scores modestly, which is different from raising the underlying ability.

What an abstract reasoning score can tell you

A score on abstract puzzles shows how you did on one kind of task, on one day, against the people the test was normed on. It carries a range of error and can differ between two tests or two sittings. It does not diagnose anything. For a decision about school, work or health, see a qualified psychologist. Are IQ tests accurate? explains where the error comes from.

If you want to try matrix puzzles, one common kind of abstract reasoning item, WhatIQ's test has 24 picture puzzles in the grid format John C. Raven published in 1938 and takes about 15 minutes (whatiq.com, read 14 October 2026). It gives an estimate on the standard IQ scale, shown with the range it most likely sits in. It is not a clinical assessment.

Frequently asked questions

What is the purpose of abstract reasoning?

It lets you take a rule from a few specific cases and use it on a case you have not seen. The APA Dictionary says inductive reasoning "underlies the process of developing hypotheses from particular facts and observations". The APA Dictionary's entry on Goldstein's abstract attitude includes the ability "to predict and plan ahead". In daily life, that might mean spotting why a bill rises each winter and planning for it.

What is the difference between logical and abstract reasoning?

Abstract and logical reasoning overlap. The APA Dictionary defines reasoning as thinking that uses "logical processes of an inductive or deductive character". Abstract reasoning uses that logic to find general rules in specific cases. Most abstract reasoning test items are inductive, where you infer a rule from examples. A few are deductive, such as the WAIS-5 Set Relations subtest (Pearson).

What is abstract reasoning in Piaget's theory?

Piaget tied abstract thinking to the formal operational stage, the fourth and last stage of his theory. It begins at about age 11 or 12 (OpenStax Psychology 2e; APA Dictionary of Psychology, read 14 October 2026), when children "can also deal with abstract ideas and hypothetical situations" (OpenStax). Piaget wrote in 1972 that young people vary in the areas where they apply it, according to their aptitudes and specializations, and several later studies support a more continuous model of development than his stages (OpenStax).

How hard is abstract reasoning?

It depends on the item. Difficulty rises with the type and the number of rules a puzzle combines (Lozano, Embretson and Revuelta, 2025), because harder items require more rules to be held in mind at once (Carpenter, Just and Shell, 1990, as summarized by von Gugelberg, 2026). Time limits also differ: the UCAT's Abstract Reasoning section, dropped in 2025, gave candidates 12 minutes for 50 questions (UCAT, 2024), while Matrix Reasoning on the adult WAIS-5 has no strict time limit (Pearson, read 14 October 2026).

Is abstract reasoning a sign of high IQ?

It is closely related to IQ but not the same. Measured across many tests, fluid reasoning and general intelligence are hard to tell apart statistically (Caemmerer and colleagues, 2024). But a single matrix test, Raven's, shares only about 50% of its variance with g across several large samples (Gignac, 2015), and a clinical IQ combines several kinds of task. A good result on abstract puzzles describes one ability, with a range of error.

What kind of people are abstract thinkers?

Abstract thinking is not a fixed type. Lally and Valentine-French (2017) note that the ability "does not apply to all situations or all adults", and many adults think abstractly in some areas but not others. Abstract reasoning in a field "requires a knowledge base", they add: a psychology student may reason abstractly about psychology but not about physics or chemistry. Piaget (1972) wrote that young people vary in where they apply formal operations, according to "their aptitudes and their professional specializations".