What is pattern recognition?

What makes a pattern hard

Difficulty is not mysterious. Carpenter, Just and Shell analyzed how people solve Raven's matrices using verbal protocols, eye-fixation patterns and errors (Carpenter, Just & Shell, 1990). Everyone used the same step-by-step strategy; what separated higher from lower scorers was mainly the ability to induce abstract relations and to manage many problem-solving goals in working memory (Carpenter, Just & Shell, 1990). Later work describing that theory says item difficulty depends on both the type of rule and the number of rules (Lozano, Embretson & Revuelta, 2025).

Commercial tests are built the same way. Items on the TONI-4 vary on eight features (shape, position, direction, rotation, contiguity, shading, size and movement), with harder items combining more of them (PRO-ED: TONI-4). The more changes an item stacks, the more a solver has to track at once. The abstract reasoning page follows this research further into matrix puzzles.

How pattern recognition is measured

Almost every IQ battery includes a pattern task. Raven's Progressive Matrices, first published in 1938, are nonverbal matrix puzzles widely used as a measure of fluid, culture-reduced reasoning (Wikipedia: Raven's Progressive Matrices). Cattell's Culture Fair test has subtests for series, classification, matrices and conditions (Zahra et al., 2022). The WAIS-5 pairs Matrix Reasoning and Figure Weights in its Fluid Reasoning index (Pearson: WAIS-5 overview). The Raven's Progressive Matrices page covers the best-known of these in detail.

No single pattern test is a pure measure of intelligence. Across several large samples, Raven's shared about 50% of its variance with g, about 10% with a separate fluid factor, and had about 25% test-specific reliable variance (Gignac, 2015). The IQ Octopus test has a Pattern Recognition section of 5 items out of 25, alongside abstract, spatial, quantitative and logical sections, and reports an estimated IQ with a confidence range.

Pattern items are popular partly because they need no reading. Culture-fair tests use nonverbal, nonacademic items such as choosing a design that completes a series (APA Dictionary: culture-fair test). That does not make them neutral. The APA Dictionary cautions that every assessment reflects some cultural norms, so no test is fully culture-free (APA Dictionary: culture-fair test), and Lohman argues that most nonverbal tests still rely on verbally mediated thinking (Lohman, 2005). Silently naming a change as you solve ("turns, then grows") is one example of how words can creep back into a wordless puzzle.

Can pattern recognition be improved?

Inductive reasoning has some of the better training evidence in this field, mostly from children. Klauer and Phye's meta-analysis of inductive reasoning training covered 74 training experiments with nearly 3,600 children and confirmed gains in both fluid intelligence and classroom learning (Klauer & Phye, 2008). Training effects on intelligence test scores grew over time, transfer to academic learning was larger than transfer to intelligence tests, and the authors reported that the results cannot be explained by placebo or test-coaching effects (Klauer & Phye, 2008). As reported by a 2022 paper, the average corrected effect on intelligence tests was d = .52 (Pásztor et al., 2022). A game-based program for fifth graders produced a corrected effect of .38 (Pásztor et al., 2022).

That is different from generic brain training. A 2016 review of studies cited by brain-training companies found extensive improvement on trained tasks, less on closely related tasks, and little on distant tasks or everyday cognition (Simons et al., 2016), and a review of several meta-analyses found minimal effects of cognitive training on domain-general skills (Sala & Gobet, 2019). The inductive training results are encouraging, but they come from children in structured programs, not from adults playing puzzle apps.

Within a single test, people also learn as they go: in one matrix test, items using some rules got easier as test takers went along, and that learning varied between people (Lozano, Embretson & Revuelta, 2025). Familiarity with a format is real, but it is a format effect, not a change in underlying ability.

Try it

Take the free pattern recognition test for a timed set of ten sequences, or work through the sequence puzzles at your own pace. The pattern recognition guide walks through the rules and traps item by item.