Matrix reasoning.
Overview
Matrix reasoning is the central item type of modern intelligence assessment. Every major culture-fair instrument — Raven's Progressive Matrices, Cattell's CFIT, Naglieri's NNAT, the matrix subtests of Wechsler and Stanford-Binet batteries — uses matrix problems as the core measure of fluid intelligence. On this test, matrix reasoning is the most heavily weighted reasoning type, contributing ten items to the Standard Test and five to the Short Assessment.
The reason for this weight is empirical: across decades of research on cognitive ability, performance on matrix reasoning items correlates more strongly with general intelligence (g) than performance on nearly any other single task. A well-designed matrix problem efficiently isolates the capacity that matters most — the capacity to detect a pattern, abstract the rule that governs it, and apply the rule correctly.
Question format
Each matrix item presents a grid of geometric figures with one cell missing. The grid may be 2×2 (four cells total, one missing) or 3×3 (nine cells total, one missing). Below the grid, three to five candidate figures are displayed. The task is to select the candidate that completes the pattern.
The rules governing the matrix range widely in complexity:
- Single-attribute rules — one property (color, shape, count, fill) changes consistently across rows or columns
- Two-attribute rules — two properties vary independently, one controlled by rows and the other by columns
- Three-attribute rules — three properties cycle independently in a Latin-square arrangement
- Transformational rules — the relationship between cells involves operations like rotation, reflection, or XOR (symmetric difference) rather than independent attributes
- Superposition rules — the third cell in a row or column is the overlay or union of the first two
On the Standard Test, matrix items are ordered by increasing rule complexity. Early items test single-attribute rules; later items require tracking three or more interacting attributes simultaneously.
What it measures
Matrix reasoning directly measures inductive reasoning — the capacity to move from specific instances to the underlying rule. The test-taker sees a few cells and must infer the pattern that connects them, then apply that inference to predict the missing cell. This is the canonical operation of fluid intelligence.
The cognitive demands include:
- Detecting which attributes vary and which stay constant
- Working out how the varying attributes are related (linearly, cyclically, logically)
- Holding multiple rules in mind simultaneously when more than one attribute varies
- Checking candidate answers against all rules, not just the one most salient
Items with multiple interacting rules require a form of systematic elimination. Candidates that satisfy one rule but violate another are common distractors — selecting them indicates that the test-taker detected part of the pattern but failed to track all of it.
Strategy and approach
Identify what varies before attempting to solve
Before studying the candidate options, look at what changes across the given cells. Color? Shape? Count? Orientation? Size? Making the variation explicit often reveals the rule faster than studying any single cell in isolation.
Check rows and columns independently
In a 3×3 matrix, both rows and columns typically carry information. A rule that explains the rows but not the columns is incomplete. A well-designed item has a rule that works in both directions.
Eliminate candidates that violate any single rule
On multi-attribute items, the correct answer satisfies every rule. A candidate that gets color right but shape wrong is a distractor designed to catch partial reasoning. Work through each rule in turn.
Do not second-guess an answer that satisfies all rules
If one candidate fits every pattern you have identified, it is the answer. Time pressure on these items is real — spending extra time on certainty you already have costs items later in the test.
Example item
Consider a 3×3 matrix where each row contains a shape (circle, square, triangle) and each column contains a color (red, blue, green). The first two rows are fully populated; the third row shows a red triangle and a blue triangle — the bottom-right cell is missing.
The distractors on such an item typically break one rule each: a fully-shaded square would violate the row rule; an outline triangle would violate the column rule. Items at the challenging end of the difficulty range preserve this structure but combine three or more attributes — color, shape, size, fill, rotation — each governed by its own independent rule, all of which must be satisfied simultaneously.
Practice
The practice guide includes 50 matrix reasoning problems across all difficulty levels, with worked explanations identifying the rules governing each item and diagnosing why each distractor is wrong. Matrix reasoning is the largest section in the guide, reflecting the item type's central importance in fluid intelligence measurement.
Begin the test to see ten matrix reasoning items scored against standard population norms.
Ten matrix items on the Standard Test.
Scored against standard population norms, with sub-scores for each of the four reasoning types. Takes about twelve minutes.