A grid of figures with one cell left empty. One rule runs across the rows, a different rule runs down the columns, and the missing figure has to satisfy both at once. Finding one rule and stopping is the single most common way to get these wrong.
No account needed. Fresh questions every time you press for five more.
Worth saying out loud the first few times. The order matters more than the speed.
Say out loud what changes from the first cell to the second, then the second to the third. One property, named.
A different property should be changing. If it looks like the same one, you have probably misread the row.
Describe the missing cell in words: "three of them, green, dotted." Prediction is what stops a plausible option from talking you out of a correct reading.
If two options fit, one of your two rules is wrong. Go back to the grid rather than choosing between them.
| CogAT | the Nonverbal battery, where it is usually the first item type children meet |
|---|---|
| NNAT | as reasoning by analogy and serial reasoning, which use the same grid logic |
| CAT4 and InView | the nonverbal reasoning sections |
| SOF olympiad papers | Logical Reasoning, where the grid is often three by three with a picture rule |
These are not guesses. They are what the wrong options in a real question are built from, which is why each one feels reasonable at the time.
The fix: Always check the column too. Options are built specifically to reward a half-solve.
The fix: Say the property out loud. "Solid, outline, dotted" is a different rule from "green, blue, amber", and mixing them up puts you one option off.
The fix: Count them again. A three-shape cell drawn in a triangle arrangement reads as two at a glance.
The fix: Nearly right is built on purpose. If your predicted description does not match exactly, it is not the answer.
A grid of figures with one cell missing. A rule governs how the figures change across each row and a different rule governs how they change down each column, and the missing figure must satisfy both. It measures reasoning with no reading and no arithmetic.
Because that is how the wrong options are built. Each one satisfies one of the two rules but not the other. When two options both look right, it is a signal that one of the rules has been misread, not that the question is unfair.
The common ones are: a property changes step by step (shading, colour), a count increases, a shape gains a side, or a figure rotates by a fixed amount. Harder items combine two of these, and the hardest apply a rule to a property you have to notice first.
The CogAT nonverbal battery, CAT4 and InView nonverbal reasoning, and SOF olympiad logical reasoning sections. The NNAT uses the same grid logic under different names, reasoning by analogy and serial reasoning.
Change what they are looking at. Ask "what is the same in every cell?" before "what changes?" Eliminating the constants usually leaves the rule visible, and it works when staring at the whole grid does not.
No, though they are often confused. Pattern completion asks which piece fits a hole in a single design. A matrix asks you to apply a rule to produce a new figure. The first is matching, the second is reasoning.
The two-by-two version with one changing property is accessible from around six. Three-by-three grids with two independent rules are a genuine step up and are where most of the difficulty in a test sits, usually from about grade 3.
One skill at a time is how you fix a weakness. A full timed test is how you find out whether it is fixed.