How to Solve Abstract Reasoning Questions Step by Step (Practice Inside)
Abstract reasoning questions appear on IQ tests, graduate admissions exams, and professional aptitude assessments — and they trip people up not because the underlying logic is difficult, but because most people have never been taught a systematic approach. They stare at a grid of shapes, try a few guesses, and either get lucky or don't. This article gives you an actual method.
The approach works best when applied to real problems, so the Matrix Reasoning Test is embedded at the bottom of this page — work through the steps here, then put them into practice immediately.
What Abstract Reasoning Questions Actually Test
Abstract reasoning questions — the kind that present grids, sequences, or matrices of shapes — are designed to measure fluid intelligence: the ability to identify rules and apply them to novel problems without relying on prior knowledge. Research has proposed that the core function in fluid intelligence is splitting a complex whole into simple, separately attended parts — which is precisely what a good abstract reasoning strategy does. The questions aren't testing what you know; they're testing how well you can extract structure from unfamiliar material.
This matters for strategy. Because there's no factual knowledge to recall, the right approach is purely procedural: you need a reliable sequence of steps that ensures you examine the right attributes in the right order, rather than hoping the answer jumps out at you.
The Attributes to Check
Every abstract reasoning question — whether it's a 3×3 matrix, a sequence of shapes, or an odd-one-out grid — varies along a limited set of visual attributes. Before applying any strategy, it helps to know what you're looking for:
Shape — What type of shape is in each cell? Does it change, cycle, or stay the same across rows and columns?
Number — How many shapes or elements appear in each cell? Is there a progression (1, 2, 3) or a constant?
Size — Are shapes getting larger or smaller? Is size constant within rows but varying across columns?
Color or shading — Is there a pattern in fill — black, gray, white — rotating across cells?
Orientation or rotation — Are shapes rotating by a fixed angle with each step?
Position — Is an element moving systematically across the grid?
Most matrix problems use two or three of these attributes simultaneously. The challenge is identifying which ones are active and what rule governs each.
The Step-by-Step Method
Step 1: Look at rows first, then columns. Start by scanning each row left to right. Ask: what changes between cell 1, cell 2, and cell 3? Then do the same for columns top to bottom. Any rule that holds across rows and columns is almost certainly the intended rule. Rules that appear in only rows or only columns are worth noting but treating with more suspicion.
Step 2: Identify what stays the same. Before hunting for what changes, note what's constant. If the shape is always a triangle in every cell, shape isn't the varying attribute — move on. Identifying constants quickly narrows the search space.
Step 3: Check one attribute at a time. Don't try to process the whole grid at once. Focus on shape only — does it follow a rule? Then switch to number. Then size. This serial approach is slower but far more accurate than trying to see everything simultaneously. Fluid reasoning tests require inductive reasoning to discover abstract rules governing figural patterns and apply them to find the correct answer — performance depends directly on the type and number of governing rules the solver can identify.
Step 4: Test your rule against all cells, not just the obvious ones. Once you think you've found a rule, check it against every cell in the grid — not just the ones that first suggested it. A rule that explains three cells but breaks on a fourth isn't the rule. This is where most mistakes happen: people find a partial pattern and stop checking.
Step 5: Use elimination on the answer options. Once you have a candidate rule, apply it to the answer options. Often you can eliminate four or five options immediately because they violate the rule on an obvious attribute. The answer that satisfies all identified rules across all attributes is correct.
Step 6: If you're stuck, change the attribute you're examining. If a rule isn't revealing itself on shape, switch to number. If number isn't working, try shading. Most questions have one "entry point" attribute that's easier to spot than the others — finding it unlocks the rest of the puzzle.
Common Rule Types to Recognize
Abstract reasoning questions draw from a predictable pool of rule types. Knowing them in advance speeds up identification considerably:
Progression rules — An attribute increases or decreases systematically across a row or column. Number of shapes going 1→2→3, size going small→medium→large, or shading going white→gray→black are all progressions. The missing cell should continue the sequence.
Constant rules — An attribute stays the same across an entire row or column. Every cell in row one has triangles; every cell in row two has circles. The missing cell must match its row or column's constant.
Distribution rules — Each row (or column) contains one of each value for a given attribute. Each row has one triangle, one circle, and one square — just in different positions. This is the "Latin square" structure common in harder matrix problems. The missing cell must complete the set.
Rotation rules — A shape rotates by a fixed angle across cells. An arrow pointing up rotates 45° or 90° with each step. Look for orientation changes that follow a consistent increment.
Combination rules — Elements from the first two cells in a row combine or cancel to produce the third. Overlapping shapes might merge, or a shape present in both cells might disappear in the third. These are common in harder questions and require a different type of inspection — comparing cells pairwise rather than looking for a progression.
Why People Get These Wrong
The most common mistake is rushing. People glance at a matrix, notice one feature that seems to follow a pattern, and select the answer that matches it — without checking whether the answer also satisfies the other attributes. An answer choice might correctly continue the shape progression but have the wrong shading, or the right number of elements but in the wrong positions.
The second common mistake is confirmation bias: once you've spotted what looks like a rule, you unconsciously stop looking for evidence against it. The step-by-step method guards against this by requiring explicit checks across all cells before committing to an answer.
The third mistake is not using elimination. When you're uncertain about the rule, applying each answer option back to the grid and asking "does this break any visible pattern?" often reveals one or two clearly wrong options immediately — narrowing the field even when you can't identify the full rule.
Does Practice Actually Improve Abstract Reasoning?
Yes — but with some nuance. Familiarity with the format and common rule types accounts for a meaningful portion of the improvement, which is why repeated practice on the same test type shows gains. Whether these gains reflect genuine improvement in fluid reasoning or just format learning is an open question in the research, but for practical purposes — exam preparation, cognitive training, professional assessments — the improvement is real and transferable within the domain of visual reasoning tasks.
The Matrix Reasoning Test on this site generates novel problems on every session, which means you're practicing rule detection on genuinely new material rather than memorizing specific patterns. For a broader look at the cognitive science behind why abstract reasoning ability matters, the article on Raven's Progressive Matrices covers what these tests actually measure in depth.
Put the Method Into Practice
The test below generates 3×3 matrix problems with a 20-second timer per question. Apply the steps above: check rows and columns, identify what's constant, test one attribute at a time, and eliminate options before committing. Ten questions — see how many you can solve systematically rather than by intuition.