Inductive Reasoning Test: Pattern Rules and Preparation

Almost nobody fails an inductive item because the pattern is beyond them. They fail because they find one rule and stop. The option they then choose matches that rule perfectly and breaks a second rule they never looked for, which is exactly what it was built to do. Difficulty here scales with the number of rules running at once, not with how clever any single rule is. A shape rotating 45 degrees per step is easy. The same rotation plus a shading cycle of period 3 plus a count that increments every other frame is hard, though each rule alone is trivial. What follows is which rule families generate these items, how the formats differ, and how the wrong options are built.

Practising for a real assessment? The Abstract & Inductive Reasoning Practice Pack runs these question types at full length — 5 timed tests, 150 questions, a worked explanation on every one.

The rule families that generate inductive items

Every item applies rules to a short, fixed list of attributes: count, shape, size, shading, orientation and position. There are only so many things a rule can do to an attribute, and the families below cover nearly all of them. Formats differ by what you do with the rule once you have it: continue a sequence, sort figures into groups, or recover an unstated rule from a majority.

Attribute progression

One attribute changes monotonically while everything else holds still. Counts go 2, 4, 6, 8; a shape gains a side per frame; a figure grows by a fixed proportion; shading steps through white, light grey, dark grey, black. Mechanically the simplest family: the rule is recoverable from any two adjacent frames and confirmable on a third. The usual error is reading a compounding step as constant, or 2, 4, 8, 16 as addition.

Rotation and reflection

A shape turns by a constant angular step, commonly 45 or 90 degrees, in one direction. Step size and direction are separate things to fix, and both need checking across two intervals, because one interval of 90 degrees clockwise is also 270 degrees anticlockwise. Reflection is not rotation: a flipped figure preserves size, shape and every proportion, so it reads as a turn to anyone matching on silhouette, yet no turning produces it. Track one asymmetric feature, not the figure.

Positional movement

An element translates around a grid or perimeter by a fixed number of cells per frame, wrapping around at the end. The rule has three parameters: step size, direction, and the path travelled, which may be a row, a column, a diagonal or the perimeter. Distractors change exactly one of them: the same direction with a step of 2 instead of 3, or the right step running anticlockwise. Count cells from frame one to two, then confirm it from two to three.

Alternation and cycling

Two or more states repeat on a fixed period, and the trap is assuming that period is 2. White, grey, black, white is not alternating; it is a cycle of length 3 caught mid-turn, and the next frame is grey. When a value repeats, count the gap between its occurrences: a four-frame stem is short enough for a period of 3 to look like noise.

Conditional and compound rules

Two independent rules act on different attributes at once: the shape rotates and the count increases, or the border thickens only when the inner figure is shaded. This is the main difficulty jump in the format, and a jump in bookkeeping rather than insight. A conditional rule links attributes, so one attribute's behaviour depends on another's state, and it surfaces only if you tabulate both across every frame.

Set membership and two-group classification

Two groups of figures, A and B, each governed by its own rule, and you assign a new figure to one. There is no next term, so sequence thinking does not apply. You need the feature that splits the sets: look for what every A has that no B has. A true description of group A is useless if it also fits B.

Odd one out

Several figures share an unstated rule and one does not, so you work backwards: find the property held by all but one. Because the rule is never given, more than one property can appear to separate a figure; the answer breaks the property shared by the most others.

Scoring and timing, and what they mean for strategy

Two scoring models are in circulation. Under raw scoring your mark is the number of correct answers, a blank scores the same as a wrong answer, and there is never a reason to leave anything unanswered. Under formula scoring a fraction of a mark is deducted for each wrong answer and nothing is deducted for an unanswered one, which exists to cancel the expected gain from random guessing. When the marking takes something off for a wrong answer but nothing for a blank, an end-of-test guessing spree moves your score down, not up. Partial elimination changes that: once two of five options are ruled out on a real reason, an answer among the rest is usually worth giving.

Timing models vary as much. A fixed question count against a fixed clock makes pacing a division problem and means every item is reachable. An indefinite bank against a clock means nobody finishes, the raw total is the measure, and 3 minutes spent proving a compound rule is 3 minutes of easier items forfeited. Per-section timers stop you borrowing minutes from the classification block to rescue the sequence block. Per-question timers remove skipping and returning, making a fast partial rule more valuable than a complete one. Adaptive versions pick each item from your previous answers, so early items carry unusual weight.

The approach that survives all of these is the same: enumerate the attributes, count, shape, size, shading, orientation, position, and test each independently across the frames rather than waiting for a pattern.

A worked example

Four frames are given and you choose the fifth. Each is a square panel with a background and a single arrow at its centre.

Frame 1: white background, arrow pointing north. Frame 2: grey background, arrow pointing north-east. Frame 3: black background, arrow pointing east. Frame 4: white background, arrow pointing south-east.

Two rules run at once. The arrow turns 45 degrees clockwise per frame, confirmed across three intervals: north to north-east, north-east to east, east to south-east. The background cycles white, grey, black on a period of 3, and frame 4 returning to white confirms that period rather than breaking the pattern. Frame 5 is an arrow pointing south on a grey background.

The options are built as follows. South on a white background applies the rotation correctly but treats the background as static, or reads frame 4 as the end of a darkening progression. South on a black background gets the rotation right and reads the shading as alternating white and black from frames 3 and 4, a period of 2 imposed on a cycle of 3. South-west on a grey background gets the shading right but steps 90 degrees instead of 45. East on a grey background keeps the step size and reverses direction. Each wrong option satisfies every rule but one, which is why checking one attribute feels conclusive and is not.

How to prepare for this type specifically

Practise the enumeration rather than the puzzles. Take any item and write the six attributes in a column, then fill in each value across all frames before looking at the options. Most columns will be constant, and the ones that are not are your rules. This is slower on easy items and much faster on hard ones, and it removes the failure mode of finding a rule and stopping: the table is not finished until every attribute is accounted for.

Then count rules before choosing. With one rule found, ask what else moved. With two, check whether one depends on the other rather than running beside it. Drill period counting: when a value repeats, measure the gap rather than assuming 2.

Treat classification and odd-one-out items as a separate skill and practise them apart. Extending a sequence and splitting a set are mechanically different tasks, and fluency in one tells you nothing about the other. For classification, test every candidate rule against both groups, not the one you formed it from.

Finally, establish which scoring model applies before you sit down, because it decides whether to answer the last three items you never read. Under formula scoring, three blanks beat three guesses.

Frequently asked questions

TestSolve is independent and not affiliated with any test provider or employer named on this page. All product names and trademarks belong to their respective owners.