Game guide 16 min read

RuleMatch

RuleMatch asks one question in three disguises: which single property is doing the work? Sometimes a figure changes and you must find another pair that changes the same way. Sometimes a group of figures shares something and you must find the one that does not, or every one that does. The figures are simple on purpose. What gets tested is how cleanly you compare them.

What you see

Every figure in RuleMatch is built from four properties, and only four. Learn them once and every trial becomes a comparison between short descriptions rather than a picture puzzle.

ShapeCountFillTurnfour shapesone to fouroutline, striped, solid45° steps, clockwise
The figure alphabet. Every RuleMatch figure is a shape, drawn one to four times, in one of three fills, turned in 45-degree steps.
  • Shape: circle, square, triangle or diamond.
  • Count: the shape drawn one to four times, in one row or two.
  • Fill: outline, striped or solid.
  • Turn: the whole figure rotated in 45-degree steps. Positive turns are clockwise. Only triangles turn: a circle looks the same at every angle, and a square or diamond turned 45 degrees would pass for the other shape, so those three are always drawn upright.

The board has two framed areas. The upper frame, labelled Example (or Examples when there are several), holds the examples. The lower frame, labelled Options, holds the options; on a desktop each option carries its number in its top-left corner. The prompt above the board names the task for this trial, and you answer by clicking a card or pressing its number key (1 up to the number of options, which is 3, 4 or 5), then pressing Check answer or Enter. Nothing counts until you check.

There are three task types.

  1. Same rule (pair mode). The prompt asks “Which option follows the same rule as the example?” The example card shows figure A on top, a down arrow, and figure B below. Exactly one property changed between them. Each option is also an A-to-B pair, drawn the same way, and exactly one option makes the same change.
  2. Find the breaker (misfit mode). The prompt asks “Which option breaks the rule the examples share?” The examples are three single figures that share one property value. All but one of the options share it too. Pick the one that breaks it.
  3. Select all that fit (fits mode). The prompt reads “Select every option that fits the rule the examples share.” The examples are three figures that share one property value. Select every option that has it. At least one option fits and at least one does not. On some trials, when only one or two fit, the line under the board says “One or two options fit.” Selecting a card again unselects it.
ExampleOptions1234
A Same rule trial. The example card shows one figure becoming another, top to bottom; option 1 repeats the change (outline to striped) and is shown selected.

In a Same rule trial, the change is always one of a small set. At the easier levels it is count up by one, count down by one, fill moved one step along the cycle outline, striped, solid (and back to outline after solid), or a 90-degree clockwise turn of a triangle. From level 5 the change is always a triangle's turn: 45 or 90 degrees, clockwise or anticlockwise. Every change is one you can see. Wrong options are built from the same menu, so a distractor can use any of these changes, including a turn at an easy level. In the two shared-property task types, the shared property is always shape, count or fill, and the examples are generated so that the other two of those vary (and, when every example is a triangle, so does the turn). That guarantees one answer: only one property is constant across the examples.

In practice mode there is no countdown. A clock in the side rail counts up while you work, and the time you take feeds a small part of the score (see below). A session starts at level 3 and runs until you press End session.

LevelTaskOptionsWhat the rule can be
1Same rule3count ±1, fill step, 90° turn
2Same rule4count ±1, fill step, 90° turn
3Find the breaker4shared shape, count or fill
4Select all that fit4shared shape, count or fill
5Same rule4turns only: 45° or 90°, either way
6Find the breaker5shared shape, count or fill
7Select all that fit5shared shape, count or fill
8Same rule4turns only
9Find the breaker5shared shape, count or fill
10Select all that fit5shared shape, count or fill

What it measures

RuleMatch is a figural inductive reasoning task: you infer a general rule from a few specific cases and apply it to new ones. On our profile it feeds the Pattern recognition facet of Reasoning, alongside Matrix Logic, and carries 10 percent of the Reasoning dimension score.

Induction

Working from particular examples to a general rule. The rule is never stated; you have to find it, then use it on cases you have not seen.

The educational psychologist Karl Josef Klauer gave one of the most detailed accounts of what induction is made of. In his view the core process is comparison: you detect similarity, difference, or both, in either the attributes of objects or the relations between them. The three RuleMatch task types line up with three of his task classes.

  • Select all that fit is generalisation: find what several items have in common, then extend it.
  • Find the breaker is discrimination: find the item that does not share the common attribute.
  • Same rule is recognising a relation: the thing to compare is not a figure but the change from one figure to another.

This matters for more than labelling. Klauer and Phye's 2008 meta-analysis pooled 74 training experiments with nearly 3,600 children and found that teaching a deliberate comparison strategy improved performance on fluid intelligence tests and on classroom learning, with larger transfer to academic learning than to intelligence tests. The trained strategy centres on systematic comparison of attributes and relations, which is the same move the method below asks you to make on every trial.

That finding is the practical heart of this lesson. Most RuleMatch errors are not failures of logic. They are encoding errors: a fill misread, a count not noticed, a turn assumed to be clockwise when it was anticlockwise.

How it is scored here

Each trial is marked by our evaluator as soon as you press Check answer. The correct option or options then turn green and a wrong pick turns red, and a line states the rule, for example “Option 3 follows the same rule: the count goes down by 1” or “Option 4 breaks the rule: every example shows 2 figures”.

  • Same rule and Find the breaker: one pick. Right is full credit, wrong is zero. There is no partial credit.
  • Select all that fit: your score is (correct selections minus wrong selections) divided by the number of options that truly fit, and it never goes below zero. Only a perfect selection counts as Correct.

Take a fits trial with two true matches. Select both and nothing else: 100 percent. Select one true match only: 50 percent. Select both true matches plus one that does not fit: (2 − 1) ÷ 2, also 50 percent. A wrong selection costs exactly as much as a missed one, so selecting generously to be safe does not pay.

Difficulty adapts trial by trial. A Correct answer moves you up one level, an Incorrect answer moves you down one, and a Partly correct select-all answer keeps you where you are. Levels run from 1 to 10.

Your RuleMatch score is on our 0–100 scale, where 50 means what the difficulty you work at predicts. Each trial's credit is its accuracy, adjusted a little for time, multiplied by the difficulty of that trial, so a correct answer at level 8 is worth far more than one at level 2. In RuleMatch accuracy carries about four-fifths of the weight and speed about one-fifth, which is more weight on speed than in most of our reasoning games. A score appears once you have enough evidence (at least four trials), small samples are pulled towards 50 until more data arrives, and recent sessions count for more than old ones.

The session report also shows three diagnostics: Selections that were right (precision), Matching figures found (recall), and Figures wrongly included. If precision is low, you are selecting too freely in fits trials. If recall is low, you are stopping too early.

Where you will meet it

The same family of task appears across aptitude batteries, usually under the label inductive or abstract reasoning.

Publicly reported uses. Formats change, so check the latest invite you receive.

  • AonInductive-logical thinking (scales cls): work out the common rules in given patterns, then assign new images to them. Inductive reasoning (scales ix): find the item in a series of 9 that does not fit.
  • SHLInductive reasoning tests that assess your ability to identify underlying patterns in information and predict outcomes.
  • CapgeminiCandidates report an inductive-logical thinking game among the four games drawn from a pool of 24 in the game-based aptitude round.

Aon's two inductive formats map closely onto our shared-property tasks: assign new images to the pattern is a fits trial, and find the item that does not fit is a misfit trial. Our Same rule trials are closer to the analogy element inside matrix and series items, where the question is how a figure changes from one cell to the next. RuleMatch is not a copy of any vendor's test. It trains the comparison habit those tests share, with a closed set of properties so you can see exactly what you missed.

The method

  1. Read the prompt first

    The prompt asks which option follows the same rule, which option breaks the rule, or asks you to select every option that fits. The first asks what changed; the other two ask what stayed the same. Decide which question you are answering before you look at a single figure.

  2. Tag every figure in four words

    Describe each figure as count, fill, shape, turn: 2 striped triangles, upright. Say it in that order every time. A fixed order is what stops you skipping the property you find least salient, which for most people is fill or turn.

  3. Same rule: diff the pair one property at a time

    Compare A and B on count, then fill, then shape, then turn. Exactly one differs. Name the change with its direction and size: count +1, outline to striped, 45° anticlockwise. A change without a direction is only half a rule.

  4. Shared property: find the constant

    Run down the three examples one property at a time. By design, of count, fill and shape exactly one is the same on all three. The first property with a single value across all three is the rule. Remember that turn is never the shared property here, so you only need to check count, fill and shape.

  5. Test options against the rule, not against the pictures

    Tag each option and check only the rule property. In Same rule trials, reject any option whose change is on a different property, then reject any whose change goes the wrong way or the wrong distance.

  6. Check the near miss before you commit

    The strongest distractor usually changes the right property the wrong way (count −1 instead of +1, a 45° turn instead of 90°). Re-read your chosen option with the direction in mind, then check.

Worked example 1: same rule

ExampleOptions1234
Worked example 1. Which option follows the same rule as two striped triangles becoming three?
Show the solution

Tag the example card: 2 striped triangles becomes 3 striped triangles. Fill, shape and turn are unchanged, so the rule is count +1.

  • Option 1: 3 outline circles become 2. Count changed, but by −1. Right property, wrong direction.
  • Option 2: 1 solid diamond becomes 2 solid diamonds. Count +1, nothing else changes. This is the answer.
  • Option 3: outline becomes striped. Fill changed, not count.
  • Option 4: the triangle turns 90° clockwise. Turn changed, not count.

Option 1 is the trap. A quick glance sees the number changed and stops. Naming the direction (+1) is what rules it out.

Worked example 2: find the breaker

ExamplesOptions1234
Worked example 2. Three examples share one property. Which option breaks it?
Show the solution

Run down the examples property by property. Counts: 1, 3, 2, so count varies. Shapes: circle, triangle, square, so shape varies. Fills: solid, solid, solid. The shared property is solid fill.

Now check only fill on the options: 4 solid diamonds, 2 solid triangles, 3 striped circles, 1 solid square. Option 3 breaks the rule.

Notice how little the other properties matter once the rule is found. Option 2's triangle points down and option 1 has four shapes, which look unusual, but neither touches fill.

Worked example 3: select all that fit

ExamplesOptions12345
Worked example 3. Select every option that shares the property the three examples have in common.
Show the solution

Examples: 2 outline squares, 2 solid triangles, 2 striped circles. Fill varies, shape varies. Count is 2 every time. The rule is count = 2.

  • Option 1: 2 solid diamonds. Fits.
  • Option 2: 3 outline squares. Same shape and fill as an example, but count 3. Does not fit.
  • Option 3: 1 striped triangle. Does not fit.
  • Option 4: 2 outline triangles, turned. Fits; the turn is irrelevant.
  • Option 5: 4 solid circles. Does not fit.

Select options 1 and 4. Option 2 is the lure: it looks like the first example, and picking it would cut your score from 100 to 50 percent.

Tips and traps

Matching by resemblance

  • Picks the option that looks most like the examples
  • Notices the most eye-catching difference and stops
  • Selects extra options in fits trials to be safe
  • Treats every change of count as the same rule

Matching by property

  • Tags every figure as count, fill, shape, turn
  • Checks all four properties before naming the rule
  • Selects only options that pass the rule test
  • Names direction and size: count +1, 45° anticlockwise

Levelling up

Difficulty climbs in two ways. The option count rises from three to five, which gives distractors more room. And from level 5, Same rule trials use turns only: 45 or 90 degrees, in either direction. Turn rules are the hardest to encode because you must track direction and size, and because shape symmetry can hide them.

To adapt at the top levels:

  • Fix a reference point for turns. For a triangle, track where its apex points: up, up-right, right and so on in 45-degree steps. For a striped figure, track the stripe angle.
  • Rehearse direction words. Clockwise and anticlockwise errors are the most common misses at levels 5 and 8. Say the direction out loud during practice until it is automatic.
  • In five-option fits trials, test every option. Do not stop at the first two matches unless the line under the board caps the count.
  1. Week 1
    • Three short RuleMatch sessions of 12 to 15 trials, tagging every figure aloud as count, fill, shape, turn
    • After each session, read the precision and recall diagnostics and note which one is lower
    • Spend one session only on accuracy: take the time you need, and never check an answer you have not tagged
  2. Week 2
    • Three sessions aiming to hold level 5 or above
    • Before each Same rule answer, say the rule with its direction (for example, 45 degrees anticlockwise)
    • Add one Matrix Logic session to practise the same comparison habit inside a grid
  3. Week 3
    • Three sessions at a steady pace: tag quickly, but keep the full comparison pass
    • Keep a tally of errors by cause: fill misread, turn direction, count, over-selection
    • Finish with one timed-feel session: aim for a decision within about 20 seconds per trial without dropping accuracy

Check yourself

5 practice questions
  1. A Same rule card shows 1 solid square becoming 1 outline square. What is the rule?
  2. In a Select all that fit trial, two options truly fit. You select both of them and one option that does not fit. What is your score for the trial?
  3. The three examples in a Find the breaker trial are 1 striped circle, 3 striped diamonds and 4 striped triangles. Which option breaks the rule?

and 2 more

Sources
  1. Klauer, K. J., & Phye, G. D. (2008). Inductive reasoning: A training approach. Review of Educational Research, 78(1), 85–123. ERIC record
  2. Pásztor, A., Magyar, A., Pásztor-Kovács, A., & Rausch, A. (2022). Online assessment and game-based development of inductive reasoning. Journal of Intelligence, 10(3), 59. PMC
  3. Sternberg, R. J. (1977). Component processes in analogical reasoning. Psychological Review, 84(4), 353–378. ERIC record
  4. Sternberg, R. J., & Rifkin, B. (1979). The development of analogical reasoning processes. Journal of Experimental Child Psychology, 27(2), 195–232. PubMed
  5. Armour-Thomas, E. (1987). Review of componential research, summarising Sternberg and Rifkin's encoding-time finding. ERIC ED286933. PDF
  6. Aon. Prepare for your online assessment: descriptions of scales cls and scales ix. aon.com
  7. SHL. SHL Direct practice tests: inductive reasoning. shl.com
  8. PrepInsta. Capgemini Exceller game-based aptitude test questions (candidate-reported). prepinsta.com

Put it into practice: RuleMatchSign in to try it