EPSO Abstract Reasoning Pattern Types: The Rule Families Worth Learning

Once you know what the EPSO abstract reasoning test looks like — five frames, choose the sixth, five options from A to E, ten items in ten minutes — the next question is the harder one. What is there to actually study?

Abstract reasoning has no syllabus. You cannot revise it the way you revise EU Knowledge. What you can do is build a vocabulary of rules, so that when a series is put in front of you, you are checking a short mental list instead of staring and hoping. This article is that list.

If you have not yet confirmed the format and the scoring, start with the companion article on what the official sources actually say about the EPSO abstract reasoning test. PassEPSO is an independent preparation platform and is not affiliated with EPSO or any EU institution.

An honest caveat before the list

EPSO publishes no official taxonomy of abstract reasoning rules. None. There is no document that says "these are the transformations we use", no category labels attached to sample items, and no worked explanations — the interactive sample test tells you only whether you were right, and the PDF solution slide is the question with a box around the correct option.

So the list below is not an official list and nobody can honestly present one as such. It is derived from two things: the item behaviour visible in EPSO's own published samples, and the standard vocabulary of the wider psychometric literature on figural series. Treat it as a working framework that covers the ground, not as a leaked answer key.

That caveat is worth taking seriously rather than skimming. Any source that presents a numbered set of "the official EPSO pattern types" is presenting an inference as a fact.

How to use a rule list without fooling yourself

The point of naming rules is not recall. Nobody passes by reciting fourteen categories. The point is that a named rule is a searchable rule.

When you have sixty seconds and the series is not resolving, an unnamed intuition gives you nothing to do next. A named list gives you an order to check things in: is anything turning? is anything moving? is anything being counted? Most items yield to the second or third question on that list.

Naming also fixes review. "I missed it" is not a diagnosis. "I missed it because I tracked the big shape and the rule was on the small marker's circuit" is a diagnosis, and after twenty of those you will know precisely which two families cost you marks.

One structural note worth holding on to: five given frames means four transitions. A rule has to hold across all four. If your candidate rule explains three transitions and not the fourth, it is the wrong rule, not a rule with an exception.

Group 1: rules about orientation

Rotation

The whole figure turns, or one element within it turns while the rest stays put. The two are genuinely different items and it is worth deciding which you are looking at before you commit. Rotation can also run at different speeds for different elements — an outer shape turning one step per frame while an inner marker turns two.

The practical tell is an asymmetry: a notch, a tail, a single filled corner, an arrowhead. Perfectly symmetric shapes hide rotation completely, which is sometimes the point.

Reflection

The figure flips across an axis — vertical, horizontal or diagonal. Diagonal reflection is the one candidates most often misread as rotation, because on many figures the two produce a similar-looking result. The distinguishing check is handedness: reflection reverses it, rotation does not. Find one chiral detail and follow it.

Takeaway

Before anything else, ask whether the figure is turning or flipping. If you cannot tell, look for the one detail that would differ between the two and check only that.

Group 2: rules about position

Translation and movement

An element travels in a consistent direction and by a consistent step. The interesting part is not the movement itself but what happens when it reaches the edge. There are three common behaviours, and they look nothing alike in the answer options:

• Wrap — it disappears off one side and reappears on the opposite side.
• Bounce — it reverses direction and comes back.
• Reset — it returns to the starting position and begins again.

Whenever an element is near an edge in frame four or five, work out which of the three the series has been using before you predict frame six. This is one of the most reliable places for a series to catch people out.

A marker advancing around a circuit

A dot, arrow or small shape moves around a fixed track — the positions of a clock face, the points of a ring, the perimeter of a square. It advances a fixed number of stations per frame and it eventually comes round again.

The trap here is a step size larger than one. A marker advancing three stations per frame does not look like it is travelling at all; it looks like it is jumping randomly. Count stations rather than watching the motion.

Takeaway

For anything that moves, establish two things: the step size, and the behaviour at the boundary. Both are needed to predict the sixth frame.

Group 3: rules about quantity

Counting and progression

Something is being counted and the count follows a progression across the frames. What gets counted is more varied than most candidates expect:

• the number of elements or shapes;
• the number of sides on a shape;
• the number of intersections where lines cross;
• the number of enclosed areas a figure contains;
• the number of axes of symmetry.

Sides, intersections, enclosed areas and symmetry axes are the ones people miss, because they require you to count a property of a figure rather than count objects in a frame. If a series looks like the shapes are simply changing arbitrarily — triangle, square, pentagon, or a scribble getting more tangled — count something about them before concluding there is no rule.

Progressions are not always +1. They can decrease, they can step by two, and two independent counts can run in opposite directions in the same item.

Takeaway

When a series resists every other reading, count. Objects first, then sides, then intersections, then enclosed areas.

Group 4: rules about appearance

Size scaling

An element grows or shrinks by a consistent factor. Easy to see when one element scales alone; much harder when everything scales together, because the frame then looks unchanged apart from the zoom. Compare an element against a fixed reference in the frame rather than against your impression of the previous frame.

Shading and fill

Elements cycle through fill states. Remember that official EPSO material is pure black-and-white line art, so this always means black, white and hatching or similar — never colour. Fill can cycle on a single element, alternate between two elements, or be tied to something else in the frame (see attribute dependency below).

Shape substitution

One shape is systematically replaced by another, either in a repeating cycle (circle, square, triangle, circle...) or by a rule that swaps a class of shapes for another. The cycle length is the thing to identify first; once you know it, the sixth frame is arithmetic.

Takeaway

Appearance rules are the easiest to spot and the easiest to over-trust. A fill cycle you have correctly identified may still be the secondary rule in the item, not the one that separates the options.

Group 5: rules about composition

Addition and removal of elements

Elements are added or taken away as the series progresses. Simple in principle, but check where new elements appear — the position of the addition is often the real rule, and the fact that something was added is just the surface.

Symmetry as a governing property

Rather than the figure changing symmetrically, symmetry itself is the property under control: a figure becomes symmetric, loses symmetry, or gains an additional axis with each frame. This one hides well, because each individual frame looks like an unrelated arrangement until you ask the symmetry question directly.

Overlay logic

Two or more components are combined, and the rule governs how they merge: what survives where they coincide, what is dropped, what is kept from each. Series items build this in more subtly than the grid formats used elsewhere in the testing industry, but the underlying question is the same — when two things sit on top of each other, what results?

Takeaway

If individual elements seem to follow no rule, ask whether the rule is about the figure as a whole: its symmetry, its total composition, how its parts combine.

Group 6: rules about relationships

Attribute dependency

One feature's value is determined by another feature. The shading depends on the number of sides; the size depends on the position; the orientation depends on the fill. Nothing progresses on its own — the relationship is what is constant.

These items feel unsolvable if you are looking for a progression, because there isn't one. The giveaway is that a feature keeps changing without any consistent direction while a second feature changes alongside it.

Interaction rules

Something happens when elements touch or overlap: a shape changes fill on contact, an element is absorbed, a marker reverses when it meets another. Between contacts, nothing seems to be happening at all, which is why these are easy to misread as broken series.

Takeaway

If a feature changes with no consistent direction, stop looking for a progression and start looking for what it is reacting to.

Composite items

Several of the families above can operate in the same item at once — a shape rotating while a count increases while fill alternates. The rules are independent, and that is the useful part: track them one at a time rather than trying to perceive the whole thing simultaneously.

The five answer options are built so that partial rule detection is not enough. An option that satisfies two of three active rules will be sitting there looking correct. Which gives you a practical test: if two options both look right, you have not found all the rules yet, and the difference between those two options tells you exactly where the missing rule lives. That is faster than starting over.

A review method that uses the list

The rule vocabulary earns its keep in review, not during the test. A method that works:

1. Do ten items under a strict ten-minute clock. No pausing, no lookups.
2. Afterwards, for every item you missed, write the rule family in your own words in one line.
3. Tag it with the family name from the list above.
4. Every few sessions, count the tags.

After a handful of sessions the tags stop being evenly distributed. Almost everyone has two families that account for most of their losses — commonly edge behaviour and attribute dependency, though it varies. Drill those two specifically for a few days, then go straight back to mixed sets, because the real test does not announce which family is coming.

The reason this works is that the official samples give you no explanations at all. If you do not write down the rule yourself, nothing else will.

How much of this to learn

A word of proportion. Abstract reasoning in the 2026 AD5 competition is a pass/fail gate: the pass mark is a combined 10/20 across the numerical and abstract tests together, and neither test contributes to the preliminary or the final ranking score. Verbal reasoning, by contrast, carries 35% of the final combined score.

So the goal is a dependable margin above the combined threshold, not mastery. Learn the families well enough to name what you missed, practise abstract in mixed sets with numerical reasoning since the pass mark is shared between them, and once you are consistently safe, redirect the time to the tests that are actually weighted.

Ten questions in ten minutes leaves one minute per item. Knowing what to look for is most of what buys that minute back.

Sources and status of this list

• Notice of Competition EPSO/AD/427/26, Official Journal C/2026/711, 5 February 2026 — for the format, timing and scoring facts.
• EU Careers — Abstract reasoning — for EPSO's own description of the test and its published sample items.

The rule families themselves are not published by EPSO in any form. They are our working framework, derived from the published sample items and from the standard vocabulary used for figural series tests. Treat the EUR-Lex notice and the EU Careers pages as the binding source for anything official.