When more than one rule is running at once
Difficulty in these puzzles comes from simultaneity, not from complexity.
There is a moment in most reasoning tests where the puzzles stop being manageable and start looking like noise. Nothing lines up, every row seems to contradict the last, and the sense arrives that you have hit the limit of what you can do.
Almost always, what has actually happened is smaller than that. The puzzle has gone from one rule to three. Each of the three is as simple as the ones you were solving a minute ago. What changed is that they are running at the same time, and your eye is trying to take them in together.
Why simultaneity is so much harder than complexity
Working memory is the constraint. Holding one rule while you test it against a grid is comfortable. Holding three, while also tracking which of them you have already checked and which cell you are on, is not — and the moment one slips you have to start over, which is why these puzzles produce the particular feeling of going round in circles.
The fix is not to hold more in your head. It is to stop trying, and to check the rules one at a time, writing each down as you fix it. Three sequential single-rule problems are dramatically easier than one triple-rule problem, and they take less time even though there are three of them.
Separate by attribute, not by cell
This is the whole technique and it is worth stating plainly. Do not work through the grid cell by cell asking 'what is going on here'. Work through the attribute list, asking one question of the entire grid each time.
- Look only at count across the whole grid. Ignore shape, shading, position, size. Write down what count does.
- Now look only at shape. Ignore everything else. Write it down.
- Now only shading. Then only orientation. Then only position.
- Read your notes. You now have three or four simple statements, and the answer is whatever satisfies all of them at once.
Deliberately ignoring most of what is in front of you feels wrong, and it is the entire skill. Each pass is easy because you have made it easy.
Pass one, count only: 1, 2, 3 across every row. The missing cell has three. Pass two, shape only: constant along each row, and row 3 is squares. Pass three, size only: constant along each row, row 3 is large. Pass four, shading only: it alternates, and row 3 went empty, shaded, so next is empty. Answer: three large empty squares. Four passes, each trivial. Attempted all at once it reads as chaos.
Rules can interact, and usually do not
Most stacked rules are independent — count does not care what shading is doing. Occasionally a puzzle makes one rule depend on another: shading changes only when the count is even, say, or a shape rotates only when it moves.
Do not go looking for interactions first. They are rare, and assuming them turns a solvable puzzle into an unsolvable one. Check for an interaction only when your independent rules produce an answer that is not among the options — that failure is the signal, and it is a reliable one.
Knowing when to stop looking
You have enough rules when they uniquely determine the missing cell. If your rules leave two options equally valid, you are missing one — and it will be an attribute you have not yet given its own pass. Go back to the checklist rather than re-examining the ones you already did.
The opposite error is rarer but real: finding more structure than is there. If a fourth rule only works by treating one cell as an exception, it is not a rule. Genuine rules hold everywhere.
Write things down. On paper, in the margin, in four words. Everyone who is good at these writes things down, and the belief that doing it in your head is more impressive costs more marks than any other single habit.
Practising the separation
Take a puzzle you have already solved and solve it again, deliberately, one attribute at a time, naming each rule aloud. It feels pointless because you know the answer. The point is not the answer — it is making the separation automatic, so that the next time a grid looks like noise your response is to start the checklist instead of staring.