Abstract Reasoning Test

A set is built on one rule and one only, and there are four of them. They are listed here because a puzzle where the rule could be anything is not a test of reasoning. It is a guess at what the author had in mind.

• The same number of filled squares. Four figures share a count and one does not.
• A mirror image down the middle. Four figures are symmetrical and one is not.
• No two filled squares touching. Four figures keep every square apart and one has a pair side by side.
• Nothing in the middle. Four figures leave the centre clear and one does not.

Within a set, only the rule varies. If the rule is the count, then every one of the five figures is equally lopsided, equally crowded, and equally busy in the middle.

Without that, an item can have two right answers. One figure could be the only symmetrical one while another is the only one with a filled centre, and both readings would be defensible.

That is the flaw in a lot of free puzzle pages, and it is invisible until you argue with one. Here the four features are checked against every figure in the set before it is shown.

It also makes the reason line honest. The rule named after your answer is the rule the set was built on, not a rule fitted to it afterwards.

Graduate schemes give you a fixed paper and a fixed number of minutes, so this page does the same. Twenty sets, five minutes by default, and a running count of how many you have right.

That makes the score two things at once. It is how many you could place, and it is how many you got through. Both are on the result card, because a score of nine from ten attempts is a different result from nine from twenty.

Turning the clock off is the practice mode. It is the honest way to learn what the four rules look like, and the number it produces cannot be compared with a timed run.

Bethell-Fox, Lohman and Snow watched where people looked during figure problems. What separated them was how they searched, not how fast they could move their eyes.

This page asks which figure is wrong. That is a different question from what comes next, and different again from what fills the gap in a grid.

The pattern recognition test runs a strip of six tiles and asks for the seventh. It ladders up through more rules at once and it has no clock.

A matrix puzzle lays out a three by three grid with one cell missing. Rules run across the rows and down the columns at once. Carpenter, Just and Shell showed the difficulty there comes from holding part-finished goals in mind.

An odd-one-out set is the flattest of the three. Every item is worth the same, there is nothing to hold between items, and the pressure comes from the clock instead.

These are our figures, our four rules and our clock. Nobody has measured a thousand people on this exact arrangement, so a percentile would be an invented number.

Test publishers do report norm groups for their own abstract reasoning papers. Those tables describe one publisher's items, sat under supervision, by the people who applied to those schemes that year. None of that transfers to a browser tab.

What you can check is the arithmetic. Five figures means a guess is right one time in five, so a run of twenty answered at random lands near four.

For a number that does have a distribution behind it, processing speed by age shows what one looks like when it is real.

It is worth what practice is always worth. You get quicker at these four rules, on these figures, in this layout. That is a real thing to gain before a test in a similar format.

It is not a measure of intelligence. Nothing here has been checked against any measure of intelligence, and no score on this page is a claim about one.

It is not a prediction about a job either. This page cannot tell you whether you will pass an employer's paper. That paper has its own items, its own clock and its own pass mark.

And it is not a screening test. Nothing here detects or rules out any condition. If something has genuinely changed in how you think or remember, that belongs with a doctor rather than a web page. How working memory works is the nearest thing to an explanation of the load these puzzles put on you.