← UX laws

Law of Prägnanz

Wertheimer, 1923PerceptionInteractive

Given a figure that could be read several ways, the eye takes the simplest reading available and hands you that one, finished.

Prägnanz is the one the other Gestalt laws hang off. Given a figure that could be read several ways, the eye takes the simplest reading available and hands you that one, finished, before you have decided anything. Simplest means fewest parts, most regular, most symmetrical. It does not mean correct.

There are four figures below. Each one is drawn from parts that are messier than the thing you will see, and each one has a slider that pulls those parts away from each other. You say what the figure is first, then you take it apart yourself.

The answers are not scored and there is no right one. Two of the three options for each figure are honest descriptions of what is on screen, and one of them is what the parts actually are. Picking the literal answer is not cleverness — most of these are recognisable if you have seen the demonstration before, and none of that changes what happens when you look at it fresh.

PRESS START BELOW
four figures
Nothing is hidden from you and nothing is timed. Every figure is drawn from its real parts at every position of the slider — at zero they are simply sitting where they were drawn, which is the only trick any of these use.

What the four are made of

The triangle is three discs. Each has a sixty-degree wedge missing, and the wedges point at each other. The three straight edges you see belong to nothing — they are the shortest way to explain why three discs would all be interrupted in the same place, and a triangle lying on top is a cheaper explanation than three coincidences. Turn the wedges and the triangle does not fade. It stops existing.

The bar is two bars. One is 28 thick and the other 30, their centre lines sit 6 apart, and there are 96 units of nothing between the ends. A panel covers the join. The eye joins them because one bar is a simpler account than two bars that happen to nearly line up, and the mismatch is small enough to be spent on that account.

The two squares are a square and an L. The dark shape has six sides. Its missing corner is exactly the part the coral square covers, which is what makes the completion so easy — there is no evidence against it anywhere in the figure, and there never is in this kind of case.

The cube is three rhombi. Flat, edge to edge, no two lines at right angles and no depth anywhere in the file. A cube is a simpler account of that hexagon than three separate shapes that happen to meet at a point, so a cube is what arrives, in a drawing that is as two-dimensional as the page it sits on.

What it costs you

The reading arrives before the label does. Whatever an arrangement of shapes resolves into is what the interface said first, and a caption underneath is a correction, not an introduction.

Alignment makes claims you did not write. Elements that line up get read as one object or one sequence, which is useful when it is true and expensive when it is a coincidence of the grid.

Simplification happens whether or not you left room for it. A busy diagram does not stay busy in the reader's head. It collapses into whatever regular shape is nearest, and the details that survive are the ones that shape had room for.

A logo made of separate parts is one part by the time anyone remembers it. This is the only law here that is reliably on your side, and it is on your side exactly once — the first simple reading is the one that sticks.

The uncomfortable case is a layout that resolves into something tidier than the content underneath it. Nobody reports that as a bug. It just quietly means the reader now believes a structure that was never in the data.

← UX laws

Where the list comes from. The collection I'm working through is Laws of UX by Jon Yablonski — the canonical version, and worth reading in full. His content is licensed CC BY-NC-ND 4.0, which permits sharing but not rewriting, so none of the wording here is his. The laws themselves belong to the psychologists who found them.

Who wrote it. Claude did. The prose, the demonstrations and the statistics, all of it, working from my direction and my edits. I picked the laws, decided what each page had to prove, and threw out the versions that did not. I would rather say that than let you assume otherwise.