Incomplete outlines get completed into whole shapes
Aliases: Gestalt closure · illusory contour · subjective contour
What it is
As long as the gap left in a shape's outline is small and the direction on either side of the gap clearly points toward the same regular shape, the visual system automatically fills the gap in and reads a complete, closed figure rather than several disconnected line segments. This cue is closure.
It has a different criterion from continuity: continuity only requires the direction to keep changing smoothly, regardless of whether the path ever meets itself; closure specifically requires the completed contour to enclose a closed region. The classic demonstration is a set of notched disks arranged at particular angles, in the middle of which a triangle appears with a crisp, visible edge even though no such shape or edge physically exists — that filled-in edge is called an illusory contour (or subjective contour), and it can be "seen" in a place with no actual difference in brightness at all.
Why it happens
Completion follows a simplicity principle: when a gap could be filled in several ways, the visual system picks whichever completion makes the overall figure most regular and simple, rather than extending the contour arbitrarily. The smaller the gap and the more clearly the remaining contour points toward one specific regular shape, the more readily completion happens; once the gap gets too large, more than one "regular and simple" completion becomes possible, and completion becomes uncertain or stops happening altogether.
Illusory contours aren't purely a cognitive-level guess — a matching neural signal has been found for them: cells in visual cortex that normally respond only to real luminance edges also fire when an observer reports "seeing" an illusory contour at that location, indicating this filled-in edge is treated as a real edge at an early stage of visual processing, rather than being reasoned out later at some higher inferential stage.
Studying it
Closure and illusory contours are typically studied by contrasting figures that do and don't induce completion: varying the notch angle, orientation, or spacing of the inducing elements and measuring the proportion of observers who report "seeing" the complete figure, or their response time, to map out how completion strength changes with these parameters. Neural evidence comes from recording visual cortex cells' responses to illusory-contour edges and comparing them to their responses to real edges.
This line of research mostly stays within basic vision science; interface design applications of closure mostly borrow the conclusion — that gaps get filled in — directly, rather than re-running this paradigm to validate a specific layout.
Where it stops holding
Completion only happens reliably when the gap is relatively small compared to the whole figure and the remaining contour clearly points to one specific regular shape; a gap that's too large, or a remaining contour that can be read as several different regular shapes at once, makes completion ambiguous or stops it from happening at all.
Whether completion happens also depends on the observer being able to clearly see the inducing gap itself in the first place — that dependence on visual capability, and the conditions under which it breaks down, needs its own separate treatment.