The mind gravitates toward the simplest, most ordered interpretation of any visual. Given a choice between complexity and simplicity, perception always resolves toward the cleanest possible form.
Prägnanz is a German word meaning “conciseness” or “precision” — the quality of being expressed in the most essential, clear form possible. The Law of Prägnanz, one of the core Gestalt principles formulated by German psychologists in the early 20th century, states that when the visual system encounters ambiguous or complex stimuli, it automatically resolves them into the simplest, most stable, most regular form available.
When you see a circle partially hidden behind a rectangle, your brain perceives a complete circle — not an arc. When you see a complex diagram, your eye finds the dominant axis first, then the secondary structure, then the details. The perceptual system is constantly performing this simplification, before any conscious thought occurs. It is not a choice users make; it is a reflex that happens to all visual information, every time.
For designers, Prägnanz has a direct prescription: the structure of your interface should match the structure the brain wants to find. When information is presented with clear hierarchy, clean grouping, and obvious visual order, the brain's simplification reflex confirms what it sees and moves on. When information is presented without clear structure — equal weight everywhere, no dominant axis, no clear grouping — the reflex works against comprehension, producing fatigue and confusion rather than clarity.
“Organisation always tends toward the simplest possible structure given the existing conditions — the law of Prägnanz.”
— Wolfgang Köhler, Gestalt Psychology, 1929
Analytics dashboards are the canonical example of Prägnanz failure. The temptation is to show every available metric — sessions, users, bounce rate, conversion rate, revenue, average order value, pages per session, time on site — at equal visual weight in a uniform grid. Every metric is present. None can be found instantly. The brain cannot resolve the grid into a simpler form because there is no simpler form: every element is equally loud.
Applying Prägnanz means establishing a visual hierarchy before choosing what to show. The three metrics that drive decisions get the top row, rendered larger and with more visual weight. Secondary metrics are grouped below in a smaller format. Related data sits in proximity. The eye has a dominant axis to follow, and it resolves the layout in a single glance rather than scanning every cell.
The good dashboard doesn't show less data. It shows the same data with a clear perceptual hierarchy: the three metrics that matter most are largest, the revenue trend is the dominant visual element, and the secondary metrics sit quietly to the right for when they're needed. The brain can resolve this layout into a simple form — “three numbers, a trend, some context” — and move on.
Onboarding screens frequently violate Prägnanz in the name of thoroughness. Every feature is explained. Every caveat is included. Every benefit is listed. The result is a screen with no dominant element — dense text blocks at equal visual weight that the brain cannot resolve into a simpler form. Users skim, miss the key point, tap through, and arrive in the product with less understanding than if they'd read nothing.
The Prägnanz-aligned onboarding screen has one dominant element — usually a large icon or illustration — and one sentence of supporting text per feature. The visual hierarchy is: icon first, title second, one-liner third. The brain resolves it immediately: “this is what this does.”
The good onboarding screen has exactly the same information density as the bad one — the same three features, the same benefits. What changed is the perceptual structure. Each feature now has a dominant visual element (the coloured icon), a short title (the primary text), and a one-liner (the supporting text). The brain can resolve each row immediately and move on.
Language models produce text — by default, flowing prose with no visual hierarchy. A well-reasoned, accurate, detailed response delivered as a single paragraph is perceptually no different from a wall of noise, because the brain has no way to resolve it into a simpler form. Every sentence has equal visual weight. The key insight is somewhere in there, but finding it requires reading everything at the same cognitive cost.
Applying Prägnanz to AI output means structuring the response with visual hierarchy: numbered steps, bold labels, section dividers, and a distinct callout for the most important point. The information doesn't change. The perceptual cost of finding and understanding it drops dramatically.
The structured response is also shorter. When structure does the organisational work, the prose doesn't have to — transitional phrases, repetition, and connective tissue disappear because the visual structure replaces them. The information compresses, the hierarchy expands, and the total cognitive load drops even as the density increases.
The practical test for Prägnanz is to squint at your interface. When squinting, all the detail blurs and only the visual structure remains. What you see when squinting is what the brain's perceptual system sees before conscious reading begins. If everything blurs together into a uniform grey mass — same weight, same size, same contrast — the interface has no Prägnanz. If a clear hierarchy is visible even through the blur — dominant elements on top, secondary elements below, clear groupings, a single visual axis — the interface is working with perception rather than against it.
Wertheimer, M. (1923). Laws of organization in perceptual forms. Psychologische Forschung, 4, 301–350. · Köhler, W. (1929). Gestalt psychology. Liveright. · Koffka, K. (1935). Principles of Gestalt psychology. Harcourt Brace.