Cognitive & perceptual design principles, explained

Two screens can offer the identical features and the identical data, and one will feel obvious while the other feels like homework. That gap isn't taste. It's perception, and a handful of laws, most of them tested in laboratories decades before the web existed, predict which side of the gap your design lands on.

The idea in depth

These five principles get filed under "UX," which makes leaders think they belong to designers. They don't. They govern how any human reads any artefact you put in front of them, a dashboard, a pricing page, a board deck. Knowing them is how you tell a designer "this is too busy" with a reason instead of a vibe.

Gestalt: people read arrangement as meaning

In 1923, the psychologist Max Wertheimer published Untersuchungen zur Lehre von der Gestalt ("Laws of Organization in Perceptual Forms"), arguing that we don't perceive a scene as loose dots and lines, we organise it into wholes by a few rules: proximity (things near each other look related), similarity (things that look alike look related), closure (we complete shapes the eye implies), continuity, and figure-ground. A 2012 review in Psychological Bulletin by Wagemans and colleagues, marking a century of the work, confirms grouping is a property of the arrangement of parts, not the parts themselves.

The move: stop styling elements and start grouping them. If two fields belong together, tighten the space between them and widen the space to everything else, proximity will do the explaining no label can. The most overlooked Gestalt lever is whitespace. It isn't empty; it's the boundary that tells the eye where one idea ends.

Where it breaks down: Gestalt describes that we group, not always which grouping wins when cues conflict. Put proximity and similarity in a tug-of-war, items close together but coloured differently, and the rules give you no referee. Test the ambiguous cases; don't assume.

Hick's law and Jakob's law: the cost of choice, and the gift of the familiar

In 1952, W. E. Hick showed in the Quarterly Journal of Experimental Psychology ("On the rate of gain of information") that the time to choose rises logarithmically with the number of options; Ray Hyman replicated and extended it the following year in the Journal of Experimental Psychology (1953). The shorthand, more choices, slower decisions, is real, but the honest version is narrower than the slogan: it was measured on simple, well-practised stimulus-response tasks, not on a user weighing a complex purchase. Treat it as a strong prompt to reduce options, not as a formula you can plug a checkout into.

What to do with it: on any high-stakes screen, count the choices and cut or defer the ones that don't serve the primary task. Progressive disclosure, show the common path, tuck the rest behind "more options", applies Hick without amputating capability.

Jakob's law, named for usability researcher Jakob Nielsen of the Nielsen Norman Group, comes at the same problem from the other side: "Users spend most of their time on other sites. This means that users prefer your site to work the same way as all the other sites they already know." People arrive carrying mental models built elsewhere, a cart in the top right, a logo that links home, an underlined link. Match those, and you spend their attention on the task instead of on relearning your interface.

So before you invent an interaction, ask whether a convention already exists. Innovate on the thing that's actually your differentiator; be boring everywhere else. The limitation: conventions ossify, and "everyone does it" once protected the hamburger menu and the carousel, both shown to hide content people never find. Familiar is a default, not an alibi.

flowchart TD
    A(["A screen, a deck, an email"]) --> B(["Group it: proximity, similarity, whitespace
(Gestalt)"])
    B --> C(["Cut the choices to the task
(Hick's law)"])
    C --> D(["Match what people already know
(Jakob's law)"])
    D --> E(["Make it fast and make it look right
(Doherty + aesthetic-usability)"])
    E --> F(["Feels effortless"])
					

Doherty and the aesthetic-usability effect: speed and beauty are not luxuries

In 1982, Walter J. Doherty and Ahrvind J. Thadani of IBM published "The Economic Value of Rapid Response Time" in the IBM Systems Journal, challenging the accepted two-second standard for computer response. Their finding, the Doherty threshold, was that productivity climbs sharply when a system responds in under roughly 400 milliseconds: people stay in a tight, fast loop with the machine instead of disengaging while it thinks.

So treat perceived latency as a product feature with a budget. Where you can't be fast, look fast, optimistic UI that shows the result before the server confirms, skeleton screens, progress that moves. The caveat: the original work measured mainframe terminals, so 400ms is a useful target, not a precise modern constant, and faster isn't infinitely better.

Finally, the most uncomfortable principle for rational leaders: the aesthetic-usability effect. In 1995, Masaaki Kurosu and Kaori Kashimura at Hitachi tested 26 ATM-interface layouts with 252 participants and found that people's ratings of aesthetic appeal correlated more strongly with perceived ease of use than the interfaces' actual ease of use did. Noam Tractinsky replicated it in 1997 across a different culture (Israel) and found the effect, if anything, stronger. The Nielsen Norman Group's summary puts it plainly: people perceive attractive products as more usable, and are more forgiving of small flaws in them.

So fund visual quality as usability spend, not decoration, it buys patience while you fix the real problems. There's a research trap hiding in it, though: a beautiful prototype can mask genuine usability failures, so people say it's easy while quietly failing the task. The limitation NN/g names: attractiveness earns forgiveness for minor issues, never for the ones that block the job.

A worked example

Picture a mid-market SaaS team whose self-serve sign-up converts at, say, 18% of trials to paid, illustrative figures, to show the mechanics, not a benchmark. Leadership's instinct is a discount. A perception audit suggests a cheaper fix. The trial-setup screen asks for eleven fields in one ungrouped column, with a "Start trial" button that triggers a four-second, silent backend provision. Walk the five laws across it:

• Gestalt: the eleven fields have equal spacing, so the eye can't tell that four are "about you," four are "about your company," and three are optional. Re-group into three labelled clusters with real whitespace between them; the form looks shorter without losing a field.
• Hick: three of the eleven aren't needed to start a trial. Defer them to first login. Fewer choices on the critical screen, faster commitment.
• Jakob: the password rules appear after a failed submit, which no familiar product does. Move them inline, where every other sign-up shows them.
• Doherty: the silent four-second wait reads as "broken." Add an optimistic confirmation and a skeleton dashboard so the perceived wait collapses toward zero.
• Aesthetic-usability: the screen is functional but visually cluttered. A clean, confident layout makes the whole flow feel trustworthy, and buys forgiveness for the one rough edge you haven't fixed yet.

None of these touches price, the feature set, or the database. They change how the same product is perceived, and perception is what the trial user is actually responding to.

flowchart LR
    P(["Eleven ungrouped fields
silent 4s wait"]) --> Q(["Group into 3 clusters
(Gestalt)"])
    Q --> R(["Defer 3 fields to later
(Hick)"])
    R --> S(["Inline rules, familiar pattern
(Jakob)"])
    S --> T(["Optimistic UI, skeleton screen
(Doherty)"])
    T --> U(["Cleaner visual layout
(aesthetic-usability)"])
    U --> V(["Same product, feels easier"])
					

Frequently asked questions

Are these "laws" actually science, or just design folklore?

It's a mix, and it pays to know which is which. Gestalt grouping, Hick's law and Doherty's threshold come from published, replicated experimental work. The aesthetic-usability effect has at least two independent studies (Kurosu & Kashimura 1995; Tractinsky 1997). Jakob's law is a practitioner principle from observed behaviour, not a controlled experiment, sound, but a different weight of evidence. Cite them accordingly.

Doesn't Hick's law just mean "fewer features"?

No, and conflating the two is how good products get gutted. Hick is about choices presented at once on a given screen, not total capability. Progressive disclosure lets you keep every feature while showing only the relevant few per moment. Reduce the choice, not the product.

If familiar always wins (Jakob), how does anything new ever succeed?

Familiarity is the default you depart from deliberately, where the novelty is the value. Spend your "newness budget" on the interaction that is your differentiator and keep everything around it conventional, so users have spare attention for the new part. Novelty everywhere is just friction.

We have no designer. Can a leader use these directly?

Yes, that's the point of learning them. They give you precise language for a review: "these two blocks are too close, they read as one group," or "this is the third novel pattern on one screen." You're not redesigning; you're naming the perception problem so whoever builds it can fix the right thing.

Isn't "make it pretty" a distraction from real usability?

The aesthetic-usability effect says appearance changes perceived usability and buys patience for minor flaws, genuinely valuable. The trap is letting it mask major ones, in testing or in production. Fund the visuals, and keep usability testing honest by watching what people do, not only what they say.

Related in the Toolkit

• Information architecture, Gestalt and Hick decide whether a single screen reads cleanly; IA decides whether the whole structure does.
• Human-centred design & empathy, perception laws are how empathy gets expressed in pixels people actually use.
• Design systems & style guides, the place to encode these principles once so every team ships them by default.
• Design thinking & the double diamond, the wider process these principles plug into at the "deliver" end.
• Customer needs identification & latent needs, knowing what to build before you decide how to make it perceptible.
• Ideation & co-creation techniques (design studios, affinity mapping, card sorting, crazy-8s), card sorting in particular tests whether your groupings match users' mental models.
• Design sprints, a fast format for prototyping and testing perception decisions in days, not quarters.
• Sales process & pipeline management, Hick and Doherty apply just as hard to a checkout or a quote flow as to a product UI.

Where to go next

• The Aesthetic-Usability Effect (Nielsen Norman Group), the clearest short write-up of the original ATM studies and what they do and don't license.
• Laws of UX (Jon Yablonski), a free, well-sourced reference site for all five of these and more; his O'Reilly book of the same name (2nd ed., 2024) is the deeper version.
• The Aesthetic Usability Effect and Prioritizing Appearance vs. Functionality (NN/g, video), a short talk on why beauty buys forgiveness but never replaces a working UI.
• A Century of Gestalt Psychology in Visual Perception (Wagemans et al., Psychological Bulletin, 2012), the rigorous review behind the Gestalt claims, if you want the primary evidence.