Mental Models, Affordances & Learnability

"Intuitive" means: matches the model already in the user's head.

Core concepts (Norman, The Design of Everyday Things)

• Mental model — the user's internal story of how the system works, built from experience with similar things. Errors are usually model mismatches, not stupidity. Design either matches the existing model or visibly teaches a new one.
• Conceptual model — the model the design communicates. Goal: the designer's conceptual model, transmitted through the system image (UI, docs, marketing), reproduces accurately in the user's head.
• Affordances & signifiers — affordances are possible actions (Gibson's term, adapted by Norman); signifiers are the perceivable cues indicating them. Flat design's core risk is stripping signifiers: NN/g documented "flat design" reducing clickability cues and slowing users. Buttons should look pressable; links should look like links; draggable things should signal grip.
• Mapping — controls should relate spatially/logically to their effects (stove-burner problem). Natural mapping removes the need for labels.
• Constraints — physical/logical/cultural limits that prevent error (can't proceed until required fields valid; date pickers that block invalid dates).
• Feedback — every action produces an immediate, interpretable result.
• Gulf of execution / gulf of evaluation — the gap between intent and knowing how to act, and between system state and understanding it. Most usability findings are one gulf or the other; name which one to fix it.

Norman's seven stages of action

Goal → plan → specify → perform → perceive → interpret → compare. Walk a flow through the stages to find where users stall (can't form a plan? can't tell if it worked?).

Error psychology: slips vs. mistakes (Norman; Reason)

The most useful error taxonomy in UX. Slips: right intention, wrong execution — typically by experts running on autopilot. Types: capture errors (a frequent action hijacks a rare one — "delete" beside "archive"), description-similarity slips (acting on the lookalike control), mode errors (right action, wrong mode — caps lock, editing the wrong document), loss-of-activation ("why did I open this menu?"). Mistakes: wrong intention — the mental model, rule, or plan itself is wrong.

The distinction dictates the remedy:

• Slips are NOT fixed by warnings or training — habituated confirmations get clicked through automatically (see Attention, Scanning & Perception). Fix with constraints, undo, spacing/visually differentiating dangerous controls, eliminating modes or making them loudly visible, and forcing-function friction for the truly irreversible.
• Mistakes are model problems: fix with a clearer conceptual model, better feedback, previews of what will happen ("this will email 400 people"), and plain-language explanations. Reason's broader point (the "Swiss cheese" model): serious failures are usually system design failures, not "user error" — blame the holes, not the human. Log and categorize real user errors; each is a free usability finding.

Learnability strategies

• Leverage transfer: reuse platform conventions and widely-known patterns (Jakob's Law) so the user's existing models apply.
• Progressive disclosure: novice surface first; advanced options behind "More" (Nielsen; also core to Apple/Material guidance). Reduces initial load without capping ceiling.
• Scaffolding, then fading: coach marks and hints that disappear with demonstrated competence; never permanent training wheels.
• Consistency compounds: internal consistency is a learnability multiplier — each learned pattern pays rent on every screen.
• Skeuomorphic bridges: when introducing genuinely new interactions, borrow familiar metaphors (files, folders, cards, shopping carts) then loosen the metaphor as conventions establish (the desktop metaphor's own history, Xerox PARC → today).

Testing mental models

Ask users to predict ("what will happen if you tap this?") and to explain ("how do you think this works?") — prediction errors locate model mismatches precisely. Card sorting and tree testing reveal category models (see Information Architecture).

Sources

• Norman, D. A. (1988; rev. 2013). The Design of Everyday Things. Basic Books — affordances, signifiers, mapping, gulfs, seven stages.
• Gibson, J. J. (1979). The Ecological Approach to Visual Perception — origin of "affordance."
• Reason, J. (1990). Human Error. Cambridge UP — slips/lapses vs. mistakes; Swiss cheese model.
• Nielsen, J. — "Progressive Disclosure"; Meyer/NN/g — "Flat Design: Its Origins, Its Problems" and flat-UI clickability studies (nngroup.com).
• Carroll, J. M. (1990). The Nurnberg Funnel — minimalist instruction, learning-by-doing.
• Cooper, A., Reimann, Cronin & Noessel (2014). About Face (4th ed.), Wiley — mental models & posture.