Affordance
Description
An affordance is the set of action-possibilities a thing offers an agent. The load-bearing structural claim is that it is relational and perception-dependent: what an object affords is fixed neither by the object alone nor by the agent alone, but by the fit between the object’s properties and the agent’s capabilities — and it steers behavior only to the extent that the agent perceives it. James Gibson coined the term in ecological psychology: the affordances of the environment are what it offers the animal, for good or ill. A rigid horizontal surface affords support to an animal of the right size; a graspable object affords grasping to an animal with the right kind of hand. Donald Norman then carried the term into design and shifted its center of gravity to perceived affordance — for a door or a control, what matters is not only what is physically possible but which possibilities the user reads as available. When the two diverge you get the everyday failure: a false affordance advertises an action that isn’t there; a hidden affordance is a real action nobody can find. The cross-domain projection is wide because the relation is substrate-independent. A branch affords perching to a bird; a handle affords pulling to a hand; a button affords clicking to a cursor; a REST verb affords an operation to a client. In every case the design question is the same: is the intended action-possibility perceptible, and is it the most perceptible one? The concept’s discipline is to stop asking “what is this thing?” and ask “what does it afford, and to whom, and can they tell?”Aliases
Norman’s later work (the 2013 revision of The Design of Everyday Things) split the concept in two to correct years of misuse: the affordance is the action-possibility itself, while the signifier is the perceptible cue that advertises it. Much of what designers colloquially call “affordances” (a shadow that says “clickable,” a bevel that says “press me”) are strictly signifiers. The catalog keeps “affordance” as the primitive and treats “perceived affordance” as the design-relevant reading; the signifier is the channel through which the affordance is made perceptible.Triggers
User-initiated: User asks how someone will know they can do something, or reports that people can’t find or figure out a capability that exists. Vocabulary cues: “affords,” “looks clickable,” “discoverability,” “intuitive,” “self-explanatory,” “how do users know they can X,” “the design should suggest.” Agent-initiated: Engine notices a capability that exists but isn’t perceivable, or a cue that advertises an action the system doesn’t actually support. Candidate inference: “this is an affordance/signifier gap — the action is possible but unsignified, or a signifier promises an action that isn’t there.” Situation-shape signals: Usability or onboarding discussions; “nobody knew that feature existed” (a hidden affordance / missing signifier); “users keep trying to click that, but it’s not a button” (a false affordance); interface redesigns that changed appearance without changing capability (flat-design discoverability regressions); API-surface design where a method is callable but undiscoverable.Exclusions
- The design goal is to foreclose the wrong action, not invite the right one — when the point is to make a mistake structurally impossible (a keyed connector, an interlock), that is poka-yoke, the opposite-polarity move. Affordance is about making the intended action perceptible; using it for constraint-design inverts its polarity.
- No agent whose capabilities the possibility is relative to — affordance is a relation, not a property of the object alone. A ledge affords perching only relative to a bird’s body and abilities. Asking what something affords with no agent fixed is under-specified: the answer is always “affords what, to whom?”
- A surface feature that signals identity or style, not a possible action — a logo, a brand color, a decorative flourish signals what a thing is or how it feels, not what you can do with it. Reading these as affordances over-extends the concept; affordance is specifically about action-possibilities.
- Bare physical possibility divorced from perception — the abstraction floor. Everything affords infinitely many physical actions (a wall affords being leaned on, painted, stared at). The concept does structural work only when a specific action-possibility is made perceptible and thereby guides behavior; stripped of the perception-guides-action claim it dissolves into “any action physically possible.”
Structure
Relationships
- poka-yoke — the matched opposite-polarity move. Affordance invites the right action by making it perceptible; poka-yoke forecloses the wrong action by making it impossible. Norman pairs them (affordances and constraints); good design uses both at once.
- surface — affordances are the action-possibility reading of what a surface exposes. Surface names the exposed boundary; affordance names what that boundary invites the agent to do.
- desire-path — affordance is the offered possibility; a desire-path is the worn trace of an afforded-but-unprescribed action actually being taken. Reading the pair together: where the sanctioned affordance and the taken action diverge, a desire-path forms.
Examples
James J. Gibson, *The Ecological Approach to Visual Perception* (Houghton Mifflin, 1979) — "The Theory of Affordances" (Ch. 8). · psychology
James J. Gibson, *The Ecological Approach to Visual Perception* (Houghton Mifflin, 1979) — "The Theory of Affordances" (Ch. 8). · psychology
Gibson gave the primitive its canonical formulation: “The affordances of the environment are what it offers the animal, what it provides or furnishes, either for good or ill.” A surface that is horizontal, flat, extended, and rigid affords support — standing, walking, running — but only relative to an animal of the right size and weight; the same surface affords neither to a creature far heavier nor far smaller. The affordance is a property of the animal–environment relation, not of the environment alone, and Gibson insists it is perceived directly: the animal sees what a surface is for, rather than measuring a neutral geometry and inferring uses from it. (Gibson introduced the coinage earlier, in The Senses Considered as Perceptual Systems, 1966; the 1979 book is the mature statement.)Inference: Gibson supplies the relational, agent-relative core of the concept — an affordance is fixed by the object–agent pair, not by the object. The transferable discipline for any designer is to stop asking “what is this thing?” and ask “what does it afford, and to whom?”, which reframes usability as a property of the fit between artifact and the user’s capabilities rather than of the artifact in isolation.
Donald A. Norman, *The Design of Everyday Things* (Basic Books; originally *The Psychology of Everyday Things*, 1988; revised and expanded edition 2013) — perceived affordances and signifiers. · architecture-and-design
Donald A. Norman, *The Design of Everyday Things* (Basic Books; originally *The Psychology of Everyday Things*, 1988; revised and expanded edition 2013) — perceived affordances and signifiers. · architecture-and-design
Norman carried Gibson’s term into design and moved its center of gravity to perceived affordance: for a door, a control, or an appliance, what matters is not only which actions are physically possible but which ones the user perceives to be possible. A flat plate on a door says “push”; a graspable bar says “pull”; mount the bar on a push-door and the mismatch produces the everyday failure now called a “Norman door.” In the 2013 revision he split the idea in two to correct years of misuse: the affordance is the action-possibility itself, while the signifier is the perceptible cue that advertises it — a door can afford pushing while signifying nothing, leaving the user to guess.Inference: Norman relocates blame from the user to the design. A wrong action is evidence that the affordance was real but unsignified, or that a false signifier advertised an action that wasn’t there. The design task is to make the intended action’s affordance perceptible and to suppress cues for actions you don’t want — the invite-the-right-action move that pairs with poka-yoke’s forbid-the-wrong-action move. This is why a grayed-out button teaches correctly and a warning label does not.
Donald A. Norman, "Affordance, Conventions, and Design," *Interactions* 6(3): 38–43 (ACM, May 1999). · computer-science
Donald A. Norman, "Affordance, Conventions, and Design," *Interactions* 6(3): 38–43 (ACM, May 1999). · computer-science
Addressing screen interfaces specifically, Norman argued that on a graphical display the designer controls almost only perceived affordances, not real ones. The physical screen affords touching and the cursor affords clicking anywhere, so what makes a particular region “clickable” is a learned or signalled convention — a highlight, a depth cue, a cursor change — not a physical possibility unique to that region. This is why stripping visual cues can hurt usability even when nothing underneath has changed: the flat-design shift that removed shadows and bevels left the underlying affordance intact but removed the perceived affordance, the signifier that told users where they could act.Inference: In software the affordance/signifier gap is at its widest, because nearly every pixel affords the same physical action and meaning is carried entirely by convention and cue. The design lever is therefore the signifier, not the affordance: discoverability failures in a UI or an API (“users can’t find the feature,” “nobody knew that method existed”) are usually missing-signifier problems, not missing-capability ones — the action was possible all along; it was simply not advertised.