Affordance

An affordance is a relationship between an object and a person: what that person can do with it. A chair affords sitting for an adult, a slider affords dragging, a text box affords typing. The term came from perceptual psychology and was brought into design by Don Norman, where it is now one of the most used, and most misused, words in the field.

Where the term comes from

Norman credits the perceptual psychologist J. J. Gibson, who used "affordance" to describe the actionable properties between the world and an actor (Norman, 1999). For Gibson it was a relationship, not a property of the object alone: a high shelf affords reaching for one person and not for another.

Norman's 1988 book The Psychology of Everyday Things introduced the word to designers. In his 1999 essay he separated real affordances (what is actually possible) from perceived affordances (what people believe is possible), and wrote that on screen-based interfaces designers mostly control the perceived ones. All screens within reach afford touching, but only some respond to it. In 2008 he went further and argued that what designers really need to provide are signifiers, the perceivable cues that tell people where and how to act (Norman, 2008).

Why it matters

People act on what they think is possible. If a control can be used but does not look usable, it is effectively missing. If something looks usable but is not, people waste clicks and lose trust. Good design lines up what the system can actually do with what users perceive it can do.

How to apply it

Do:

• Make every interactive element look interactive, and keep static content visually distinct.
• Match the control to the action: a toggle for on/off, a slider for a range, a text field for free input.
• Test whether new users can tell what they can do on a screen without instructions.

Don't:

• Style plain text like a link or a button. Nielsen Norman Group warns that flat designs that strip out 3D cues remove one of the strongest clickability signals.
• Hide key actions behind hover states on touch devices, where hover does not exist.
• Say "I added an affordance" when you mean you added a visual hint. That hint is a signifier.

Common mistakes

• Using "affordance" for any visual cue. Norman himself complained that the term was widely used and misused. The cue is a signifier; the affordance is the possible action.
• Assuming it lives in the object. What a design affords depends on the user: their skills, device, language and abilities.
• Clickable everything. On a screen you can click anywhere. The question is whether people perceive that clicking a given spot is meaningful.

In AI products

A chat box affords almost anything: you can type any request. That is both its strength and its problem, because the interface gives no hint which of those requests the model handles well. A blank prompt field has a huge real affordance and a weak perceived one. Products close the gap with signifiers such as prompt suggestions, example tasks in the empty state, attachment and voice buttons that reveal other input types, and labels that say what the assistant can and cannot do.