“Affordance” is one of those terms that has come to be used for so many different things that I’ve recently just about given up using it in order to avoid confusion.
Affordance as possible physical interactions
In the original definition put forth by Donald Norman (1988) in The Design of Everyday Things:
The term affordance refers to the perceived and actual properties of a thing… that determine just how the thing could possibly be used (p9).
Note the use of the word “possibly.” Affordance is not a strong clue about the way you can interact with things. It all the ways it is physically possible to interact with things. So, by this definition, a doorknob does afford twisting with one hand. It also affords pushing, pulling, lifting, pressing, and slapping with the hand. It also affords licking with the tongue, kicking with the foot, and lifting with the knee. You can hang your hat on it. You can just look at it. There are typically many affordances for a given object.
Some of these interactions are part of the “correct” way to interact with the door knob/handle (twist and push/pull with the hand), some are not, but that’s okay. The affordances still rule out certain other ways you could interact with something. The doorknob doesn’t afford sitting –you butt physically won’t stay on it. It doesn’t afford swallowing –it won’t fit down your throat or even in your mouth. It doesn’t afford throwing (when screwed to the door). It doesn’t afford inserting your finger.
Affordance is a perceptual property developed through personal experience. For example, we learn at a very early age concepts like two objects cannot occupy the same space at once, and we learn to recognize hard stiff materials like metal, so that’s how we know, just by looking, that we can’t insert our finger in a door knob. However, affordance by this definition is pretty much culturally neutral: everyone everywhere learns pretty much the same things about the physical properties of shapes and materials.
So by this definition, door knobs and your door handles have the same affordances. Pretty much everything I said about door knobs applies to door handles. If you were to bring in a bunch of people who have never seen a door or cupboard before, they’d be equally likely to twist and pull both the door knob and door handle. Or hang their hats on them.
The lesson of affordances to designers is that, if you want users to interact with something in certain way, then make sure you give it a physical shape and structure with the right material to afford that interaction. It seems obvious, but there are certain cases where designers have failed by perceptually hiding clues of possible interactions.
Affordances as conventions
The term “affordances” took on a new meaning when it was applied to UIs presented on computer screens. By the old definition, every point on the computer screen has the same affordance. It doesn’t matter what it looks like or what it says –it’s physically possible to click, touch, lick, head butt, etc. any place on the screen. Affordances, by this old definition, is not a very useful concept to modern UI designers.
So a new much looser and less formal definition emerged, which is basically affordances are the interactions a thing communicates. Blue underlined text, for example “affords” clicking or tapping –more so than plain black text, or empty white space for that matter.
But what does that really mean? It means users have learned an arbitrary cultural convention –blue means "click this," black does not. This is the definition of affordance that you are using for door knobs versus door handles. By this definition, if a knob in you culture means pull, and a lever in your culture mean twists, then a door knob does not “afford “ twisting.
I’ve come to the conclusion that this definition of affordance is a disservice (even though I’ve used it myself). First of all it’s unnecessary. You can simply say “that’s the convention” or “that’s what users expect to do to that” or "that's a familiar idiom," or “that’s a consistent UI” (i.e., things that look the same as what the user saw before act the same as they did before). There’s no point in getting all jargony and saying “that’s the affordance.”
Second of all, I believe the new loose definition of affordance breeds in designers a misperception that certain visual properties on the screen inherently communicate interactions, the way physical objects inherent communicate possible physical interactions. With very few exceptions related to Gestalt principles, visual properties on the screen don’t have any inherent meaning. It’s all about what the particular user is used to. If we forget that, we can design the wrong thing for the wrong users.