CHI-WEB Archives

ACM SIGCHI WWW Human Factors (Open Discussion)


Options: Use Forum View

Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
"Donald A. Norman" <[log in to unmask]>
Reply To:
Mon, 19 Aug 2002 10:57:42 -0500
text/plain (91 lines)
> "An affordance refers to the actual properties of an object.*"

Sigh.  This is why I have decided to give up on the meaning of the term
(despite all the many people who have written to say, essentially, "it
is wrong to give up. Fight!")

An affordance is a RELATIONSHIP.   It is not a property of anything --
it holds between an agent and an object or the environment.

Let me illustrate by showing how the same items can have or can fail to
have the affordances of support, depending upon the agent that is to be

Take the affordance of support. I am a human being. For me, a tabletop,
a chair, or a stair all afford support. But a match does not. Nor does a
piece of paper folded into box.   The affordance of support applies to
the relationship between me and the tabletop, chair, or stair.

If I were an ant, a match or the paper box would afford support. The
affordance of support applies to the relationship between an ant and a
match or paper folded into a box, even if it doesn't for me.

If I were a giant (or even an elephant), none of those items would
support. So those same objects would not have any affordances of

Affordance is a relationship.  You cannot talk about an object having an
affordance unless you also specify for whom.

Notice that affordances to not have to be known by the agent. The
relationship is a result of the physical/biological properties of the
agent and the object, whether or not either item is aware of the fact.
I can eat the paper or match, so they have the affordance of eatability.
I can throw them. A paperclip has the affordance of, when unfolded,
being stuck into tiny holes, perhaps to eject my diskette. An affordance
I would not have known about unless told.

Finally, the whole concept of affordance is bizarre when applied to the
design of graphical interfaces. We say that a hyperlink affords
clicking. Well, yes, but the entire screen is clickable.  Or if you
like, none of the screen is clickable -- only the mouse button is
clickable. I can click the mouse button anytime I want, no matter where
the cursor is located. The affordance is always there.

When the cursor is over some graphically depicted objects, then the
clicking of the mouse button has a meaning.  But this has nothing to do
with affordance. A hyperlink is no more or no less clickable than an
empty part of the screen. So the utility of affordance disappears in the
GUI world unless the meaning of the term is modified.

Note that when I discussed affordances in the design of Everyday Things,
it was in the context of physical products. That domain is much closer
to the concepts that Gibson addressed, so there, the concept works.

Bill Gaver tried hard to work out these "technological affordances." I
don't think he succeeded, not because he wasn't capable, but because it
is not a domain handled well by the concept of affordance. I know this
area well, and I despair of getting it right. (I had numerous long
discussions with Gibson, fights and debates, and long semi-drunk nights
in bars with him.  I believe I understand his work. Similarly with Bill
Gaver (who got his PhD under my direction).

So I give up, and it isn't for lack of trying.

Someone quoted Cooper's attempt to explain my definition of affordance.
Bah. That was Cooper speaking, not me.

It seems that it is very very very difficult for people to understand
properties that are relationships. Properties should go with objects,
seems to be the belief. A property that goes with a relationship is just
too complex for most.

There is a need for a term in the design community.  The technical,
formal definition of "affordance" is wrong, but nonetheless, affordance
has been drafted. So be it.

Don Norman

        Tip of the Day: Email mailto:[log in to unmask]
               with any comments, questions or problems
     CHI-WEB: POSTINGS: mailto:[log in to unmask]
              MODERATORS: mailto:[log in to unmask]