Cognition: Mental Representations, Problem Solving, and Decision Making - Foundations for Designing User-Centered Systems - page 183

An old tale from Dick Pew (and documented in Norman's 1988 POET book) is
that in a nuclear power plant the controls to raise and lower the rods were not clear,
so the operators put beer taps on them to differentiate between them—with Loe-
wenbrau on one (low) and Heineken on the other (high). This may be an urban
myth, but it provides another clear illustration of what we mean by stimulus-
response compatibility.
Probably the most powerful version of compatibility is the spatial compatibility
that exists in the elevator example (and in Payne's example in Fig. 1.1 ). This effect
can influence computer interfaces in many other places that use spatial and met-
aphorical content (e.g., that use relationships like up, down, left, right, higher,
lower, more, less). Designs that violate this compatibility are likely to lead to a
greater number of errors and take longer to use.
Two major operating systems have both included actions that violate this
principle and rightly have been criticized for doing so. The Macintosh operating
system originally had users drag the floppy disc icon onto the trash can icon in
order to eject the disc: users just wanted their disc back, they did not want to throw
it into the trash can. Also, both Windows and the Macintosh operating systems
have users press the start button to shut down the operating system.
Finally, Fig. 6.14 gives an example of poor semantic mapping. OK normally
means something is good. Cancel normally means stop doing something. In this
example, OK means to cancel the email and cancel means do not cancel the email.
A better design would be to have buttons that said 'cancel email' and 'continue
with email'.
6.4.4 Known Influences on Decision Making
There are often differences between what people decide to do, and what an outside
observer thinks that they rationally should decide to do. These differences are often
predictable based on known biases and difficulties in reasoning. There are
explanations for some of these effects showing how different assumptions by the
user can lead to these effects. In other cases, these effects appear to be biases
between otherwise equivalent choices, and sometimes they represent powerful
tendencies that are not helpful most of the time.
Understanding people's biases—both your own and your users'—is important
when designing software. It is important, albeit difficult, to counteract these biases
when making decisions. Generally, these biases lead you to overestimate how
good you are at something, and how much control you have over situations. You
should design your system in such a way that it helps your users reason more
accurately by highlighting information that they would otherwise naturally over-
look or discount. The inherent biases described below are unique and have dif-
ferent causes, although in real-world settings several may apply at once, and even
combine to make reasoning more complicated.