Information Technology Reference
In-Depth Information
but they represent useful and interesting differences about users that help with
interface design.
Declarative learning is learning facts (declarations). The ''power button is on
the keyboard'' and ''the computer manager's office is in 004A'' are two examples.
Declarative learning can be separated into two subtypes of recognition and recall.
Recognition memories are easier to build than recall memories. That's why
multiple choice tests seem easier—you just have to recognize the answer. This
corresponds to the first stage of declarative learning.
Procedural learning is learning how to do procedures. Using an interface, and
playing a computer game are examples of this. Procedural memories are probably
more complex than declarative memories in that they generally support the skill
being performed in a wide variety of environments and slightly different situations,
which thus represent more knowledge. These memories have to come after the
declarative representations are available to create them.
These two types of learning have different regularities associated with them.
Declarative learning can, by definition, be described and reported. Procedural
memories cannot be directly reported (Ericsson and Simon 1993 ). You cannot
directly describe the knowledge you use to ride a bike. You can, however, accu-
rately report the declarative knowledge that you use to generate your procedures
(like keep your weight balanced) and what is in your working memory as you are
doing the task (there is a parked car ahead). You can also watch yourself do a task
and attempt to describe much of what you were paying attention to as you did it.
What you think you were paying attention to when doing the task will, however,
be dependent on your mental model of the task and how demanding the task is.
This approach is called introspection.
There are fundamental problems with introspecting like this. While introspec-
tion can lead to useful and helpful insights, it does not lead to complete and valid
theories of human thinking (Ericsson and Simon 1993 ). Just as you can't program
a computer to write out the instructions as it does them (the instruction to write out
replaces the instruction it copies), you can't think about thinking very accurately
while thinking. Mainstream psychology has rejected introspection as a true rep-
resentation of how people think, but you may find it useful for inspiration for ideas
that can be later validated by other approaches. In our experience it is sometimes
useful inspiration, but it is sometimes just way off because of our biases about how
we would like to think we think.
As an example of this, users think that they cannot learn new key bindings
between keys and commands (e.g.,\Ctrl-s[to search vs.\Ctrl-f[to search). When
key bindings in an editor were changed on users on the second day of a study about
learning to use a text editor, for the first hour or two the users felt much, much slower.
They were slower than they were at the end of the first day, but faster than when they
started. They regained their skill level by the end of the second day (Singley and
Anderson 1989 ). This study illustrates three important things about users. The first is
that they are always learning. The second is that introspection often leads to incorrect
conclusions. They rather disliked the new interface, but they adapted more quickly
than they thought they did. The third is that the users were able to transfer much of
Search WWH ::




Custom Search