Page 114 -

3.2 What is cognition? 83

years. He suggests that it is profitable to view this process as involving two memory
processes: recall-directed, followed by recognition-based scanning. The first refers
to using memorized information about the required file to get as close to it as possi-
ble. The more exact this is, the more success the user will have in tracking down the
desired file. The second happens when recall has failed to produce what a user
wants and so requires reading through directories of files.
To illustrate the difference between these two processes, consider the following
scenario: a user is trying to access a couple of websites visited the day before that
compared the selling price of cars offered by different dealers. The user is able to re-
call the name of one website: "alwaysthecheapest.com". She types this in and the
website appears. This is an example of successful recall-directed memory. However,
the user is unable to remember the name of the second one. She vaguely remembers
it was something like 'autobargains.com'; but typing this in proves unsuccessful. In-
stead, she switches to scanning her bookmarks/favorites, going to the list of most re-
cent ones saved. She notices two or three URLs that could be the one desired, and on
the second attempt she finds the website she is looking for. In this situation, the user
initially tries recall-directed memory and when this fails, adopts the second strategy
of recognition-based scanning-which takes longer but eventually results in success.
Lansdale proposes that file management systems should be designed to opti-
mize both kinds of memory processes. In particular, systems should be devel-
oped that let users use whatever memory they have to limit the area being
searched and then represent the information in this area of the interface so as to
maximally assist them in finding what they need. Based on this theory, he has
developed a prototype system called MEMOIRS that aims at improving users'
recall of information they had encoded so as to make it easier to recall later
(Lansdale and Edmunds, 1992). The system was designed to be flexible, provid-
ing the user with a range of ways of encoding documents mnemonically, includ-
ing time stamping (see Figure 3.6), flagging, and attribution (e.g., color, text,
icon, sound or image).
More flexible ways of helping users track down the files they want are now be-
ginning to be introduced as part of commercial applications. For example, various
search and find tools, like Apple's Sherlock, have been designed to enable the user
to type a full or partial name or phrase that the system then tries to match by listing
all the files it identifies containing the requested nametphrase. This method, how-
ever, is still quite limited, in that it allows users to encode and retrieve files using
only alphanumericals.