Cognition: Memory, Attention, and Learning - Foundations for Designing User-Centered Systems - page 142

Goals and needs

Knowledge-based
control
(model-based)

Analysis and
identification

Evaluation and choice
of goal and task

Analysis of means
and planning

Reading symbols

Rule-based
control
(know-how)

Stored rules
for activities

Recognition of cues Association cue / task

Looking for cues

(signs)
Seeing changes

Skill-based
control
(automatic)

Automated sensorimotor
movements and routines

Feature formation

Sensory information
(multiple channels)

Signals Movements

Fig. 5.8 A schematic of Rasmussen's theory of performance. (Adapted from Rasmussen 1983 )

With learning and practice, the user progresses to the associative stage, or rule-
based stage (as shown in the middle of Fig. 5.8 ). The learner can solve problems
more routinely and with less effort. The declarative knowledge has been compiled
into procedures relevant to the task domain that can be performed directly. At this
stage users may often be able to just recognize what needs to be done. For users of
complex systems, behavior becomes a conscious activity and is based on familiar
rules, either dictated or acquired, such as changing lanes when driving a car, or
formatting a paragraph in a Word document using styles.
The final stage, called skills or tuning, tunes the knowledge that is applied. At
this point the user still gets faster at a task, but the improvements are much smaller,
as they are smaller adjustments. New declarative information is rarely learned, but
small adjustments to rule priorities happen. At this point users consider them-
selves, if not experts, to be real users. Rasmussen calls this level of performance
skill-based. Anderson calls it the autonomous stage.
At the skill-based level, performance is much more automatic, as shown at the
bottom of Fig. 5.8 . Skill-based performance usually occurs when users are per-
forming routine functions in their normal operating environment. Much of their
behavior is no longer available to conscious thought, or available for verbalization
etc. Users not only perform the task faster, they may also appear to have more
attention to provide to other tasks. Examples include changing gear in a car with a
manual shift gearbox, and cutting and pasting text.
These stages of learning have been noticed in several areas of formal reasoning.
Formal reasoning involves problems with known goals (like solve for x), and
equations or rules that can transform representations (such as adding 2 to each side
of an equation). Formal reasoning is used in areas such as physics problem solving
(Larkin 1981 ; Larkin et al. 1980a , b ), geometrical proofs and solving algebraic
problems.