Page 211 - Deep Learning

194 Adaptation

division is nevertheless a useful device for organizing a review of mechanisms
for skill acquisition.

Stage 1: Getting started
The first stage begins when the learner encounters the practice task and ends
when he completes the task correctly for the first time. in this stage, the
main theoretical problem is to understand how the leaner can start practic-
ing. How does he know what to do before he has learned what to do? There
are five distinct types of information that might be available at the outset of
practice: direct instructions; declarative knowledge about the task; strategies
for analogous tasks; demonstrations, models and solved examples; and out-
comes of unselective search. Consequently, researchers have hypothesized
five different types of learning mechanisms that operate primarily in this
stage.
A coach, teacher or tutor who already knows how to perform the tar-
get task can tell the novice how to perform the target skill. For example, a
parent might tell a child: If you want the elevator to come, then press the but-
ton. The cognitive processes that operate on such instructions are presumably
the same as for discourse comprehension generally, with a twist: to benefit,
a novice must not only understand the utterance in the linguistic sense but
also turn it into a piece of executable practical knowledge. That is, the instruc-
tion has to be transformed into one or more rules. Anderson and david M.
neves introduced the convenient term proceduralization to refer to the mech-
52
anism responsible for this step. in the elevator example, proceduralization is
straightforward: The “if” part becomes the rule condition and the “then” part
becomes the action. For other instructions, proceduralization is less straight-
forward (e.g., be cautious!). Language is notoriously elliptical, so the result-
ing rules are likely to be incomplete (which button?), but they nevertheless
constitute a kernel for the strategy-to-be-learned, to be improved by other
mechanisms.
A second hypothesis claims that people might be able to figure out what
to do if they possess sufficient declarative knowledge about the task. This
53
view is often heard among mathematics educators. For example, if a teacher
tells a student that two triangles in which all three sides are congruent are
themselves congruent, he would want the student to infer that one way to
prove two triangles congruent is to prove that all their sides are pairwise con-
gruent, thus deriving a proof procedure from a descriptive statement. A good
term for this type of process is practical reasoning. The most consequential
practical inference in history is perhaps that if the Earth is round, you can