[The Central Role of Learning in Cognition, 1981, with Pat Langley]
p.103 The mind is an adaptive system whose biological function is to enable the organism to behave
effectively and, hence, to survive in a complex, changing, and often unpredictable environment. Adaptation takes
place on several different time scales.
p.107 The most direct kind of learning is the accumulation of a knowledge base
p.108 At the heart of the General Problem Solver (GPS) is a so-called table of connections. This table associates
with each of the differences between present situation and goal situation that GPS can detect one or more actions that might
be relevant for eliminating or reducing that difference. [JLJ - only one problem here. In a competitive
environment, your opponent will seek a position where there is no obvious, direct route to a goal. Worse, he will be assuming
an adaptive position full of resilience. What now?]
p.109 When problem solving is carried out by heuristic search, evaluation functions of some kind are needed
to assess the promise of different branches of the search tree and hence to control the continuation of the search. [JLJ -
what if we have a complex adaptive system, where cause and effect are distant in time and space?]
p.111 The learning mechanism must be able to try alternate behaviors, hence must
be able to generate them in some manner... all the learning programs described in Anderson (1981) are capable of
generating alternatives, primarily in the course of their problem-solving searches.
[The Theory of Learning by Doing, 1979]
p.116 Learning takes place in a wide variety of situations and probably by a number of different processes.
p.116-117 Problem-solving tasks of any complexity allow considerable leeway in solution strategy because,
generally, considerable numbers of alternative strategies are available, all of which are adequate for reaching a solution...
This is not to say that all strategies are equivalent. Some may be far more efficient than others in terms of speed in arriving
at the solution... we might expect a person to employ a sequence of strategies as he or she gains skill in performing the
task. Initially, the solver might hit on one of the "obvious" strategies, and then gradually progress to more efficient ones
[Problem Solving and Education, 1980]
p.286 there is a small arsenal of general problem-solving procedures that have been identified
from research in psychology and artificial intelligence... these procedures need to be made evident for students,
and then they need to be practiced, and practiced again.
p.267 I believe that a strong case can still be made for teaching problem solving explicitly,
as an important component of professional education. It must be taught, of course, in the context of a rich environment
of problems - mostly but not entirely drawn from the professional field in question.
In the teaching of problem solving, major emphasis needs to be directed toward extracting,
making explicit, and practicing problem-solving heuristics - both general heuristics, like means-end analysis, and
more specific heuristics
[Perception]
p.385 Like many other people, especially scientists and painters, I have a hard time believing that all,
or even most, thinking is done verbally. If not in words and propositions, then what? What is or are the alternatives?
[The information-Processing Explanation of Gestalt Phenomena, 1987]
p.491 compare the program, at as fine a level of detail as possible, with human behavior in identical tasks.
If the programs are in fact able to behave just like intuitive, insightful, understanding human beings, then we have no choice
but to conclude that they provide successful theories of the phenomena of intuition, insight, and understanding. [JLJ - a
machine, even when simulating a human thinking, still has no idea of what it is doing. The strategic plan of the programmer
is responsible for whatever successful behavior is achieved, or not achieved, as the case may be. A machine exhibits
intelligent behavior much as a sock puppet exhibits intelligent behavior
