p.7 Concretely, human thinking is to be explained in terms of precisely
specified simple mechanisms called elementary information processes.
p.8 There is every reason to suppose that simple information processes are
performed by quite different mechanisms in computer and brain
p.9 How is the "right" program discovered--the one that explains the behavior?
Proceed the same way that you would to find correct theory for any phenomena.
p.10 computers... are, in fact, extremely general devices for manipulating
symbols of any kind; and the elementary processes required to simulate human thinking could be performed by a computer that...
could do no more than simple counting.
p.12 the trace that the computer prints out while it is attempting
to solve the problem can be compared, line by line, with the tape recording of the human thinking-aloud protocol. If the stream
of words produced by the two processes is almost the same, then the computer program that produced the trace is an explanation
of the thought process of the human subject in every significant sense of the word.
p.14 If the program makes the same analysis as the humans, notices the same
features of the board, overlooks the same traps, then we will infer, and properly, that down to some level of detail, the
program provides an explanation of the human processes.
p.15 A program that solves large problems by relying substantially
on the arithmetic speed and "brute force" of the computer in performing systematic routine calculations is
certainly not a simulation of the program that humans use in solving similar problems.
p.16 The programs that simulate human thinking tend to rely on less systematic,
more selective search for paths to the solution. Their selectivity is based on relatively unsystematic rules of thumb, which
seldom guarantee a solution to the problem, but which frequently yield solutions with relatively little processing... We call
such procedures heuristics.
p.18 the attraction of chess for these researches
lies in the fact that it is a game of sufficient complexity and irregularity that a heuristic rather
than an algorithmic approach is almost certainly required for strong play.
p.36 There are evidences that the programs of the subjects change--that
they learn--in the course of problem solving.
p.38 Perhaps the most striking characteristic of this program
is that it selects the paths it explores by first determining the functions that have to be performed, and then finding
course of action relevant to those functions.
