The Road To Excellence: the Acquisition of Expert Performance in the Arts and Sciences, Sports, and Games by

Language: eng
Format: azw3
ISBN: 9781317779056
Publisher: Taylor and Francis
Published: 2014-03-05T05:00:00+00:00

The Problem-Solving Model

EPAM’s indexed memory, the body of knowledge stored in it, and its processes for acquiring new knowledge by learning can account for a considerable part of the expert’s superior abilities in memory retrieval and problem solving in the domain of expertise (and the absence of that superiority outside the domain). However, to complete the story, we must look at problem solving that requires more than recognition processes—situations in which more or less extensive heuristic search is also required. Here again, existing models of problem-solving processes like GPS and Soar provide most of the answers we need; but to see this we have to discuss the search process in a little more detail (Newell, 1990; Newell & Simon, 1972).

Solving a simple problem, say the Tower of Hanoi, is usually described as a search through the space of possible (“legal”) arrangements of the disks on the pegs, from the starting arrangement to the arrangement specified as the goal. Successive states in the problem space are reached by moves, each of which, in this puzzle, amounts to changing the location of a single disk. The search is almost never random, but is guided by various heuristic rules that seek to guide the selection of the proper moves. The heuristic rules may be more or less complete, more or less correct.

When we come to more complex tasks, however, even tasks like finding the concept an experimenter has in mind to distinguish one set of objects from another, the search becomes more complex, usually involving interaction between two or more distinct problem spaces: the space of “in stances” and the space of “hypotheses” (Simon & Lea, 1974). Suppose, for example, that a subject is presented with a succession of objects and asked to designate each one as belonging or not belonging to the concept by which the experimenter classifies them. The succession of objects constitutes the instance space, the hypotheses that the subject generates as possible classifiers constitute the hypothesis space. When the subject is told that he or she has made an incorrect judgment, the current hypothesis is usually rejected and a move made to another point in the hypothesis space. Again, this move may be guided by heuristics that are based on information collected from the previous choices and reinforcements. It has been shown that the behavior of subjects in concept attainment experiments can be explained in terms of search in the dual instance and hypothesis spaces.

Finding a scientific law that describes data obtained in a similarly cumulative fashion has been modeled in the same way by computer programs like BACON and others, using a dual search in the space of possible laws and the space of possible data observations. But why limit the process to two spaces? A scientist may search in a space of instruments, a space of experiments, a space of possible descriptive laws, a space of explanatory mechanisms, a space of problem representations, a space of research problems, and perhaps others (Langley et al., 1987).

Krebs, for example,

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.