Cogprints

Skill acquisition can be regarded as program synthesis: An integrative approach to learning by doing and learning by analogy

Schmid, Ute and Wysotzki, Fritz (1998) Skill acquisition can be regarded as program synthesis: An integrative approach to learning by doing and learning by analogy. [Book Chapter]

Full text available as:

[img]Postscript
330Kb

Abstract

In this paper we propose an approach to skill acquisition which is based on a technique for inductive program synthesis developed in the domain of automatic programming. This approach enables us to model skill acquisition as generalization on three levels: In a first step, learning by doing is performed by generalizing over problem states which were explored when solving a given problem. This process is similar to compilation or chunking of production rules. But in contrast to these approaches, we represent procedural knowledge as conditional programs. In a second step, descriptive generalization of the initial conditional program is performed. A recursive program scheme is constructed which generalizes over recursive enumerable problem spaces. In a third step, learning by analogy is performed by abstracting from the concrete semantic of the operation symbols contained in a recursive program scheme. The abstract scheme represents the class of structurally identical problems. By describing, how problem schemes can be constructed as generalization over knowledge gained during solving concrete problems, our approach gives an unifying framework for describing learning by doing and learning by analogy. Additionally, we consider the acquisition of some types of motor and process control behavior as a special variant of the acquisition of problem solving skills, and demonstrate, how acquisition of behavioral skills can be integrated in our framework.

Item Type:Book Chapter
Keywords:learning by doing, skill acquisition, analogicalproblem solving, program synthesis, computer simulation
Subjects:Psychology > Cognitive Psychology
Computer Science > Artificial Intelligence
Computer Science > Artificial Intelligence
ID Code:484
Deposited By:Schmid, Ute
Deposited On:01 Jul 1998
Last Modified:11 Mar 2011 08:53

Metadata

Repository Staff Only: item control page