Cognitive Mechanisms of Learning

2Definition and Historical Overview

2.1. Definition

Let us start by defining what we mean by “learning”. In the first collective work on learning in AI, entitled Machine Learning, an Artificial Intelligence Approach¹, Herbert Simon (1983) defined a learning process as a sequence of events following which a system undergoes a modification, such that its future behavior is better suited to attaining the desired results. Simon also defines the notion of “adaptation”, stating that the result of learning is to render the system capable of carrying out a task, or class of tasks of the same type, in a more efficient manner².

We should add that the degree to which learning is successful is dependent on three prerequisite capacities: a) perceptive capacity, b) capacity for comprehension and c) reasoning capacity.

The “efficiency” present in Simon’s definition (1983) may be permanent or temporary in nature³. Nevertheless, the outcome of every learning situation is beneficial, to some extent, for the three prerequisite capacities.

In order to evaluate the results of learning, we must define one or more means of measuring a subject’s progress⁴. Two of the most common methods are 1) to measure progress over the course of the learning process, and 2) to place the subject in a situation where they must solve problems of a similar type, both during and after learning. In the latter case, the second problem situation is known as the transfer situation or post-test. The transfer may be qualified ...