Cognitive Mechanisms of Learning

3Learning to Solve a Problem

Solving the same problem multiple times results in a reduction in both the time taken by the learner to provide an answer and the number of wrong answers. This result may also be referred to as “increased fluency”.

Following Newell and Rosenbloom (1981), we consider that this progress is the result of both qualitative and quantitative changes. Several theories of knowledge acquisition have been put forward, and certain aspects have been modeled. The two dominant theories are SOAR (Newell, 1994) and ACT-R (Anderson and Lebiere, 1998).

Despite the fact that these different theories relate to the same level of explanation, considering the human cognitive system as an information processing system (IPS), each proposes a different paradigm. For computer formalization specialists, these paradigms are not incompatible (see, for example, Anderson and Lebiere, 2003). For our target readership, however, we consider that the presentation of a single “school” is sufficient to introduce the research work described in this section.

In our view, ACT-R does not yet constitute a single unifying theory, but it provides a functional theoretical framework that enables it to evolve in response to research findings. In this chapter, we shall present three recent research projects issuing from the “ACT-R school”. All three cases focus on learning by solving problems of the same type, with the aim of providing an increasingly fine analysis of the main cognitive mechanisms ...