Conceptual models of software artifacts

According to constructivism, learning takes place by constructing cognitive structures based upon current, perhaps naive, knowledge and new experiences. In the case of software artifacts like programming languages and applications, current knowledge is not a solid base upon which to build viable new knowledge. Therefore, we conjecture that explicit conceptual models constructed by educators should be able to improve the performance of users in their interaction with software artifacts, in our case the popular word processor MS-Word. The experiments described in this paper support this idea: we found that learners who used our conceptual models were able to analyze and solve problems conceptually, while learners who used task-oriented learning materials of equivalent scope employed aimless trial and error.