Learning in Man-Machine Systems: the Measurement of Behavioural and Cognitive Complexity
Abstract:
A learning experiment was carried out to investigated the development of mental structures. Six subjects were carefully instructed to operate a commercial database management system (DBMS). The long-term knowledge about general DBMS- and computer experience was measured with a questionnaire once at the beginning of the investigation. On three weeks in a row all subjects had to solve the same task twice repeated in an individual session, overall there are six solutions of the same task. At the beginning of each of the three individual sessions the short-term knowledge about the task and the tool was measured with a short questionnaire. For each task solving process all keystrokes were recorded with a time stamp in a logfile. With a special analysing program the logical structure of each empirically observed task solving process was extracted. This logical structure is given as a Petri net. The behavioural complexity (BC) of this net structure can be measured with the McCabe-measure. With some special assumptions the cognitive complexity (CC) can be derived from the empirically gained BC. The main results are: (1) The time structure and BC measure different aspects of the learning process; (2) the time structure is–overall–positively correlated with BC and negatively correlated with CC; and (3) as well the long-term- as the short-term knowledge has an increasing predictive power with the time structure, but not with BC and CC. 1.
