抄録
The aim of this research was to construct a linear model for motor skill learning, especially in inverted printing task, considering response sequences. After constructing the model, the experiments were conducted to test the utility of this model.
Method. 48 Japanese letters, Hiragana, were used in inverted printing task. The rank ordered difficulties of 48 letters were determined by the number of letters printed per 30 seconds, and three kinds of learning list (List-A: low degree of difficulty, List-B: middle degree of difficulty, List-C: high degree of difficulty) were constructed according to the rank ordered difficulty. 24 kinds of letters were used in each list, and letters were arranged in random order. The conditions were 3 (List-A, B & C)×3 (32, 16 & 8 trial sessions). One trial session took 60 seconds, and each condition was conducted in massed practice.
First, a linear model was constructed, then three parameters, A1 (initial response value), lambda; (asymptotic value) and α (learning rate), were estimated by using 32 trial sessions' data, and after that expected values were obtained. Comparison between expected and observed values showed the satisfactory goodness of fit. A short trial sessions' data (16 and 8 trial sessions), in which subjects did not reach the asymptote, were used to test the utility and stability of the parameter α. Using the parameter α which was obtained in 32 trial sessions' condition, the parameters A1 and lambda; were estimated again. The expected values were obtained in both conditions of trials. The goodness of fit between expected and observed values was again satisfactory. Several possible hypotheses, which were suggested by the above results, were discussed.
