We present a new learning model which accounts for changing values of b, a phenomenon found in the learning process. It is shown that this new model can be expressed in terms of two parameters, given the standard time for the task. The Power Curve turns out to be a particular case of the dynamic b model.
Checking the model for empirical data yields better fits than the Power Curve. Two methods for finding the model's parameters are presented.
We believe the dynamic-b model can provide important insights into the mechanics of the learning process and improve predictive capability of completion times. All this is achieved by building a model which is compatible with empirical findings such as a changing b and a nonzero standard time.