Quantification and Analysis of the Efficiency of Iterative Learning by Using Event Related Potentials

Abstract

In order to discuss quantitatively the effect and progress of learning, we introduce a definition of learning effect. We intend to estimate quantitatively the proficiency of learning tasks by suitable measurement. The learning effect has been usually evaluated by the percentage of correct answers in a test/examination at the end of a learning course. However, human intellectual actions are also controlled by their cerebral nervous systems, and the learning effect can be evaluated by the measured data from their brain activities. In this paper, we adopt a change of the correct answer ratios as the definition of the learning effect. We estimate the change of the ratios by observing the changes of the electrical signals from brains. The learning effect reflects the proficiency of the learner during the process of a learning task. We first propose a model for estimating the learning effect from the brain activities. Next we analyze the experimental data of the tasks selecting one from three choices. From the data analysis we determine parameters that are used in the model. EEGs (ElectroEncephaloGrams) are electrical signals from brains that are activated by stimuli. They have a fine time resolution. An ERP (Event-Related Potential) is the average of EEGs. We use ERPs to measure brain activities caused by the tasks. Then we evaluate the usefulness of the parameters adopted in the model. This approach is an attempt to quantify the learning effect by physiological parameters.