Prediction of recall accuracy in contextual understanding tasks using features of oculo-motors

Contextual understanding, which consists of memory, reasoning, and recall, is a key process of human---computer interactions and interfaces which provide universal accessibility. To determine the possibility of predicting the recall accuracy of reading and memorizing tasks using features of oculo-motors, eye movements, and pupillary changes, a contextual understanding task experiment, which was composed of definition and question statements, was conducted. The existence of a relationship between the features of oculo-motors during memorization and recall performance was hypothesized. The features of viewer's oculo-motors while reading and judging a statement were observed, and significant differences were identified in most features according to the correctness of responses. The possibility of predicting the verification accuracy of question statements using these features for definition statements and the support vector regression technique was established, and there was a significant correlational relationship between the predicted accuracy and the experimental accuracy. Also, the estimation performance for judging question statements was assessed using the features and support vector machines. Estimation accuracy was the highest when all extracted features of eye movements were used. The number of definition statements, which were experimental conditions used to control the task difficulty, affected the accuracy of both prediction and estimation. These results provide evidence that features of oculo-motors during the reading, memorizing, and recalling of statements can be used as an index of contextual understanding.