Eye Blinks Describing the State of the Learner Under Uncertainty

Abstract

Adaptive systems are able to support the human-machine interaction in a great manner. However, the question arises which parameter are useful to gain insights into the user and can be easily implemented in the adaptive system. Eye blinks are frequent and most of the time automatic actions that reflect attentional and cognitive processes. They have not gained much attention in the context of adaptive systems until now. Thus, the current experiment investigated the number of blinks as an indicator of the state of the user while interacting with a technical system. Participants had to perform a dynamic visual spatial search task while their eye blinks were tracked. The task is to predict the appearance of target objects and thereby to learn a probability concept in order to improve the prediction. Results showed that eye blinks could distinguish between good and poor learner and increased parallel to the increasing task performance. Further, eye blinks reflected the information processing during the performance of a trial and the completion of the task. Thus, eye blinks might inform about the needs of the user with regard to the amount and detailedness of new information as well as additional help. However, the individual variability necessitates a separate baseline to be determined for each user. Further research is needed to foster the results also in more applied settings.