LNCS 9359 - Symbiotic Adaptive Interfaces: A Case Study Using BrainX 3

Abstract

Modern symbiotic and adaptive HCI paradigms include real-time algorithms capable of inferring user's states. We present an exper-imental interface which aims to understand the information processing capacity of a human user and use this information to improve interaction in a database exploration task. We collected the electrodermal activity and pupillometry signals in tasks of increasing difficulty and used their features to infer whether the subject was performing the task correctly or not. By combining principal component analysis and logistic regression methods, we successfully inferred the accuracy of users' responses from the signal after the response was made. This study provides a quantita-tive framework for modelling user internal states and evaluates it in a practical human computer interaction task.