Can EEG-devices differentiate attention values between incorrect and correct solutions for problem-solving tasks?

Abstract

The affective state of an individual can be determined using physiological parameters; an important metric that can then be extracted is attention. Looking more closely at compact EEGs, algorithms have been implemented in such devices that can measure the attention and other affective states of the user. No information about these algorithms is available; are these feature classification algorithms accurate? An experiment was conducted with 23 subjects who utilized a pedagogical agent to learn the syntax of the programming language Java while having their attention measured by the NeuroSky MindWave Mobile 2. Using a concurrent validity approach, the attention values measured were compared to band powers, as well as measures of task performance. The results of the experiment were in part successful and supportive of the claim that the EEG device’s attention algorithm does in fact represent a user’s attention accurately. The results of the analysis based on raw data captured from the device were consistent with previous literature. Inconclusive results were obtained relating to task performance and attention.

Keywords

• physiological computing
• attention
• eeg
• learning analytics