Investigating hierarchical and ensemble classification approaches to mitigate the negative effect of varying stress state on EEG-based detection of mental workload level - and vice versa. Issue 1 (3rd April 2021)

• Record Type: Journal Article
• Authors: Bagheri, Mahsa; Power, Sarah D.
• Is Part Of: Brain-computer interfaces. Volume 8:Issue 1/2(2021)
• Page Start: 26
• Page End: 37
• Publication Date: 2021-04-03
• Subjects: passive brain–computer interface -- electroencephalography (EEG) -- mental workload -- stress -- classification; Brain-computer interfaces -- Periodicals; Neurology -- Periodicals; 616.800285
• DOI: 10.1080/2326263X.2021.1948756
• Languages: English
• ISSNs: 2326-263X

Abstract

ABSTRACT: Studies on EEG-based mental workload detection generally consider workload exclusively in terms of task demands, as manipulated through the variation of task difficulty. However, mental workload cannot be estimated precisely with the properties of the task alone because different aspects of the user's state could affect the mental effort needed to perform the task. We argue that affective states, specifically stress/anxiety, is of particular relevance. Thus, the overall objective of this work is to develop a passive brain–computer interface that detects mental workload and affective state simultaneously. Recently, we found that variation in affective state negatively affects EEG-based classification of mental workload, and vice versa. Thus, such variation should be explicitly considered when developing detection algorithms for use in real-world applications. Here, we investigated five classification approaches for detecting mental workload and affective state which explicitly consider variation in the other state. Significant improvements in classification accuracy were achieved.