Decoding the cognitive states of attention and distraction in a real-life setting using EEG

Abstract

Lapses in attention can have serious consequences in situations such as driving a car, hence there is considerable interest in tracking it using neural measures. However, as most of these studies have been done in highly controlled and artificial laboratory settings, we want to explore whether it is also possible to determine attention and distraction using electroencephalogram (EEG) data collected in a natural setting using machine/deep learning. 24 participants volunteered for the study. Data were collected from pairs of participants simultaneously while they engaged in Tibetan Monastic debate, a practice that is interesting because it is a real-life situation that generates substantial variability in attention states. We found that attention was on average associated with increased left frontal alpha, increased left parietal theta, and decreased central delta compared to distraction. In an attempt to predict attention and distraction, we found that a Long Short Term Memory model classified attention and distraction with maximum accuracy of 95.86% and 95.4% corresponding to delta and theta waves respectively. This study demonstrates that EEG data collected in a real-life setting can be used to predict attention states in participants with good accuracy, opening doors for developing Brain-Computer Interfaces that track attention in real-time using data extracted in daily life settings, rendering them much more usable.

Using the body to think: an analysis of the cognitive mechanisms underlying Thinking at the Edge and Tibetan monastic debate

“…Indeed, we found that Tibetan monastic debate is associated with an increase in frontal midline theta oscillations (van Vugt et al 2020), which are also found in other meditation practices and thought to reflect attentional focus. Moreover, we found that more experienced monks spend less time distracted than less experienced monks during monastic debate (Kaushik et al 2022).…”

Mind wandering state detection during video-based learning via EEG

“…We separately tested decoding performance using different frequency bands, which may subserve different functional roles during mind wandering ( Kam et al, 2022 ), and we observed that beta band gave the highest prediction performance among the 4 frequency bands in both classification pipelines. Kaushik et al (2022) reported theta and alpha gave the highest performance. Dhindsa et al (2019) compared alpha, theta, and beta band (beta1:13–18 Hz, and beta2: 19–30 Hz), and their results showed that beta band (both beta1 and beta2) had the highest performance in a portion of the participants.…”

“…The 2-s window was based on a trade-off between better feature extraction with longer windows versus the need for real-time decoding with shorter windows. The choice of a 2-s window duration is consistent with the approach taken by Kaushik et al (2022) . The segments from the focused learning condition were labeled as non-mind wandering and segments from the future planning condition were labeled mind wandering (2-class).…”

“…The frequently used machine learning methods in related works ( Dong et al, 2021 ; Chen et al, 2022 ; Kaushik et al, 2022 ) were (1) linear support vector machine [SVM(linear)], (2) radial basis function kernel support vector machine [SVM(rbf)], (3) random forest (RF) and (4) logistic regression (LR). We evaluated and compared their classification performance, as measured by the area under the receiver operating characteristic curve (AUC).…”

“…The authors reported classification accuracy of 76.82% with a support vector machine (SVM) classifier. Kaushik et al (2022) conducted a study with 24 participants performing Tibetan monastic debate tasks, and the attentional state of the participants was rated by 3 observers of video recordings. The group level decoding performance of attention vs. distraction was high when using a long short-term memory neural network classifier (95.86% accuracy); the authors acknowledged that a limitation of the second-observer method was that only clear instances could be annotated.…”