Continuous decoding of cognitive load from electroencephalography reveals task-general and task-specific correlates

Abstract

Objective. Algorithms to detect changes in cognitive load using non-invasive biosensors (e.g. electroencephalography (EEG)) have the potential to improve human–computer interactions by adapting systems to an individual’s current information processing capacity, which may enhance performance and mitigate costly errors. However, for algorithms to provide maximal utility, they must be able to detect load across a variety of tasks and contexts. The current study aimed to build models that capture task-general EEG correlates of cognitive load, which would allow for load detection across variable task contexts. Approach. Sliding-window support vector machines (SVM) were trained to predict periods of high versus low cognitive load across three cognitively and perceptually distinct tasks: n-back, mental arithmetic, and multi-object tracking. To determine how well these SVMs could generalize to novel tasks, they were trained on data from two of the three tasks and evaluated on the held-out task. Additionally, to better understand task-general and task-specific correlates of cognitive load, a set of models were trained on subsets of EEG frequency features. Main results. Models achieved reliable performance in classifying periods of high versus low cognitive load both within and across tasks, demonstrating their generalizability. Furthermore, continuous model outputs correlated with subtle differences in self-reported mental effort and they captured predicted changes in load within individual trials of each task. Additionally, alpha or beta frequency features achieved reliable within- and cross-task performance, suggesting that activity in these frequency bands capture task-general signatures of cognitive load. In contrast, delta and theta frequency features performed considerably worse than the full cross-task models, suggesting that delta and theta activity may be reflective of task-specific differences across cognitive load conditions. Significance. EEG data contains task-general signatures of cognitive load. Sliding-window SVMs can capture these signatures and continuously detect load across multiple task contexts.

The task pre-configuration is associated with cognitive performance evidence from the brain synchrony

“…The experience-dependent changes in rest functional connection, demonstrating a certain plausibility for task driven mechanism (Hearne, Cocchi, Zalesky, & Mattingley, 2017;Millar et al, 2021). The brain's functional network architecture during task performance was shaped primarily by an intrinsic network architecture that is also present during rest, and secondarily by evoked task-general and task-specific network changes (Boring, Ridgeway, Shvartsman, & Jonker, 2020;Messel, Raud, Hoff, Skaftnes, & Huster, 2019;Raichle, 2010). Overall, the brain has a task-general architecture, but there are still specific task configurations in different tasks (Cole et al, 2014).…”

Gaze dynamics are sensitive to target orienting for working memory encoding in virtual reality

“…Given that a task and environment can substantially alter gaze patterns, it is possible that any model built on gaze data within a single task is in fact capturing task-specific gaze patterns. This is indeed a critique that could be made for any neuro- or bio-sensing device and model (for a similar argument, see Boring, Ridgeway, Shvartsman, & Jonker, 2020 ). For this reason, we applied our gaze-dynamics model to a novel dataset to explore its generalizability and found that gaze dynamics contain valuable markers of task-general target orienting for encoding behaviors, a finding that has both theoretical and practical value in demonstrating that gaze dynamics can generally capture target orienting for encoding in rich, naturalistic settings.…”

Semantic scene-object consistency modulates N300/400 EEG components, but does not automatically facilitate object representations

“…Second, to track the impact of consistent or inconsistent objects on scenes representations, we performed decoding analyses to discriminate between the eight scene categories separately for consistent and inconsistent conditions at each time point from -100 ms to 1800 ms relative to the onset of the scene (-1100 ms to 800 ms relative to the onset of the object). For all decoding analyses, we adopted two approaches: standard timeseries decoding (Boring et al, 2020;Kaiser and Nyga, 2020), using data from a sliding time window, and cumulative decoding (Ramkumar et al, 2013;Kaiser et al, 2020a), using aggregated data from all elapsed time points. The two approaches are detailed in the following paragraphs.…”

“…In both analyses, we used all available trials, including those in participants responded incorrectly. For each decoding analysis, we adopted two approaches: standard timeseries decoding (Boring et al 2020;Kaiser and Nyga 2020), using data from a sliding time window, and cumulative decoding (Ramkumar et al 2013;Kaiser, Häberle, et al 2020), using aggregated data from all 12 elapsed time points. The two approaches are detailed in the following paragraphs.…”