Toward neuroadaptive support technologies for improving digital reading: a passive BCI-based assessment of mental workload imposed by text difficulty and presentation speed during reading

Andreessen, Lena M.; Gerjets, Peter; Meurers, Detmar; Zander, Thorsten O. · 2020 · Crossref

DOI: 10.1007/s11257-020-09273-5

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This study investigates whether a passive brain–computer interface (pBCI) can assess mental workload during digital reading to support the development of neuroadaptive technologies. The authors address the challenge of creating adaptive reading systems that adjust text difficulty and presentation speed in real-time based on a reader’s cognitive state. While previous pBCI models for mental workload were often task-specific, this research tests a task-independent predictive model, calibrated using arithmetic tasks, to determine if it can reliably distinguish between different levels of mental workload induced by text readability and reading speed. The experimental design involved thirteen subjects (twelve included in final analysis) who underwent two phases. First, a calibration session used a mental workload paradigm where subjects alternated between low-workload relaxation and high-workload arithmetic subtraction tasks, with EEG data recorded via a 64-channel system. Second, subjects read twelve texts using a rapid serial visual presentation (RSVP) interface. The texts were categorized as easy or difficult based on Flesch-Kincaid readability metrics. Each category was presented at two speeds: a self-adjusted comfortable speed and a speed increased by 40%. EEG data were recorded during reading, and the task-independent predictive model, trained on the calibration data using filter bank common spatial patterns and linear discriminant analysis, was applied to one-second epochs aligned with each word’s onset. The results demonstrated that the task-independent predictive model successfully distinguished between workload levels during reading. Predictive values for mental workload were significantly higher for difficult texts compared to easy texts. Additionally, texts presented at the increased speed yielded higher predictive workload values than those presented at the self-adjusted normal speed. These findings held true at the single-subject level, indicating that the model could reliably detect individual variations in cognitive load imposed by text complexity and presentation rate. The study also analyzed the cumulative predictive values over word positions, showing that the model could discriminate between text categories after processing a sufficient number of words. The significance of this work lies in its validation of task-independent pBCI models for real-world applications like digital reading. By demonstrating that mental workload can be assessed online without task-specific recalibration, the study supports the feasibility of neuroadaptive systems. Such systems could continuously monitor brain activity to dynamically adjust text difficulty or presentation speed, keeping readers within an optimal cognitive load range. This approach aims to enhance reading efficiency and comprehension by preventing cognitive overload or under-stimulation, thereby fostering more effective text-based learning and interaction with digital media.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-nvidia on 2026-08-10; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• neuro workload indices