Understanding mental workload has never been more accessible with the advancements in mobile EEG technology. This study researched brain responses under different mental workloads in the real-world scenarios.
Mental workload measurements in real-world settings
In today’s dynamic work environment, it’s imperative to understand the brain’s response under varying mental workloads.
This is especially relevant when tasks need high attention. These include multi-monitor setups or complex tasks. The key research question is: How can EEG measure changes in memory load in a real-world setting?
Study Methods
The study combined both well-established laboratory procedures and a novel real-world task. The mental workload was measured through EEG.
Participants performed a number-copying task on either a single or dual-monitor setup. Single setup represents high memory workload, while dual setup represents low memory workload. Alongside this, data from a previous Sternberg task was also utilized for comparison.
The EEG data was processed with three feature spaces:
band power (BP), modulation index (MI), and coherence, reflecting different cognitive processes.
Researchers focused on short-term epochs to observe real-time workload changes.
Findings
The research highlighted the following:
Effective EEG Features
BP, MI, and coherence were found effective in distinguishing between high and low memory workloads. Using a grouping (Ng) of 8 epochs provided a balanced accuracy of around 70%. Yet, shorter epochs like 1 or 2 didn’t yield plausible results.
Outcomes from the number-copying study. Bar diagrams display balanced accuracy for varying values of Ng. On each graph’s x-axis, distinct feature domains are marked: (1) Band Power (bp), (2) Mutual Information (mi), (3) Coherence (coh), and (4) Comprehensive combination (all).Comparison with Previous Data
The Sternberg task data showed a higher classification accuracy compared to the number-copying task. It’s hypothesized that the Sternberg task’s controlled environment and the differentiation between high memory workload and a distinct eye-open resting period contributed to this.
Real-world vs. Laboratory Data
The study found differences between data collected in a controlled environment Vs. a real-world setting. Despite the differences in accuracy between the two data sets, EEG features consistently distinguished between high and low memory workload states.
Feature Analysis
Most features were useful for classifying high and low memory workload states. The peak performance observed when 100% of the features were preserved. Some redundancy was observed, but no negative impact on performance was recorded.
Signatures Across Conditions
The data-driven signatures derived from EEG data were consistent across both tasks. The significance was observed even when differentiating between hard and easy tasks.
Interpretation and Implications
The study underscores the efficacy of EEG in measuring workload in real-world scenarios. Results provide a foundation for the development of more practical, plug-and-play workload-monitoring systems.
The research implies that while EEG can capture stress or high cognitive load, there’s a nuanced difference between these states. A high memory workload state in EEG does not directly equate to stress, but rather the demands placed on memory.
The observed variation in feature values across participants suggests a promising avenue for transfer learning methods to reduce variations, creating more applications.
It’s essential to acknowledge the study’s limitations. The research had a relatively small dataset for the number-copying task. This was due to challenges brought about by the Covid-19 pandemic.
In Conclusion
In a world increasingly interested in understanding the human brain and its reactions to various stimuli, this study provides vital insights.
It affirms the utility of EEG as an objective measure for human factors. This study shows its potential usage in determining cognitive workload.
This research bridges the gap between laboratory settings and real-world environments. It offers a glimpse into the future of neuroergonomics and how it can shape our understanding of the brain at work.
Original publication source: https://iopscience.iop.org/article/10.1088/1741-2552/accbed/meta
