Can EEG Be Adopted as a Neuroscience Reference for Assessing Software Programmers’ Cognitive Load?

Medeiros, Júlio; Couceiro, Ricardo; Duarte, Gonçalo; Durães, João; Castelhano, João; Duarte, Catarina; Castelo-Branco, Miguel; Madeira, Henrique; de Carvalho, Paulo; Teixeira, César · 2021 · Crossref

DOI: 10.3390/s21072338

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 electroencephalography (EEG) can serve as a neuroscience reference for assessing the cognitive load of software programmers during code comprehension tasks. The research is motivated by the high cost of software defects, which are often linked to human error factors like cognitive overload and fatigue. While previous studies have used wearable biosensors measuring autonomic nervous system (ANS) signals (e.g., heart rate variability, pupillometry) to infer cognitive states, these methods suffer from indirect measurement, latency, and noise. The authors propose using EEG, which records direct brain activity, to establish a "ground truth" for validating the accuracy of less intrusive wearable devices. The researchers conducted a controlled experiment involving 26 programmers who performed three code comprehension tasks using Java snippets of varying complexity levels (Code 1 being simplest, Code 3 most complex). Each trial included a baseline period, a control text-reading task, and the code comprehension task. Data acquisition included EEG signals, eye-tracking data, and subjective assessments via the NASA-TLX survey to measure perceived mental effort. The study aimed to determine if EEG-derived features correlate with subjective complexity, if classic software complexity metrics accurately reflect mental effort, and if EEG combined with eye-tracking can pinpoint specific code lines causing high cognitive load. The results demonstrated that EEG features related to Theta, Alpha, and Beta brain waves possessed the highest discriminative power for identifying code lines requiring higher mental effort. The EEG data revealed evidence of mental effort saturation as code complexity increased. Crucially, the study found that classic software complexity metrics (such as McCabe Cyclomatic Complexity) did not accurately represent the mental effort involved in code comprehension; programmers’ perceived complexity and EEG-measured load deviated significantly from these traditional metrics. Furthermore, the combination of EEG biomarkers with eye-tracking information allowed for the accurate identification of specific code regions corresponding to peaks in cognitive load. The significance of this work lies in its proposal of EEG as a reference standard for evaluating wearable biosensors in software engineering. By establishing a direct neural measure of cognitive load, the study provides a method to calibrate and validate less intrusive ANS-based devices. The findings challenge the reliance on traditional software complexity metrics as proxies for human cognitive effort, suggesting that industry practices should incorporate measures of perceived complexity. Ultimately, this approach supports the development of biofeedback-augmented software engineering tools that can predict error-prone scenarios and improve software quality by monitoring developers' real-time cognitive states.

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
• workload measurement

Information type

What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).

• Empirical Findings: physiological data