Next Article in Journal
Sustainability Meets Society: Public Perceptions of Energy-Efficient Timber Construction and Implications for Chile’s Decarbonisation Policies
Previous Article in Journal
LSH-YOLO: A Lightweight Algorithm for Helmet-Wear Detection
Previous Article in Special Issue
Analysis of Cognitive Biases in Construction Health and Safety in New Zealand
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Research on Cognitive Load of Tunnel Construction Workers in Different Environments Based on EEG

1
China Railway 14th Bureau Group Third Engineering Co., Ltd., Jinan 250101, China
2
Sanming Puyan Expressway Co., Ltd., Sanming 365000, China
3
College of Civil Engineering, Fuzhou University, Fuzhou 350116, China
*
Author to whom correspondence should be addressed.
Buildings 2025, 15(16), 2920; https://doi.org/10.3390/buildings15162920
Submission received: 27 May 2025 / Revised: 4 July 2025 / Accepted: 16 July 2025 / Published: 18 August 2025
(This article belongs to the Special Issue Human Factor on Construction Safety)

Abstract

The tunnel construction environment is complex, and workers’ cognitive load directly affects safety and efficiency, making a dynamic assessment urgently needed. This study aims to explore the cognitive load of tunnel construction workers under different working environments using EEG technology. In the experimental design, subjects adapted to the virtual reality (VR) environment and received instructions before wearing a wireless EEG system and VR equipment to begin the formal experiment. Each subject underwent four rounds of experiments, corresponding to four different scenarios: control, night shift, noise, and confined space. Each round included three tasks of low, medium, and high difficulty. Analysis of EEG data showed that tunnel construction tasks in different environments significantly affected cognitive load, especially during night shifts and in confined spaces, with cognitive load increasing significantly with task difficulty. The results provide a theoretical basis for optimizing the management of tunnel construction environments and task design.

1. Introduction

Tunnel engineering is an important part of modern transportation infrastructure, characterized by a complex and high-risk working environment [1]. Such conditions significantly affect workers’ physiological state and pose severe challenges to their cognitive abilities and safety performance. Studies show that harsh working environments can increase workers’ cognitive load, thus raising the probability of safety risks [2]. Therefore, scientifically assessing cognitive load variations in tunnel workers under different environments is of great significance for improving working conditions, optimizing operations, and enhancing construction safety.
Cognitive Load Theory (CLT) suggests that when individuals face excessive cognitive load, their cognitive resources become over-consumed, leading to distraction and decision errors, increasing accident risks. Measurement methods for cognitive load mainly include subjective scales, task performance, and physiological indicators [3]. In the context of tunnel construction, CLT assumes critical importance due to the industry’s inherently high-risk operational environment. Construction workers must simultaneously process multiple information streams while maintaining situational awareness under challenging environmental conditions. When cognitive resources become over-consumed due to excessive load, workers experience attention narrowing, reduced hazard detection capability, and compromised safety decision-making [4]. The tunnel construction environment presents unique cognitive challenges that amplify CLT’s relevance. Workers must navigate complex spatial configurations, interpret acoustic and visual safety signals amid ambient noise, coordinate with team members, and make time-critical decisions while operating in confined, poorly lit spaces. These environmental stressors create additional extraneous cognitive load that competes with task-essential cognitive resources, potentially pushing workers beyond their cognitive capacity limits [5].
EEG is an effective bioelectrical signal reflecting real-time brain activity. Due to its objectivity and real-time nature, EEG-based physiological measurement has been widely used in cognitive load assessment [6] especially in detecting critical cognitive states like alertness and attention [7]. Zhang [8] and Wei [9] have applied EEG to assess mental load, fatigue, alertness, and emotional states in task settings. Saedi [10] and Cheng [11] reviewed EEG applications in construction safety, focusing on cognitive states, experiment design, and data analysis methods.
Special characteristics of tunnel construction environments include lighting, noise levels, and space constraints. Lu found that low illumination affects workers’ visual sensitivity and alertness, reducing hazard recognition efficiency [12]. Ke [13] and Ou [14] both conducted field experiments and VR experiments have shown that noise interference leads to participant distraction, increases cognitive fatigue, and reduces information processing capacity. However, most existing studies focus on single environmental factors, lacking systematic analyses of cognitive load variations under multiple factors, particularly using objective physiological indicators.
Traditional cognitive load measurement methods are often hard to implement on-site, whereas virtual reality (VR) offers a new research approach by simulating complex construction scenarios [15]. VR can precisely control environmental parameters and create realistic settings while ensuring participant safety [16]. Combined with wireless EEG systems, EEG data can be collected in real time during simulated tasks, providing objective evidence for cognitive load assessment [17]. Previous studies show that α, β, and θ EEG bands are significantly correlated with cognitive load. Alpha power is related to relaxation, beta to brain activity, and theta to attention and memory load [18].
In view of this, the present study used virtual reality technology to construct a simulated tunnel construction environment and designed tasks of varying difficulty levels. By measuring the EEG activity of construction workers under four working conditions—control, night shift, noise, and confined space—the study systematically analyzed the interactive effects of environmental factors and task difficulty on cognitive load. The results provide a theoretical basis and practical guidance for optimizing tunnel construction environment design, developing scientific operating procedures, and designing targeted safety training.
This research first introduces the research methods, including the setup of the simulated work environment, selection of participants, experimental equipment, task design, and experimental procedures; it then analyzes the variations in alpha, beta, and theta EEG power and their composite indicators under different working environments and task difficulties; finally, based on the experimental results, it discusses the mechanisms by which environmental factors influence cognitive load and proposes specific suggestions for improving the work environment and reducing cognitive load.

2. Experimental Methodology

2.1. Simulated Work Environment

The experimental scene simulated the tunnel face at a construction site, with humidity, temperature, and ventilation conditions controlled to minimize confounding variables. The lighting intensity and noise levels were set with reference to Chinese national standards, local standards, and existing studies [19,20]. Four working environments were established for tunnel excavation tasks. For the control group, illumination was set at 300 lux and noise at 75 dB, based on Chinese national and regional standards for tunnel construction (JTG/T3660-2020; DGJ32/TJ 216-2016) [19,20], representing standard daytime tunnel working conditions. No additional stressors were introduced in this condition, and spatial constraints were removed. This setting serves as a baseline environment against which the effects of reduced lighting, increased noise, and spatial confinement could be isolated and evaluated, as detailed in Table 1. The VR scene was developed using Unreal Engine 4 software. Participants rated the realism of the experimental scene on a 10-point Likert scale, with an average realism score of 7.14, indicating a high level of realism.

2.2. Subjects

A total of 18 participants were recruited for this experiment, with an age range of 25–38 years (M = 32.74, SD = 3.23) and a male-to-female ratio of 2:1. All participants had at least six months of tunnel construction experience, were in good health, had no color blindness, and had no history of 3D motion sickness. Before the experiment, all participants received standardized training to familiarize themselves with tunnel construction procedures and safety regulations, ensuring their understanding and ability to operate the virtual tasks.

2.3. Equipment

2.3.1. VR Equipment

The experiment used the HTC Vive VR device (manufacturer: HTC & Valve, Barcelona, Spain), with a resolution of 1080 × 1200 pixels per eye (combined resolution of 2160 × 1200 pixels), a refresh rate of 90 Hz, and a field of view of 110°. The tunnel worker risk awareness measurement system was developed based on the UE4 engine and operated within the Steam VR environment (Figure 1).

2.3.2. Enobio Wireless EEG System

The EEG acquisition was conducted using the Enobio wireless EEG system, which is capable of transmitting 24-bit EEG data and accurately reconstructing the original EEG signals. The system has a bandwidth of 0–250 Hz, a sampling rate of 500 samples per second, 24-bit resolution, and a sensitivity of 0.05 μV, with noise levels below 1 μV RMS.

2.4. Task Design

In the tunnel construction tasks, based on task complexity, attention demands, decision-making difficulty, and multitasking requirements, tasks were classified into low-difficulty (R1), medium-difficulty (R2), and high-difficulty (R3) levels.
Low-difficulty tasks involved observing designated targets in the tunnel scene (such as scattered rebar or blue gas cylinders) and confirming them by nodding or gazing. These tasks required low cognitive resources, with low levels of complexity, attention demands, decision-making difficulty, and multitasking needs. Medium-difficulty tasks involved identifying and reporting hazard signs, requiring a certain degree of observation and judgment. These tasks had moderate demands on cognitive resources, complexity, attention, decision-making, and multitasking. High-difficulty tasks combined hazard monitoring with a timed evacuation decision task, involving complex environmental monitoring, verbal judgment, and countdown pressure. These tasks placed high demands on cognitive resources, complexity, attention, decision-making, and multitasking.
Low-difficulty tasks: Observe and confirm targets. In the virtual tunnel scene, participants observed designated areas (e.g., a material storage area, a piece of equipment, or an exit sign). The virtual environment presented typical tunnel construction scenes as shown in Figure 2, which displays the tunnel face during construction with various equipment and materials visible. When prompted by the system (e.g., “Please confirm the blue gas cylinder”), participants confirmed by verbal response or nodding. The success rate was recorded in real time and synchronized with EEG data to ensure data alignment.
Medium-difficulty tasks: Identify and report hazard signs. The virtual environment included preset hazard signals (e.g., scattered rebar, Figure 3; wall cracks and water seepage, Figure 4; and improper loading of transport vehicles, Figure 5) consistent with real-world tunnel construction hazards. Participants were required to observe environmental information and immediately report upon identifying hazards. Report times and accuracy were recorded in real time and aligned with EEG data timestamps for subsequent analysis.
High-difficulty tasks: hazard monitoring and timed decision-making dual task. Participants monitored potential safety risks (as in the medium-difficulty task) and made quick decisions during timed evacuation tasks. For example, when a 10 s countdown appeared, the system would ask, “Which evacuation route is closer?” Participants had to respond verbally with “Route A” or “Route B” before the countdown ended. Abnormal recognition performance, decision accuracy, and timeout occurrences were recorded, while EEG data were simultaneously collected.

2.5. Experimental Procedure

After the participants adapt to the VR environment and receive experimental instructions, they begin the formal trial by wearing a wireless EEG system and VR equipment. During the experiment, the virtual reality scenes being viewed by the participants are also monitored via a laptop screen. Each participant completes four rounds of experiments, each corresponding to a different experimental scenario. Each round consists of three tasks, each representing a low, medium, and high difficulty level. Each difficulty level corresponds to a task duration of approximately 2 min, with a 1 min rest interval between tasks of different difficulty. A 5-min break is provided between each group of tasks, and a 10-min interval is scheduled between each round of experiments to reduce the impact of fatigue and other factors. The participants’ task completion status is recorded in real time throughout the experiment.

3. Method

3.1. Indicator Selection

Previous studies have shown that electroencephalography (EEG) is an effective tool for assessing workers’ cognitive load and fatigue levels [10]. Commonly used EEG signals include delta (1–4 Hz), theta (4–8 Hz), alpha (8–13 Hz), and beta (13–30 Hz) waves [5]. In this study, α, β, and θ waves were selected as the primary experimental indicators, while the α/β ratio and (α + θ)/β ratio were used as composite indices. Research has indicated that alpha waves are associated with the brain’s relaxation and resting state; for example, Ke et al. confirmed a negative correlation between α wave power and work-related stress in studies monitoring construction workers’ emotional states [21]. β waves are related to the brain’s active state and high levels of concentration, whereas θ waves are linked to memory and learning processes, reflecting memory load and learning effort. Wang et al. found that when workers are under high stress or fatigue, θ wave activity increases while β wave activity decreases [22]. Additionally, the α/β ratio and (α + θ)/β ratio were selected as composite indicators because the α/β ratio can comprehensively reflect alertness and concentration, while the (α + θ)/β ratio reflects overall alertness as well as memory and learning load, which helps to more thoroughly assess tunnel workers’ cognitive load under different tasks [23].

3.2. Data Collection and Processing

3.2.1. Experimental Data

Based on the participants’ task completion status, three participants did not complete the tasks or ended the test prematurely, resulting in EEG data collected from fifteen participants, yielding a total of 120 EEG data samples.

3.2.2. Electrode Selection

The electrodes selected for EEG signal extraction are channels Fz, F3, F4, F7, F8, T3, T4, CZ, C3, and C4. Specifically, Fz, F3, F4, F7, and F8 belong to the frontal region, which is related to cognitive functions, attention, and decision-making [24]; T3 and T4 are located in the temporal region and are generally associated with auditory processing, language comprehension, and memory [25]; Cz, C3, and C4 are positioned in the central region, which mainly involves the primary motor cortex and the somatosensory cortex and have been used to assess cognitive load during mentally demanding activities [26]. These channels focus on the frontal, parietal, and temporal lobes of the brain, areas associated with attention and visual-spatial abilities. They are closely linked to danger perception and the brain activation patterns under varying cognitive load conditions.

3.2.3. Data Preprocessing

Using the EEG toolbox in MATLAB 2019b, the data undergo noise reduction and artifact removal. The raw EEG data files (easy format) are first converted to the set format using a batch converter. Then, EEG frequencies below 0.5 Hz and above 40 Hz are filtered to remove interference from other signals. Re-referencing is performed based on the arithmetic mean of the overall data. Independent Component Analysis (ICA) is applied for artifact removal, accurately extracting valid EEG signals while manually removing artifacts.

3.2.4. Band Power Processing

Finally, the α, β, and θ band power values for all participants across different scenarios are extracted, box plots are generated, and a two-way repeated measures ANOVA is performed using IBM SPSS Statistics 27.

4. Results and Discussion

4.1. Alpha Wave Analysis

Through the processing and statistical analysis of the experimental data, the absolute α wave power for 15 participants across 4 experimental scenarios and 3 task groups was obtained, and a box plot was generated. As shown in Figure 6, in all work environments, the α wave power decreases as cognitive load increases from R1 to R3, with the night shift group showing a slight increase from R2 to R3. α wave power is typically associated with relaxation or rest states, and the continuous decline from R1 to R3 suggests that as task difficulty increases, the brain’s relaxation level decreases, thereby increasing the cognitive load on the participants.
A two-way repeated measures ANOVA was conducted to examine the significance of α wave differences under varying task difficulties across different work environments. The results are presented in Table 2. Significant within-subject effects were observed in all four types of work environments (p < 0.05). As shown in Table 2, the results of the analysis of variance were significant. However, the interaction between the two factors was not significant, F(6, 112) = 0.170, p = 0.318. To further investigate the differences in α wave activity across different task difficulties within the same environment, a Tukey HSD post-hoc test was conducted.
As shown in Table 2, the results of the variance analysis are significant, so a Tukey HSD post-hoc test was conducted. The comparison between R2 and R1 showed significant differences in all work environments (p < 0.05). The largest difference was observed in the confined space group (−139.6773), and the smallest difference was in the control group (−120.4402). For R3 and R1, significant differences were found in all work environments (p < 0.001). The largest difference occurred in the confined space group (−208.4045), while the smallest was in the night shift group (−130.7502). For R3 and R2, no significant differences were found in any environment (p > 0.05).
In the control group, α wave power was highest during R1 (960.621), indicating a higher relaxation level. It significantly decreased during the high-difficulty task, suggesting that higher task difficulty increases cognitive load. In the night shift group, α wave power was lower during R1 (885.549), indicating a higher cognitive load, potentially caused by sleep deprivation. The small F-value (9.086) suggests that the differences in cognitive load were not significant. In the noise and confined space groups, α wave power during R1 was lower than that of the control group (872.913 and 897.461), possibly due to environmental stressors (noise, spatial constraints). The larger F-value indicates a stronger impact of environmental factors on cognitive load.
As task difficulty increases, α wave power decreases, reflecting an increase in cognitive load. This suggests that high cognitive load in noisy and confined spaces may lead to mental fatigue, affecting work performance and safety. In the night shift environment, α wave power slightly increased during the medium-difficulty task, which could be due to the brain’s adaptive adjustment when transitioning from a low cognitive load task to a medium cognitive load task in the night shift environment.

4.2. Beta Wave Analysis

According to Figure 7 and Table 3, the mean β wave power increases with task difficulty in all work environments, and the p-values are all less than 0.05, indicating significant results. β waves are typically associated with active thinking, focus, and problem-solving. The continuous increase from R1 to R3 suggests that as cognitive load increases, brain activity and attention levels rise.
To examine whether there were significant differences in β wave absolute power under varying cognitive load task difficulties across different work environments, a two-way repeated measures ANOVA was performed (Table 3). The results indicated that the interaction effect between work environment and task difficulty on β wave power was not statistically significant, F(5.262, 98.228) = 0.17, p > 0.05 (as the assumption of sphericity was violated, Greenhouse-Geisser correction was applied). This suggests that the effect of cognitive load on β wave power was consistent across different work environments. To further explore the differences in cognitive load within the same work environment, a Tukey HSD post-hoc test was conducted.
The Tukey HSD post-hoc test results are as follows:
R2 and R1: Significant differences were observed in the control group (p = 0.0477) and noise environment (p = 0.0415), but not in the night shift (p = 0.5113) and confined space (p = 0.1522) environments. The largest difference was found in the noise group (64.8763), and the smallest difference was in the confined space group (50.6255).
R3 and R1: Significant differences were observed in all work environments. The largest difference occurred in the noise group (141.9135), and the smallest difference was in the confined space group (115.807).
R3 and R2: Significant differences were found in the night shift (p = 0.0414), noise (p = 0.013), and confined space (p = 0.0486) environments, while the control group showed no significant difference (p = 0.0698). The largest difference was observed in the night shift group (82.9541), and the smallest difference was in the control group (61.1245).
During R1 in the night shift, β wave power was relatively high (524.9311413), possibly due to additional psychological stress associated with the night shift itself. The F-value was small (6.895), but the difference from R1 to R3 was significant, suggesting that high-difficulty tasks under night shift conditions have a greater impact on brain activity. The trends in β wave power for the noise and confined space environments were similar, with the noise group showing the largest difference from R1 to R3 (141.9135), indicating that cognitive load in noisy environments has the strongest impact on brain activity.

4.3. Theta Wave Analysis

According to Figure 8, the θ wave power in the control, noise, and confined space environments gradually increases with task difficulty. However, in the night shift environment, θ wave power is higher during R1, decreases during R2, and rises again during R3. This result may be due to the fact that night shift work often leads to physical and mental fatigue, making it difficult to maintain attention. In a fatigued state, cognitive load increases, but at the same time, brain excitability and information processing capacity decrease, resulting in a reduction of θ wave activity during medium-difficulty tasks. The increase in θ wave power with increasing task difficulty suggests that high difficulty may lead the brain to excessively mobilize memory and attention resources, particularly in noise and confined space environments, which may increase the risk of mental fatigue.
A two-way repeated measures ANOVA was conducted on the absolute power of θ waves under different cognitive load task difficulties (Table 4). The results showed that the interaction between work environment and task difficulty had a statistically significant effect on θ wave power, F(5.2, 97.058) = 4.663, p < 0.05 (as the assumption of sphericity was violated, Greenhouse-Geisser correction was applied). To further investigate this interaction, a simple effects analysis was conducted.
The simple effects analysis results are as follows:
R1 and R2 and R1 and R3: The results are similar, with significant differences observed in the control group, noise, and confined space environments (p < 0.005). In contrast, the night shift group showed no significant differences (p = 0.7896 for R1 and R2).
R2 and R3: Significant differences were found in all work environments (p < 0.05). The largest difference occurred in the noise group (−177.8748), and the smallest difference was found in the control group (−88.0501).

4.4. α/β Analysis

According to Figure 9 and Table 4, in all work environments, the α/β ratio shows a decreasing trend as cognitive load increases across task levels, with significant differences in the α/β values (p < 0.05). The decrease in the α/β ratio indicates that as task difficulty increases, the relative contribution of α waves decreases, while the relative contribution of β waves increases, reflecting a shift in the brain from a relaxed state to an active state, signifying an increase in cognitive load. The differences from R2 to R3 in the night shift group and the confined space group are relatively small, which may suggest that under high-difficulty tasks, the balance between relaxation and activation becomes more stable, leading to a more balanced cognitive load.
To examine whether there were significant differences in the absolute power of the α/β ratio under different cognitive load task difficulties across various work environments, a two-way repeated measures ANOVA was conducted (Table 5). The results indicated that the interaction effect between work environment and task difficulty on the α/β ratio power was not statistically significant, F(6, 112) = 0.17, p > 0.05, suggesting that the variation trend of the α/β ratio remained consistent across different work environments. To further explore the differences in cognitive load within the same environment, a Tukey HSD post-hoc test was performed.
The Tukey HSD post-hoc test results are as follows:
R2 and R1: Significant differences were observed in the control group, night shift, and noise environments (p < 0.005), but not in the confined space group (p = 0.0829). The largest difference occurred in the noise group (−0.4919), and the smallest difference was in the confined space group (−0.2562).
R3 and R1: Significant differences were observed in all work environments (p < 0.001).
R3 and R2: Significant differences were observed in the control group (p = 0.0393) and noise environment (p = 0.009), but not in the night shift (p = 0.3077) or confined space (p = 0.086) environments. The largest difference occurred in the noise group (−0.2691), and the smallest difference was in the night shift group (−0.1527).
In the control group, the α/β ratio was highest during R1 (2.361), and the F-value (21.587) was relatively high, indicating a significant difference in cognitive load. In the night shift group, the α/β ratio was moderate during R1 (1.745), with an F-value of 15.546. The difference between R2 and R3 was not significant, which may suggest that cognitive load stabilizes under high-difficulty tasks.

4.5. (α + θ)/β Analysis

According to Table 6 and Figure 10, significant differences in the (α + θ)/β ratio across cognitive load levels were observed in the control group, night shift, and noise environments (p < 0.05), while no significant difference was found in the confined space group (p = 0.563). In the significant groups (control, night shift, and noise), the (α + θ)/β ratio decreased as task difficulty increased, indicating that the relative contribution of relaxation and memory-related states decreased compared to active states, reflecting a marked increase in cognitive load. The lack of significant change in the confined space group may suggest that the impact of environmental constraints on cognitive load has unique characteristics.
To examine whether there were significant differences in the absolute power of the (α + θ)/β ratio under varying cognitive load task difficulties across different work environments, a two-way repeated measures ANOVA was conducted (Table 6). The results showed that the interaction effect between work environment and task difficulty on the (α + θ)/β ratio power was not statistically significant, F(6, 112) = 1.434, p > 0.05, indicating that the effect of cognitive load was consistent across different work environments. To further investigate the differences in cognitive load within the same environment, a Tukey HSD post-hoc test was conducted.
The Tukey HSD post-hoc test results are as follows:
R2 and R1: Significant differences were found in the night shift (p = 0.0139) and noise environments (p = 0.0076), while no significant differences were observed in the control group (p = 0.1601) or confined space group (p = 0.9924). The largest difference was in the night shift group (−0.6156), and the smallest was in the confined space group (0.0228).
R3 and R1: Only the confined space group showed no significant difference (p = 0.6618). The largest difference was in the noise group (−0.6388), and the smallest in the control group (−0.7904).
R3 and R2: No significant differences were observed across all environments (p > 0.05), with the night shift group showing the smallest difference (−0.1312) and the control group the largest (−0.3363).
The absence of significant changes from R2 to R3 in the night shift group may be due to fatigue, which weakens the effect of cognitive load changes. The confined space group showed the lowest F-value (0.583) and no significant differences, possibly indicating that environmental constraints limit the brain’s ability to modulate cognitive resources.

4.6. Cognitive Load Patterns Analysis

4.6.1. EEG Band-Specific Responses

The experimental results demonstrate clear patterns in the relationships between EEG frequency bands and cognitive load across different environments.
Specifically, the α wave power consistently decreased as task difficulty increased, reflecting a reduction in the brain’s relaxation level and indicating that workers need to maintain higher vigilance and information processing under greater cognitive demands. This finding aligns with the meta-analysis by Chikhi et al. [27], which established that alpha power suppression is a robust indicator of increased cognitive workload across various domains. The confined space environment exhibited the most pronounced alpha reduction (−208.4045 μV2 from R1 to R3), suggesting that spatial constraints amplify cognitive demands through additional attentional resources required for spatial awareness and movement planning.
In contrast, β wave power increased significantly with higher task difficulty, showing that more complex tasks require intensified active thinking, concentration, and problem-solving, which is consistent with the findings of Al-Hosani’s study [28]. The night shift environment’s unique pattern of elevated baseline beta activity (524.931 μV2) suggests that circadian disruption chronically elevates cognitive control demands even during low-difficulty tasks, revealing a persistent state of neural vigilance that may contribute to accelerated mental fatigue.
The θ wave, which is associated with memory and learning processes, also showed an upward trend in most environments as tasks became more challenging, implying that workers mobilize more memory and attention resources to cope with high cognitive load. This is consistent with findings from Wang et al. [29], who demonstrated that theta power increases during complex cognitive tasks in construction workers. But the night shift environment’s anomalous theta pattern, characterized by high baseline activity (843.741 μV2) followed by a decrease during medium-difficulty tasks (811.880 μV2) before rising again during high-difficulty tasks (942.501 μV2), reveals fatigue-related disruptions in memory processing systems.

4.6.2. Environment-Dependent Load Modulation

This study reveals how distinct environmental conditions modulate the degree and pattern of cognitive load increase or decrease, as reflected in specific brainwave frequency bands and composite indices.
In the night shift environment, workers exhibited higher baseline theta power and lower alpha power, indicating that even at low task difficulty, the brain maintains a heightened state of cognitive strain due to circadian rhythm misalignment and accumulated fatigue. This finding corroborates the neurophysiological basis of circadian rhythm disruption effects documented by Chellappa et al. [30], who showed that cortical excitability exhibits robust circadian dynamics correlated with cognitive performance. As task difficulty increases, the expected proportional rise in cognitive load is partially suppressed, producing a ceiling effect: the brain’s ability to further increase neural activation is limited, resulting in relatively smaller differences in band power and lower F-values compared to other conditions. Thus, although absolute cognitive load remains high throughout, the incremental increase with task difficulty is reduced.
In the noise environment, a clear and consistent pattern of cognitive load increase is observed. Beta and theta power rise substantially with task complexity, while the alpha/beta ratio decreases significantly (from 1.920 to 1.159). This indicates that noise exposure amplifies the rate at which cognitive load increases, because workers must continuously allocate extra attentional and memory resources to suppress irrelevant auditory input while processing task-relevant information. This finding builds upon the work of Astuti et al. [31], who used EEG to monitor construction worker cognitive performance caused by noise, by providing quantitative measures of the cognitive load escalation mechanism under noise interference. Consequently, mental effort rises more steeply as task demands grow, reflecting an enhanced cognitive load escalation mechanism under noise interference.
In the confined space condition, the absolute alpha power shows the largest decrease as task difficulty rises, suggesting a significant drop in relaxation and a corresponding increase in baseline cognitive tension. However, the composite ratios (α/β and (α + θ)/β) show relatively small or non-significant changes across tasks. This suggests that the confined space environment imposes a consistently high baseline cognitive load from the start, leaving less capacity for additional load increases as task complexity grows. In other words, the environmental constraint itself elevates the starting point of cognitive burden and flattens the typical load–task relationship, resulting in less pronounced incremental increases but a persistently high cognitive strain level throughout. This pattern has not been previously characterized in occupational neuroscience literature, representing a novel insight into how spatial constraints affect cognitive resource allocation.
Overall, the findings indicate that
Night shift work maintains a high but less adaptable cognitive load with a limited increase range as tasks become more difficult;
Noise exposure accelerates and amplifies the rate of cognitive load increase, reflecting more rapid neural resource mobilization;
Confined spaces elevate baseline cognitive load, producing a smaller incremental increase yet sustaining high mental tension regardless of task level.
These differentiated patterns of cognitive load increase and decrease highlight the importance of tailored environment management strategies. Understanding whether an environment mainly raises baseline load, amplifies task-related increments, or both, provides critical insights for scheduling, task allocation, and hazard control to protect workers’ mental well-being and ensure safety.

5. Conclusions

This study analyzed the α, β, and θ frequency bands as well as the α/β and (α + θ)/β ratios of tunnel construction workers under four work environments—control, night shift, noise, and confined space—across different task difficulty levels. The following conclusions were drawn.

5.1. Work Environment Management

In the night shift environment, baseline θ power was relatively high (R1: 843.741), while both the α/β ratio (R1: 1.745) and (α + θ)/β ratio (R1: 3.423) were also elevated. In contrast, baseline β power was relatively low (R1: 524.931), and the θ wave exhibited a decreasing-then-increasing trend across the three tasks. This suggests that night shifts have a substantial impact on cognitive load, possibly due to fatigue and circadian rhythm disruptions that impair attention and require increased neural resource mobilization under high load. It is recommended that work tasks during night shifts be reasonably scheduled, avoiding prolonged continuous work and ensuring adequate rest periods for construction personnel.
Under low-difficulty tasks in the confined space environment, the α/β ratio was relatively high (R1: 1.635) but showed a marked decrease under high-difficulty tasks (down to 1.124), while the (α + θ)/β ratio exhibited no significant change (R1: 2.731 to R3: 2.563). This indicates that spatial constraints may attenuate the effects of task-induced cognitive load. It is recommended that task allocation in confined space environments be carefully planned, avoiding prolonged high-load work to safeguard workers’ safety and health.
In the noise environment under high-difficulty tasks, the α/β ratio decreased significantly (R1: 1.920 to R3: 1.159). This change was consistent with a decline in α power (872.913 to 684.799) alongside increases in β power (459.796 to 601.710) and θ power (659.024 to 936.099), indicating that noise may differentially influence cognitive load by amplifying demands on activation and memory. Effective noise-reduction measures, such as the use of earplugs or soundproofing equipment, are recommended to mitigate the impact of noise on construction workers.

5.2. Task Management

It is recommended to appropriately increase the frequency of low-difficulty tasks to help workers gradually adapt to the work environment, improve work efficiency, and maintain cognitive load at a lower level during task completion. For medium-difficulty tasks, it is important to reasonably schedule working hours, avoid prolonged continuous work, and ensure that workers have sufficient rest time. Since high-difficulty tasks impose the greatest impact on cognitive load, it is advised to carefully allocate such tasks, avoid extended high-load work periods, and ensure the safety and health of workers. Implementing a shift system can help guarantee adequate rest time for workers following demanding tasks.

5.3. Integrated Management

Provide training on cognitive load management to enhance workers’ awareness and self-regulation capabilities, enabling them to better cope with varying work environments and task demands. Regularly monitor workers’ cognitive load using EEG data and other methods to promptly detect cases of excessive load and take appropriate corrective actions. Optimize the work environment to reduce adverse factors that contribute to cognitive load, such as improving lighting, ventilation, and noise reduction, thereby enhancing workers’ comfort and efficiency. For example, in night shift environments, implement a maximum of 4 h of continuous work time with mandatory 30-min cognitive recovery breaks every 2 h; in noisy environments, mandate the use of active noise-cancelling personal protective equipment when ambient noise exceeds 80 decibels, establish low-noise zones for complex work tasks, and implement scheduled worker rotation (at 60–90 min intervals) in environments exceeding 85 decibels; in confined spaces, implement dedicated supervision with one worker specifically assigned to cognitive support/monitoring responsibilities.

5.4. Limitations and Future Directions

In the results for α/β and (α + θ)/β ratios, certain groups (e.g., the confined space group, F = 0.583, p = 0.563) did not show significant effects. Future studies will consider increasing the sample size or further refining the experimental design to enhance statistical power. Although the present study employed absolute EEG band power values to assess cognitive load variations, we acknowledge that normalized metrics—such as relative band power—can enhance robustness against inter-subject variability and improve the generalizability of findings. The repeated-measures design used in this study helped mitigate individual differences to some extent; however, future research could benefit from incorporating normalized EEG indicators to further strengthen the stability and cross-population applicability of the results. Additionally, incorporating more indicators (such as eye-tracking data) into future analyses could improve the precision of predictive outcomes and provide a more comprehensive assessment of the effects of cognitive load on attention and fatigue. Future research may also explore the real-time monitoring of workers’ cognitive load using EEG, eye-tracking, and other data streams, dynamically extracting features. The conclusions of this study can serve as theoretical support for dynamic feature extraction and provide a basis for real-time interventions.

Author Contributions

Conceptualization, Z.G., C.X., S.H. and Y.Y.; Methodology, Z.G.; Formal analysis, H.T.; Investigation, H.T.; Data curation, H.T. and S.H.; Writing—original draft, Z.G. and C.X.; Writing—review & editing, Y.Y.; Visualization, C.X.; Project administration, Y.Y. All authors have read and agreed to the published version of the manuscript.

Funding

This research received no external funding.

Institutional Review Board Statement

Ethical review and approval were waived for this studydue to Article 32 of the “Ethical Review Measures for Life Science and Medical Research InvolvingHumans” (in China), which exempts studies that do not cause harm to individuals, do not inyolvesensitive personal information or commercial interests, and are conducted through observationwithout interfering with public behavior, or research using anonymous data.

Informed Consent Statement

Informed consent was obtained from all subjects involved in the study.

Data Availability Statement

The datasets generated and analyzed during the current study areavailable from the corresponding author upon reasonable request.

Conflicts of Interest

Authors Zongyong Guo and Huadi Tao were employed by the company China Railway 14th Bureau Group Third Engineering Co., Ltd. Author Chengming Xia was employed by the company Sanming Puyan Expressway Co., Ltd. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

References

  1. Li, J.; Zhu, J.; Guan, C. Assessing illumination fatigue in tunnel workers through eye-tracking technology: A laboratory study. Adv. Eng. Inf. 2024, 59, 102335. [Google Scholar] [CrossRef]
  2. Chen, J.; Song, X.; Lin, Z. Revealing the “Invisible Gorilla” in construction: Estimating construction safety through mental workload assessment. Autom. Constr. 2016, 63, 173–183. [Google Scholar] [CrossRef]
  3. Moray, N. Mental Workload: Its Theory and Measurement; Springer Science & Business Media: Berlin/Heidelberg, Germany, 2013; Volume 8. [Google Scholar]
  4. Alotaibi, A.; Gambatese, J. Impact of Task Demand and Physical Energy on Mental Workload Experienced by Construction Workforce: Insights of Construction Professionals. Pract. Period. Struct. Des. Constr. 2024, 29, 04024006. [Google Scholar] [CrossRef]
  5. Jiang, T.; Fang, Y.; Zheng, N.; Chen, J. Understanding construction workers’ cognitive processes under risky scenarios through electroencephalography. Autom. Constr. 2024, 166, 105674. [Google Scholar] [CrossRef]
  6. Bell, M.A.; Cuevas, K. Using EEG to study cognitive development: Issues and practices. J. Cogn. Dev. 2012, 13, 281–294. [Google Scholar] [CrossRef]
  7. Ding, Z.; Xiong, Z.; Ouyang, Y. A Bibliometric Analysis of Neuroscience Tools Use in Construction Health and Safety Management. Sensors 2023, 23, 9522. [Google Scholar] [CrossRef]
  8. Zhang, Y.; Zhang, M.; Fang, Q. Scoping review of EEG studies in construction safety. Int. J. Environ. Res. Public Health 2019, 16, 4146. [Google Scholar] [CrossRef]
  9. Wei, B.a.; Yang, B.; Zhang, W.; Liu, P.; Fu, H.; Lv, Z.; Wang, F. Construction site hazard identification and worker adverse reaction monitoring using electroencephalograms: A review. Buildings 2024, 14, 180. [Google Scholar] [CrossRef]
  10. Saedi, S.; Fini, A.A.F.; Khanzadi, M.; Wong, J.; Sheikhkhoshkar, M.; Banaei, M. Applications of electroencephalography in construction. Autom. Constr. 2022, 133, 103985. [Google Scholar] [CrossRef]
  11. Cheng, B.; Fan, C.; Fu, H.; Huang, J.; Chen, H.; Luo, X. Measuring and computing cognitive statuses of construction workers based on electroencephalogram: A critical review. IEEE Trans. Comput. Social Syst. 2022, 9, 1644–1659. [Google Scholar] [CrossRef]
  12. Lu, M.; Hu, S.; Mao, Z.; Liang, P.; Xin, S.; Guan, H. Research on work efficiency and light comfort based on EEG evaluation method. Build. Environ. 2020, 183, 107122. [Google Scholar] [CrossRef]
  13. Ke, J.; Zhang, M.; Luo, X.; Chen, J. Monitoring distraction of construction workers caused by noise using a wearable Electroencephalography (EEG) device. Autom. Constr. 2021, 125, 103598. [Google Scholar] [CrossRef]
  14. Ouyang, Y.; Huang, G.; He, S. Impacts of adverse environmental factors on construction workers’ attention allocation during hazard identification: A study of noise and heat exposure. Eng. Constr. Archit. Manag. 2024. [Google Scholar] [CrossRef]
  15. Hwang, S.; Jebelli, H.; Choi, B.; Choi, M.; Lee, S. Measuring workers’ emotional state during construction tasks using wearable EEG. J. Constr. Eng. Manag. 2018, 144, 04018050. [Google Scholar] [CrossRef]
  16. Wang, P.; Wu, P.; Wang, J.; Chi, H.-L.; Wang, X. A critical review of the use of virtual reality in construction engineering education and training. Int. J. Environ. Res. Public Health 2018, 15, 1204. [Google Scholar] [CrossRef]
  17. Jeon, J.; Cai, H. Wearable EEG-based construction hazard identification in virtual and real environments: A comparative study. Saf. Sci. 2023, 165, 106213. [Google Scholar] [CrossRef]
  18. Schapkin, S.; Raggatz, J.; Hillmert, M.; Böckelmann, I. EEG correlates of cognitive load in a multiple choice reaction task. Acta Neurobiol. Exp. 2020, 80, 76–89. [Google Scholar] [CrossRef]
  19. DGJ32/TJ 216-2016; Technical Specification for Noise Control in Urban Tunnels. Phoenix Publishing: Nanjing, China, 2016.
  20. JTGT3660-2020; Technical Specification for Highway Tunnel Construction. People’s Transportation Press: Beijing, China, 2013.
  21. Ke, J.; Chen, J.; Luo, X. Monitoring distraction of construction workers using a wearable electroencephalography (EEG) device. In Proceedings of the Creative Construction Conference, Budapest, Hungary, 29 June–2 July 2019; pp. 383–390. [Google Scholar]
  22. Wang, D.; Li, H.; Chen, J. Detecting and measuring construction workers’ vigilance through hybrid kinematic-EEG signals. Autom. Constr. 2019, 100, 11–23. [Google Scholar] [CrossRef]
  23. Xing, X.; Zhong, B.; Luo, H.; Rose, T.; Li, J.; Antwi-Afari, M.F. Effects of physical fatigue on the induction of mental fatigue of construction workers: A pilot study based on a neurophysiological approach. Autom. Constr. 2020, 120, 103381. [Google Scholar] [CrossRef]
  24. Stuss, D.T.; Knight, R.T. Principles of Frontal Lobe Function; American Chemical Society: Washington, DC, USA, 2013. [Google Scholar]
  25. Sukeerthi, S.; Veezhinathan, M.; Geethanjali, B. Analysing the Cognitive load for Monolinguals using EEG. In Proceedings of the 6th 2020 International Conference on Advanced Computing and Communication Systems (ICACCS), Coimbatore, India, 6–7 March 2020; pp. 253–260. [Google Scholar]
  26. Okunola, A.; Abiola Akanmu, P.D.; Jebelli, H. Experimental Analysis of Cognitive Load Risks in Employing Active Back-Support Exoskeleton in Construction. In Proceedings of the 60th Annual Associated Schools, Auburn, AL, USA, 3–5 April 2024; Volume 5, pp. 894–902. [Google Scholar]
  27. Chikhi, S.; Matton, N.; Blanchet, S. EEG power spectral measures of cognitive workload: A meta-analysis. Psychophysiology 2022, 59, e14009. [Google Scholar] [CrossRef]
  28. Al-Hosani, F.I.A.I. Use of Machine Learning to Establish the Role of Cognitive Stress in Postural Sway Based on Multimodal Physiological; Khalifa University of Science: Abu Dhabi, United Arab Emirates, 2024. [Google Scholar]
  29. Wang, Y.; Huang, Y.; Gu, B.; Cao, S.; Fang, D. Identifying mental fatigue of construction workers using EEG and deep learning. Autom. Constr. 2023, 151, 104887. [Google Scholar] [CrossRef]
  30. Chellappa, S.L.; Morris, C.J.; Scheer, F.A. Effects of circadian misalignment on cognition in chronic shift workers. Sci. Rep. 2019, 9, 699. [Google Scholar] [CrossRef]
  31. Astuti, R.D.; Suhardi, B.; Laksono, P.W.; Susanto, N. Investigating the relationship between noise exposure and human cognitive performance: Attention, stress, and mental workload based on EEG signals using power spectrum density. Appl. Sci. 2024, 14, 2699. [Google Scholar] [CrossRef]
Figure 1. HTC Vive.
Figure 1. HTC Vive.
Buildings 15 02920 g001
Figure 2. Tunnel face during construction.
Figure 2. Tunnel face during construction.
Buildings 15 02920 g002
Figure 3. Scattered rebar.
Figure 3. Scattered rebar.
Buildings 15 02920 g003
Figure 4. Rock layer seepage.
Figure 4. Rock layer seepage.
Buildings 15 02920 g004
Figure 5. Improper loading of transport vehicle.
Figure 5. Improper loading of transport vehicle.
Buildings 15 02920 g005
Figure 6. Absolute power of α.
Figure 6. Absolute power of α.
Buildings 15 02920 g006
Figure 7. Absolute power of β.
Figure 7. Absolute power of β.
Buildings 15 02920 g007
Figure 8. Absolute power of θ.
Figure 8. Absolute power of θ.
Buildings 15 02920 g008
Figure 9. α/β ratio.
Figure 9. α/β ratio.
Buildings 15 02920 g009
Figure 10. (α + θ)/βratio.
Figure 10. (α + θ)/βratio.
Buildings 15 02920 g010
Table 1. Parameters of Different Work Environments.
Table 1. Parameters of Different Work Environments.
EnvironmentIllumination (lux)Noise (dB)Space Size
Control group3007510 m × 10 m
Night shift1507510 m × 10 m
Noise3008510 m × 10 m
Confined space300753 m × 3 m
Table 2. ANOVA Results of α Wave Absolute Power.
Table 2. ANOVA Results of α Wave Absolute Power.
Work
Environment
TaskMeanSDF-Value (p)Tukey HSD
ComparisonMean
Difference
p-ValueCI
control groupR1960.621108.66316.108 (0.000)R2 and R1−120.44020.0049(−208.065, −32.816)
R2840.181112.22416.108 (0.000)R3 and R1−203.5750(−291.200, −115.950)
R3757.04669.76316.108 (0.000)R3 and R2−83.13480.0661(−170.760, 4.490)
night shiftR1885.549125.0699.086 (0.001)R2 and R1−137.31380.0014(−225.596, −49.032)
R2748.23589.5769.086 (0.001)R3 and R1−130.75020.0024(−219.032, −42.468)
R3754.79977.7399.086 (0.001)R3 and R26.56360.9822(−81.718, 94.846)
noiseR1872.91391.09721.522 (0.000)R2 and R1−133.85670.0001(−205.565, −62.149)
R2739.05672.52221.522 (0.000)R3 and R1−188.11380(−259.822, −116.406)
R3684.79977.73921.522 (0.000)R3 and R2−54.25710.1697(−125.965, 17.451)
confined spaceR1897.461112.93221.487 (0.000)R2 and R1−139.67730.0003(−218.391, −60.964)
R2757.78374.86721.487 (0.000)R3 and R1−208.40450(−287.118, −129.691)
R3689.05672.52221.487 (0.000)R3 and R2−68.72720.0977(−147.440, 9.986)
Table 3. ANOVA Results of β Wave Absolute Power.
Table 3. ANOVA Results of β Wave Absolute Power.
Work
Environment
TaskMeanSDF-Value (p)Tukey HSD
ComparisonMean
Difference
p-ValueCI
control groupR1413.10451.44611.192 (0.000)R2 and R165.7030.0477(0.561, 130.845)
R2478.80760.38311.192 (0.000)R3 and R1126.82750.0001(61.686, 191.969)
R3539.93199.41411.192 (0.000)R3 and R261.12450.0698(−4.017, 126.266)
night shiftR1524.93199.4146.895 (0.003)R2 and R136.77880.5113(−43.473, 117.031)
R2561.71079.0656.895 (0.003)R3 and R1119.73290.0022(39.481, 199.985)
R3644.66491.7416.895 (0.003)R3 and R282.95410.0414(2.702, 163.206)
noiseR1459.79653.22415.113 (0.000)R2 and R164.87630.0415(2.087, 127.666)
R2524.67377.10215.113 (0.000)R3 and R1141.91350(79.124, 204.703)
R3601.71079.06515.113 (0.000)R3 and R277.03720.013(14.248, 139.827)
confined spaceR1507.79048.5499.463 (0.000)R2 and R150.62550.1522(−14.217, 115.468)
R2558.41666.3999.463 (0.000)R3 and R1115.8070.0003(50.964, 180.650)
R3623.59796.2399.463 (0.000)R3 and R265.18150.0486(0.339, 130.024)
Table 4. ANOVA Results of θ Wave Absolute Power.
Table 4. ANOVA Results of θ Wave Absolute Power.
Work
Environment
TaskMeanSDF-Value (p)Simple Effects Analysis
ComparisonMean
Difference
p-ValueCI
control groupR1599.72597.37228.162 (0.000)R1 and R2−141.2990.0014(−234.810, −47.788)
R2741.02451.74128.162 (0.000)R1 and R3−229.3490.0000(−311.026, −147.672)
R3829.07496.08128.162 (0.000)R2 and R3−88.0500.0049(−153.511, −22.589)
night shiftR1843.741206.9373.753 (0.032)R1 and R231.8610.7896(−61.650, 125.732)
R2811.8878.5763.753 (0.032)R1 and R3−98.7610.0129(−180.438, −17.084)
R3942.50181.3443.753 (0.032)R2 and R3−130.6220.0000(−196.083, −65.161)
noiseR1659.02451.74151.036 (0.000)R1 and R2−99.2010.0343(−192.712, −5.690)
R2758.22597.18151.036 (0.000)R1 and R3−277.0760.0000(−358.753, −195.399)
R3936.09972.50151.036 (0.000)R2 and R3−177.8750.0000(−242.336, −112.414)
confined spaceR1609.963110.85730.500 (0.000)R1 and R2−147.2620.0008(−240.772, −53.7512)
R2757.22597.18130.500 (0.000)R1 and R3−270.1370.0000(−351.814, −188.46)
R3880.09972.50130.500 (0.000)R2 and R3−122.8750.0001(−188.336, −57.413)
Table 5. ANOVA Results of α/β ratio.
Table 5. ANOVA Results of α/β ratio.
Work
Environment
TaskMeanSDF-Value (p)Tukey HSD
ComparisonMean
Difference
p-ValueCI
control groupR12.3610.41721.587 (0.000)R2 and R1−0.56120.0008(−0.903, −0.219)
R21.8000.44021.587 (0.000)R3 and R1−0.9180(−1.260, −0.576)
R31.4430.28021.587 (0.000)R3 and R2−0.35690.0393(−0.699, −0.015)
night shiftR11.7450.42315.546 (0.000)R2 and R1−0.4020.0009(−0.652, −0.152)
R21.3430.14215.546 (0.000)R3 and R1−0.55470(−0.804, −0.305)
R31.1900.19615.546 (0.000)R3 and R2−0.15270.3077(−0.403, 0.097)
noiseR11.9200.28540.037 (0.000)R2 and R1−0.49190(−0.702, −0.282)
R21.4280.18840.037 (0.000)R3 and R1−0.76110(−0.971, −0.551)
R31.1590.22540.037 (0.000)R3 and R2−0.26910.009(−0.479, −0.060)
confined spaceR11.6350.4669.601 (0.000)R2 and R1−0.25620.0829(−0.539, 0.027)
R21.3780.2379.601 (0.000)R3 and R1−0.51030.0002(−0.793, −0.227)
R31.1240.1799.601 (0.000)R3 and R2−0.25410.086(−0.537, 0.029)
Table 6. ANOVA Results of (α + θ)/β ratio.
Table 6. ANOVA Results of (α + θ)/β ratio.
Work
Environment
TaskMeanSDF-Value
(p)
Tukey HSD
ComparisonMean
Difference
p-ValueCI
control groupR13.832 0.659 5.335 (0.009)R2 and R1−0.45410.1601(−1.044, 0.136)
R23.377 0.694 5.335 (0.009)R3 and R1−0.79040.0062(−1.380, −0.200)
R33.041 0.641 5.335 (0.009)R3 and R2−0.33630.3578(−0.926, 0.254)
night shiftR13.423 0.868 7.324 (0.002)R2 and R1−0.61560.0139(−1.122, −0.109)
R22.808 0.268 7.324 (0.002)R3 and R1−0.74690.0025(−1.253, −0.241)
R32.676 0.389 7.324 (0.002)R3 and R2−0.13120.8046(−0.637, 0.375)
noiseR13.375 0.462 9.761 (0.000)R2 and R1−0.47880.0076(−0.844, −0.113)
R22.896 0.372 9.761 (0.000)R3 and R1−0.63880.0003(−1.004, −0.273)
R32.736 0.396 9.761 (0.000)R3 and R2−0.160.5416(−0.526, 0.205)
confined spaceR12.731 0.736 0.583 (0.563)R2 and R10.02280.9924(−0.446, 0.492)
R22.753 0.404 0.583 (0.563)R3 and R1−0.16810.6618(−0.637, 0.301)
R32.563 0.365 0.583 (0.563)R3 and R2−0.19080.5884(−0.660, 0.278)
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Guo, Z.; Xia, C.; Tao, H.; Huang, S.; Yang, Y. Research on Cognitive Load of Tunnel Construction Workers in Different Environments Based on EEG. Buildings 2025, 15, 2920. https://doi.org/10.3390/buildings15162920

AMA Style

Guo Z, Xia C, Tao H, Huang S, Yang Y. Research on Cognitive Load of Tunnel Construction Workers in Different Environments Based on EEG. Buildings. 2025; 15(16):2920. https://doi.org/10.3390/buildings15162920

Chicago/Turabian Style

Guo, Zongyong, Chengming Xia, Huadi Tao, Shoujie Huang, and Yanqun Yang. 2025. "Research on Cognitive Load of Tunnel Construction Workers in Different Environments Based on EEG" Buildings 15, no. 16: 2920. https://doi.org/10.3390/buildings15162920

APA Style

Guo, Z., Xia, C., Tao, H., Huang, S., & Yang, Y. (2025). Research on Cognitive Load of Tunnel Construction Workers in Different Environments Based on EEG. Buildings, 15(16), 2920. https://doi.org/10.3390/buildings15162920

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop