Next Article in Journal
Gradient vs. Non-Gradient-Based Model Free Control Algorithms: Analysis and Applications to Nonlinear Systems
Next Article in Special Issue
Spatial and Temporal Distribution Pattern of Pre-Mining Grouting-Induced Microseismicity and Prediction of Water Inrush
Previous Article in Journal
A Simulation Study on Pressure Control in Oil Well Drilling Using Gain-Scheduled PID Controllers
Previous Article in Special Issue
Study on the Optimization of Coal Pillar Width in Goaf-Side Roadway Under the Synergistic Effect of Mining and Seepage
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

The Impact of Task Interruptions on the Unsafe Behavior of Coal Mine Tunneling Machine Operators: The Moderating Role of Fatigue

1
College of Safety Science and Engineering, Xi’an University of Science and Technology, Xi’an 710054, China
2
Xi’an Key Laboratory of Human Factors & Intelligence for Emergency Safety, Xi’an University of Science and Technology, Xi’an 710054, China
3
College of Management, Xi’an University of Science and Technology, Xi’an 710054, China
*
Authors to whom correspondence should be addressed.
Appl. Sci. 2025, 15(5), 2764; https://doi.org/10.3390/app15052764
Submission received: 15 January 2025 / Revised: 21 February 2025 / Accepted: 3 March 2025 / Published: 4 March 2025
(This article belongs to the Special Issue Novel Technologies in Intelligent Coal Mining)

Abstract

:
Task interruptions and fatigue in high-risk environments such as coal mining significantly affect the safety behavior of coal mine tunneling machine operators. Understanding the cognitive mechanisms underlying unsafe behaviors is crucial for improving workplace safety. This study, based on the executive control and resource allocation theory in the ACT-R cognitive architecture, investigates the impact of task interruptions under fatigue on unsafe behavior and cognitive neural mechanisms in these operators. A dual-perspective analysis of behavioral performance and event-related potentials (ERPs) was employed. The behavioral analysis revealed that fatigue exacerbates the negative effects of task interruptions. Under non-fatigued conditions, individuals compensated for interruptions with longer reaction times, maintaining accuracy without significant decline. However, under fatigue, task interruptions notably reduced accuracy, especially during recovery trials. ERP analysis showed that fatigue impaired cognitive and neural mechanisms that are critical for task performance. Following interruptions, an increase in P200 amplitude and prolonged latency indicated reduced task switching efficiency. Under fatigue, a decline in frontal P300 amplitude over time reflected weakened executive control, while an increase in central P300 amplitude suggested compensatory control mechanisms’ efforts. However, these compensatory control mechanisms were insufficient to counteract the negative impact of fatigue. In conclusion, fatigue-induced impairments in attention shifting, response inhibition, and the imbalance between facilitation and inhibition further exacerbated performance declines after task interruptions. Although compensatory control mechanisms attempted to mitigate these effects through resource reallocation, they were unable to fully counteract fatigue’s negative impact. This study underscores the moderating role of fatigue in the relationship between task interruptions and unsafe behaviors, and highlights the limitations of brain compensatory control mechanisms. These findings offer valuable theoretical insights and practical guidance for optimizing workflow and task design for tunneling machine operators in coal mining operations.

1. Introduction

In recent years, China has actively advanced the intelligent development of coal mines [1], significantly enhancing the mechanization of coal mining operations [2]. As frontline workers at excavation sites [3], coal mine tunneling machine operators are tasked with extensive human–machine interaction, such as operating roadheaders, under hazardous conditions, including coal and gas outbursts, high temperatures, dust, noise, and high humidity [4]. Additionally, these operators must maintain a heightened level of vigilance to monitor abnormal conditions in the roof, floor, and coal walls, and the presence of surrounding personnel near the roadheader [5]. This requires operators to frequently perform or switch between multiple tasks simultaneously [6], with task interruptions being both common and often urgent [7]. Working in such harsh environments, roadheader operators are particularly susceptible to fatigue due to heavy workloads, shift work, and sleep deprivation, which disrupt their circadian rhythms [8]. Evidence from case studies highlights that task interruptions under fatigued conditions can lead to unsafe behaviors among workers. Research further indicates that 90% of accidents in human–machine interaction processes are attributable to unsafe human behaviors [9]. As a result, worker behavior plays a pivotal role in ensuring the safety and efficiency of coal mining operations [10].
Previous studies have independently examined the associations between task interruptions, fatigue, and unsafe behaviors; however, few have considered the combined effects of both factors on unsafe behaviors in coal mining environments. This study aims to fill this gap by investigating the behavioral and neural mechanisms underlying task interruptions and fatigue states. Unsafe behaviors are assessed through behavioral performance, with a focus on analyzing the impact of task interruptions on operators’ safety behaviors in a fatigued state. In existing research, Hakim et al. [11] explored the impact of task-unrelated interruptions on the neural representation of working memory, observing that task interruptions lead to shifts in attention and adjustments in information retrieval mechanisms. Kalgotra et al. [12] found that task interruptions trigger changes in multiple brain regions, with the frontal, temporal, and insular lobes showing the most pronounced associations with interruptions. These findings reveal how task interruptions, by affecting different brain regions, in turn influence operators’ work efficiency and decision-making abilities. This study draws on the ACT-R (Adaptive Control of Thought–Rational) cognitive architecture [13], particularly its concepts of executive control and resource allocation. ACT-R is of significant relevance in studying unsafe behaviors in high-risk environments, as it suggests that task interruptions increase cognitive load, making it difficult for individuals to switch between tasks and slowing down information processing [14]. Furthermore, the theory posits that task interruptions affect attention mechanisms and executive control processes [15]. Building on the ACT-R framework, this study employs the spatial 2-back task as the primary interruption task, alongside simple mathematical problems as interruptive tasks. Additionally, while previous studies have primarily focused on the effects of secondary task interruptions on subsequent performance, less attention has been given to pauses in task flow. This study expands the scope by incorporating research on pause-based interruptions. Furthermore, by using the AX-CPT (Continuous Performance Test) task to induce fatigue in participants, this study explores the effects of task interruptions on brain activity under both fatigued and non-fatigued states, while also investigating the moderating role of fatigue in the task interruption process.

2. Materials and Methods

2.1. Participants

The participants in this experiment were 40 male tunneling machine operators selected from a coal mine in Shaanxi, China. All participants were right-handed, had binocular vision of 1.0 or better, and had no eye diseases such as color blindness or color weakness. They had no behavioral or cognitive impairments, no history of brain injuries, and no hereditary diseases. The average age of the participants was 34.24 ± 1.98 years, and their average work experience was 10.69 ± 1.36 years. All participants voluntarily took part in the experiment and signed an informed consent form after fully understanding the details of the experiment.

2.2. Experimental Design

(1)
Experimental paradigm
The experimental paradigm for this study was developed using E-Prime 3.0 (Psychological Software Tools (PST), Pittsburgh, PA, USA).To induce fatigue in participants, the Continuous Performance Task AX-CPT was employed [16]. This paradigm involves cue stimuli and probe stimuli, using 24 letters as materials, excluding “K” and “Y” due to their similarity to “X”. Each trial presents four letters: the first is a red cue (only “A” is valid), the last is a red probe letter (only “X” is valid), and the middle two are random white distractor letters. The task includes four types of trials. A target trial is defined as A–distractor–distractor–X (70% probability). Three non-target trials (each with a 10% probability) are as follows: an invalid cue–distractor–distractor–valid probe, an invalid cue–distractor–distractor–invalid probe, and a valid cue–distractor–distractor–invalid probe. Each letter is displayed for 200 ms, with a 1000 ms inter-trial interval for responses. Participants press the “f” key for target trials and the “j” key for non-target trials.
The interruption experimental task was designed based on the spatial 2-back working memory paradigm. Tunneling machine operators often need to respond to multiple equipment alarms and sudden environmental changes during their actual work. These task interruptions require them to quickly switch attention and allocate resources. The N-back paradigm, as a classic method for measuring working memory load [17], simulates the cognitive load encountered by tunnel drivers in their work. The 2-back task stimuli consisted of small black squares randomly appearing in one of the eight positions (excluding the center) of a 3 × 3 grid against a white background. Each stimulus was displayed for 500 ms, with an interstimulus interval of 1500 ms [18]. Participants were required to compare the currently presented stimulus with the one presented two steps earlier, to determine whether they matched. Responses were made by pressing the “f” key for “match” and the “j” key for “non-match”. The random location of each stimulus was treated as a “target” memory, requiring participants to maintain and retrieve these memories in the short term to make match decisions. During the execution of the primary 2-back task, an immediate interruption strategy [19] was employed to introduce interruption tasks, which disrupting the continuity of the main task. Two types of interruptions were designed: task interruptions and pause interruption. Task interruptions included two subtypes: simple and complex, illustrated in Figure 1. The simple task interruption involved arithmetic problems within 100, such as determining whether “85−27 < 60” was correct. The complex task interruptions involved arithmetic problems with mixed operations of multiplication, division, addition, and subtraction, such as judging whether “8 × (2 + 12) > 300” was correct. Each task interruption trial was displayed for 2500 ms, with an inter-trial interval of 2000 ms. Pause interruption, on the other hand, required no active operation, but involved prolonged fixation. The pause duration was matched with the total duration of the arithmetic task interruptions, to facilitate comparative analysis of the impact of irrelevant information interference during interruptions on task performance and ERP components. Each interruption included twelve pre-interruption and two recovery trials. In total, there were 312 pre-interruption trials and 48 recovery trials under both task and pause interruption conditions.
(2)
Experimental Procedure
Before the formal experiment, participants were required to thoroughly practice the three task types (task interruption, pause interruption, and baseline non-interruption) to avoid learning effects during the formal experiment and ensure familiarity with all task scenarios. The formal experiment was divided into a fatigue phase and a non-fatigue phase, conducted on separate days in a randomized order. During the fatigue phase, to reliably induce a state of mental fatigue, participants were instructed to continuously perform the AX-CPT task for 100 min [20]. The accuracy of the AX-CPT task was monitored every 20 min, and if the task accuracy within a time segment exceeded 85%, that segment was considered valid for fatigue induction. Before starting the interruption tasks, participants were asked to complete the Stanford Sleepiness Scale (SSS) [21] to assess their current fatigue state. In each phase, participants completed the three task types: task interruption, pause interruption, and baseline non-interruption. This study adopted a within-subjects design. Although no control group was set, potential biases were controlled by ensuring consistency in experimental conditions and conducting pre- and post-measurements. Additionally, all participants performed the experiment under similar environmental conditions to minimize interference from external variables. The experiment adopted a Latin square balanced design to minimize sequence effects. After completing each set of tasks, participants also completed the NASA Task Load Index (NASA-TLX) [22] to assess subjective mental workload. The complete experimental procedure is shown in Figure 2.

2.3. Data Processing

(1)
Data Collection and Processing
The data in this experiment included questionnaire scores, behavioral data, and ERP data. Behavioral data consisted of reaction times (RTs) and accuracy (ACC), recorded by E-prime 3.0 software. Behavioral data were statistically analyzed using SPSS 27.0 (IBM Corporation, Armonk, NY, USA) and E-Prime 3.0.
EEG signals were recorded using the international 10–20 system with a Neuroscan Quick-cap electrode cap containing 64 Ag/AgCl electrodes (NeuroScan, El Paso, TX, USA). The sampling rate was set to 1000 Hz, and a real-time bandpass filter of 0.1–30 Hz was applied [23]. The analysis epoch was set to 1000 ms, including 200 ms before stimulus onset and 800 ms after stimulus onset. Independent component analysis (ICA) was used to remove ocular artifacts, and trials containing motion artifacts were excluded using a 75 µV threshold. EEG data preprocessing was conducted using EEGLAB2020 (Swartz Center for Computational Neuroscience, University of California San Diego, CA, USA). First, unnecessary electrodes (“CB1”, “CB2”, “HEO”, “VEO”, “EKG”, “EMG”, and “TRIGGER”) were removed. For segmented signals, M1-M2 referencing was applied. Signal segments containing artifacts from eye movements (EOG amplitude greater than 50 µV), muscle activity (amplitude greater than 100 µV at other electrodes), or erroneous responses were excluded. ICA was then performed on the remaining signal segments, retaining 30 independent components. Twelve electrodes of interest were selected, including the frontal region (F3, FZ, F4), central region (C3, CZ, C4), parietal region (P3, PZ, P4), and occipital region (PO7, OZ, PO8) electrodes. The P200 time window was selected as 180–280 ms after stimulus presentation, and the P300 time window was selected as 300–450 ms after stimulus presentation, ensuring that the peak points of the waveforms fell within the specified time windows. For the same experimental condition (task interruption group, pause interruption group, and baseline group), trials of the same type were averaged. After artifact removal, the average acceptance rate for the trials used for averaging was 90.32%.
(2)
Data Analysis
For behavioral data, the relationship between task interruption, pause interruption, and fatigue was first examined using a repeated-measures ANOVA with a 2(fatigued, non-fatigued) × 3(simple task interruption, complex task interruption, pause interruption) × 2(pre-interruption trials, recovery trials) design, to evaluate the cumulative effects of fatigue. Subsequently, the relationship between fatigue and task interruption complexity was analyzed using a 2(fatigued, non-fatigued) × 2(simple task interruption, complex task interruption) × 3(pre-interruption trials, interruption trials, recovery trials) repeated-measures ANOVA.
For ERP data, P200 and P300 mean amplitudes and peak latencies were analyzed. A 2(fatigued, non-fatigued) × 3(simple task interruption, complex task interruption, pause interruption) repeated-measures ANOVA was conducted, followed by a 2(fatigued, non-fatigued) × 3(simple task interruption, complex task interruption, pause interruption) ×3(pre-interruption trials, interruption trials, recovery trials) repeated-measures ANOVA, to analyze the differential effects of various interruption types on neural responses before and after interruptions across the frontal, central, and parietal regions.
Two participants’ data were excluded due to excessive eye movement and motor artifacts during the experiment, as well as poor task accuracy. Data from the remaining 38 participants were included in the final analysis. For data that did not meet the sphericity assumption in ANOVA, Greenhouse–Geisser corrections were applied. The significance level was set at p = 0.05, and effect size was reported using partial eta squared ( η p 2 ).

3. Experimental Results

3.1. Analysis of Subjective Scale Scores

(1)
Validation of Fatigue Induction via AX-CPT Task: To verify the effectiveness of fatigue induction in this experiment, the Stanford Sleepiness Scale (SSS) was used. As normality testing indicated that the data did not follow a normal distribution, the Wilcoxon signed-rank test, a non-parametric method, was employed to analyze the pre- and post-task SSS scores. Results showed that the post-AX-CPT task scores (mean = 6.13, SD = 1.12) were significantly higher than the pre-task scores (mean = 2.49, SD = 1.23) (p < 0.001), indicating that the 90 min AX-CPT task successfully induced fatigue in participants.
(2)
Analysis of NASA-TLX Scale Scores: A two-factor repeated-measures ANOVA was conducted on NASA-TLX scores, with the factors of fatigue state (non-fatigue vs. fatigue) and task type (task interruption vs. pause interruption). The results revealed a significant main effect of fatigue (F(1, 48) = 17.59, p < 0.001), with the subjective mental workload significantly higher in the fatigue state than in the non-fatigue state. The main effect of task type was also significant (F(2, 80) = 32.15, p < 0.001), showing that the mental workload during task interruptions was higher than during pause interruptions.

3.2. Behavioral Data Analysis

Descriptive statistics for behavioral performance are shown in Table 1. In the non-fatigue state, the mean RT for the baseline task without interruptions was 692.09 ± 114.10 ms, while the mean RT for the baseline task with interruptions was 813.81 ± 230.64 ms. A significant difference in RT was observed between the two baseline tasks (F(1, 83) = 8.302, p = 0.005, η p 2 = 0.091). For ACC, the mean ACC for the baseline task without interruptions was 94.1 ± 0.5%, compared to 92.8 ± 2.5% for the baseline task with interruptions. However, no significant difference in ACC was found between the two baseline tasks (F(1, 83) = 2.348, p = 0.131). Under fatigue conditions, the mean RT for the non-interrupted baseline task was 751.16 ± 61.11 ms, while that for the interrupted baseline task was 889.38 ± 227.01 ms, with a significant difference between the two tasks (F(1, 83) = 9.234, p = 0.003, η p 2 = 0.10). The mean ACC for the non-interrupted baseline task was 92.3 ± 2.8%, compared to 90.0 ± 1.6% for the interrupted baseline task, also showing a significant difference (F(2, 82) = 17.92, p < 0.01, η p 2 = 0.30).
A 2(fatigued, non-fatigued) × 3(simple task interruption, complex task interruption, pause interruption) × 2(pre-interruption trials, recovery trials) repeated-measures ANOVA was conducted. The results (see Table 2) revealed a significant interaction between fatigue state and interruption type for RT (F(1, 48) = 5.389, p = 0.002, η p 2 = 0.147). Significant interactions were also found between fatigue state and trial type, as well as between interruption type and trial type. Further analysis showed that, in the non-fatigued state, RTs differed significantly in the recovery trials (F(1, 28) = 4.829, p = 0.038, η p 2 = 0.147). Significant differences in RTs were observed in recovery trials across the three interruption types in the fatigued state.(F(1, 28) = 3.449, p = 0.004, η p 2 = 0.11), as well as in ACC (F(1, 28)=20.255, p < 0.01, η p 2 = 0.420). However, only ACC (F(1, 48) = 9.839, p = 0.003, η p 2 = 0.057) showed significant differences in the pre-interruption trials.
To further explore the relationship between fatigue and task interruption complexity, a 2(fatigued, non-fatigued) × 2(simple task interruption, complex task interruption) × 3(pre-interruption trials, interruption trials, recovery trials) repeated-measures ANOVA was performed. The results (see Table 3) showed that, in the non-fatigue condition, task interruption type had a significant main effect on RT (F(1, 36) = 27.143, p < 0.01, η p 2 = 0.24), but no significant effect on ACC (F(1, 36) = 5.714, p = 0.54). This indicates that, despite longer RTs for complex tasks, sufficient cognitive resources and processing capacity maintained high ACC levels. In the fatigue condition, task interruption type significantly affected both RT (F(1, 36) = 15.490, p = 0.04, η p 2 = 0.054), and ACC (F(1, 36) = 12.822, p = 0.007, η p 2 = 0.170), reflecting the detrimental impact of fatigue on cognitive resource allocation and performance. In recovery trials, there was no significant difference in RT (F(1, 36) = 1.739, p = 0.224), but ACC showed a significant decline (F(1, 36) = 48.000, p < 0.01, η p 2 = 0.420).

3.3. Event-Related Potential (ERP) Analysis

(1)
P200
The descriptive statistics for P200 mean amplitude and peak latency are shown in Table 4. A 2(fatigued, non-fatigued) × 3(simple task interruption, complex task interruption, pause interruption) repeated-measures ANOVA revealed that fatigue state significantly affected P200 mean amplitude in the frontal (F(1, 446) = 50.225, p < 0.01, η p 2 = 0.101), central (F(1, 446) = 28.275, p < 0.01, η p 2 = 0.060), and parietal regions (F(1, 446) = 48.817, p < 0.01, η p 2 = 0.099). Interruption type significantly increased P200 amplitude in the frontal region (F(2, 429) = 4.817, p = 0.009, η p 2 = 0.022), while no significant effects were observed in the central (F(2, 445) = 2.813, p = 0.114) or parietal regions (F(2, 429) = 2.867, p = 0.058). A significant interaction between fatigue state and interruption type was observed in all three regions, highlighting the modulation of cognitive resource allocation under fatigue conditions. For P200 peak latency, fatigue state significantly affected the frontal region (F(1, 446) = 5.979, p = 0.015, η p 2 = 0.021), but had no significant effects in the central (F(1, 446) = 16.529, p = 0.784) or parietal regions (F(1, 446) = 0.985, p = 0.322).
To analyze the differences in the mean amplitude and peak latency of the P200 component under varying fatigue states and interruption task types across different trial types, a 2(fatigued, non-fatigued) × 3(simple task interruption, complex task interruption, pause interruption) × 3(pre-interruption trials, interruption trials, recovery trials) repeated-measures ANOVA was conducted for the frontal, central, and parietal regions. The trial type significantly affected the intensification of the mean amplitude of P200 in the frontal region, but the main effects were not significant for the central region (F(2, 445) = 2.813, p = 0.114) or the parietal region (F(2, 445) = 1.585, p = 0.206). An interaction effect between fatigue state and trial type was observed for mean P200 amplitude in the frontal (F(5, 442) = 5.766, p < 0.01, η p 2 = 0.061), central (F(2, 445) = 7.539, p < 0.01, η p 2 = 0.063), and parietal regions (F(2, 429) = 10.723, p < 0.01, η p 2 = 0.048). Further analysis of simple effects revealed that the trial type primarily amplified the P200 amplitude in the frontal region. In the fatigued state, significant differences in P200 amplitude were also noted among trial types in the central and parietal regions. These findings suggest that the frontal region plays a relatively prominent role in cognitive tasks. Additionally, an interaction effect between trial type and interruption type on P200 amplitude was significant in the frontal region (F(7, 440) = 6.530, p < 0.01, η p 2 = 0.094), but not in the central (F(7, 440) = 1.266, p = 0.265) or parietal regions (F(7, 440) = 1.117, p = 0.351). This further underscores the importance of the frontal region in processing these tasks. For P200 peak latency, an interaction effect between fatigue state and trial type was observed in the frontal region (F(2, 429) = 0.911, p = 0.033, η p 2 = 0.014), but not in the central (F(2, 445) = 2.813, p = 0.114) or parietal regions (F(2, 429) = 1.257, p = 0.286). Analysis indicated that differences in peak P200 latency across trial types were significant only in the fatigued state.
(2)
P300
The descriptive statistics for the mean amplitude and peak latency of the P300 component are presented in Table 5. A 2(fatigued, non-fatigued) × 3(simple task interruption, complex task interruption, pause interruption) repeated-measures ANOVA was conducted, and the results are summarized as follows: The main effect of fatigue state on P300 mean amplitude was not significant (frontal: (F(1, 446) = 0.158, p = 0.692); central: (F(2, 445) = 0.042, p = 0.837); parietal: (F(1,446) = 0.927, p = 0.336). However, a significant interaction between fatigue state and interruption type was observed, particularly in the frontal (F(1, 446) = 6.002, p < 0.01, η p 2 = 0.013) and central regions (F(1, 446) = 2.373, p = 0.038, η p 2 = 0.005). Descriptive statistical analysis indicated that under fatigue conditions, the P300 amplitude in the frontal region gradually decreased following task and pause interruptions as the experiment progressed, while the P300 amplitude in the central region increased after task interruptions. For P300 peak latency, the fatigue state significantly affected the frontal region (F(1, 446)=9.573, p = 0.002, η p 2 = 0.021) but had no significant effects on the central (F(1, 446) = 1.154, p = 0.283) or parietal regions (F(1, 446) = 0.038, p = 0.854). The interruption type had a significant effect on P300 peak latency in the central (F(1, 446) = 4.031, p = 0.018, η p 2 = 0.035) and parietal regions (F(1, 446) = 3.226, p = 0.041, η p 2 = 0.017), but not in the frontal region (F(1, 446) = 0.658, p = 0.518).
To further analyze the effects of trial type on P300 mean amplitude and peak latency, a 2(fatigued, non-fatigued) × 3(simple task interruption, complex task interruption, pause interruption) × 3(pre-interruption trials, interruption trials, recovery trials) three-way repeated-measures ANOVA was conducted for the frontal, central, and parietal regions. In the frontal region, trial type had a significant main effect on P300 mean amplitude, and showed a significant interaction with fatigue state (F(2, 445) = 6.002, p < 0.01, η p 2 = 0.163). Further analysis revealed significant differences in P300 amplitudes across trial types under fatigue (F(1, 222) = 10.058, p = 0.004, η p 2 = 0.069), with a notable decline in the recovery trials. Similarly, P300 peak latency showed significant differences across trial types in the fatigue condition (F(1, 222) = 5.264, p = 0.014, η p 2 = 0.026), indicating that fatigue inhibited the redistribution of attention and cognitive resources between trials. In the central region, a significant three-way interaction was found between fatigue state, interruption type, and trial type for P300 mean amplitude (F(1, 222) = 6.254, p = 0.044, η p 2 = 0.027). Further analysis showed that in the non-fatigue condition, P300 mean amplitude in the recovery trials during pause interruption was significantly lower than in the pre-interruption trials (F(1, 222) = 4.589, p = 0.022, η p 2 = 0.022). However, this effect was not observed under fatigue conditions.
The heatmaps and waveform plots for different electrode regions under the fatigued state are shown in Figure 3 and Figure 4. This is consistent with previous studies, such as Chen’s [24], which utilized brain imaging techniques to reveal that the fronto-parietal network plays a crucial role in working memory and attention. As attention shifts and the amount of information in working memory increases, activity in the fronto-parietal network also rises.
Time Window: 180–280 ms: In this time window, the scalp potential distribution for simple task interruption displayed lower electrical activity, especially in the frontal, central, and parietal regions, with reduced negative potentials represented by light blue areas. This indicates that simple task interruptions impose a relatively low load on early attention allocation and perceptual processing. In contrast, complex task interruption showed a broader range of negative potential distributions in the same time window, particularly in the prefrontal and central regions. This suggests that under complex task interruptions, the brain requires more cognitive resources to handle task switching, resulting in larger P200 amplitudes. These findings align with the ANOVA results.
Time Window: 300–450 ms: In this later window, the P300 distribution for simple task interruptions showed reduced positive potential distribution, primarily concentrated in the frontal region. This reflects relatively lower demands on working memory updating and cognitive control. Conversely, complex task interruptions demonstrated higher positive potential distributions, particularly in the central and parietal regions. This corresponds to a larger P300 amplitude, suggesting that complex task interruptions require more cognitive resources for information processing and decision-making. These findings support the ANOVA results, which highlight the significant increase in cognitive load caused by complex task interruptions.
Electrode Region Sensitivity: Under both task interruption conditions, the frontal region (FZ) demonstrated higher sensitivity to complex tasks, particularly within the P300 time window. This aligns with the ANOVA findings, which highlighted significant amplitude changes in the frontal region during complex cognitive tasks. The central region (CZ) and parietal region (PZ) also showed stronger responses under complex task interruption conditions. In particular, the P300 response in the parietal region was significantly enhanced, suggesting that these regions bear a higher cognitive load during information categorization and task recovery. The heatmaps and waveform plots corroborate the ANOVA results, further validating the influence of interruption type and fatigue state on cognitive resource allocation and neural activity in the brain.

4. Discussion

This study explored the moderating effects of fatigue on task and pause interruption by comparing behavioral performance and ERP data in both fatigued and non-fatigued states. From both behavioral and neuroscientific perspectives, it provides an in-depth investigation into the mechanisms through which fatigue impacts behavioral performance during task interruptions. At the behavioral level, the results indicated that task interruptions significantly increased RTs, while their impact on ACC was contingent on the fatigue state. Specifically, ACC was more adversely affected under fatigued conditions compared to non-fatigued states. ERP data revealed that in the fatigued state, participants exhibited significantly reduced P300 amplitudes in the frontal region, alongside a significant increase in P200 amplitudes in the same region. This reflects the impact of task interruption on cognitive resource processing and working memory, which is consistent with previous research [25]. However, further studies have found that fatigue negatively moderates behavioral performance by exacerbating the adverse effects of task interruptions on working memory and overall behavioral performance.
(1)
Fatigue and Task Interruptions: Impacts on Behavioral Performance
From the behavioral performance data, a significant interaction between fatigue state and trial type was observed for RTs, indicating that interruptions led to a decline in post-interruption performance, with fatigue amplifying this negative effect. Subtle differences were noted between task interruptions and pause interruptions, where fatigue had a slightly greater negative impact during pause interruptions than simple task interruptions. These phenomena can be understood through the mediating effects of fatigue and the compensatory control mechanisms.
In the ACT-R cognitive architecture, executive control primarily coordinates and manages the activities of various cognitive modules (e.g., perception, cognition, memory, and motor modules) [26]. Its core functions include regulating attention, managing working memory, suppressing impulsive and erroneous responses, and adapting task switching strategies to meet changing task demands. Executive control achieves optimal cognitive resource allocation and task switching through the coordination of the facilitation system and the inhibition system [27]. During task interruptions, the facilitation and inhibition systems collaborate to regulate cognitive and behavioral responses. The facilitation system accelerates responses to new tasks by enhancing arousal levels and attention focus, possibly via neurotransmitter release. Meanwhile, the inhibition system reduces interference and protects working memory for the primary task.
When an interruption occurs, executive control’s coordination of cognitive resources is disrupted, and the balance between the facilitation and inhibition systems is compromised, reducing the efficiency of reactivating the primary task, and thereby impairing behavioral performance. Previous studies have shown that enhancing system performance can lead to the neglect of relevant cues, resulting in longer response times and increased error rates in completing the main task [28]. Fatigue further diminishes allocation efficiency, weakening the activation of the facilitation system and making the inhibition system more active. This overactivation hinders the ability to filter irrelevant stimuli, impairing information processing, which leads to longer RTs and decreased ACC after interruptions.
In the non-fatigued group, “compensatory control mechanisms” [29] were observed. When the executive control system is overloaded and cognitive resources such as working memory capacity are impaired, individuals may unconsciously activate compensatory control mechanisms to counteract this damage. These compensatory control mechanisms work by adjusting execution strategies, prioritizing important tasks, or adopting more effort-efficient approaches to maintain behavioral performance [30]. Under non-fatigued conditions, significant differences in RTs between task interruptions and pause interruptions were only observed in the recovery trials. The compensatory control mechanisms effectively redistributed cognitive resources in response to the attentional shifts caused by interruptions. This redistribution and accelerated processing resulted in longer RTs, but participants were able to maintain relatively high ACC. This phenomenon demonstrates that the executive control system can sustain cognitive capabilities at a relatively high level through adaptive resource allocation strategies, although its capacity for regulation is finite.
In the fatigued state, the facilitation system was weakened, and complex task interruptions induced additional stress and frustration, further diminishing the system’s effectiveness. Behavioral performance was more severely impaired under these conditions. Complex interruptions, which involve more information and interference factors, required participants to allocate more inhibitory resources to suppress distractions. Fatigue increased the load on the inhibition system, reducing its efficiency and impairing its ability to filter irrelevant information. This supports the conclusion that fatigue primarily disrupts the balance between the facilitation and inhibition systems. Moreover, it highlights the limited capacity of the executive control system to effectively adjust between primary and interrupting tasks, which in turn affects behavioral performance.
During pause interruptions under fatigue, participants were in a relatively task-free state, which more readily activated the inhibition system. Extended pauses, often characterized by monotony and a lack of challenge, induced boredom and drowsiness [31], further diverting attention from the primary task. Combined with accumulated fatigue, the inhibition system increasingly suppressed cognitive engagement. This coordination mechanism may represent a rest signal to avoid overexertion and maintain homeostasis. As fatigue accumulated, the inhibition system’s role became more pronounced, leading to further declines in task performance. Consequently, when the primary task resumed, participants required more time to reactivate relevant cognitive resources, resulting in a larger increase in RTs.
In contrast, during simple task interruptions, although the interruptions themselves might have increased cognitive load, participants were able to maintain a certain level of cognitive alertness to respond to the interruption [32]. This alertness partially activated the facilitation system, helping individuals to sustain or improve task performance under resource-constrained conditions. This mitigated the negative impact of mental fatigue on cognitive abilities, allowing participants to more quickly recover primary task performance after interruptions. Thus, despite the effects of fatigue, the increase in RTs after simple task interruptions was relatively small.
(2)
ERP Evidence of Fatigue-Induced Cognitive Decline: P200 and P300 as Key Indicators
The ERP analysis revealed that fatigue significantly affects the amplitude and latency of P200 and P300, leading to a decline in task processing efficiency. Following task interruptions, the increase in P200 amplitude and prolongation of its latency reflect the depletion of cognitive resources. Under fatigue conditions, the decrease in frontal P300 amplitude over the course of the experiment indicates impaired executive control, reduced attentional allocation, and diminished working memory updating capacity. Although the increase in central P300 amplitude may reflect a compensatory control mechanism, it fails to effectively offset the negative effects of fatigue.
Fatigue state significantly affected the average amplitude and peak latency of the P200 component. The effects of trial type and task type on the P200 amplitude and latency were primarily observed in the frontal region. Trial type significantly altered the P200 amplitude in the frontal area, and changes in P200 amplitude are associated with early attentional allocation and active maintenance during working memory updating in the prefrontal cortex (PFC) [25], indicating the role of the executive control system in coordinating cognitive resources after an interruption occurs [33]. Under non-fatigued conditions, the P200 amplitude in the frontal region increased after both task and pause interruptions, suggesting a quick response of the facilitation system to task switching [34]. Notably, pause interruptions provide individuals with a brief “cognitive buffer”, allowing the facilitation system to respond rapidly to task switching, resulting in a slightly higher P200 amplitude compared to simple task interruptions [35]. Under fatigue, when cognitive resources are nearly depleted, during pause interruptions, the inhibition system further reduces cognitive resource allocation, making it difficult for individuals to reallocate attention to the task. As a result, the P200 amplitude decreases, which is consistent with the behavioral data and aligns with previous research findings.
Fatigue and trial type showed a significant interaction effect on P200 latency in the frontal region. Generally, P200 latency is a measure of stimulus classification speed, and is used as an indicator of neural efficiency and processing speed. Under fatigued conditions, increased P200 latency suggested a slight decrease in information processing speed [36]. Functionally, the frontal region is involved in generating attention-switching control signals, and shorter P200 latencies reflect stronger task switching abilities [37]. Task interruptions inherently involve switching between primary and interrupting tasks. Under non-fatigued conditions, P200 latency in the central and parietal regions decreased slightly after pause interruptions compared to pre-interruption trials, likely because participants could prepare and rehearse for the subsequent task during the pause. However, this preparatory advantage was not observed under fatigued conditions. The significant increase in post-interruption latency under fatigue aligns with the increase in RTs observed in the behavioral data. The increase in P200 amplitude and latency under fatigue indicates that the executive control system was actively adjusting strategies to address the challenges posed by fatigue, expending additional effort to maintain task stability and ACC. However, due to resource depletion and reduced efficiency of the facilitation system, behavioral performance further deteriorated.
Fatigue did not significantly affect the mean amplitude of P300 overall, but its interaction with task type resulted in distinct neural response patterns under specific tasks. Fatigue significantly affected the peak latency of P300 in the frontal region, but not in other regions. The P300 component reflects attention reallocation to task-relevant events and their evaluation [38]. In the frontal region, fatigue, task type, and trial type demonstrated significant interaction effects for P300 amplitude. Under fatigued conditions, P300 amplitude decreased as the experiment progressed, following task and pause interruptions. Decreased frontal P300 amplitude indicates impaired executive control, with reduced capacity for working memory updating and attentional resource allocation [39]. Fatigue exacerbated these impairments [40], weakening inhibition functions and the ability to suppress irrelevant information [41], which is a typical top-down process requiring active executive control. This is consistent with the earlier findings.
In the central region, P300 amplitude showed a significant interaction between fatigue and task type. Under fatigued conditions, P300 amplitude increased with experimental progression in both types of task interruptions. This suggests that as fatigue accumulated, the frontal region failed to allocate sufficient resources to meet task demands [42], leading to a decline in active control ability [43]. The brain compensated by shifting more attention to the central region to maintain task performance. Previous research has shown that the central region is associated with perceiving task-relevant information and processing automatic responses [44]. Therefore, under fatigued conditions, individuals may rely more on automation to handle tasks and meet lower-level task demands. Fatigue and trial type also showed a significant interaction effect for P300 latency. Under non-fatigued conditions, P300 latency decreased slightly post-interruption, but this phenomenon was absent under fatigue. P300 latency reflects attentional focus during the information processing stage, and is functionally used to evaluate attention and memory capabilities [45].
Consistently with previous research, under fatigued conditions, decreased P300 amplitude post-interruption indicates impaired executive control, slower decision-making adjustments, and increased risk of critical failures. Further studies have found that the differing changes in P300 amplitude between the frontal and central regions reveal the complex effects of task interruptions on response inhibition under fatigue. Decreased P300 amplitude in the frontal region reflects weakened inhibition function, while increased P300 amplitude in the central region represents a compensatory control mechanism, where the brain reallocates attention to maintain task execution. However, this compensation does not always effectively counteract the negative effects of fatigue. Changes in cognitive resource allocation strategies under fatigue may impair efficiency and precision, increasing the likelihood of unsafe behaviors.
(3)
Mechanisms of Unsafe Behaviors Under Fatigue: the limitations of the compensation mechanism.
Task interruptions under fatigue increase the risk of unsafe behaviors by impairing the functioning of the executive control system and disrupting the coordination of cognitive resources. As analyzed earlier, after a task interruption, individuals must exert greater cognitive effort to resume task execution. This involves the facilitation system supporting task switching, and the inhibition system reducing interference from irrelevant information and errors. In non-fatigued states, increases in P200 and P300 amplitudes provide evidence for these mechanisms. From the observed increase in P200 latency and RTs, it is apparent that the compensatory control mechanisms made efforts to mitigate the negative effects of task interruptions through strategies such as increased cognitive effort and adjustments in execution strategies, partially offsetting the adverse impacts. However, fatigue exacerbates the cognitive load required for task switching, demanding more cognitive resources to re-engage in the task—resources that are nearly depleted in the fatigued state. The difficulty in reallocating cognitive resources after task interruptions directly impacts the timeliness and ACC of decision-making, as evidenced by significant changes in RTs and ACC under fatigue. Changes in neural components reflect the detrimental effects of fatigue on the executive control system. The increase in P200 amplitude and prolongation of its latency indicate a slower response to task switching, while the reduction in the frontal region P300 amplitude and the extension of its latency further reveal weakened executive control. This is particularly pronounced during complex task interruptions, where overactivation of the inhibition system impairs its ability to effectively suppress irrelevant information and distractions, further affecting individuals’ judgment and response efficiency during tasks. Meanwhile, the increased P300 amplitude in the central region suggests that the brain attempts to compensate for this imbalance. However, this compensatory control mechanism fails to adequately counteract the lack of cognitive resources, leading to reduced efficiency in information processing and task execution. Consequently, this inefficiency increases the likelihood of unsafe behaviors.
In this study, the experimental conditions were relatively controlled, whereas task interruptions in actual tunneling operations are more complex and diverse. The characteristics of interruptions may vary significantly depending on task requirements and stress intensity. Future research could expand to multitasking or high-risk scenarios to investigate the impact of fatigue on different occupational groups and enhance the ecological validity of the findings. Fatigue is typically a cumulative process, while this study primarily focused on the effects of short-term fatigue on task interruptions. Longitudinal studies could be conducted to examine the dynamic changes in behavior and neurophysiological mechanisms associated with long-term fatigue (e.g., sleep deprivation or chronic fatigue syndrome), and explore its sustained impact on executive control and resource allocation systems. Although this study primarily employed ERP data to analyze neurophysiological mechanisms, fatigue may also induce other physiological and psychological changes, such as heart rate variability (HRV), electrodermal activity (EDA), or subjective fatigue ratings. Integrating multimodal data could provide a more comprehensive depiction of the effects of fatigue on individual cognition and behavior. In addition, future research could further explore the moderating effects of non-cognitive factors, such as work environment and social support, on miner fatigue. Based on the neurophysiological and behavioral characteristics revealed in this study, fatigue state prediction models could be developed. Coupled with machine learning algorithms, intelligent monitoring systems could be designed to monitor fatigue states in real time and issue alerts in high-risk tasks, thereby reducing the risks associated with fatigue-induced accidents.

5. Conclusions

This study, based on the executive control and resource allocation theory in the ACT-R cognitive architecture, examined the moderating effects of fatigue on task interruptions from both behavioral and neuroscientific perspectives, and explored the mechanisms through which task interruptions influence unsafe behavior in coal mine tunneling machine operators. The results show that fatigue amplifies the adverse effects of task interruptions on performance by impairing proactive and reactive control, as evidenced by decreased ACC and prolonged RTs, particularly during complex interruptions. ERP analysis revealed that fatigue enhanced P200 amplitude and prolonged its latency, while P300 amplitude decreased in the frontal region and increased in the central region. These findings indicate fatigue-induced impairments in attentional shifting, response inhibition, and an imbalance between facilitation and inhibition systems, contributing to post-interruption performance declines. Although compensatory control mechanisms attempted to reallocate resources to sustain task performance, they were insufficient to fully counteract these effects. These findings provide practical guidance for workflow and shift scheduling in the mining industry, particularly in reducing task interruptions and optimizing the work environment. At the same time, they support the development of fatigue monitoring systems and strategies to mitigate risks in high-stakes environments. Furthermore, future research could further explore the moderating effects of non-cognitive factors, such as work environment and social support, on miner fatigue.

Author Contributions

Conceptualization, G.S. and S.T.; methodology, G.S.; software, G.S. and F.T.; validation, F.T. and L.C.; formal analysis, G.S. and Y.Z.; data curation, G.S.; writing—original draft, G.S. and S.T.; writing—review and editing, G.S., F.T., Y.Z., and L.C.; visualization, G.S. and F.T.; supervision, Y.Z.; project administration, S.T. and L.C. All authors have read and agreed to the published version of the manuscript.

Funding

This work was supported by the National Natural Science Foundation of China (grant numbers 51874237, U1904210), the National Social Science Foundation of China (grant number 20XGL025).

Institutional Review Board Statement

This study was conducted in accordance with the ethical principles outlined in the [Declaration of Helsinki/other relevant guidelines], and was approved by the Ethics Committee of the 521 Hospital, China North Industries Group Corporation (Approval Number: 2023-56487, Date: 25 October 2023).

Informed Consent Statement

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

Data Availability Statement

The data used to support the findings of this study are available from the corresponding author upon request (18120089018@stu.xust.edu.cn).

Conflicts of Interest

The authors declare that there are no conflicts of interest regarding the publication of this paper.

References

  1. Tian, S.C.; Mao, J.R.; Li, H.X. Disaster-Causing Mechanism of Hidden Disaster-Causing Factors of Major and Extraordinarily Serious Gas Explosion Accidents in Coal Mine Goafs. Sustainability 2022, 14, 12018. [Google Scholar] [CrossRef]
  2. Chen, Y.; Lu, C.W.; Tian, S.C.; Zhang, H. Monitoring and Detecting Coal Miners’ Fatigue Status Using MPA-LSSVM in the Vision of Smart Mine. Process Saf. Environ. Prot. 2023, 179, 774–783. [Google Scholar] [CrossRef]
  3. Guo, J.; Wang, X.D.; Deng, Q.G. Development and Practice of Comprehensive Safety Protection Facilities for Tunneling Machine Drivers. J. Henan Polytech. Univ. Nat. Sci. 2020, 39, 24–29. [Google Scholar]
  4. Li, H.X.; Fan, H.Z.; Zhang, J.Q.; Chen, L.; Tian, F.Y. Study on Influencing Factors of Human Error in Smart Mines. J. Xi’an Univ. Sci. Technol. 2021, 41, 1090–1097. [Google Scholar]
  5. Kvarekova, E.; Tomakova, M.; Sabadka, D.; Šofranko, M.; Zelenák, Š. Evaluation and Risk Factors of Roadheaders in Coal Mines. Manag. Syst. Prod. Eng. 2021, 29, 242–250. [Google Scholar]
  6. Liebowitz, J. Interruption Management: A Review and Implications for IT Professionals. IT Prof. 2011, 13, 44–48. [Google Scholar] [CrossRef]
  7. Wilson, M.D.; Farrell, S.; Visser, T.A.W.; Loft, S. Remembering to Execute Deferred Tasks in Simulated Air Traffic Control: The Impact of Interruptions. J. Exp. Psychol. Appl. 2018, 24, 360–379. [Google Scholar] [CrossRef]
  8. Tian, F.Y. fNIRS Brain Functional Connectivity Characteristics and Classification Identification of Coal Miners’ Situational Awareness. Master’s Thesis, Xi’an University Of Science And Technology, Xi’an, China, 2022. [Google Scholar]
  9. Tian, S.C.; Shao, G.T.; Li, H.X.; Yang, P.; Dang, Q.; Yuan, F. Research on Management and Control of Miners’ Unsafe Behavior Based on Gray Theory. Shock Vib. 2021, 2021, 12. [Google Scholar] [CrossRef]
  10. Chen, L.; Li, H.X.; Zhao, L.; Zhang, H. The Effect of Job Satisfaction Regulating Workload on Miners’ Unsafe State. Sci. Rep. 2022, 12, 12345. [Google Scholar] [CrossRef]
  11. Hakim, N.; Feldmann-Wüsterfeld, T.; Awh, E.; Vogel, E.K. Perturbing Neural Representations of Working Memory with Task-Irrelevant Interruption; MIT Press: Cambridge, MA, USA, 2020; Volume 3. [Google Scholar]
  12. Kalgotra, P.; Sharda, R.; McHaney, R. Don’t Disturb Me! Understanding the Impact of Interruptions on Knowledge Work: An Exploratory Neuroimaging Study. Inf. Syst. Front. 2019, 21, 1019–1030. [Google Scholar] [CrossRef]
  13. Stewart, T.C.; West, R.L. Deconstructing and Reconstructing ACT-R: Exploring the Architectural Space. Cogn. Syst. Res. 2007, 8, 227–236. [Google Scholar] [CrossRef]
  14. Lebiere, C.; Anderson, J.R.; Bothell, D. Behavioral Representation, Multi-Tasking and Cognitive Workload in an ACT-R Model of a Simplified Air Traffic Control Task. In Proceedings of the Tenth Conference on Computer Generated Forces and Behavioral Representation, Norfolk, VA, USA, 15–17 May 2001. [Google Scholar]
  15. Miyake, A.; Friedman, N.P. The Nature and Organization of Individual Differences in Executive Functions: Four General Conclusions. Curr. Dir. Psychol. 2012, 21, 8–14. [Google Scholar] [CrossRef]
  16. Linden, D.V.D.; Massar, S.A.A.; Schellekens, A.F.A.; Ellenbroek, B.A.; Verkes, R.-J. Disrupted Sensorimotor Gating Due to Mental Fatigue: Preliminary Evidence. Int. J. Psychophysiol. 2006, 62, 168–174. [Google Scholar] [CrossRef] [PubMed]
  17. Owen, A.M.; McMillan, K.M.; Laird, A.R.; Bullmore, E. N-back Working Memory Paradigm: A Meta-Analysis of Normative Functional Neuroimaging Studies. Hum. Brain Mapp. 2010, 25, 46–59. [Google Scholar] [CrossRef]
  18. Brouwer, A.M.; Hogervorst, M.A.; Van Erp, J.B.F.; Heffelaar, T.; Zimmerman, P.H.; Oostenveld, R. Estimating Workload Using EEG Spectral Power and ERPs in the N-back Task. J. Neural Eng. 2012, 9, 045008. [Google Scholar] [CrossRef] [PubMed]
  19. McFarlane, D. Comparison of Four Primary Methods for Coordinating the Interruption of People in Human-Computer Interaction. Hum.-Comput. Interact. 2002, 17, 63–139. [Google Scholar] [CrossRef]
  20. Marcora, S.M.; Staiano, W.; Manning, V. Mental Fatigue Impairs Physical Performance in Humans. J. Appl. Physiol. 2009, 106, 857–864. [Google Scholar] [CrossRef]
  21. Hoddes, E.; Dement, W.C.; Zarcone, V. The Development and Use of the Stanford Sleepiness Scale (SSS). Psychophysiology 1972. [Google Scholar] [CrossRef]
  22. Hart, S.G.; Staveland, L.E. Development of NASA-TLX (Task Load Index): Results of Empirical and Theoretical Research. Adv. Psychol. 1988, 52, 139–183. [Google Scholar]
  23. Hruby, T.; Marsalek, P. Event-Related Potentials—The P3 Wave. Acta Neurobiol. Exp. 2003, 63, 55–63. [Google Scholar] [CrossRef]
  24. Chen, Y.N. Working Memory in N-back Tasks: ERP Studies. Ph.D. Dissertation, University of Warwick, Coventry, UK, 2007. [Google Scholar]
  25. Lenartowicz, A.; Escobedo-Quiroz, R.; Cohen, J.D. Updating of Context in Working Memory: An Event-Related Potential Study. Cogn. Affect. Behav. Neurosci. 2010, 10, 298–315. [Google Scholar] [CrossRef] [PubMed]
  26. Tang, G.Z.; Hu, Y.J.; Zhou, X.M.; Yang, G. Theory and Application of ACT-R Cognitive Architecture. J. Comput. Sci. Explor. 2014, 8, 10. [Google Scholar]
  27. Braver, T.S.; Kizhner, A.; Tang, R.; Freund, M.C.; Etzel, J.A. The Dual Mechanisms of Cognitive Control Project. J. Cogn. Neurosci. 2021, 33, 1990–2015. [Google Scholar] [CrossRef] [PubMed]
  28. Pero, C.L.; Valacich, J.S.; Vessey, I. The Influence of Task Interruption on Individual Decision Making: An Information Overload Perspective. Decis. Sci. 2010, 30, 337–360. [Google Scholar]
  29. Hockey, G.R. Compensatory Control in the Regulation of Human Performance Under Stress and High Workload: A Cognitive-Energetical Framework. Biol. Psychol. 1997, 45, 73–93. [Google Scholar] [CrossRef]
  30. Trafton, G.J.; Monk, C.A. Task Interruptions. Rev. Hum. Factors Ergon. 2007, 3, 111–126. [Google Scholar] [CrossRef]
  31. Speier, C.; Vessey, I.; Valacich, J.S. The Effects of Interruptions, Task Complexity, and Information Presentation on Computer-Supported Decision-Making Performance. Decis. Sci. 2003, 34, 771–797. [Google Scholar] [CrossRef]
  32. Chen, Y.Y.; Fang, W.N.; Guo, B.Y.; Bao, H. The Impact of Task Interruptions on Performance and the Moderating Role of Mental Fatigue. Acta Psychol. Sin. 2023, 55, 14. [Google Scholar] [CrossRef]
  33. Gábor, S.; Jan, K.E.; István, C. Visual Mismatch Negativity: A Predictive Coding View. Front. Hum. Neurosci. 2014, 8, 666. [Google Scholar]
  34. Potts, G.F. An ERP Index of Task Relevance Evaluation of Visual Stimuli. Brain Cogn. 2004, 56, 5–13. [Google Scholar] [CrossRef]
  35. Luck, S.J. An Introduction to the Event-Related Potential Technique; MIT Press: Cambridge, MA, USA, 2005. [Google Scholar]
  36. Wongupparaj, P.; Sumich, A.; Wickens, M.; Kumari, V.; Morris, R.G. Individual Differences in Working Memory and General Intelligence Indexed by P200 and P300: A Latent Variable Model. Biol. Psychol. 2018, 139, 96–105. [Google Scholar] [CrossRef] [PubMed]
  37. Kliegl, O.; Pastötter, B.; Bäuml, K.H.T. Contributions of Encoding and Retrieval Processes to Proactive Interference. J. Exp. Psychol. Learn. Mem. Cogn. 2015, 41, 1631–1639. [Google Scholar] [CrossRef] [PubMed]
  38. Cui, T.; Wang, P.P.; Liu, S.; Zhang, X. P300 Amplitude and Latency in Autism Spectrum Disorder: A Meta-Analysis. Eur. Child Adolesc. Psychiatry 2017, 26, 523–534. [Google Scholar] [CrossRef]
  39. Xiao, Y.X. P300 and Cognitive Processing: Methods, Mechanisms, and Applications. Chin. J. Health Psychol. 2015, 23, 6. [Google Scholar]
  40. Fan, X.L.; Zhao, C.Y.; Luo, H.; Zhang, W. Objective Evaluation of Mental Fatigue Based on ERP Characteristics Under 2-back Task. J. Biomed. Eng. 2018, 35, 837–844. [Google Scholar]
  41. Zickerick, B.; Kobold, S.O.; Thnes, S.; Küper, K.; Wascher, E. Don’t Stop Me Now: Hampered Retrieval of Action Plans Following Interruptions. Psychophysiology 2021, 58, e13725. [Google Scholar] [CrossRef]
  42. Zhang, Y. Effects of Mental Fatigue on Attention Characteristics. Master’s Thesis, Fourth Military Medical University, Xi’an, China, 2009. [Google Scholar]
  43. Faber, L.G.; Maurits, N.M.; Lorist, M.M. Mental Fatigue Affects Visual Selective Attention. PLoS ONE 2012, 7, e48073. [Google Scholar] [CrossRef]
  44. Boksem, M.A.S.; Meijman, T.F.; Lorist, M.M. Effects of Mental Fatigue on Attention: An ERP Study. Cogn. Brain Res. 2005, 25, 107–116. [Google Scholar] [CrossRef]
  45. Pitt, K.M.; Cole, Z.J.; Zosky, J. Promoting Simple and Engaging Brain-Computer Interface Designs for Children by Evaluating Contrasting Motion Techniques. J. Speech Lang. Hear. Res. 2023, 66, 1234–1245. [Google Scholar] [CrossRef]
Figure 1. Experimental paradigm of task interruption based on 2-back experimental paradigm.
Figure 1. Experimental paradigm of task interruption based on 2-back experimental paradigm.
Applsci 15 02764 g001
Figure 2. Experimental procedure for experiment.
Figure 2. Experimental procedure for experiment.
Applsci 15 02764 g002
Figure 3. Brain topographic maps of P200 and P300 during fatigue state for each task.
Figure 3. Brain topographic maps of P200 and P300 during fatigue state for each task.
Applsci 15 02764 g003
Figure 4. Waveform charts of P200 and P300 during fatigue state for each task.
Figure 4. Waveform charts of P200 and P300 during fatigue state for each task.
Applsci 15 02764 g004
Table 1. Descriptive statistics analysis results of behavioral performance under different conditions (mean ± standard deviation).
Table 1. Descriptive statistics analysis results of behavioral performance under different conditions (mean ± standard deviation).
Task TypeRTs (ms)ACC (%)Trial TypeRTs (ms)ACC (%)
Non-
Fatigue State
Simple749.53 ± 182.6393.1 ± 2.0Pre-Interruption655.65 ± 84.8994.5 ± 0.8
Interruption1078.80 ± 61.8792.0 ± 1.0
Recovery701.92 ± 119.9190.0 ± 1.0
Complex878.09 ± 258.1592.5 ± 2.9Pre-Interruption724.33 ± 78.9294.7 ± 0.7
Interruption1350.74 ± 46.6290.0 ± 1.5
Recovery866.74 ± 88.4788.6 ± 1.1
Pause692.09 ± 114.1094.3 ± 3.0Pre-Interruption687.24 ± 143.7495.6 ± 0.4
Recovery698.56 ± 59.1892.3 ± 1.0
Fatigue StateSimple841.7 ± 213.4890.0 ± 1.5Pre-Interruption739.76 ± 81.5291.6 ± 0.6
Interruption1220.37 ± 30.9389.8 ± 0.2
Recovery769.10 ± 156.308801 ± 0.2
Complex937.0 ± 234.3589.0 ± 1.0Pre-Interruption806.78 ± 50.7490.8 ± 0.3
Interruption1355.67 ± 70.3788.3 ± 0.9
Recovery909.05 ± 178.5886.9 ± 2.7
Pause782.6 ± 156.3292.0 ± 2.5Pre-Interruption784.24 ± 46.7493.1 ± 1.2
Recovery806.56 ± 20.4991.0 ± 4.8
Table 2. Results of 2 × 3 × 2 three-way ANOVA analysis for behavioral performance.
Table 2. Results of 2 × 3 × 2 three-way ANOVA analysis for behavioral performance.
AnovaReaction Time (RT)Accuracy (ACC)
EffectFpFp
Fatigue State11.646<0.01 *28.385<0.01 *
Interruption Type3.9440.0590.0090.926
Trial Type5.2920.023 *67.743<0.01 *
Fatigue State × Interruption Type5.3890.002 *13.9390.675
Fatigue State × Trial Type4.2550.049 *49.2310.03 *
Interruption Type × Trial Type4.6760.004 *34.516<0.01 *
Three-Way Interaction3.8480.35640.7570.256
Note: * indicates that the effect is significant, i.e., p < 0.05.
Table 3. Results of 2 × 2 × 3 three-way ANOVA analysis for behavioral performance.
Table 3. Results of 2 × 2 × 3 three-way ANOVA analysis for behavioral performance.
AnovaReaction Time (RT)Accuracy (ACC)
EffectFpFp
Fatigue State2.7260.10242.193<0.01 *
Task Interruption Type6.1510.015 *2.4860.118
Trial Type67.821<0.01 *59.002<0.01 *
Fatigue State × Task Interruption Type3.0780.031 *15.578<0.01 *
Fatigue State × Trial Type76.848<0.01 *129.311<0.01 *
Task Interruption Type × Trial Type106.703<0.01 *26.864<0.01 *
Three-Way Interaction6.0430.2459.3530.511
Note: * indicates that the effect is significant, i.e., p < 0.05.
Table 4. Descriptive statistics results of mean P200 amplitude and peak latency (mean ± standard deviation).
Table 4. Descriptive statistics results of mean P200 amplitude and peak latency (mean ± standard deviation).
Task TypeTrial TypeP200 Mean Amplitude (μV)P200 Peak Latency (ms)
Frontal RegionCentral RegionParietal RegionFrontal RegionCentral RegionParietal Region
Non-
Fatigue State
SimplePre-Interruption1.966 ± 1.691.706 ± 1.471.422 ± 1.12170 ± 12172 ± 16171 ± 12
Interruption2.242 ± 1.891.896 ± 1.671.618 ± 1.56171 ± 15176 ± 14172 ± 14
Recovery2.363 ± 1.991.957 ± 2.261.639 ± 1.30176 ± 14176 ± 14176 ± 15
ComplexPre-Interruption2.871 ± 2.501.914 ± 1.771.529 ± 1.33175 ± 14173 ± 14170 ± 12
Interruption3.052 ± 2.492.178 ± 2.141.657 ± 1.13175 ± 16178 ± 13176 ± 14
Recovery3.037 ± 2.412.488 ± 1.921.715 ± 1.13176 ± 16181 ± 16176 ± 14
PausePre-Interruption2.173 ± 1.631.791 ± 1.201.175 ± 1.12171 ± 16173 ± 15174 ± 15
Recovery2.392 ± 1.661.755 ± 1.541.249 ± 0.93172 ± 14172 ± 14174 ± 15
Fatigue StateSimplePre-Interruption2.826 ± 1.162.147 ± 1.452.972 ± 2.41173 ± 14167 ± 10174 ± 13
Interruption4.320 ± 2.692.675 ± 1.472.650 ± 1.69178 ± 16175 ± 14175 ± 14
Recovery4.333 ± 1.872.792 ± 1.893.366 ± 2.28176 ± 15176 ± 10178 ± 15
ComplexPre-Interruption4.216 ± 2.922.769 ± 2.582.106 ± 1.16178 ± 15178 ± 14175 ± 16
Interruption5.304 ± 3.882.947 ± 1.652.765 ± 1.67178 ± 15181 ± 14181 ± 11
Recovery5.812 ± 3.933.392 ± 2.992.765 ± 1.25182 ± 12181 ± 15181 ± 15
PausePre-Interruption4.316 ± 2.992.046 ± 2.232.037 ± 1.98172 ± 15173 ± 15173 ± 13
Recovery3.948 ± 1.972.273 ± 1.882.038 ± 1.78176 ± 14173 ± 16169 ± 14
Table 5. Descriptive statistics results of mean P300 amplitude and peak latency (mean ± standard Deviation).
Table 5. Descriptive statistics results of mean P300 amplitude and peak latency (mean ± standard Deviation).
Task TypeTrial TypeP300 Mean Amplitude (μV)P300 Peak Latency (ms)
Frontal RegionCentral RegionParietal RegionFrontal RegionCentral RegionParietal Region
Non-
Fatigue State
SimplePre-Interruption4.442 ± 2.702.753 ± 2.242.668 ± 2.19381 ± 14377 ± 15380 ± 15
Interruption5.137 ± 4.242.796 ± 2.613.421 ± 2.61378 ± 15381 ± 14387 ± 12
Recovery5.706 ± 3.612.922 ± 2.893.794 ± 3.73381 ± 14386 ± 12390 ± 10
ComplexPre-Interruption5.807 ± 4.423.195 ± 3.363.350 ± 3.28388 ± 13379 ± 15381 ± 14
Interruption6.253 ± 5.323.884 ± 3.593.709 ± 2.68379 ± 16379 ± 15385 ± 13
Recovery6.400 ± 5.874.743 ± 4.294.643 ± 2.97379 ± 16385 ± 11388 ± 14
PausePre-Interruption3.706 ± 3.613.363 ± 3.423.907 ± 3.72384 ± 13385 ± 14386 ± 14
Recovery3.584 ± 2.952.999 ± 2.713.420 ± 2.57384 ± 13387 ± 12389 ± 13
Fatigue StateSimplePre-Interruption6.162 ± 5.582.790 ± 3.003.475 ± 3.05385 ± 13379 ± 16378 ± 15
Interruption5.887 ± 5.533.422 ± 2.503.735 ± 3.05376 ± 16385 ± 14387 ± 13
Recovery5.221 ± 3.754.355 ± 2.453.882 ± 1.78382 ± 15386 ± 12391 ± 8
ComplexPre-Interruption8.138 ± 8.913.460 ± 3.003.916 ± 3.73386 ± 12379 ± 14380 ± 16
Interruption7.486 ± 5.305.004 ± 3.883.912 ± 2.55383 ± 14382 ± 13388 ± 11
Recovery7.311 ± 4.705.070 ± 3.363.981 ± 2.39381 ± 14386 ± 14388 ± 13
PausePre-Interruption5.176 ± 3.822.796 ± 6.673.427 ± 2.99384 ± 14385 ± 13380 ± 16
Recovery4.148 ± 3.765.753 ± 3.192.991 ± 5.41382 ± 15386 ± 15390 ± 12
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

Shao, G.; Tian, S.; Tian, F.; Chen, L.; Zhao, Y. The Impact of Task Interruptions on the Unsafe Behavior of Coal Mine Tunneling Machine Operators: The Moderating Role of Fatigue. Appl. Sci. 2025, 15, 2764. https://doi.org/10.3390/app15052764

AMA Style

Shao G, Tian S, Tian F, Chen L, Zhao Y. The Impact of Task Interruptions on the Unsafe Behavior of Coal Mine Tunneling Machine Operators: The Moderating Role of Fatigue. Applied Sciences. 2025; 15(5):2764. https://doi.org/10.3390/app15052764

Chicago/Turabian Style

Shao, Guangtong, Shuicheng Tian, Fangyuan Tian, Lei Chen, and Yifan Zhao. 2025. "The Impact of Task Interruptions on the Unsafe Behavior of Coal Mine Tunneling Machine Operators: The Moderating Role of Fatigue" Applied Sciences 15, no. 5: 2764. https://doi.org/10.3390/app15052764

APA Style

Shao, G., Tian, S., Tian, F., Chen, L., & Zhao, Y. (2025). The Impact of Task Interruptions on the Unsafe Behavior of Coal Mine Tunneling Machine Operators: The Moderating Role of Fatigue. Applied Sciences, 15(5), 2764. https://doi.org/10.3390/app15052764

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