Understanding mental fatigue and its detection: a comparative analysis of assessments and tools

Mental fatigue

Mental fatigue and physical fatigue are two separate entities. Physical fatigue is a form of tiredness caused by repeated muscle movements (Mizuno et al., 2011). In contrast, mental fatigue is defined as a psychobiological state of tiredness caused by prolonged periods of performing demanding, cognitive-load-inducing activities, and it reduces efficiency in cognitive performance (Craik, 2014; Tanaka, Ishii & Watanabe, 2014). The concept of mental load can be explained by perceptual load and processing capacity. First, perceptual load is associated with the amount of information involved in processing a relevant task (Forster & Lavie, 2016). High perceptual load results in inattentional blindness leading to impairments of memory (Murphy, Groeger & Greene, 2016; Murphy & Greene, 2016). Perceptual load is one of the causes of high mental load; however, higher mental load can also occur from tasks involving executive functioning skills irrespective of whether or not they involve higher perceptual load (Blair, 2016). Second, processing capacity can be defined as the amount of information an individual can process at any given time (Hancock et al., 2021; van Elburg et al., 2021). Once the capacity is reached, the ability to process new information becomes difficult, resulting in mental overload. Experiencing long durations of mental overload leads to mental fatigue (Mizuno et al., 2011).

Cognitive performance

Research has shown that mentally fatigued individuals tend to have decreased performance in cognitive tasks that require vigilance, sustained attention, and divided attention (Smith et al., 2019). On the one hand, this observation could be partially attributed to fatigue-related motivation. Motivation has been found to play a role in the declining cognitive performance observed in individuals that are mentally fatigued. A study comparing cognitive performance found that individuals that were not being rewarded showed less cognitive effort, decreased frequency of saccades and increased blinks compared to participants that were expecting a reward when both groups reported being mentally fatigued (Herlambang, Taatgen & Cnossen, 2019). Similarly, it was found that individuals with higher levels of intrinsic motivation performed better in a cognitive task compared to individuals with low levels of intrinsic motivation regardless of being mentally fatigued or not (Herlambang, Cnossen & Taatgen, 2021). Nevertheless, there are limited objective measures to firmly associate motivation declination caused by mental fatigue to be the core factor of a poorer cognitive performance (Jimura, Locke & Braver, 2010).

On the other hand, Hopstaken et al. (2016) and Gergelyfi et al. (2015) stated that a depletion in performance is not due to reduced motivation in an individual but due to the lack of mental resources available to efficiently process and engage in the task. The prefrontal cortex has a critical role in producing goal-directed signals for attention switching (Rossi et al., 2009). It has been shown that when an individual is mentally fatigued, an overactivation occurs in the frontal cortex that results in an inhibition to process information (Tanaka, Ishii & Watanabe, 2014; Ishii, Tanaka & Watanabe, 2014; Goodman & Marino, 2020). This explains the cause of decreased executive control (e.g., inhibition responses) and an increase in response time when performing a task (Gantois et al., 2020; Guo et al., 2018; Jaydari Fard, Tahmasebi Boroujeni & Lavender, 2019). As a result, individuals with mental fatigue show decreased cognitive performance (Sasahara et al., 2015) and are less likely to be able to detect hazards which could risk occupational safety (Rocha, Marziale & Robazzi, 2007). Individuals experiencing mental fatigue find it difficult to suppress irrelevant information (Faber, Maurits & Lorist, 2012; Csathó et al., 2012). Goal-directed action is inhibited when a person is mentally fatigued, resulting in stimulus-driven actions to be more pronounced (Boksem, Meijman & Lorist, 2005). To be brief, it is harder to inhibit stimulus-driven actions when an individual is mentally fatigued, resulting in an increase in reaction times and error rates when performing tasks that require an individual to actively suppress irrelevant information. This phenomenon reflects the existence of decreased selective visual attention due to mental fatigue (Guo et al., 2016).

Physical performance

Mental fatigue has also shown to impair physical performance such as balance control when attentional processing is compromised (Webster, Richter & Kruglanski, 1996). Pires et al. (2018) showed that mentally fatigued individuals have impaired physical performance (e.g., balance control, physical strength) that was also associated with an increased activation in the prefrontal cortex, as reflected in increased slow theta brain waves, resulting in increased bottom-up responses and decreased top-down processes. Similarly, Marcora, Staiano & Manning (2009) stated that mentally fatigued individuals were found to have higher perceptions of effort when performing physical activities which resulted in individuals reaching their perceived exertion limits faster, and therefore ending their physical activities. In support of this finding, Penna et al. (2018a) also showed that a decline in physical performance in mentally fatigued individuals was not attributed to physiological variables such as oxygen uptake, heart rate or blood lactate, but can be attributed to higher perceived exertion which is an effect of mental fatigue. Mental fatigue does not affect the maximum strength and power that can be exerted by an individual but rather it affects the perception (i.e., effort exerted, or effort required to complete the task) of an individual which leads to a decrease in physical performance (Van Cutsem et al., 2017).

Self-reporting questionnaires

Various self-assessments have been developed in the pursuit of mental fatigue detection. The diverse types of questionnaires that have been developed to measure mental fatigue are summarized in Table 1 along with their individual strengths and weaknesses. In general, subjective measures such as self-reporting questionnaires have often been criticized to be biased as the honesty (i.e., being truthful instead of answering based on socially acceptable behaviors) and introspective abilities (i.e., to accurately assess themselves) of participants can be easily compromised. Humans are also unable to accurately make judgments regarding their cognitive states (Schmidt et al., 2009). This is why experts strongly suggest that these questionnaires be used hand in hand with other objective measures (Tempelaar, Rienties & Nguyen, 2020). This brings to surface a limitation as it does not provide a quick and simple method of detecting transient states of mental fatigue. These questionnaires are unable to provide moment-to-moment fluctuations of mental fatigue and often these methods are affected by cognitive ability, mood, and anxiety levels.

Attention Network Test (ANT)

Another method proposed to detect mental fatigue is the use of the ANT (Fan et al., 2002). This test basically examines the attentional network of an individual in three fundamental areas such as alerting, orienting, and executive control (Federico et al., 2013). The alerting network is examined by the changes in reaction times insinuated by a warning signal. The orienting network is evaluated by the changes in reaction times resulting from cues indicating where the target will occur. The efficiency of the executive network is examined by requiring the subject to respond by pressing a key indicating the direction of a central arrow surrounded by congruent, incongruent, or neutral flankers. This method proposes that individuals that are mentally fatigued will perform worse than a non-fatigued individual, as their attention is compromised by mental fatigue (Lou, 2009). Research has shown that mentally fatigued individuals display significantly lower executive and attention network efficiency and accuracy compared to controls (Pauletti et al., 2017; Togo et al., 2015).

The implementation of the ANT in health monitoring technologies are not promising as Jones et al. (2016) found that a practice effect was common in healthy adults when the test is repeated after a 24-h interval, suggesting that test-scores can be affected if this detection tool were to be used as a daily method of assessing mental fatigue. Moreover, attention systems comprise various independent components making it too complicated to be evaluated by a single task (Wang et al., 2015). For example, attention consists of four independent systems (i.e., endogenous, exogenous, spatial, objective) that vary in behavioral effects and have specific neural substrates (Chica, Bartolomeo & Lupiáñez, 2013; Weinbach & Henik, 2012). Furthermore, this detection tool is yet to be validated against objective and/or subjective measures (e.g., salivary cortisol levels, self-reporting questionnaires, saccadic eye movements, etc.) of mental fatigue. The testing duration of ANT is also relatively long as it takes an average of 25 to 30 min, depending on the length of the test. The lengthy testing period raises potential compliance and burden issues (Kas, Serretti & Marston, 2019). Also, although this method of detection is not as subjective as questionnaires’ responses and able to provide objective measurements (e.g., quantifiable data from pressing a key), the ANT task is more of a cognitive assessment which is mainly used for inferring purposes (Kayser et al., 2020). This may lack clarity and direct observation that potentially lead to questions about its validity in detecting mental fatigue.

EEG

Mental fatigue can also be measured using EEG spectral changes in the theta, alpha and beta waves as summarized in Table 2 (Cheng & Hsu, 2011). In a recent study, theta wave activity was found to be a promising primary biomarker for mental fatigue and the alpha wave activity was considered a second line biomarker (Tran et al., 2020). Studies have shown that the use of EEG recordings alongside electrooculography (EOG; used to assess the cornea-retinal standing potential that exists between the front and the back of the eye) increases the accuracy in detecting mental fatigue (Jia & Tyler, 2019). This method has been reported to provide most accurate measures of mental fatigue. In contrast, Smith et al. (2019) found that cognitively demanding tasks used to induce mental fatigue in individuals had unclear effects on EEG results. Moreover, the major limitation of using EEG as a mental fatigue detection tool is that it is too sophisticated and requires a laboratory setting to minimize signal to noise ratio (Read & Innis, 2017). A robust assessment method would be needed in order to ensure this method is able to handle the variations in electric signal between individuals that occur due to the EEG signal being highly sensitive (Monteiro et al., 2019). EEG also requires the use of electrodes which needs to be placed specifically on an individual’s head in order to get accurate measurements of the alpha, beta, and theta wave rhythms, making it perplexed to integrate into a simple health monitoring device such as a smartphone and surveillance devices that would be able to monitor individuals’ mental fatigue continuously whilst they carry out their daily tasks. This problem has been addressed in several studies, whereby researchers developed a simple measurement technique that is just as effective as the entire EEG setup with the use of only one EEG sensor to detect mental fatigue, thus decreasing the complexity of this detection method (Resalat & Saba, 2015; Rohit et al., 2017; Hu, 2017). A recent study on the impact of mental fatigue on brain activity showed that when using the entire EEG setup, the EEG relative power index was better at assessing mental fatigue when individuals were in a resting state compared to when they were actively performing a task (Li et al., 2020). Overall, this method has the ability to reflect the changes in the brain associated with mental fatigue, however in light of recent findings (Smith et al., 2019), it has been shown that this method may not be able to accurately detect the changes that occur in the brain whilst a mentally fatigued individual is carrying out a cognitively demanding task which would limit the use of this detection tool. As individuals tend to go for long hours before taking a break, and to avoid prolonged mental fatigue go unnoticed, it would be more beneficial to have mental fatigue detected during task engagement as an indicator to needing a break before continuing to the next task.

Heart rate variability (HRV)

Apparies, Riniolo & Porges (1998) reported this cardiovascular index to be a sensitive measure of fatigue and mental workload. Heart rate is controlled by the autonomic nervous system (e.g., SNS) whereby the release of hormones (e.g., Estrogen) causes fluctuations in heart rate (Leicht, Hirning & Allen, 2003). These fluctuations occur according to the ratio between the absolute low frequency and high frequency power measured via HRV (Goldberger, 1999). HRV is a simple, non-invasive measure of the changes between each heartbeat in respect to time, using an electrocardiogram (ECG) (ChuDuc, NguyenPhan & NguyenViet, 2013). HRV also reflects the increase of the body’s self-regulatory abilities (e.g., behavioral, cognitive and emotional process) when performing mentally fatiguing tasks. The body’s self-regulatory processes (specifically the cognitive processes) play an important role in changing or suppressing information (Barutchu et al., 2013; Heatherton, 2011). Segerstrom & Nes (2007) found an increase in HRV during high self-regulatory efforts. Various studies have reported similar findings, that an increase in heart rate is strongly correlated with mental fatigue (Huang et al., 2018; Delliaux et al., 2019; Melo, Nascimento & Takase, 2017). These findings were supported by EEG data—a decrease in P300 amplitude indicative of a decrease in attention correlated with an increase in heart rate (Zhang & Yu, 2010; Zhao et al., 2012).

The HRV detection tool has been integrated into smart devices (i.e., smart watch, fitness bracelets) to be used as health monitoring technologies (Naranjo-Hernández et al., 2017). It is mainly used to measure heart rate in real time for the purpose of cardiovascular health monitoring but has not been successfully adapted to specifically detect mental fatigue. Nelesen et al. (2008) investigated the relationship between fatigue and heart rate, finding that fatigue was not associated with heart rate but rather significantly associated with decreased cardiac index and stroke index when at rest. Accordingly, the reliability of detecting mental fatigue via HRV is heterogeneous with various studies demonstrating no link between mental fatigue and physiological variables such as heart rate (Penna et al., 2018b; Thompson et al., 2019).

In addition, HRV can be influenced by too many confounding variables such as movements made by individuals during the assessment resulting in false readings, pathological conditions that affect the nervous system (e.g., degenerative diseases, brain damage), the heart (e.g., arrhythmia), or other organs (e.g., renal failure), mental health disorders (e.g., depression), smoking, caffeine, and medication (Acharya et al., 2006; Hayano et al., 1990; Monda et al., 2013; Massaro & Pecchia, 2016). Although it would make sense for physical fatigue to be linked with heart rate, there are no direct links of heart rate to mental fatigue. Gonzalez et al. (2017) stated that the current algorithm used for HRV detection via smartwatches and simple heart rate detection devices are unable to distinguish the differences between mental fatigue and physical fatigue. Thus, HRV may not be sufficient for the use of mental fatigue detection.

Cortisol levels

Mental fatigue has been found to modulate the autonomic and endocrine responses in the body (Moreira et al., 2018). Fatigued individuals precipitate an increase in cortisol levels in the body released by the hypothalamic-pituitary-adrenal axis (Torres-Harding et al., 2008; Thompson et al., 2016). The increase of cortisol levels has been associated with a decrease in cognitive functions, specifically attention, after a fatigue-inducing task (Bohnen et al., 1990; Lee et al., 2007). Salivary cortisol levels have been shown to be a reliable indicator of mental fatigue because they reflect blood cortisol levels, which rise when an individual is stressed (Kirschbaum & Hellhammer, 2007).

In contrast, Powell et al. (2013) found that total cortisol output was scarcely associated with fatigue. Another limitation of this method is that not all of the studies mentioned above had examined the relationship between cortisol level and mental fatigue specifically. Most of these studies evaluated physical and mental fatigue as a whole, rather than studying mental fatigue separately. For example, in Torres-Harding et al. (2008), where a relationship between cortisol levels and the symptoms of chronic fatigue syndrome was observed, the fatigue associated with the salivary cortisol levels was not specifically related to mental fatigue but more so to fatigue in general. As mentioned above, physical and mental fatigue are two different forms of fatigue that need to be evaluated separately as one can experience fatigue physically but not mentally and vice versa. Hence, there is limited data that can specifically link changes in cortisol levels to mental fatigue. Furthermore, a recent review reported a lack of evidence that salivary cortisol is correlated with stress and fatigue, suggesting that the relationship between mental fatigue and cortisol levels remains inconclusive (de Assis, de Resende & Marziale, 2018).

Nonetheless, a combination of salivary cortisol measures and health monitoring technology seems to be remarkably successful in determining fatigued drivers. To do that, smartphones, and similar devices (e.g., smartwatches and surveillance devices) have been demonstrated to have potential in providing a platform for mental health detection tools to be integrated into. For example, a smartphone-based cortisol detection technique has been introduced to estimate driver fatigue (Russell et al., 2022). This “Smart Fatigue Phone” is a health monitoring technology developed to detect real-time fatigue using immunosensors and a smartphone-linked fluorescence signal reader. The immunosensor functions by measuring the concentration of salivary cortisol levels of an individual. This device has been validated by EEG measurements whereby decreased beta rhythms resulting from mental fatigue correlated with high cortisol levels as measured by the smart fatigue phone (Shin et al., 2019).

However, to improve the accuracy of this device, fatigue related biomarkers (e.g., alpha-amylase, cortisol levels in the blood, and lactate) should be correlated with the salivary cortisol levels. In the sporting environment, it was found that salivary biomarkers were not found to be a good objective indicator of mental fatigue as the changes observed in cortisol and alpha-amylase were not related to mental fatigue (Russell et al., 2022). Another limitation to using salivary cortisol as an assessment tool is that saliva is easily contaminated by blood and food particles in the mouth which can provide inaccurate results of cortisol levels (Kamodyová et al., 2015). This makes testing a little less convenient as individuals would need to rinse their mouths at least 10 min before testing themselves for mental fatigue. On top of that, studies have shown that gender affects the temporal changes in salivary cortisol levels whereby females were observed to have higher cortisol levels than males (Nakajima et al., 2012). According to Montero-López et al. (2018) menstrual cycle phase was found to increase cortisol responses to laboratory induced mental stress. Contrastingly, it was found that in healthy women, the basal cortisol levels were not affected by menstrual cycles (Wilson, Lorenz & Heiman, 2018). These gender effects do not pose as a major limitation to using salivary cortisol as an assessment tool for mental fatigue, but it does make this testing method more complicated, as the assessment tool will need to be calibrated specifically for each gender. Overall, there is no evidence linking this method to a decline in executive functions, specifically inhibition, which studies have shown to be a strong indication of mental fatigue in individuals (Guo et al., 2018).

Saccadic eye movements

In recent years, characteristics of saccadic eye movements have gained popularity in identifying various neurodegenerative-related cognitive deficits such as those found in Alzheimer’s Disease (AD) and Parkinson’s disease (Anderson & MacAskill, 2013; Opwonya et al., 2021). Saccadic eye movements can be described as the rapid movement of the eyes from one point to another (Purves et al., 2001a). Saccades play a fundamental role in the ability to attend to the visual environment. For example, we need saccades to rapidly scan the environment (e.g., before crossing the road) and to be able to identify sudden changes in the environment to avoid danger (Purves et al., 2001b). Reflexive saccades are stimulus driven and involves bottom-up processes, whereas voluntary saccades are goal driven and involves higher-level cortical functions to be executed (Mulligan, Stevenson & Cormack, 2013; Goffart, 2009; Walker et al., 2000). Voluntary saccades involve top-down processing, they are vulnerable to natural aging and less resilient, hence deteriorates as a person ages (i.e., from adulthood to old age) or encounters illnesses or disorders (e.g., migraine and Tourette’s Syndrome) that affect higher cortical brain areas (LeVasseur et al., 2001; Filippopulos et al., 2021; Bafna & Hansen, 2021; Chen et al., 2021; Chen & Machado, 2016).

As saccadic eye movements are related to the neural activity in the frontal lobe, the area of the brain that is mostly affected by mental fatigue (Ishii, Tanaka & Watanabe, 2014; Purves et al., 2001b), it is very likely that the reliability and validity of using saccadic eye movements in detecting mental fatigue would be high (Yamada & Kobayashi, 2018). As mentioned previously, the overactivation that occurs in the frontal lobe when an individual is mentally fatigued results in inability to inhibit responses (Gantois et al., 2020; Guo et al., 2018; Jaydari Fard, Tahmasebi Boroujeni & Lavender, 2019). Hence, inhibition responses can be used as a marker to assess mental fatigue. The “antisaccade task” is designed to assess an individual’s ability to suppress reflexive saccades in order to produce a voluntary saccade (as shown in Fig. 1). It involves high levels of executive control and has been shown to be reliable to assess an individual’s executive functions (e.g., inhibition responses) (Chehrehnegar et al., 2021).

There are various physiological measurement tools that can be used to assess saccadic eye movement characteristics such as EOG. EOG is more commonly known to be used in measuring saccades and fixations by analyzing the electrical signals on the skin around the eyes produced by the eye movement (Wang et al., 2019). The electrodes are able to detect changes in the orientation of the eyes based on the amount of muscle movements towards or against the positive or negative electrode poles (Jia & Tyler, 2019). It was proposed that placing the electrodes on the forehead is best for measuring eye movements related to fatigue as this placement detects the electrical signals on the skin that is produced by eye movements with the use of fewer electrodes and less interference with facial activities, compared to when electrodes are placed in the conventional method, that is around the eyes (as shown in Fig. 2) (Zhang et al., 2015; Zheng & Lu, 2017; Heo, Yoon & Park, 2017). Furthermore, electrodes placed at the forehead can be integrated into many wearable devices such as headbands and caps, and thus increasing the usability for the user (Heo, Yoon & Park, 2017).

Moreover, there are other tools that have been developed to track saccadic eye movements such as infrared based eye tracking techniques. With the advancement of technology, these infrared sensors might be able to be integrated into webcams, handphone cameras and even surveillance cameras in the near future. Infrared sensors can detect saccades in all directions (e.g., horizontal and vertical saccades), however, horizontal saccades have been shown to be sufficient for mental fatigue detection (Jia & Tyler, 2019). It has been found that individuals experiencing mental fatigue presents a specific pattern of eye movement metrics which include increased saccade latency, blink frequency, and decreased pupil dilation range (Yamada & Kobayashi, 2018; Bafna, Bækgaard & Hansen, 2021; Zargari et al., 2018). These characteristics occur as mental fatigue causes deficits in inhibition of response as well as a decrease in information processing speed. The Wearable Transparent Eye Detection System is an example of a health monitoring technology that uses saccades to detect mental fatigue (Sampei et al., 2016). This device measures the movement of the pupil and blinking from light reflected from the eye. This light is detected by dye-sensitized photovoltaic cells. In the same study, the device has also been validated using the NASA Task Load Index (NASA-TLX), a self-reporting questionnaire used to assess an individual’s estimated mental load that results in mental fatigue. However, this validation study was only conducted on a small sample of five subjects, with no correlations reported between pupil movement, blinking, and mental load for two subjects (De Vries et al., 2004). Therefore, further studies are necessary to identify specific saccadic eye movement characteristics that correlate strongly with other measures of mental fatigue.

Recently, many studies have been carried out to develop a method of tracking saccadic eye movements outside of the clinical environment via smartphone cameras and surveillance devices (Lai et al., 2020; Ansari, Kasprowski & Obetkal, 2021). Appearance-based gaze estimation with the help of convolutional neural networks was found to be successful in accurately decoding saccadic eye movements from facial images and desktop cameras (Ansari, Kasprowski & Obetkal, 2021). This model advances the mental fatigue detection method via saccades as it allows for individuals to be tested for mental fatigue whilst they carry out their day-to-day living. This method, unlike other methods previously mentioned, increases the accuracy and convenience for dynamic saccadic eye movement testing for mental fatigue. Future replication and validation studies would be necessary to validate these saccadic eye movement characteristics with other mental fatigue measures as the data would be useful to integrate this detection method into modern health monitoring technologies.