Preventive effect of one-session brief focused attention meditation on state fatigue: Resting state functional magnetic resonance imaging study

Introduction: The extended practice of meditation may reduce the influence of state fatigue by changing neurocognitive processing. However, little is known about the preventive effects of one-session brief focused attention meditation (FAM) on state fatigue in healthy participants or its potential neural mechanisms. This study examined the preventive effects of one-session brief FAM on state fatigue and its neural correlates using resting-state functional MRI (rsfMRI) measurements. Methods: We randomly divided 56 meditation-naïve participants into FAM and control groups. After the first rsfMRI scan, each group performed a 10-minute each condition while wearing a functional near-infrared spectroscopy (fNIRS) device for assessing brain activity. Subsequently, following a second rsfMRI scan, the participants completed a fatigue-inducing task (a Go/NoGo task) for 60 min. We evaluated the temporal changes in the Go/NoGo task performance of participants as an indicator of state fatigue. We then calculated changes in the resting-state functional connectivity (rsFC) of the rsfMRI from before to after each condition and compared them between groups. We also evaluated neural correlates between the changes in rsFC and state fatigue. Results and discussion: The fNIRS measurements indicated differences in brain activity during each condition between the FAM and control groups, showing decreased medial prefrontal cortex activity and decreased functional connectivity between the medial prefrontal cortex and middle frontal gyrus. The control group exhibited a decrement in Go/NoGo task performance over time, whereas the FAM group did not. These results, thus, suggested that FAM could prevent state fatigue. Compared with the control group, the rsFC analysis revealed a significant increase in the connectivity between the left dorsomedial prefrontal cortex and right superior parietal lobule in the FAM group, suggesting a modification of attention regulation by cognitive effort. In the control group, increased connectivity was observed between the bilateral posterior cingulate cortex and left inferior occipital gyrus, which might be associated with poor attention regulation and reduced higher-order cognitive function. Additionally, the change in the rsFC of the control group was related to state fatigue. Conclusion: Our findings suggested that one session of 10-minute FAM could prevent behavioral state fatigue by employing cognitive effort to modify attention regulation as well as suppressing poor attention regulation and reduced higher-order cognitive function.


Introduction:
The extended practice of meditation may reduce the influence of state fatigue by changing neurocognitive processing.However, little is known about the preventive effects of one-session brief focused attention meditation (FAM) on state fatigue in healthy participants or its potential neural mechanisms.This study examined the preventive effects of one-session brief FAM on state fatigue and its neural correlates using restingstate functional MRI (rsfMRI) measurements.Methods: We randomly divided 56 meditation-naïve participants into FAM and control groups.After the first rsfMRI scan, each group performed a 10-minute each condition while wearing a functional near-infrared spectroscopy (fNIRS) device for assessing brain activity.Subsequently, following a second rsfMRI scan, the participants completed a fatigue-inducing task (a Go/NoGo task) for 60 min.We evaluated the temporal changes in the Go/NoGo task performance of participants as an indicator of state fatigue.We then calculated changes in the resting-state functional connectivity (rsFC) of the rsfMRI from before to after each condition and compared them between groups.We also evaluated neural correlates between the changes in rsFC and state fatigue.Results and discussion: The fNIRS measurements indicated differences in brain activity during each condition between the FAM and control groups, showing decreased medial prefrontal cortex activity and decreased functional connectivity between the medial prefrontal cortex and middle frontal gyrus.The control group exhibited a decrement in Go/NoGo task performance over time, whereas the FAM group did not.These results, thus, suggested that FAM could prevent state fatigue.Compared with the control group, the rsFC analysis revealed a significant increase in the connectivity between the left dorsomedial prefrontal cortex and right superior parietal lobule in the FAM group, suggesting a modification of attention regulation by cognitive effort.In the control group, increased connectivity was observed between the bilateral posterior cingulate cortex and left inferior occipital gyrus, which might be associated with poor attention regulation and reduced higher-order cognitive function.Additionally, the change in the rsFC of the control group was related to state fatigue.Conclusion: Our findings suggested that one session of 10-minute FAM could prevent behavioral state fatigue by employing cognitive effort to modify attention regulation as well as suppressing poor attention regulation and reduced higher-order cognitive function.

Introduction
Mindfulness is the process of consciously and openly participating in one's present experience (Creswell, 2017).Higher dispositional mindfulness, characterized by mindfulness personality traits, is associated with positive psychological effects such as reduced emotional reactivity and improved behavioral regulation (Keng et al., 2011).Meditation is commonly used to cultivate mindfulness, beginning with focused attention meditation (FAM) (Chiesa and Malinowski, 2011;Lutz et al., 2008).In FAM, participants focus on a particular object, such as their breath, and return their focus to that object whenever their mind wanders (Lippelt et al., 2014;Lutz et al., 2008).Recently, the potential impact of a one-session brief FAM (less than 30 min) has been highlighted due to its adaptability for daily routines and applicability in the workplace (Hafenbrack, 2017;Howarth et al., 2019;Jiménez et al., 2020;Mantzios and Giannou, 2019).For instance, studies have suggested that one-session brief FAM can enhance response inhibition processes (Colzato et al., 2016) and working memory (Ma et al., 2021;Yamaya et al., 2021).
Our previous study suggested a new potential effect of one-session brief FAM on state fatigue (Yamaya et al., 2023).In particular, we investigated the duration of the effect of one-session brief FAM on response inhibition processes.Participants completed the Stroop task before and after a one-session brief FAM (10 min) or a 10-min control condition and 20-, 40-, and 60-min post-condition.We found no differences between conditions immediately after, 20-, or 40-min post-condition; however, performance at 60-min post-condition was significantly higher among participants performing FAM than that among those performing the control condition (Yamaya et al., 2023).These results indicated that participants performing FAM maintained response inhibition processes for 60 min.In contrast, participants performing the control condition showed decreased response inhibition processes over time, possibly due to state fatigue.Nevertheless, to our knowledge, previous studies have yet to explicitly examine the preventive effect of FAM on state fatigue.As such, whether one-session brief FAM can prevent state fatigue remains unclear.
State fatigue can be experienced by prolonged cognitive demands, such as fatigue-inducing tasks (Kato et al., 2012;Kato 2009), and expressed as a decline in task performance over time (DeLuca, 2005;Mackworth, 1948).It can also be indicated by decreases in parasympathetic nervous activity and increases in sympathetic nervous activity (Mizuno et al., 2011;Tanaka et al., 2009) and subjective fatigue (Shigihara et al., 2013).Given the complexity of state fatigue, multiple assessments, including behavioral, psychophysiological, and subjective evaluations, are useful (Matthews et al., 2017;Solomon and Manea, 2022).However, as far as we know, no study has yet investigated the preventive effect of one-session brief FAM on state fatigue by employing a fatigue-inducing task and multiple assessments.
The extended practice of meditation and greater dispositional mindfulness have been associated with alterations in functional connectivity (FC) among the default, frontoparietal/executive control, and salience networks (Ganesan et al., 2022;Mooneyham et al., 2017Mooneyham et al., , 2016;;Sezer et al., 2022).The default network, which comprises the posterior cingulate cortex (PCC), the medial prefrontal cortex (mPFC), the inferior parietal lobule (IPL), and other temporal and occipital regions, contributes to the generation of spontaneous thought (Christoff et al., 2016;Dixon et al., 2017;Raichle et al., 2001).The frontoparietal network, which is centrally supported by the bilateral dorsolateral prefrontal cortex (DLPFC), is involved in sustained attention and top-down attention regulation (Vincent et al., 2008).The salience network, comprising the insular cortex and dorsal anterior cingulate cortex (dACC), is related to immediate responses to stimuli and processing errors (Menon et al., 2001;Seeley et al., 2007).Increased resting-state functional connectivity (rsFC) through meditation between the default network and frontoparietal network regions may be related to improving attention regulation (Sezer et al., 2022).In addition, increased rsFC within default network regions may be involved in emotion regulation (Buckner et al., 2008;Sezer et al., 2022).Further, increased connectivity between the salience and default networks (de la Cruz et al., 2019) or within the default network (Beissner et al., 2013) may be related to increased parasympathetic nervous activity (Tang et al., 2012;Y.-Y. 2009).Because a one-session brief FAM had positive effects on cognitive performance (Ma et al., 2021;Yamaya et al., 2023Yamaya et al., , 2021)), emotional regulation (Arch and Craske, 2006), and parasympathetic nervous activity (Azam et al., 2015), one-session brief FAM may induce the higher connectivity between these networks, resulting in a brain state inducing efficient effortless attentional control and regulation of parasympathetic nervous activity before a fatigue-inducing task.This brain state could lead to the prevention of performance decline, the prevention of decreased parasympathetic nervous system activity, and the prevention of increased subjective fatigue (McCormick et al., 2020;Tang et al., 2012;Thomson et al., 2015).However, the effect of one-session brief FAM on rsFC (in nonexperts and meditation-naïve individuals) and the neural correlates between changes in rsFC resulting from one-session brief FAM and their potential preventive effect on state fatigue remain unexplored.
This study examined the preventive effect of one-session brief FAM on state fatigue and rsFC compared with those performing a control condition.We also investigated the neural correlates of the alteration of rsFC evoked by one-session brief FAM and state fatigue.To do so, we tested two hypotheses.First, we predicted that state fatigue would be prevented more in the FAM than in the control group, with less performance decrement over time, less decrease in parasympathetic nervous activity, and less increase in subjective fatigue feelings after a fatigue-inducing task.Second, we posited that the FAM group would show increases in rsFC between the default network and frontoparietal network regions, between the salience network and default network regions, and within default network regions, which would correlate with the prevention of state fatigue.FAM requires attention regulation for reducing spontaneous thought, supported by the deactivation of the mPFC and activation of the middle frontal gyrus (MFG) (Brewer et al., 2011;Hasenkamp et al., 2012;Scheibner et al., 2017).Whereas, the control condition could give rise to spontaneous thoughts, which would induce the activation of both regions (Fox et al., 2015).Therefore, in addition to self-reported assessments of the degree of spontaneous thought, we expected that measurement of the brain activity during each condition using functional near-infrared spectroscopy (fNIRS) could detect group differences.Multiple assessments of changes in state fatigue would allow us to elucidate the effect of FAM on state fatigue in greater detail.Moreover, neural activity during and immediately after FAM could reveal the underlying mechanism of the effect of FAM on subsequent fatigue tasks.Therefore, this study may contribute to the accumulation of scientific evidence on the effectiveness of one-session brief FAM and propose new measures against state fatigue.

Participants
This study was approved by the Institutional Review Board of the Smart-Aging Research Center of Tohoku University, Japan (acceptance number: 2011-04).Written informed consent was obtained from each participant, in accordance with the Declaration of Helsinki.Seventy right-handed Japanese university undergraduate and postgraduate students (42 men and 28 women, mean age: 22 ± 1.5 years) were recruited using campus flyers, bulletin boards, and e-mail invitation.The sample size was 66, which was calculated with G* Power (Version 3.1.9.2), given a repeated measures ANOVA design, a small to medium effect on cognitive task performance (f = 0.16) (Cásedas et al., 2020;Norris et al., 2018), a type I error rate of 0.05, a level of statistical power of 0.95, two groups, and six measurement sessions (for more details, see section 2.3.4 -Hybrid Go/NoGo Task Analysis).To prepare for some participants N. Yamaya et al. failing to meet the inclusion criteria (described below), we recruited a total sample size of 70 (35 in each group).No participant was pregnant, and none had a history of psychiatric or neurological illness.Furthermore, all participants reported being unexperienced in meditation or that they had not meditated for the previous two years (Atchley et al., 2016).
We divided the participants into two groups, FAM or control condition, using blocked random assignment.The inclusion criteria were not having chronic fatigue or sleep problems and not using supplements or medications to treat fatigue; these criteria were assessed through questionnaires administered before the experiments.Three participants were excluded because they scored 77 or higher on the Japanese version of the Checklist Individual Strength Questionnaire (CIS-J), which is indicative of probable chronic fatigue cases (Bültmann et al., 2000;Nakagawa et al., 2013).Another three participants were excluded because they scored 5.5 or higher on the Japanese version of the Pittsburgh Sleep Quality Index (PSQI-J), which is an indicator of poor sleep quality (Doi et al., 2000).In addition, one participant taking supplements related to fatigue or anti-fatigue effects was excluded.In total, six participants were excluded because they failed to meet the inclusion criteria (One of the excluded participants scored higher than the criteria in both the CIS-J and PSQI-J).

Procedures and experimental tasks
Prior to the initiation of the experiments, participants attended a briefing and were instructed to 1) sleep for at least 7 h one day prior to the experimental day; 2) avoid caffeine intake, alcohol consumption, and strenuous exercise 12 h prior to the start of the experiment; and 3) finish breakfast at least 1 h before the start of the experiment and consume nothing but water during the experiment.The experimental procedure (shown in Fig. 1) was conducted 1 or 2 d after the briefing (from 9:00am to 12:00pm or from 9:30am to 12:30pm).Participants engaged in FAM or the control condition for 10 min while wearing an fNIRS instrument.They underwent rsfMRI twice: before (first scan) and after (second scan) each condition.T1-weighted anatomical images were obtained immediately after the first scan.After the second scan, participants completed a fatigue-inducing task for 60 min, consisting of 5 min per block and 10 s intervals.Autonomic nervous function (ANF) and subjective assessments were measured three times: immediately before (coded as A1 in Fig. 1) and after (A2) each condition, and after the fatigue-inducing task (A3).Participants were also asked to rate the degree of spontaneous thought they experienced during each condition and their adherence to the content of the instructions at A2.

Focused attention meditation and control condition
All participants wore an fNIRS instrument and maintained an upright posture on their chairs in front of a laptop during each condition.They were instructed to rest for 1 min and perform either of the conditions for 10 min.The duration of each condition was based on the findings of our previous study (Yamaya et al., 2023).Both the onset and end points of each condition were indicated by changing the content of the laptop screen from "rest" to a cross mark at the center of the screen with a brief ring sound.The FAM group engaged in Su-soku meditation, which is suitable for beginners because it requires no special training and allows for easier attention regulation (Chiesa and Malinowski, 2011;Hanh, 2016;Kubota et al., 2001).The FAM group was instructed to open their eyes to look at the cross mark on the computer, pay attention to their breathing, and repeatedly count the number of breaths up to 10 (Yamaya et al., 2023(Yamaya et al., , 2021)).The full instructions are provided in the Supplemental Material.
The control group received the same instructions as the FAM group, except for those related to attention regulation (see the Supplemental Material) (Yamaya et al., 2023).Thus, this group could freely think about anything.This mental state corresponds to spontaneous thought, which includes daydreaming, mind-wandering, and creative thinking (Christoff et al., 2016).Spontaneous thoughts account for 30-50 % of the time in the daily lives of people (Killingsworth and Gilbert, 2010).The control condition was, therefore, similar to the mental state observed naturally in the real world, facilitating the connection of the study results to everyday life (Noone and Hogan, 2018).
As the core component of FAM was attention regulation, the difference between the FAM and control groups was the engagement of participants in attention regulation (Yamaya et al., 2023).To allow for comparisons under the same conditions, participants were informed that each condition was an authentic meditation (Yamaya et al., 2023).

Assessment of differences in each condition state
2.2.2.1.Adherence to instructions and spontaneous thought.To confirm Fig. 1.Overview of the experimental procedure.Both the focused attention meditation (FAM, upper part of the figure) and control groups performed the respective condition for 10 min wearing a functional near-infrared spectroscopy (fNIRS) instrument.One minute of rest (REST) was set just before participants began each condition to assess their brain activity at baseline.Participants underwent resting-state functional magnetic resonance imaging (rsfMRI) before (first scan) and after (second scan) each condition.T1-weighted anatomical images (T1) were obtained immediately after the first scan.After the second scan, they completed a fatigueinducing task for 60 min, consisting of 5 min per block and 10 s intervals.Autonomic nervous function and subjective fatigue assessments were performed before (A1) and after (A2) each condition, and after the fatigue-inducing task (A3).Participants were also asked to rate the degree of spontaneous thought they experienced and their adherence to the content of the instructions at A2.
N. Yamaya et al. whether participants adhered to the instructions, we asked them to rate their degree of adherence by answering the following question: "How well did you follow the instructions?"(Brewer et al., 2011;Garrison et al., 2015Garrison et al., , 2014;;Yamaya et al., 2023).Further, to confirm whether participants reduced their spontaneous thought during FAM as intended, they were asked to retrospectively rate the degree of spontaneous thought during each condition immediately afterward by answering the following question: "How much did you imagine or think about something?"(Brewer et al., 2011;D'Argembeau et al., 2005;Garrison et al., 2015Garrison et al., , 2014;;Yamaya et al., 2023).The questions were rated on an 11-point Likert scale (0= not at all, 10= completely, or always).
2.2.2.2.Functional near-infrared spectroscopy measurement.To assess the differential brain state between the FAM and control groups, we evaluated prefrontal activity during each FAM or control condition by using a wearable two-channel continuous-wave fNIRS (HOT-2000, NeU, Tokyo, Japan) and a data acquisition system (HOT Measure Ver 3.0, NeU) (Supplementary Fig. 1).The device consists of a single light source and two optical detectors located 1.0 cm and 3.0 cm from the light source, respectively.The near-infrared light wavelength was 800 nm, and changes in total hemoglobin were measured at a sampling rate of 10 Hz.Scalp blood flow was automatically excluded using the real-time scalp signal separating method (Kiguchi and Funane, 2014).The measurement was continuously recorded for the 1-min rest and 10-min of each FAM or control condition.The two channels covered the prefrontal cortex according to the international 10-20 method, with one channel located on Fpz, suggested to be on the mPFC (Derosière et al., 2014), and the other channel located on the right side 5.5-6.0 cm away from Fpz, suggested to be the right MFG (Supplementary Fig. 1).

Hybrid Go/NoGo task
We employed the Hybrid Go/No-Go task (Fig. 2) for 60 min to induce and evaluate the state fatigue of participants (Kato et al., 2012;Kato 2009).This task was implemented on a laptop PC using the PsychoPy software (version 2020.2.4.).The task comprised 12 blocks, with 120 trials (5 min) per block, totaling 1440 trials over 60 min.A 10-s break was included between each block.To become habituated to the task, participants performed 12 practice trials.A white fixation cross was continuously displayed in each block at the center of a black background.Participants were required to quickly push a key for a white triangle (Go stimulus), withhold their response for a white circle (No-Go stimulus), and use their left or right index finger depending on the position of the Go stimulus.Of the 120 trials in each block, 96 were Go (80 %), whereas 24 were No-Go (20 %).Both stimuli were presented in equal proportions on the right and left sides of the laptop screen.Each stimulus was presented for 100 ms, with the stimulus-stimulus interval being 2400 ms.The reaction time (RT) for a correct response to the Go stimulus and the number of correct rejections for the No-Go stimulus (correct rejection) were assessed.RT values faster than 150 ms were excluded because they were considered anticipatory responses (Egly et al., 1994;Pilz et al., 2012).

Autonomic nervous function assessments
In addition to performance decrement during a fatigue-inducing task, state fatigue can be measured by changes in psychophysiological conditions, assessed using ANF.Thus, we also complementarily measured ANF to evaluate the state fatigue of participants.A VM302 system (Fatigue Science Laboratory, Inc., Osaka, Japan) was used to assess the changes in cardiac ANF.The VM302 system simultaneously conducted electrocardiography and photoplethysmography from both sides of the fingertips at a sampling rate of 600 Hz (Supplementary Fig 2).The measurement duration was 3 min, during which participants were required to sit quietly with their eyes closed (Mizuno et al., 2017).The built-in firmware in the VM302 system detects wave peaks from electrocardiography and photoplethysmography and transmits them to an external application on a laptop that generates R-R interval and a-a interval variations (Kume et al., 2017).In the system, the low-frequency (LF) and high-frequency (HF) component power, and the LF to HF ratio (LF/HF) were automatically calculated via frequency analysis using the maximum entropy method.LF was the power within the frequency range of 0.04-0.15Hz, whereas HF was the power within the frequency range of 0.15-0.4Hz.Of note, HF is assumed to reflect parasympathetic Fig. 2. Hybrid Go/No-Go task employed as a fatigue-inducing task.A white fixation cross was continuously displayed in one block at the center of a black background.Participants were required to quickly push a key for a white triangle (Go stimulus), withhold their response for a white circle (No-Go stimulus), and use their left or right index finger depending on the position of the Go stimulus.The ratio of Go to No-Go stimuli was 96 Go trials (80 %) and 24 No-Go trials (20 %) per block, respectively.The presentation time for each stimulus (Go or NoGo) was 100 ms, while the stimulus-stimulus interval was 2400 ms.Each block consisted of 120 trials for a total of 300 s (5 min).Participants completed 12 blocks (1440 trials in 60 min) with a 10-s break between each block.For the analysis, two blocks were combined into a session.
N. Yamaya et al. nervous activity, whereas LF and LF/HF reflect sympathetic nervous activity (Hayano and Yuda, 2021).However, because both LF and HF are determined mainly by the parasympathetic nervous system (Akselrod et al., 1981;Reyes del Paso et al., 2013), and given that parasympathetic nervous activity is related to attention regulation supporting the fatigue-inducing task (Barber et al., 2020;Suess et al., 1994), HF might be an appropriate index for state fatigue.Thus, we focused on HF as a measure of state fatigue.

Subjective assessments of state fatigue, motivation, and sleepiness
Three subjective assessments (state fatigue, motivation, and sleepiness) were conducted at three times (A1-A3 in Fig. 1) to confirm whether these factors influenced state fatigue as measured by performance.Subjective fatigue was assessed using a visual analog scale (VAS) that ranged from 0 (no fatigue) to 100 (complete exhaustion) (Japanese Society of Fatigue Science, 2011).Motivation can also influence performance decrements (Müller and Apps, 2019).As meditation can increase motivation (Donald et al., 2020), we evaluated motivation using the VAS, ranging from 0 (lowest level of motivation) to 100 (highest level of motivation) (Shigihara et al., 2013).Furthermore, if participants became sleepy during the experiment, their sleepiness may have reduced their task performance.However, the psychological mechanisms of state fatigue might differ from those of sleepiness in that state fatigue does not rapidly change over the course of a few seconds, whereas sleepiness does (Hu and Lodewijks, 2020).In addition, rest reduces state fatigue but promotes sleepiness (Hu and Lodewijks, 2020).Thus, sleepiness was assessed using the Japanese version of the Karolinska Sleepiness Scale (KSS-J), which ranges from 0 (very clearly awake) to 9 (very sleepy) (Kaida et al., 2006).

Data analysis
Data from 16 of the 70 participants in the sample were excluded from the main analyses: six did not meet inclusion criteria (i.e., probability of chronic fatigue and low sleep quality), four fell asleep during the fatigue-inducing task, three dropped out of the experiment due to discomfort during the FAM or rsfMRI measurement, or due to conflicting personal schedules, one failed to undertake the fatigue-inducing task correctly, one had undetectable fNIRS data, and one had fNIRS data that did not meet the analysis criteria (explained in Section 2.3.3).Accordingly, data from 56 participants were used for the fatigue assessment analyses (i.e., fatigue-inducing task, ANF, and subjective assessments), subjective assessments for each condition (i.e., adherence and spontaneous thought), and rsfMRI.The sample size for the fNIRS analysis was 54.
All statistical analyses were performed using R, version 4.3.0.All linear mixed-effects modeling was conducted using the "lmer" function of the "lmerTest" package with the Kenward-Roger approximation, which is recommended for complex covariance structures combined with small sample sizes (Schaalje et al., 2002).The "lsmeansLT" function was used to produce least square means with Kenward-Roger approximation (Kuznetsova et al., 2017).Within-and between-group post-hoc comparisons were conducted using the "difflsmeans" function with Kenward-Roger corrections.The obtained p-value of certain comparisons from the analysis using the "difflsmeans" function, which depended on the results of linear mixed-effects modeling, was corrected via false discovery rate (FDR) correction using the "p.adjust" function from the "stats" package (Kuznetsova et al., 2017).The 95 % confidence intervals (95 % CI) of the standardized model coefficients (β) were calculated using the "confint.merMod"function.The significance level was set at p < .05.

Demographic data analysis
We used two-sample t-tests for normally distributed data and Wilcoxon rank sum tests for data that violated the normality assumption to identify between-group differences in age, sex, chronic fatigue scores (CIS-J), and sleep quality scores (PSQI-J).Factors with significant between-group differences were introduced as covariates in the main analysis.

Analysis of adherence and spontaneous thought assessments
To identify between-group differences in the degree of adherence and confirm the reduction of spontaneous thought during FAM, we ran between-group comparisons on the adherence and spontaneous thought scores of participants using the Wilcoxon rank-sum test with Cliff's delta effect sizes.

Functional near-infrared spectroscopy analysis
We preprocessed the data from fNIRS using Python (version 3.8.8).First, the missing signals were linearly interpolated.Next, after detrending, a low-pass filter with order 3 and a cut-off frequency of 0.09 Hz was performed on the 11-min signal to remove artifacts, such as heart rate (1.0 -1.5 Hz), breath (0.2 -0.5 Hz), and Mayer waves (0.1 Hz) (Franceschini et al., 2006;Naseer and Hong, 2015;Pinti et al., 2019).As raw fNIRS data cannot be directly averaged and compared across subjects or channels (Matsuda and Hiraki, 2006;Moriguchi and Hiraki, 2009), after preprocessing the fNIRS signals, we converted the signal of each channel into a z-score using the following formula: z = (x − μ)/σ, where x was the raw value during each condition period (i.e., 10 min), μ was the mean of the last half (30 s) in the rest period, and σ was the standard deviation of the last half in the rest period (Matsuda and Hiraki, 2006;Moriguchi and Hiraki, 2009).We removed the first half (30 s) in the rest period to eliminate the potential effects of scanning instability and mental conditions that occurred before the onset of the rest period.For the z-score, we calculated functional connectivity (FC) for each participant in terms of the temporal correlation of the mPFC and MFG using the Pearson correlation coefficient (Duan et al., 2012); therefore, the correlation coefficients were calculated between 6000 time points of data (i.e., 10 Hz sampling rate during 10 min examination) in each region.For group comparisons in each region, we calculated the z-score average for each channel for each participant.Participants with a z-score of mean±3SD were excluded.
To confirm differences in brain activity between the FAM and control groups, we compared the z-score average for each channel and FC between each group using two-sample t-tests, or Wilcoxon rank-sum tests if the data violated the normality assumption.

Hybrid Go/NoGo task analysis
We combined two blocks to construct a session (e.g., Session 1 = 1st block + 2nd block; Session 2 = 3rd block + 4th block), resulting in six sessions, to improve the signal-to-noise ratio by increasing the number of trials (i.e., 240 trials= 120 + 120).We calculated the average RT of correct responses for the Go stimulus for each session.We also calculated the correct rejection ratio for each session.Considering the tradeoff between reaction time and correct rejection, we calculated the inverse efficiency score (IES) for every session using the following formula: IES i = the average of RT i /the ratio of the correct rejection i (i= Session 1, 2, 3, 4, 5, and 6) (Quaglia et al., 2019; Townsend and Ashby, N. Yamaya et al. 1978).A higher IES indicated lower performance.First, we compared IES at Session 1 between groups to confirm group differences at baseline.Next, IES was analyzed using linear mixed effects modeling to consider variability within and between participants (Brown, 2021).The model included group (FAM or control), number of sessions (1 to 6), and a group × session interaction as a fixed effect.We also included the number of participants in the model as random intercepts and the centered chronic fatigue scores (i.e., CIS-J) as a covariate.It should be noted that when we included the number of participants in the model as random slopes, the model failed to converge.Post-hoc comparisons were conducted depending on the results of linear mixed effects modeling analysis.For example, if a statistically significant group × session 6 interaction was observed, we conducted group comparisons at session 6 and session comparisons between sessions 1 and 6 in each group with FDR correction.
To investigate the neural correlate of the preventive effect of FAM on performance in the fatigue-inducing task, we calculated the change in the Go/NoGo task scores using the following formula: ΔIES = IES k − IES 1 , where, IES 1 was the session 1 IES, and k denoted the number of sessions with significant group × session interactions.

Autonomic nervous function analysis
To examine temporal changes in ANF, we mainly analyzed HF and additionally analyzed LF and LF/HF before and after each condition and after the fatigue-inducing task using piecewise linear mixed modeling (Naumova et al., 2001;Pang and Ruch, 2019).Owing to differences in the degree of influence on ANF between each condition and the fatigue-inducing task, we assumed that each participant had a two-piece linear growth curve with a knot after each condition (Supplementary Fig. 3) (Naumova et al., 2001;Pang and Ruch, 2019).Thus, the time variable was split into two phases: (1) from A1 to A2 (Phase 1), and (2) from A2 to A3 (Phase 2).The time variable was dummy coded into two variables: Phase 1 (0, 1, 1) and Phase 2 (0, 0, 1) to represent the different periods.We included group (FAM and control), Phase 1, Phase 2, a group × Phase 1 interaction, and a group × Phase 2 interaction in the model as fixed effects.The LF and HF model also included the number of participants in the model as random intercepts and random slopes for Phase 2. The centered chronic fatigue scores (i.e., CIS-J) were included as a covariate because of statistically significant between-group differences.Note that the LF and HF model failed to converge when we included the number of participants in the model as random slopes for Phase 1.The LF/HF model included the number of participants as random intercepts and centered chronic fatigue scores as a covariate.This model also failed to converge when we included the number of participants in the model as random slopes for Phase 1 or Phase 2. Post-hoc comparisons were conducted depending on the results of piecewise linear mixed effects modeling analysis.For example, if a statistically significant group × Phase 2 interaction was observed, we conducted group comparisons at A3 and time comparisons between A2 and A3 in each group with FDR correction.
To investigate the neural correlates of the preventive effect of FAM on ANF, we focused on changes in LF, HF, and LF/HF from A2 to A3 due to the influence of the fatigue-inducing task on ANF.We calculated the ANF change score using the following formula: ΔV = V A3 − V A2 , where V indicated that LF, HF, and LF/HF showed significant group × Phase 2 interactions.The A2 and A3 denoted after each condition and after the fatigue-inducing task, respectively (Fig. 1).

Analysis of state fatigue, motivation, and sleepiness subjective assessments
To examine temporal changes in subjective fatigue, motivation, and sleepiness, we analyzed each score before and after each condition and after the fatigue-inducing task using piecewise linear mixed modeling, as in the ANF analysis (Naumova et al., 2001;Pang and Ruch, 2019).We dummy coded the time variable into Phase 1 (0, 1, 1) and Phase 2 (0, 0, 1) to represent the different time periods (Supplementary Fig. 3).The model included group (FAM and control), Phase 1, Phase 2, a group × Phase 1 interaction, and a group × Phase 2 interaction as fixed effects.All models also included the number of participants as random intercepts and random slopes for Phase 2. The centered chronic fatigue scores were included as a covariate.All models failed to converge when we included the number of participants in the model as random slopes for Phase 1. Post-hoc comparisons were conducted as in the ANF analysis.

Resting-state functional MRI analysis
We preprocessed the rsfMRI data using the Data Processing Assistant for Resting-State fMRI, Advanced Edition (Yan and Zang, 2010;Yan et al., 2016) on SPM 12 and MATLAB R2020a (MathWorks, Natick, MA, USA).Following our previous study (Hashimoto et al., 2022), the preprocessing methods included despiking (Clip threshold= 4) and a 17-tap high pass filtering from the ArtRepair toolbox in SPM (Mazaika et al., 2005), realignment to the first volume, slice timing correction, T1 image coregistration to fMRI data, T1 image segmentation with a diffeomorphic anatomical registration through an exponentiated lie (DARTEL), the algebraic registration process, normalization to the Montreal Neurological Institute (MNI) space by DARTEL, detrending, nuisance covariates regression, smoothing (FWHM 6 mm), and temporal filtering (band pass, 0.01 to 0.1 Hz).Nuisance covariate regression was conducted using the CompCor method with five principal components (Behzadi et al., 2007;Chai et al., 2012).In addition, we used the Friston-24 model to regress out nuisance covariates, including six head motion parameters, six head motion parameters at one time point before, and 12 corresponding squared items.The global mean signal was not regressed because global signal regression removes true neuronal signals and can diminish the connectivity-behavior relationship (Murphy and Fox, 2017).The exclusion criteria of mean frame-wise displacement (FD Power > 0.3 mm) were used to account for excessive head movement (Lu et al., 2017;Power et al., 2012); however, no participants had a mean FD > 0.3 mm.
For the first level, we performed seed-to-voxel analyses using the 19 seed ROIs to separately calculate rsFC for the first and second scans with Pearson's correlations.The 19 seeds were defined from 20 seeds obtained by a systematic review of FC studies of mindfulness (no coordinates of one seed) (Mooneyham et al., 2016) (Supplementary Table 1).After calculating rsFC for the first and second scans separately, we also calculated the changes in rsFC from the first to the second scan (ΔrsFC) for each group.
For the second level, to compare ΔrsFC between each group, we conducted t-tests with contrasts: FAM group > control group and FAM group < control group.FD, age, and sex were included as covariates.The statistical threshold was set to family-wise error (FWE) correction p < .05 at the cluster level with uncorrected p < .001at the voxel level.
To explore the neural correlates of the preventive effect of FAM on performance in the fatigue-inducing task (explained in Section 2.3.4) and ANF responses (explained in Section 2.3.5),we calculated the Pearson correlation coefficient between the ΔrsFC that showed significant group differences and each of the indices, including chronic fatigue scores (i.e., CIS-J) as a covariate.For data that violated the normality assumption, Spearman's rank correlation coefficient was calculated.
The data and code are publicly available via the Open Science Framework at https://osf.io/vr6kn/.

Demographic analysis
A statistically significant between-group difference was detected in CIS-J scores (t= − 2.29, df= 41.6, p= .027,Hedges's g = − 0.62), indicating that chronic fatigue was significantly higher in the FAM group (M = 49.7,SD= 14.6) than in the control group (M = 42.1,SD= 9.4).However, no statistically significant between-group differences were N. Yamaya et al. observed for age, sex, or sleep quality score (for detail, see Supplementary Analysis and Results 1).Thus, the CIS-J score was included as a covariate in the fatigue assessment analysis to minimize any potential confounding influence of chronic fatigue.

Adherence and spontaneous thought assessments
The Wilcoxon rank sum test showed no statistically significant between-group differences in adherence to the instructions (Z = 1.72, p= .086,Cliff's Delta= 0.26), indicating that the adherence of the FAM group (M = 7.5, SD= 1.4) was similar to that of the control group (M = 8.2, SD= 1.2).However, a statistically significant difference in spontaneous thoughts was observed between groups (Z = 3.30, p < .001,Cliff's Delta= 0.51), with the spontaneous thoughts in the FAM group (M = 3.3, SD= 1.9) being significantly reduced compared with those in the control (M = 5.2, SD= 2.3).

Fatigue-inducing task performance
The results of the descriptive data (least square mean: LSM and 95 % confidence interval: 95 % CI) are shown in Supplementary Table 2, while the linear mixed effects model results are shown in Table 1.No statistically significant difference in IES at Session 1 was found between groups (for detail, see Supplementary Analysis and Results 2).The linear mixed effects model analysis revealed statistically significant simple effects in Sessions 3, 4, 5, and 6, indicating that the performance in the control group was decreased at these sessions.Importantly, a statistically significant FAM × Session 6 interaction (β = − 0.04, df= 270.0, t= − 2.12, 95 % CI= [− 0.07, − 0.00], p= .035)was observed, which indicated that compared with Session 1, the performance of the FAM group in Session 6 was improved more than that of the control group.Followup comparisons with FDR correction showed that the performance of the control group in Session 6 was significantly lower than that in Session (Session 1 minus Session 6; β = − 0.06, df= 270.0, t= − 4.60, 95 % CI= [− 0.08, − 0.03], p = 1.96 × 10 − 5 ).Whereas, no statistically significant difference was observed in the performance of the FAM group in Session 6 compared with that in Session 1 (Session 1 minus Session 6; β = − 0.02, df= 270.0, t= − 1.39, 95 % CI= [− 0.04, 7.7 × 10 − 3 ], p= .25).A betweengroup difference in performance in Session 6 was not statistically significant (control group minus FAM group; β = 0.01, df= 83.5, t = 0.41, 95 % CI = [− 0.04, 0.06], p= .68).These results indicated that the control group showed a gradual performance decrement, whereas the FAM group did not (Fig. 4).Given the significant group × Session interaction, we calculated the Go/NoGo task change score using the following formula: ΔIES = IES 6 − IES 1 , where IES 6 referred to Session and IES 1 referred to Session 1.
Sleepiness (Supplementary Table 6) showed no statistically significant FAM × Phase 1 interaction effect (β = 0.3, df= 54.0, t = 0.87, 95 % CI= [− 0.4, 1.1], p= .39)or FAM × Phase 2 interaction effect (β = 0.5, df= 66.7, t = 0.85, 95 % CI= [− 0.7, 1.7], p= .40).In addition, neither a simple effect of Phase 1 was significant nor that of Phase 2. These results  IES can be interpreted as reaction time, indicating that higher IES means lower performance.The FAM group maintained its performance over the sessions, whereas the control group showed a gradual decrement in its performance over the sessions.No statistically significant difference in IES at Session 1 was observed between groups.The linear mixed effects model results showed a statistically significant FAM × Session 6 interaction (p < .05).The control group showed a statistically significantly higher IES in Session 6 than that in Session 1 (p with FDR correction < 0.05).Note.FAM: focused attention meditation.Phase 1 (before to after each condition) indicates the contrast before and after each condition, and after the fatigueinducing task, specified as 0, 1, and 1, separately; whereas, in Phase 2 (before to after the task) they were specified as 0, 0, and 1, separately.CIS-J: the Japanese version of the Checklist Individual Strength Questionnaire (chronic fatigue score).HF: the high-frequency component power, which indicates parasympathetic nervous function activity.Estimates: partial regression coefficient.df: degrees of freedom.CI: confidence interval.Positive FAM × Phase 1 interaction coefficients indicate that the HF was increased in the FAM group after the condition compared with before more than that in the control group.Positive FAM × Phase 2 interaction coefficients indicate that the HF was increased in the FAM group after the fatigue-inducing task compared with before the fatigueinducing task more than that in the control group.
N. Yamaya et al. indicated that FAM had no significant effect on these subjective experiences immediately after the condition or after the fatigue-inducing task.

Effect of one-session focused attention meditation on resting-state functional connectivity
Among 19 seeds, we found statistically significant between-group differences in three seeds of ΔrsFC: right PCC (8, − 56, 30) (Fig. 5a), left PCC (− 5, − 49, 40) (Fig. 5b), and left dmPFC (− 3, 27, 51) (Fig. 5c).Compared with the FAM group, the control group showed significantly higher ΔrsFCs between the right PCC seed and left inferior occipital gyrus (IOG), and between the left PCC seed and left IOG and right occipital fusiform gyrus (Table 4).Whereas, the FAM group showed a significantly higher ΔrsFC between the left dmPFC seed and right superior parietal lobule (SPL) than that in the control group (Table 4).
To explore the neural correlates between the ΔrsFC that showed significant group differences and changes in performance in the fatigueinducing task, we calculated the Pearson correlation coefficient between each of the three ΔrsFC, which was shown as a significant group difference, and ΔIES (i.e., IES in Session 6 minus Session 1) while controlling CIS-J in each group.In the FAM group, a statistically significant negative correlation was found in ΔrsFC between the right PCC and left IOG and ΔIES (r= − .43,p= .032)(Fig. 6a).No significant correlations were found in ΔrsFC between the left PCC and left IOG and ΔIES (r= − .12,p= .57)(Fig. 6b), or in ΔrsFC between the left dmPFC and right SPL and ΔIES (r= 0.065, p= .76)(Fig. 6c).In the control group, a Note.FAM: focused attention meditation.Phase 1 (before to after each condition) indicates the contrast before and after each condition, and after the fatigueinducing task, specified as 0, 1, and 1, separately; whereas in Phase 2 (before to after the task) these were specified as 0, 0, and 1, separately.CIS-J: the Japanese version of the Checklist Individual Strength Questionnaire (chronic fatigue score).Estimates: partial regression coefficient.df: degrees of freedom.CI: confidence interval.Positive FAM × Phase 1 interaction coefficients indicate that the subjective fatigue of FAM participants was increased after condition compared with that before more than that in the control group.Positive FAM × Phase 2 interaction coefficients indicate that the subjective fatigue of FAM participants was increased after the fatigue-inducing task compared with that before more than that in the control group.statistically significant positive correlation was found in ΔrsFC between the left PCC and left IOG and ΔIES (r= 0.40, p= .030)(Fig. 6b).No significant correlations were found in ΔrsFC between the right PCC and left IOG and ΔIES (r= 0.13, p= .51)(Fig. 6a), or in ΔrsFC between the left dmPFC and right SPL and ΔIES (r= − .14, p= .46)(Fig. 6c).
To explore the neural correlates between ΔrsFC that showed significant group differences and the change in ANF, we calculated the Spearman's rank correlation coefficient between each of the three ΔrsFC that was shown as a significant group difference and ΔHF, while controlling CIS-J in each group.As we could not obtain appropriate HF data from some participants due to the VM302 system showing warnings that indicated reduction of reliability in the data, we excluded participants who received the system warning at A2 or A3 in our calculations.Consequently, we included 19 participants from the FAM group (excluding seven participants) and 24 participants from the control group (excluding six participants).In the FAM group, statistically significant positive correlations were found in ΔrsFC between the right PCC and left IOG and ΔHF (rho =0.59, p= .011)(Fig. 7a), and in ΔrsFC between the left PCC and left IOG and ΔHF (rho= 0.60, p= .0078)(Fig. 7b).No significant correlation was found in ΔrsFC between the left dmPFC and right SPL and ΔHF (rho= 0.061, p= .81)(Fig. 7c).In the control group, no statistically significant correlations were found in ΔrsFC between the right PCC and left IOG and ΔHF (rho= 0.24, p= .26)(Fig. 7a), in ΔrsFC between the left PCC and left IOG and ΔHF (rho= 0.18, p= .41)(Fig. 7b), and in ΔrsFC between the left dmPFC and right SPL and ΔHF (rho= − .36,p= .088)(Fig. 7c).

Discussion
This is the first study to reveal that one-session brief FAM has a preventive effect on state fatigue.We evaluated state fatigue based on performance in a 60-min fatigue-inducing task.In addition, we evaluated state fatigue using ANF and subjective assessments.We also explored the effect of one-session brief FAM on rsFC.To examine the neural mechanism of the preventive effect of FAM on state fatigue, we investigated between-group differences in ΔrsFC to identify the effects of FAM on rsFC and its neural correlates with state fatigue.Finally, we assessed brain activity during each experimental condition using an fNIRS instrument to clarify differences in brain state during the performance of each condition between groups.The results of the fatigueinducing task showed that the 10-minute FAM could prevent state fatigue.This preventive effect of FAM might stem from the greater ΔrsFC, which is related to the modification of attention regulation, and the suppression of ΔrsFC, which is related to poor attention regulation and reduced higher-order cognitive function.

Effects of focused attention meditation on behavioral, autonomic, and subjective measures of state fatigue
The FAM group maintained its performance in the fatigue-inducing task, whereas the task performance of the control group was gradually decreased over time.These results suggested that the FAM group did not experience state fatigue, whereas the control group did.These findings were consistent with those of our previous research, which showed that one-session FAM could prevent a decline in performance when cognitively demanding tasks were repeated intermittently (Yamaya et al., 2023).However, motivation and sleepiness did not explain the differences between the groups.No apparent increment or reduction in motivation might be attributed to differences in forms of motivation (Donald et al., 2020) or in the methods of one-session FAM (Hafenbrack and Vohs, 2018).Our results suggested that one-session brief FAM could prevent the state fatigue caused by prolonged cognitive demands.
Despite decreases in task performance, neither the HF or LF, or LF/ HF, changed from before to after the fatigue-inducing task in the control group.Thus, the ANF assessment suggested that the control group did not exhibit state fatigue despite the appearance of state fatigue in the behavioral assessment.The reason for the discrepancy in the development of state fatigue between the behavior and ANF assessments might be the extent to which fatigue-inducing tasks influence behavior and ANF as a result of the intensity of demand of the task (Rosa et al., 2022).It is possible that the effect of the fatigue-inducing task used in this study on ANF may be smaller than its effect on behavior.This small effect might have caused greater dispersion in this sample size, resulting in no significance.In contrast, the FAM group showed increased HF from before to after the fatigue-inducing task (i.e., from A2 to A3 in Fig. 1), suggesting enhanced parasympathetic nervous activity.As state fatigue is characterized by reduced parasympathetic nervous activity (Melo et al., 2017;Mizuno et al., 2011;Tanaka et al., 2011), FAM might have had a potential preventive effect on state fatigue.
In contrast to the behavioral and ANF assessments, no statistically significant effect of FAM on subjective fatigue was found.The discrepancy between the behavioral and subjective assessments in the FAM group might be due to the small effect of the one-session brief FAM.Subjective fatigue might occur before performance decrements become apparent (Benoit et al., 2019;Kanfer, 2011); therefore, a larger effect might be needed for FAM to prevent subjective fatigue.In summary, the one-session brief FAM could prevent behavioral state fatigue (task performance), but did not prevent subjective fatigue.Thus, we partially

Effect of focused attention meditation on a resting-state functional MRI
The rsFCs in the control group might have changed as a result of more spontaneous thought.The control group showed higher ΔrsFCs between the default network and secondary visual cortex.Increased FC between the default and visual networks has been suggested to be associated with spontaneous thought (Zhou and Lei, 2018).The default network contributes to internally oriented cognition such as spontaneous thought (Andrews-Hanna et al., 2010;Christoff et al., 2016).The IOG plays a role in internal visual representations during spontaneous thought (Groot et al., 2022;Kosslyn et al., 2001;Petro et al., 2017).As the control group was not required to regulate or suppress spontaneous thought, the higher ΔrsFCs between the default network and secondary visual cortex in this group might have been induced by uncontrolled spontaneous thought for 10 min.
The FAM group showed a higher ΔrsFC between the default network and right SPL.The SPL, which is one of the dorsal attention network regions (Corbetta and Shulman, 2002;Fox et al., 2005), plays a role in controlling attentional focus size for increasing attentional processing efficiency (Zeng et al., 2017) and attention orienting (Corbetta and Shulman, 2002;Fan et al., 2005).The ΔrsFC between the default network and SPL might reflect modified attention regulation, similar to the rsFC between the default and frontoparietal network (Sezer et al., 2022).A previous study showed that the longitudinal effects of FAM increased rsFC between the SPL and certain default network regions, suggesting that FAM enhances the ability to rapidly switch between spontaneous thought and a concentration state and the ability to maintain this concentration state (Zhang et al., 2021).Thus, a higher ΔrsFC between the default network and SPL might reflect a modification of attention regulation for maintaining the concentration state induced by FAM.
One of the reasons for the lack of changes in some rsFCs might be due to the short FAM duration.Although the frontoparietal network has a close relationship with the dorsal attention network in the network topology, the frontoparietal network may be more related to higher-order cognitive control (executive function) than the dorsal attention network (Corbetta and Shulman, 2011;Dixon et al., 2018;Sezer et al., 2022).Considering the hierarchical relationship between attention and executive function (Glisky, 2007), we suggested that only one brief session FAM could change rsFC related to attention regulation (i.e., rsFC between the default and dorsal attention network), instead of rsFC related to executive function (i.e., rsFC between the default and frontoparietal network).Thus, to achieve changes in rsFC between the default and frontoparietal networks, participants might need more practice, because developing the ability to regulate attention and cognition requires more training (Fell et al., 2010;Lutz et al., 2008).Similarly, participants might require more time for engaging in FAM to increase rsFC between the salience network and default network regions, as well as within the default network regions.Although further research is needed to confirm this premise, our second hypothesis was partially confirmed.

Neural correlates of the fatigue-inducing task and autonomic nervous function
In the control group, the higher ΔrsFC between the left PCC and left IOG was related to a decline in task performance.A previous study suggested that a greater anticorrelation between the default network and visual cortex is associated with better attention regulation (Barber et al., 2015).In addition, increased rsFC between the default network and visual cortex may reflect visual network failure, resulting in a reduction of higher-order cognitive functions such as object recognition and reading in schizophrenia and at-risk mental states (Javitt, 2009;Sasabayashi et al., 2023).Thus, the increased ΔrsFC between the default network and visual cortex stemming from greater spontaneous thought might reflect the brain state that induced poor attention regulation and reduced higher-order cognitive function, resulting in a decrease in task performance over time (Thomson et al., 2015;Yamaya et al., 2023).
Though the FAM group exhibited higher ΔrsFC between the left dmPFC and right SPL than that of the control group, no relationship was observed between the ΔrsFC and task performance.We speculated that other factors might mediate or moderate the relationship between the ΔrsFC and maintaining steady task performance.For example, the activation of networks related to cognitive control, such as the DLPFC, and the deactivation of the default network during the fatigue-inducing task may be a possible mediator/moderator for preventing performance decline (Fiene et al., 2018;Hanken et al., 2016;Ishii et al., 2014;Salihu et al., 2022).Consequently, the ΔrsFC related to the modification of attention regulation may not be directly associated with the maintenance of task performance.Another possibility is that the maintained task performance in the FAM group might stem from the suppression of increased rsFC between the default network and left IOG.The control group showed significantly higher ΔrsFC between the default network (left PCC) and left IOG, as well as a positive relationship with performance decrement.In contrast, no significant relationship with the same rsFC was observed in the FAM group.As noted above, increased rsFC between the default network and visual cortex might be related to reduced cognitive function (Javitt, 2009;Sasabayashi et al., 2023).Thus, the absence of increased rsFC between the left PCC and left IOG might have allowed participants to avoid the brain state that would induce reduced higher-order cognitive function, resulting in the maintenance of steady task performance.
In addition, under the FAM condition, greater spontaneous thought might be related to the maintenance of task performance.In the FAM group, the ΔrsFC between the right PCC and left IOG was negatively related to ΔIES.This result suggested that the participants in the FAM group showing higher ΔrsFC between the right PCC and left IOG could improve their task performance even more as time passed, which could be interpreted as preventing state fatigue.As mentioned above, positive change in ΔrsFCs between the bilateral PCC and left IOG might stem from more spontaneous thought.Thus, in contrast to the control group, more spontaneous thoughts might be related to improved task performance in the FAM group.FAM requires attention regulation for switching between focused and spontaneous thoughts, and this switch itself, rather than dwell time in the focused mind, might be a factor in modifying attention regulation ability (Hasenkamp et al., 2012;Lim et al., 2018;Malinowski, 2013).As such, during FAM, which required the suppression of spontaneous thought, participants who experienced spontaneous thoughts more frequently might have been required to switch between focused and spontaneous thoughts.This frequency of attentional switching during FAM might have resulted in a modification of attention regulation when implementing fatigue-inducing tasks.Collectively, we suggest that the ΔrsFC between the right PCC and left IOG in the FAM group might be related to an improvement in task performance.
Regarding the neural correlates of ANF, we suggest that the neural correlates of parasympathetic nervous activity differed from those of task performance.In the control group, a significant association was observed between ΔrsFC and task performance, whereas no significant relationships were found between ΔrsFCs and parasympathetic nervous activity.These results indicated that ΔrsFCs induced by control condition might not be related to changes in parasympathetic nervous function.Another interpretation was that the small effect of the fatigueinducing task used in this study prevented us from detecting a significant relationship.As discussed above, the effect of the fatigue-inducing task on ANF in this study might be smaller than that on behavior and subjective feeling.This small effect might have caused greater dispersion in this sample size, resulting in no significance.
As with the neural correlate of task performance in the FAM group, the neural correlates of parasympathetic nervous activity may be interpreted from a similar perspective.Although the ΔrsFCs between the bilateral PCC and left IOG were increased more in the control than the FAM group, significantly positive relationship between ΔrsFCs and parasympathetic nervous activity were only observed in the FAM group.These results suggested that parasympathetic nervous activity might have been increased more in participants in the FAM group showing higher ΔrsFCs between the bilateral PCC and left IOG.
As mentioned above, because positive change in ΔrsFCs between the bilateral PCC and left IOG might stem from the time participants engaged in spontaneous thought, the positive relationship between ΔrsFCs and parasympathetic nervous activity can be interpreted as increased parasympathetic nervous activity being more likely among participants in the FAM group who experienced spontaneous thought.Considering that the core component of FAM is the frequency of attentional switching between focused and spontaneous thoughts, the participants who experienced spontaneous thoughts during FAM engaged more in this component.Thus, the positive relationship between ΔrsFCs and the parasympathetic nervous system may reflect the mechanism of the effect of FAM on parasympathetic nervous activity.This interpretation is supported by a previous study showing that FAM could increase rsFC between the PCC and visual cortex (Zhang et al., 2021).Additionally, another previous study showed that the PCC in the default network is involved in parasympathetic nervous system activity (Beissner et al., 2013).The PCC and visual cortex are also involved in the regulation of parasympathetic nervous activity (de la Cruz et al., 2019).Overall, we suggest that the ΔrsFCs between the bilateral PCC and left IOG in the FAM group might be related to an increase in parasympathetic nervous activity.

Brain activity during focused attention meditation
The differences in each condition state between the FAM and control groups might reflect differences in the instructions that participants were given.In the subjective assessments of the degree of spontaneous thought, the FAM group reported significantly lower levels of spontaneous thoughts than those in the control group.In addition, the FAM group showed lower mPFC activity and weaker FC between the mPFC and MFG in the fNIRS results than those in the control group.FAM participants were required to focus on their breath, suppress spontaneous thoughts, and return their attention to their breath whenever it wandered.This attention regulation is related to the deactivation of mPFC and activation of MFG (Brewer et al., 2011;Hasenkamp et al., 2012;Scheibner et al., 2017).A previous fNIRS study showed that spontaneous thought increases mPFC activation (Durantin et al., 2015).Thus, the fNIRS results might reflect the difference between the FAM and control groups in terms of attention regulation for suppressing spontaneous thoughts, corresponding to differences in the instructions they were provided with.Consequently, the FAM group may have been more likely to implement attention regulation based on the provided instructions for performing FAM.To confirm the adherence of participants to the instructions given for each condition, previous studies have evaluated the consequence of attention regulation during FAM by assessing the degree of spontaneous thoughts retrospectively (Brewer et al., 2011;D'Argembeau et al., 2005;Garrison et al., 2015Garrison et al., , 2014;;Yamaya et al., 2023).However, individuals sometimes engage in spontaneous thoughts that evade self-reported detection, even in conditions where they attempt to suppress such thoughts (Baird et al., 2013).Thus, it may be useful to compensate for the degree of spontaneous thoughts during FAM in real-time, in addition to retrospectively, to confirm that participants conduct FAM as intended (Gearing et al., 2011).

Effects of one-session exercise in meditation naïve participants
The dynamics of the meditation progress have been suggested, and novices need effort to sustain intended focus (Fell et al., 2010;Lutz et al., 2008).Meditation-naïve subjects in this study might not have been able to focus and regulate attention as required by the FAM style.The changes observed across various parameters might also be related to increased cognitive effort to do so rather than by really doing so (Lutz et al., 2008;Yamaya et al., 2023).

Limitations
This study had four limitations.First, the degree of chronic fatigue differed between groups.Even though we randomly divided the participants into groups, we found statistically significantly higher chronic fatigue in the FAM than in the control group.Therefore, although we controlled for the differences in chronic fatigue using statistical methods, these differences might have reduced the preventive effect of FAM on fatigue.
Second, we could not detect any obvious fatigue in the ANF assessment of the control group.In addition to the small effect of the fatigueinducing task on ANF discussed above, this might have resulted from differences in sensitivity to time-on-task influences on ANF indices.A previous study suggested that frequency-domain indices (i.e., LF, HF, and LF/HF) are less sensitive to time-on-task influences than timedomain indices such as pNN50, which is a percentage of consecutive RR intervals that are greater than or equal to 50 ms (Melo et al., 2017).Furthermore, one systematic review indicated that frequency-domain measures demonstrate more variation between studies than time-domain measures (Nunan et al., 2010).Thus, a fatigue-inducing task might have affected ANF in the control group; however, the lower sensitivity of those indices did not allow for statistical significance.Further studies are required to assess ANF using time-domain indices.
Third, although we interpreted the relationship of ΔrsFCs between the PCC and IOG and ΔIES in the FAM group as the frequency of switching mental states, we did not confirm this frequency.Our previous study showed a positive correlation between the degree of spontaneous thoughts during FAM and the effort participants exerted to conduct FAM, indicating that they may have switched mental states frequently (Yamaya et al., 2023).Therefore, participants in the current study may have similarly switched their mental state between a focused mind and spontaneous thoughts.Further studies are required to confirm this assumption.
Last, a considerable number of participants were excluded from analysis.This reduction in the sample size might have prevented the generalization of our results.

Conclusions
Our findings suggested that one session of 10-minute FAM could prevent behavioral state fatigue in meditation-naïve participants.This preventive effect of FAM could stem from greater ΔrsFC, which could be related to the modification of attention regulation by cognitive effort, and the suppression of ΔrsFC, which could be associated with poor attention regulation and reduced higher-order cognitive function.The findings of this study contribute to the accumulating knowledge regarding the efficacy of one-session brief FAM for preventing state fatigue and scientifically advance the understanding of FAM.Our study also offers novel insights for the development of nonpharmacological methods for the treatment of fatigue.

Fig. 3 .
Fig. 3. Group comparison of brain activity during focused attention meditation (FAM) or control condition.(a): The normalized functional near-infrared spectroscopy (fNIRS) signal in the medial prefrontal cortex (mPFC).A two-sample t-test showed statistically significant lower mPFC activity in the FAM group compared with that in the control group (p < .05).(b): The normalized fNIRS signal in the middle frontal gyrus (MFG).The Wilcoxon rank sum test revealed no statistically significant between-group difference in the MFG.(c): The normalized Pearson correlation coefficient representing functional connectivity (FC) between the mPFC and MFG.The Wilcoxon rank sum test revealed significantly lower FC in the FAM group than that in the control group (p < .05).Each semi-transparent dot indicates individual participant data.The centerline in each box plot shows the median, and the box extent shows the interquartile range (IQR; 25th-75th percentile).Individual whiskers extend to 1.5 × IQR.

Fig. 4 .
Fig. 4. The mean values of each group at each time point of the fatigueinducing task.The red line represents the focused attention meditation (FAM)group.The blue line represents the control group.The x-axis represents the six sessions, each consisting of two blocks, and indicates approximately every 10min progress from the beginning of the task (e.g., session 6 indicates approximately 60 min progress from the beginning of the task).The y-axis represents the inverse efficiency score (IES) of the fatigue-inducing task.IES can be interpreted as reaction time, indicating that higher IES means lower performance.The FAM group maintained its performance over the sessions, whereas the control group showed a gradual decrement in its performance over the sessions.No statistically significant difference in IES at Session 1 was observed between groups.The linear mixed effects model results showed a statistically significant FAM × Session 6 interaction (p < .05).The control group showed a statistically significantly higher IES in Session 6 than that in Session 1 (p with FDR correction < 0.05).

Fig. 5 .
Fig.5.Group differences in changes in resting-state functional connectivity (rsFC).After calculating rsFC separately before and after each condition, we calculated changes in rsFC before and after each condition for each group (ΔrsFC) and compared them between groups.(a): The control group showed a higher ΔrsFC between the right posterior cingulate cortex (PCC) and left inferior occipital gyrus than that in the focused attention meditation (FAM) group.(b): The control group showed a higher ΔrsFC between the left PCC seed and bilateral occipital areas than that in the FAM group.(c): The FAM group showed a higher ΔrsFC between the left dorsomedial prefrontal cortex (dmPFC) and the right superior parietal lobule than that in the control group.The voxel-level p-value threshold was 0.001, and the FWE-corrected cluster-level p-value was 0.05.The color bar shows the range of t-values.

Fig. 6 .
Fig. 6.Correlations between changes in resting-state functional connectivity before and after each condition and changes in performance in the fatigue-inducing task.The task performance was the inverse efficiency score (IES), and change (Δ) IES was calculated using the IES in Session 6 minus that in Session 1. (a): Pearson correlation coefficient between changes in resting-state functional connectivity (ΔrsFC) between the right posterior cingulate cortex (PCC) and left inferior occipital gyrus (IOG) and ΔIES.Higher ΔrsFC was only associated with improved task performance from Sessions 1 to 6 in the focused attention meditation (FAM) group.(b): Pearson correlation coefficient in ΔrsFC between the left PCC and left IOG and ΔIES.Higher ΔrsFC was only associated with decreased task performance from Sessions 1 to 6 in the control group.(c) Pearson correlation coefficient in ΔrsFC between the left dorsomedial prefrontal cortex and right superior parietal lobule and ΔIES.No statistically significant relationship was found for either group.Each dot indicates individual participant data.Shading: 95 % confidence interval.

Fig. 7 .
Fig. 7. Correlations between changes in resting-state functional connectivity before and after each condition and changes in autonomic nervous function.The autonomic nervous function was represented by high-frequency component power (HF), with the change (Δ) in HF being calculated as HF at A3 minus that at A2. (a): Spearman's rank correlation coefficient of changes in resting-state functional connectivity (ΔrsFC) between the right posterior cingulate cortex (PCC) and left inferior occipital gyrus (IOG) and ΔHF.Higher ΔrsFC was only associated with increased HF from after the condition to after the task in the focused attention meditation (FAM) group.(b): Spearman's rank correlation coefficient in ΔrsFC between the left PCC and left IOG and ΔHF.Higher ΔrsFC was only associated with increased HF from after the condition to after the task in the FAM group.(c): Spearman's rank correlation coefficient in ΔrsFC between the left dorsomedial prefrontal cortex (dmPFC) and right superior parietal lobule and ΔIES.No statistically significant relationship was found for either group.Each dot indicates individual participant data.Shading: 95 % confidence interval.

Table 1
Linear mixed effect model tests of performance in the fatigue-inducing task by session and group.Session means the combination of every two blocks depicted in Fig.1.CIS-J: the Japanese version of the Checklist Individual Strength Questionnaire (chronic fatigue score).IES: inverse efficiency score of the fatigue-inducing task.Estimates: partial regression coefficient.df: degrees of freedom.CI: confidence interval.Negative coefficients of the FAM × session interaction indicate that compared with Session 1, the FAM group decreased its IES (i.e., improvement of performance) more than that of the control group.

Table 2
Piecewise linear mixed effect model tests of high-frequency component power by phase and group.

Table 3
Piecewise linear mixed effect model tests of subjective assessments for fatigue by phase and group.

Table 4
Summary of cluster-level statistics for functional connectivity showing significant differences between focused attention meditation and control condition.