Exploring the impact of transcutaneous vagus nerve stimulation in subjects with and without burnout: Potential benefits for executive function and neural processing

Abstract

Introduction

Many brain disorders and conditions, including occupational burnout, are linked with challenges in executive function (EF). Yet, there is a lack of treatments geared at restoring them. We have previously demonstrated that VNS enhances EF in patients with epilepsy and that transcutaneous VNS (tVNS) modulates the underlying neural processes in healthy subjects. In this study, we investigated the immediate impact of tVNS on EF in subjects with and without occupational burnout.

Material and methods

We hypothesized that tVNS enhances EF, as reflected in both behavioral and neural levels. Subjects with (n = 27) and without burnout (n = 24) performed an integrated computer-based test of EF, the Executive Reaction Time (RT) test. At the same time, EEG was recorded and tVNS or sham stimulation was delivered to each subject in an alternating manner. Event-related potentials, N2 and P3, peak and interpeak amplitudes, and latencies were used to evaluate the speed and allocation of neural processes linked with EFs and errors and RTs to assess performance.

Results

Frontal N2-P3 interpeak latency (IPL) in the NoGo condition was shorter with active tVNS (m = 157.6 ms (IQR = 59.2 ms)) than with sham (m = 169.6 ms (IQR = 88.4 ms), p = 0.01). Further, active tVNS reduced total errors in healthy subjects.

Conclusion

Active tVNS resulted in partly accelerated neural processing in the context of response inhibition in both groups and enhanced EF performance in subjects without burnout. We suggest that tVNS enhances neural processes underlying EFs in specific situations. Even though caution is warranted, tVNS shows some promise as a potential cognitive enhancer.

1. Introduction

Transcutaneous vagus nerve stimulation (tVNS) is a relatively recent, safe, and affordable neuromodulation method designed as a non-invasive alternative to invasive vagus nerve stimulation (VNS) (Redgrave et al., 2018). While VNS is already FDA-approved for therapeutic use in drug-resistant epilepsy and depression, the therapeutic potential of tVNS for a broader range of diseases and cognitive impairments remains under active investigation. Despite the growing prevalence of conditions compromising cognitive function, especially executive functions (EF), such as depression (Rock et al., 2014), and burnout (Pihlaja et al., 2022), pharmacological therapies for cognitive dysfunction are limited. TVNS has been associated with a variety of cognitive benefits (Fischer et al., 2018; Beste et al., 2016; Borges et al., 2020; Miyatsu et al., 2024). To that end, it may offer potential for restoring and enhancing cognitive functions in individuals experiencing occupational burnout. This, in turn, could help prevent and alleviate work incapacity, facilitating a quicker return to work for those affected by occupational burnout.

We have previously shown that invasive VNS improves working memory performance and enhances neural processes related to attention in subjects with drug-resistant epilepsy (Sun et al., 2017a). Our findings showed that non-invasive VNS (tVNS), compared to sham stimulation, reduced the frontal N2 event-related potential (ERP) without impairing performance on an EF task (Pihlaja et al., 2020). This suggests that tVNS may enhance cognitive control-related neural efficiency, requiring fewer cognitive resources to achieve the same performance level. Supporting this interpretation, Fischer et al. (2018) also reported a reduction in the N2 component during a task involving cognitive flexibility—a distinct aspect of EF that relies on cognitive control. Additionally, Wang et al. (2022) found improved cognitive control performance following tVNS.

Beyond these effects, tVNS has also been shown to positively influence a range of cognitive functions, including attentional processing (Ventura-Bort et al., 2018), associative memory (Jacobs et al., 2015), speech category learning (Llanos et al., 2020), vocabulary acquisition (Miyatsu et al., 2024), and inhibitory control processes dependent on working memory (Beste et al., 2016). Notably, a recent meta-analysis suggests that the cognitive effects of tVNS are particularly observed within the domain of executive functions (EF) (Ridgewell et al., 2021). Given that EF are higher-order cognitive processes responsible for regulating other cognitive functions (Diamond, 2013), it is plausible that the observed improvements across various cognitive domains are mediated by enhancements in EF.

In addition to its cognitive benefits, tVNS has been found to exert positive effects on emotional functioning in clinical populations, such as depression (Liu et al., 2016; Koenig et al., 2021) aligning with the established use of invasive VNS for treating depression. EF and affective processes are closely intertwined. Emotionally salient, particularly threat-related, stimuli readily capture attention through bottom-up mechanisms, thereby drawing on EF resources (Hartikainen et al., 2000, 2010a, 2012a, 2012b; Hartikainen, 2021). Conversely, EF facilitates top-down regulation of emotional processes and their interaction with attentional processes, helping to prevent excessive or prolonged focus on negative emotional stimuli (Mäki-Marttunen et al., 2014), a phenomenon observed in depression (Armstrong and Olatunji, 2012), a frequent concomitant to burnout.

Individuals with occupational burnout frequently report challenges in cognitive and affective functions, such as attention, working memory and emotional control. Recently, we reported that occupational burnout is linked with subjective challenges in EF in daily life (Pihlaja et al., 2022, 2023). Although objective EF performance remained intact, ERPs indicated that maintaining these performance levels required greater neural resource allocation (Pihlaja et al., 2023). Additionally, we found a significant correlation between EF and burnout severity, with more severe burnout associated with weaker EF (Pihlaja et al., 2022).

Despite occupational burnout becoming a global crisis, effective treatment options remain scarce, and its impacts on brain function and brain health remain largely unexplored. In addition to its potential beneficial effects on EF, tVNS has been suggested to have a positive influence on stress physiology. Consequently, tVNS demonstrates potential as a therapeutic intervention for stress-related disorders, including those characterized by cognitive and affective dysfunctions such as depression, post-traumatic stress disorder (Bremner et al., 2020) and burnout. Further, McIntire et al. (2021) demonstrated that tVNS enhances arousal and reduces subjective fatigue in healthy sleep-deprived individuals, while also improving multitasking performance, closely tied to EF.

To enhance the sensitivity of detecting subtle changes in EF, whether improvements or decrements, it may be beneficial to simultaneously engage affective circuits and incorporate EF tasks with threat-related distractors that tend to capture attention and EF resources (Hartikainen, 2021). In addition to engaging both affective and cognitive circuits, simultaneously measuring both behavior and electrophysiology allows for sensitive detection of alterations in EF due to neuromodulation (Sun et al., 2017b).

P3 is a vastly studied event-related potential (ERP) component, that has been applied in both tVNS and burnout research. De Taeye et al. (De Taeye et al., 2014) suggested that P3 could serve as a possible biomarker of VNS treatment in epilepsy. P3 is thought to reflect the activation of the locus coeruleus-noradrenalin (LC-NA) system, which is known to modulate cognition at different levels (Poe et al., 2020), and is found to activate in response to both VNS and tVNS (Frangos et al., 2015; Badran et al., 2018; Villani et al., 2022). However, the findings on the impact of tVNS on P3 amplitude are inconsistent, with some studies indicating no alteration in the P3 component due to tVNS (Fischer et al., 2018; Warren et al., 2019), while others report significant changes (Ventura-Bort et al., 2018; Rufener et al., 2018; Warren et al., 2020). While considerable variability in stimulation protocols and study designs is likely to account for some of the inconsistency, interindividual variability in P3 component may also contribute to it. To control for some of the interindividual variability, we have used ERP measures such as N2-P3 peak-to-peak amplitude and N2-P3 interpeak latency, where the N2 amplitude or latency is subtracted from that of P3 (Pihlaja et al., 2023; Mäki-Marttunen et al., 2015).

In our previous study, which the current research is a part of, we found that subjects with burnout had longer N2-P3 inter-peak-latency (IPL) and higher P3 amplitudes compared to control subjects, while their level of cognitive performance was comparable (Pihlaja et al., 2023). We suggested that subjects with burnout have slower transitions between neural processes related to cognitive control, and due to inefficient processing they need to allocate more neural resources to achieve the same cognitive performance level as healthy controls (Pihlaja et al., 2023).

This study explores previously unpublished effects of tVNS on EF, examining its impact at the behavioral level by measuring reaction times (RTs) and error rates during an EF task, and at the neurophysiological level using EEG and ERPs, in participants with and without burnout. We expected tVNS to improve EF performance, enhance the efficiency or speed of neural processing underlying EF, and/or restore alterations in neural processes observed due to burnout (Pihlaja et al., 2023). We expected improved performance due to tVNS to be reflected in reduced errors or faster RTs in the Executive RT test (Hartikainen et al., 2010b; Erkkilä et al., 2018), with enhanced neural processes detected as either reduced N2, P3, or N2-P3 (interpeak) ERP amplitudes or latencies. We also assessed whether tVNS has an impact on emotion-attention interaction, similar to what was previously observed by VNS (Sun et al., 2017a).

2. Methods

2.1. Study protocol

This study was conducted in accordance with the Declaration of Helsinki and received ethical approval from the Ethics Committee of Tampere University Hospital (approval number: R20094). All participants provided written informed consent. The study utilized the same subject groups and protocol as the research conducted by Pihlaja et al., in 2023 (Pihlaja et al., 2023). Participants were teaching professionals employed by the city of Tampere, Finland. An email newsletter was sent to approximately 1300 teaching professionals, resulting in 54 volunteers who met the study criteria. Exclusion criteria included a history of neurological diseases or injuries, psychiatric or cardiac conditions, or the use of medication to treat these conditions. All participants were right-handed.

Participants completed the Bergen Burnout Indicator 15 (BBI-15) (Näätänen et al., 2003), Beck's Depression Inventory 21 (BDI-21) (Beck et al., 1996), and the Behavioral Rating Inventory of Executive Function Adult version (BRIEF-A) (Roth et al., 2005). The BBI-15 comprises 15 items that measure exhaustion, cynicism, and reduced professional self-esteem, with scores categorized by severity using normative percentiles. Based on their BBI-15 scores, participants were classified into burnout (percentile >75) and non-burnout (percentile <75) groups (Näätänen et al., 2003). The BDI-21 consists of 21 items rated on a 0–3 scale to classify depressive symptoms from none to severe, giving a total score from 0 to 63, with a higher score indicating more severe depressive symptoms (Beck et al., 1996). The BRIEF-A evaluates executive function in daily life using 75 items across nine domains, summarized into Metacognition, Behavior Regulation, and Global Executive Composite scores; responses are converted to T-scores, with values above 65 indicating clinically significant executive dysfunction (Roth et al., 2005). Detailed information about the questionnaire results and the intake process can be found in the Methods section of the Pihlaja et al. (2023) article.

2.2. Transcutaneous vagus nerve stimulation and the executive reaction time test

Subjects performed a computer-based reaction time (RT) test, Executive RT test (Hartikainen et al., 2010b), for objective measurement of EF, while EEG was recorded and tVNS and sham stimulation were alternately delivered. Integrating EEG recording with a computerized cognitive task and affective distractors, featuring rapid stimulus presentation and a demanding task, enables the effective regulation of overall attention and arousal levels. This approach enables the repeated administration of the test across multiple cycles of active and sham stimulation, providing a precise and dependable method for evaluating the immediate impact of neuromodulation on cognitive and affective brain functions (Sun et al., 2015, 2017a; Hartikainen et al., 2014). Behavioral measures, such as reaction times (RTs) and errors, allow for the assessment of cognitive performance and its alterations due to neuromodulation. Meanwhile, EEG and event-related potentials (ERPs) enable the evaluation of the impact of transcranial vagus nerve stimulation (tVNS) on the underlying neural processes. EEG is a suitable method to assess rapid mental events with temporal resolution within ms and it allows for assessing neural processing even in the absence of a behavioral response such as in case of a NoGo condition.

Executive RT Test(Hartikainen et al., 2010b), a Go/NoGo test, is an integrated test of EF, assessing controlled attention, working memory, shifting, response inhibition, and emotional control. A schematic presentation of the stimulation protocol and the Executive RT Test is presented in Fig. 1.

Fig. 1.

Schematic presentation of the stimulation protocol and trial structure in the Executive RT test, as developed by Hartikainen et al. (Hartikainen et al., 2010b). (a) The stimulation protocol involves continuous transcutaneous vagus nerve stimulation (tVNS) for four blocks, followed by a 4-min washout period without stimulation, allowing participants to rest, and then four blocks of sham stimulation. (b) Each trial begins with the presentation of a triangle pointing either up or down for 150 ms in the center of the screen. This is followed by a 150 ms presentation of a traffic light displaying either a red or green light, along with a neutral or emotional distractor. The distractor is a black line drawing that resembles either a spider (an emotional distractor) or a flower (a neutral distractor), composed of identical line elements arranged in different configurations. Participants are instructed to respond to the orientation of the previously displayed triangle during Go-trials by pressing a button with their middle finger if the triangle points upwards, and with their index finger if it points downwards. In NoGo trials, participants are instructed to refrain from responding. The traffic light color serves as the Go or NoGo signal. At the beginning of each block, participants are informed of the response rule, indicating which color (red or green) corresponds to Go-trials and which to NoGo-trials. The Go-NoGo response rule alternates between blocks, as presented in Fig. 1a. Figure was modified from (Pihlaja et al., 2023).

TVNS was applied using a CE-approved tVNS device (Salustim Group, Kempele, Finland). The device has two clip electrodes, one of which is attached to the tragus and the other to the earlobe. The electrode affixed to the tragus stimulated its inner surface, targeting the sensory region of the auricular branch of the vagus nerve. Conversely, the electrode placed on the earlobe served as a sham stimulation, as the earlobe lacks branches of the vagus nerve. The round electrode had a diameter of 5 mm. Prior to electrode attachment, the skin was cleaned, and an EEG electrode gel was applied to minimize skin resistance. Upon attachment of both electrodes, the stimulation level for each electrode was individually adjusted in a stepwise manner until the subject experienced a mild tingling sensation without pain in the electrode area. In both locations, the level of stimulation exceeded the individual sensation threshold but remained below the pain threshold. The device utilized in the study had 10 intensity levels, with levels 2–5 being employed. Due to variations in skin resistance, the exact output current could not be determined. However, assuming the skin resistance with the gel was below 5 kΩ (the threshold resistance between EEG electrode and scalp), the output current ranged from 1.6 mA to 4 mA. The stimulation was administered in constant mode using a symmetric, biphasic rectangular waveform with a pulse width of 250 μs and a pulse rate of 30 Hz, consistent with the frequency used in the VNS study by Sun et al. (2017a). This stimulation protocol was also used in our previous study (Pihlaja et al., 2020).

This was a single-blind, sham-controlled study. The participants were unaware of the differences between tVNS and sham stimulation locations, as well as any hypotheses related to the study. During the recordings, neither the researcher nor the participants knew which group they would be assigned to, as the assignment to burnout and non-burnout groups was determined post-recording based solely on the BBI-15 questionnaire score.

Participants alternated between beginning with active tVNS and sham stimulation. The stimulation types alternated throughout the test, with half of the participants performing the first and third task blocks under tVNS and the second and fourth task blocks under sham stimulation, while the other half performed the opposite sequence. Detailed stimulation times are provided in Fig. 1. No side effects were reported by the participants during or immediately after the test.

2.3. Electroencephalography and preprocessing

EEG was recorded using 64 Ag/AgCl active electrodes (actiCAP, Gilching, Germany), a QuickAmp amplifier, and Brain Vision Recorder software (Brain Products, GmbH) and sampled at a rate of 500 Hz. The impedance of all electrodes was maintained below 5 kΩ throughout the recordings. EEG preprocessing was performed using Brain Vision Analyzer software (version 2.1, Brain Products GmbH).

The EEG signals were re-referenced to linked mastoid electrodes and band-pass filtered between 0.1 and 40 Hz. Blinks and other artifacts were removed using independent component analysis (ICA)-based correction. After ICA correction, signals with peak-to-peak voltage differences exceeding 80 μV were excluded from the analysis. EEG data were segmented into 2000 ms epochs, starting 200 ms before and ending 1800 ms after trial onset. Additionally, EEG segments were carefully examined visually to include only those with distinct ERPs free from stimulation-induced noise, as such noise could disrupt the semi-automatic identification of ERPs. EEG segments were then averaged for each condition (Go or NoGo), stimulator type (active or sham), and distractor type (emotional or neutral), resulting in eight distinct ERP conditions for each subject (neutral active, neutral sham, emotional active, and emotional sham in both Go and NoGo conditions). Based on established conventions (Boudewyn et al., 2018; Luck, 2005), prior experience with this paradigm, and visual inspection of individual ERP averages, a minimum of 50 artifact-free EEG epochs per ERP condition were required for a subject to be included in the analysis. The number of trials across different conditions was comparable for each subject in the analysis.

We focused on the frontal N2 amplitude during the NoGo condition, thought to reflect cognitive control required in response inhibition (Folstein and Van Petten, 2008), and centroparietal P3 during Go conditions, thought to reflect attentional resource allocation required in target detection (Kok, 2001; Polich, 2007). The N2 component after the Go/NoGo signal was identified from the averaged waveforms using semiautomatic peak detection and visual inspection. N2 was identified as the most negative peak occurring between 200 ms and 350 ms, while P3 was defined as the subsequent positive peak appearing between 300 ms and 500 ms after the traffic light, corresponding to 500–650 ms and 600–800 ms from the trial onset, respectively. See schematic presentation of the timing of the N2 and P3 ERP components and N2P3 IPL in Fig. 2. Amplitudes were measured from the averaged baseline of 200 m prior to trial onset. Due to the frontal distribution of the N2 peak (Folstein and Van Petten, 2008), we used Fz, F1, F2, F3, and F4 to measure N2 peak amplitude as well as NoGo N2-P3 IPL. Similarly, due to centroparietal distribution of the Go P3, CPz, CP1, CP3, CP2, and CP4 were used to measure P3 amplitude and centroparietal N2-P3 IPL in Go trials.

Fig. 2.

Illustration of ERP waveform. A triangle, a traffic light, and N2, P3 and IPL are presented in the figure to demonstrate the relationship between measured ERPs and the Executive RT-test.

2.4. Statistical analyses

Statistical analysis was performed using R statistics v. 3.1.1 (R Core Team, 2022). Repeated measures ANOVA was conducted using the with EZ package version 4.2–2 (Lawrence, 2016).

For statistical analyses, the average N2 amplitude and latency were calculated for channels F1-F4 and Fz, and the average P3 amplitude and latency were calculated for channels CP1-CP4 and CPz. The N2-P3 IPL was determined by subtracting the N2 latency from the P3 latency for the same individual channels. These values (peak amplitudes, latencies and IPLs) were then averaged across the frontal and CP channels. Statistical data were manually checked for missing values, and one subject from each group was excluded from the analysis of the centroparietal N2-P3 Go IPL, while two subjects from the non-burnout group were removed from the analysis of the centroparietal N2-P3 NoGo IPL due to missing scenarios (previously excluded due to poor quality of EEG in these segments).

Repeated-measures analysis of variance (ANOVA) was used to analyze reaction times and event-related potentials (ERPs). The ERP data were not normally distributed; however, they were normalized before performing the ANOVA. Normality was checked using the Shapiro-Wilk test. “Stimulator status” (active vs. sham) and “emotional valence” (emotional vs. neutral) were within-subject factors, and “group” (burnout vs. non-burnout) was the between-subjects factor in reaction time, error, and ERP analyses. ERP analysis was conducted separately for Go and NoGo conditions. Significant interaction effects were analyzed further with post hoc ANOVAs. Based on our prior knowledge of differences in EF processing between the two groups from our previous study, we expected the impact of tVNS to differ between these groups. Thus, we planned to analyze the groups separately even in the case of a lack of interaction effect. The Benjamini-Hochberg method for False Discovery Rate (FDR) correction was employed to account for multiple comparisons in the analysis.

Errors were analyzed with a generalized mixed-effects logistic regression (Dixon, 2008; Jaeger, 2008). Each error type was analyzed using a separate model predicting the probability of making that kind of error. “Subject” was used as a random effects predictor and “group”, “stimulator status”, and “emotional valence” were used as a fixed effect predictors. The logistic regression trial outcomes were dichotomized into binary classes: for total errors, the classes were ‘correct’ (correct button press in Go trial or no response in NoGo trial) or ‘error’ (incorrect or missing button press in Go trial or any button press in NoGo trial); for incorrect responses, the classes were ‘incorrect’ (incorrect button press) or ‘other’ (correct or missing button press); for missed responses, the classes were ‘miss’ (missing button press) or ‘other’ (correct or incorrect button press); and for commission errors, the classes were ‘commission error’ (button press in NoGo trial) or ‘correct’ (no button press in NoGo trial).

3. Results

3.1. Subjects

A total of 27 subjects (25 females and two males) from the non-burnout group and 24 subjects (22 women and two males) from the burnout group met the quality criteria for cognitive performance data (RT and errors less than 3 SDs from the whole group mean) and were included in behavioral data analyses. Fifteen females and two males from the non-burnout group, and sixteen females and two males from the burnout group met the quality criteria for ERP data. Groups did not differ in age or years of education.

Groups differed significantly in BDI-21 scores, as well as in BRIEF-A indices, global executive composite (GEC), metacognition index (MI) and behavior regulation index (BRI), with greater scores (i.e., more depression-related symptoms and more challenges in EF) in the burnout group, as presented in the previous article by Pihlaja et al. (2023) and in Table 1.

Table 1.

Age, education, and questionnaire scores of burnout and non-burnout groups.

	Group	Mean (SD)	p
Age (years)	Burnout	44.6 (9.4)	0.84
	Non-burnout	44.0 (10.4)
Education (years)	Burnout	19.0 (2.3)	0.28
	Non-burnout	18.3 (2.2)
BBI score	Burnout	60.0 (11.3)	<0.00
	Non-burnout	33.2 (7.5)
BDI score	Burnout	15.9 (7.1)	<0.00
	Non-burnout	5.0 (3.8)
BRI	Burnout	58.9 (10.9)	<0.00
	Non-burnout	46.3 (9.9)
MI	Burnout	61.4 (10.5)	<0.00
	Non-burnout	48.5 (10.3)
GEC	Burnout	60.9 (10.3)	<0.00
	Non-burnout	47.2 (9.6)

3.2. Behavioral results

There were no significant main effects of stimulator status. There was, however, an interaction effect between the group and stimulator status on the total errors in the Executive RT-test (Table 2). Post-hoc analysis indicated that tVNS significantly reduced total errors in subjects without occupational burnout (Table 2, Table 3). No significant main or interaction effects were found for RTs or different error types such as incorrect responses or commission errors. The number of misses was too low for reliable statistical analysis.

Table 2.

Odds ratios, confidence intervals, and p-values for total errors.

Variable	OR	95 % CI	p	adjusted p
Group	1.01	0.61–1.66	0.97	0.97
Stimulator	1.09	0.86–1.37	0.50	0.97
Emotion	0.93	0.73–1.19	0.57	0.97
Group: stimulator	0.67	0.46–0.96	0.03∗	0.21
Group: emotion	1.28	0.90–1.82	0.17	0.60
Stimulator: emotion	1.05	0.75–1.47	0.77	0.97
Group: stimulator: emotion	1.03	0.62–1.71	0.90	0.97
Post hoc
Burnout: stimulator	1.11	0.94–1.32	0.22	0.22
Non-burnout: stimulator	0.76	0.63–0.92	0.008∗∗	0.016∗

Table 3.

Medians and IQRs for behavioral results.

Executive RT -test	Group (n)	Stimulator status	Median (IQR)
Reaction time (ms)	Burnout (27)	Active	356.47 (87.75)
		Sham	361.76 (77.65)
	Non-burnout (24)	Active	389.62 (103.54)
		Sham	402.59 (108.78)
Total errors (%)	Burnout (27)	Active	1.76 (1.95)
		Sham	1.56 (2.43)
	Non-burnout (24)	Active	0.98 (1.17)
		Sham	1.56 (1.22)
Incorrects (%)	Burnout (27)	Active	1.17 (1.77)
		Sham	1.17 (1.77)
	Non-burnout (24)	Active	0.78 (1.86)
		Sham	0.78 (1.83)
Misses (%)	Burnout (27)	Active	0.39 (0.78)
		Sham	0.39 (0.70)
	Non-burnout (24)	Active	0.00 (0.39)
		Sham	0.00 (0.70)
Commission errors (%)	Burnout (27)	Active	1.56 (1.86)
		Sham	0.78 (1.95)
	Non-burnout (24)	Active	0.78 (0.09)
		Sham	0.89 (1.46)

3.3. Event-related potentials

3.3.1. N2 and P3 amplitudes and latencies

There were no significant main or interaction effects of stimulator status on the amplitudes or single-peak latencies of frontal N2 or centroparietal P3. Emotional distractor significantly increased frontal N2 NoGo amplitude compared to neutral distractor (emotional: 3.67 μV (3.89 μV), neutral: 3.15 μV (3.72 μV), p = 0.004). Further, centroparietal P3 in Go condition was significantly larger in the burnout group than in the non-burnout group (burnout: 10.28 μV (4.58 μV), non-burnout: 7.19 μV (4.71 μV), p = 0.025) as presented in our previous article (Pihlaja et al., 2023). Medians and IQRs for amplitudes and single-peak latencies are presented in Table 4.

Table 4.

Medians and IQRs for amplitudes and single peak latencies of frontal N2 and CP P3 between groups and stimulator statuses.

ERP	Group (n)	Stimulator status	Amplitudes (μV)	Latencies (ms)
Frontal N2 Go	Burnout (18)	Active	−3.54 (6.28)	256.4 (40.2)
		Sham	−5.46 (6.08)	259.2 (29.2)
	Non-burnout (17)	Active	−3.17 (3.53)	266.0 (38.6)
		Sham	−3.37 (3.74)	262.0 (49.2)
Frontal N2 NoGo	Burnout (18)	Active	−3.74 (7.25)	256.4 (52.0)
		Sham	−4.18 (6.57)	260.8 (83.2)
	Non-burnout (17)	Active	−1.84 (4.15)	250.4 (71.6)
		Sham	−2.80 (4.50)	251.2 (79.2)
CP P3 Go	Burnout (18)	Active	10.24 (4.81)	396.5 (85.8)
		Sham	10.74 (4.93)	398.1 (114.8)
	Non-burnout (17)	Active	7.22 (5.22)	378.4 (54.4)
		Sham	7.51 (3.91)	378.4 (53.8)
CP P3 NoGo	Burnout (18)	Active	7.10 (4.34)	402.0 (64.0)
		Sham	6.14 (6.28)	394.8 (55.5)
	Non-burnout (17)	Active	6.95 (3.62)	398.0 (44.2)
		Sham	6.04 (4.23)	404.8 (37.0)

3.3.2. N2-P3 inter-peak latencies

There was a main effect of stimulator status on the frontal N2-P3 IPL in the NoGo condition. The frontal N2-P3 IPL was significantly shorter with active stimulation (m = 157.6 ms (IQR = 59.2 ms)) than with sham stimulation (m = 169.6 ms (IQR = 88.4 ms), p = 0.01, ηG2 = 0.02). Medians and IQRs for active and sham stimulation, separately for each group, are presented in Table 5. Emotional valence or group had no significant effect on the frontal N2-P3 IPL. Furthermore, the main effect of the group on the centroparietal Go N2-P3 IPL was found. This result has been reported in our previous study (Pihlaja et al., 2023). The centroparietal N2-P3 IPL was significantly shorter in the non-burnout group compared to the burnout group. Medians and IQRs for centroparietal and frontal N2-P3 IPLs in Go and NoGo conditions are presented in Table 6.

Table 5.

ANOVA for frontal NoGo N2-P3 inter-peak latency.

Effect	df	F	p	ges	adjusted p
Group	33	0.15	0.70	0.00	0.72
Stimulator	33	7.17	0.01∗∗	0.02	0.07
Emotion	33	0.97	0.33	0.00	0.46
Group: stimulator	33	1.15	0.29	0.00	0.46
Group: emotion	33	0.13	0.72	<0.00	0.72
Stimulator: emotion	33	0.98	0.32	0.00	0.46
Group: stimulator: emotion	33	1.28	0.27	0.00	0.46

Table 6.

Medians and IQRs for N2-P3 interpeak latencies.

	Group (n)	Stimulator status	IPL (ms)
Frontal N2-P3 Go	Burnout (18)	Active	180.4 (77.2)
		Sham	184.8 (107.4)
	Non-burnout (17)	Active	145.2 (98.2)
		Sham	143.6 (95.2)
Frontal N2-P3 NoGo	Burnout (18)	Active	162.4 (76.0)
		Sham	186.4 (100.8)
	Non-burnout (17)	Active	145.2 (98.2)
		Sham	174.0 (49.4)
CP N2-P3 Go	Burnout (17)	Active	194.0 (73.9)
		Sham	198.0 (84.2)
	Non-burnout (16)	Active	135.6 (46.8)
		Sham	158.4 (64.8)
CP N2-P3 NoGo	Burnout (16)	Active	193.2 (73.8)
		Sham	180.8 (60.1)
	Non-burnout (17)	Active	134.8 (53.8)
		Sham	158.4 (60.4)

4. Discussion

We found both behavioral and neural-level evidence that tVNS enhances EF in specific situations. TVNS reduced the frontal N2-P3 IPL in the NoGo condition. Further, tVNS enhanced cognitive performance in the Executive RT-test in the non-burnout group by reducing the total number of errors. The decrease in total errors attributed to tVNS indicates enhancements in attention and EFs. Additionally, the shortened frontal N2-P3 IPL in the NoGo condition suggests a faster transition between specific brain processes during the cognitive control task. These findings align with a previous meta-analysis by Ridgewell et al., which suggested that tVNS could particularly enhance executive function (Ridgewell et al., 2021).

To the best of our knowledge, this is the first study to report a reduction in interpeak latency between the frontal N2 and P3 components with tVNS. While interpeak latencies have been studied less compared to single-peak latencies, Kumar et al. demonstrated that moderate physical training shortened the N2-P3 IPL in male subjects when measured immediately after training during the auditory oddball paradigm, suggesting an enhancement of cognitive processes (Kumar et al., 2023). In another study focused on P3, Rufener et al. found that tVNS led to increased P3 amplitude and shorter P3 latency during the auditory oddball task, without affecting performance (Rufener et al., 2018). Additionally, Gurtubay et al. observed similar results, with 7 min of tVNS leading to increased P3 amplitude, decreased P3 latency, and shorter reaction times during the auditory oddball paradigm (Gurtubay et al., 2023). The effects lasted for 28 min following the stimulation and then gradually reverted to their pre-stimulus state. A shorter P3 latency likely indicates an acceleration in cognitive processes comparable to what we observed. However, IPL might be more sensitive in depicting the acceleration as it minimizes intersubject variability by subtracting the latency of the preceding peak, N2, from that of P3 within the same subject.

Previously, we have shown that the centroparietal N2-P3 IPL in the Go condition was prolonged in subjects with occupational burnout compared to those without. IPL predicted the severity of burnout and subjective metacognitive problems. A longer IPL was associated with higher BBI-15 and MI scores (Pihlaja et al., 2023). While burnout slowed down, tVNS sped up transitions between brain processes related to cognitive control. This finding aligns with our previous study, which suggested that tVNS may lead to more efficient cognitive control in young, healthy subjects (Pihlaja et al., 2020). Although no performance improvement was detected in the earlier study, we observed a reduction in frontal N2 amplitude. We proposed that the reduced N2 observed during tVNS indicated a lower demand for cognitive control resources to achieve comparable performance levels with sham stimulation (Pihlaja et al., 2020). In the current study, no significant effect of tVNS on N2 amplitude was observed. It is worth noting, however, that an increase in frontal N2 amplitude was observed in the context of threat-related distractors, independent of stimulation. As threat-related stimuli automatically capture attentional (Hartikainen, 2021) and cognitive control resources (Hartikainen et al., 2012b) to minimize danger, the increased N2 amplitude observed in the current study in the context of threat-related distractors aligns with the general cognitive control resource allocation model.

Interestingly, tVNS enhanced cognitive performance in the non-burnout group, but not in those experiencing burnout. This discrepancy may be due to the altered autonomic nervous system (ANS) activity in individuals with occupational burnout (Lennartsson et al., 2016; Olsson et al., 2010), linked with either hypo- or hyperarousal. It is feasible that the effect of tVNS on EF tasks may be linked to arousal and depend on the subject's baseline level of arousal. However, similar results were found in the study by Koenig et al., where tVNS improved the EF of healthy controls but impaired response inhibition in the group of depressed adolescents (Koenig et al., 2021). It is plausible that in chronic conditions such as depression and burnout, long-term alterations in brain physiology may attenuate the effects of short-term stimulation, resulting in differential responses compared to those observed in healthy individuals.

In line with the accelerated IPL due to tVNS, there was a slight tendency towards faster RT with tVNS; however, this tendency disappeared when corrected for multiple comparisons. A statistically significant difference, which survived correction for multiple comparisons, was observed only in total errors. Total errors are the sum of all the different error types: incorrect responses, commission errors, and misses. To that end, the number of total errors is higher than the individual error types. In individual error types, the number of errors was too small to detect significant differences. In other words, the test was too easy, resulting in a ceiling effect in the individual error types.

Heart rate variability (HRV) is often used to assess ANS balance, and previous studies have shown that tVNS increases HRV (Machetanz et al., 2021), shifting ANS activity towards parasympathetic dominance. Our previous research indicated that individuals with more severe burnout had lower HRV and greater difficulties with EF in daily life. Additionally, weaker metacognitive functions were associated with lower HRV (Pihlaja et al., 2022). These findings suggest that burnout may increase sympathetic tone, leading to suboptimal arousal levels for optimal EF performance. TVNS, on the other hand, could potentially help optimize ANS activity and arousal in certain situations to achieve optimal cognitive performance.

While vagus nerve activation may lead to increased parasympathetic tone, the immediate impact of tVNS on the brain is paradoxically linked with increased arousal due to activation of LC-NA system (Trifilio et al., 2023; Ross and Van Bockstaele, 2021). In addition to the effects of tVNS on LC-NA system, there are widespread effects via afferent pathways from the nucleus tractus solitarius (NTS) to other brain regions than LC and other neurotransmitters than NA essential for cognition (Murphy et al., 2023) and emotion (Sun et al., 2017a). Thus, the impact of tVNS on LC-NA system may explain some of the observed effects of tVNS on neuronal signaling underlying EF. Like tVNS, occupational burnout presumedly has an impact on the LC-NA system, crucial for stress response, arousal, attention, and EF. However, there is limited knowledge about the overall effects of burnout on the brain. Additionally, to the best of our knowledge, there are no previous studies evaluating the impact of tVNS on the cognitive functions of individuals with burnout.

As demonstrated in our previous study (Pihlaja et al., 2023), individuals with burnout required a greater allocation of neural resources to achieve a comparable performance level to those without burnout. Thus, they may have already reached their maximum performance level, with no room for improvement through neuromodulation. According to the Yerkes-Dodson law, the relationship between arousal and cognitive performance follows an inverted U-shaped curve, where the optimal performance occurs at a moderate level of arousal at the peak of the inverted U, while both hypoarousal on the left of the curve and hyperarousal on the right of the curve lead to suboptimal cognitive performance (Dodson, 1915). If subjects with burnout were already at the top of the inverted U-curve, increasing arousal with tVNS would not result in improved performance. In contrast, if the healthy subjects were positioned to the left of the inverted U-curve peak, tVNS likely shifted them rightward on the curve towards optimal performance. Where subjects are initially located on the inverted U-curve regarding their arousal and cognitive performance levels may be a critical factor in whether improvements are detected with tVNS and might explain the distinct effects of tVNS between burnout and non-burnout groups in the current study as well as the variability in research results more generally.

The effect of tVNS is also likely dependent on the time course of the stimulation and its impact. In the current study, we focused only on the immediate effects of stimulation and observed that in healthy subjects, tVNS acutely enhanced the accuracy of cognitive performance. In contrast, individuals with occupational burnout did not show any acute benefits in performance. However, at the neural level, there seemed to be an immediate processing speed benefit in the whole group independent of burnout status.

One of the limitations of the current study was that the burnout group was heterogeneous, comprising of individuals with varying severity levels of burnout within the relatively small group. The heterogeneity of the burnout group may result in greater variability and potentially mask some effects. To that end, future research should focus on examining the effects in clinical or severe burnout groups, where challenges in EF tend to be more pronounced. In the current study, burnout and depression were assessed exclusively through self-report questionnaires. A clinical diagnosis, incorporating an interview and a more thorough assessment, would provide deeper insights into specific burnout subtypes and potential comorbidities with depression, factors that could not be fully explored within the study's limitations.

Another factor to consider in this study is the gender distribution of participants, with the majority being female. This is attributed to the recruitment of subjects from teaching professionals in the city of Tampere, where women make up the predominant workforce. Given that the effects of burnout on EF and the ANS are not well understood and may differ between sexes, future research could benefit from exploring the impact of tVNS in balanced groups or separately by sex.

Another aspect that warrants caution is the relatively short stimulation washout period in the current study. A recent study by Gurtubay et al. demonstrated that the effects of tVNS on ERPs persisted even up to 28 min (Gurtubay et al., 2023). It is therefore plausible that our stimulation protocol, with its 4-min washout periods, may have masked some effects, as tVNS effects could still be observed during sham stimulation.

Furthermore, technical challenges in recording high-quality EEG during active electrical stimulation reduced our sample size, thereby decreasing the statistical power. With these technical challenges, the study was partly underpowered to reliably detect interaction effects and fully the impact of tVNS, especially in subjects with burnout who seemed to have more variability in performance measures than the healthy control subjects. The inability to reliably detect interaction effects hindered the observation of potential stimulation-related influences on emotional processes. The immediate impact of invasive VNS in heightening vigilance to threat, as noted by Sun et al. (2017a), was not replicated in the present study with tVNS. However, given the study's slightly limited power, future research is needed to determine whether tVNS exhibits similar effects. Another challenge faced in the current study pertains to multiple comparisons, which were addressed using FDR-corrected p-values. Not all p-values remained significant after applying this correction.

The primary strength of the current study lies in its simultaneous measurement of ERP and cognitive performance. ERP is particularly valuable as it can detect changes in neural processing related to executive functions, even when there is no observable change in performance. ERP reflects rapid mental functions and the underlying brain physiology, as well as their alterations, in burnout or due to neuromodulation. Performance measures, such as RT and errors, are coarser measures and have more limited temporal precision than ERP, with temporal resolution in the ms range. In addition, in a successful NoGo situation ERP reflects the neural processes needed to withhold a response even though there is no measurable behavioral response. Further strength relates to assessing the impact of stimulation within each subject in this mixed-design, single-blinded, sham-controlled study. During the test, stimulation status alternated between active and sham, thereby minimizing the impact of various confounding factors and individual differences on ERPs and cognitive performance. This approach facilitated an accurate assessment of the immediate effects of tVNS on cognition and the associated neural processes (Folstein and Van Petten, 2008).

To conclude, there is a significant demand for cognitive enhancers for both healthy individuals and those with cognitive challenges. Methods like tVNS, which have minimal side effects or risks, could be beneficial in various conditions and situations. While current results indicate improved EF and enhanced underlying neural processing, given the small study population, skewed gender distribution, heterogeneity in burnout severity, exclusion of many subjects due to inadequate EEG quality, and the use of an infrequently studied ERP measure (IPL), caution is warranted. Future studies with a larger sample size are needed to confirm the preliminary results of the current research on the effects of tVNS on neural processes and cognitive performance. Additionally, future research is necessary to evaluate the impact of tVNS on the interaction between emotion and attention, as well as emotion and EF, across various clinical populations, such as those with depression and burnout. These populations are expected to exhibit alterations in cognitive and affective functions, as well as the interplay between these two domains and it will be of interest to asses weather these alterations can be ameliorated with neuromodulation such as tVNS.

In summary, tVNS accelerated transitions in neural processes during tasks requiring cognitive control and response inhibition. TVNS also enhanced EF performance in subjects without burnout. This acceleration of neural processing underlying EF was reflected in reduced IPL. Previously, we proposed EEG-derived indices as biomarkers of neuromodulation (Sun et al., 2017a) and mild traumatic brain injury (Kuusinen et al., 2021). Our earlier research demonstrated that N2-P3 IPL during an EF task is sensitive to decelerating effects of burnout on EF (Pihlaja et al., 2023). Now we show that tVNS can accelerate these processes, associated with improved EF. The current study introduces IPL as a potential measure of neural processes underlying EF and their modulation through neuromodulation.

5. Conclusion

The findings suggest that tVNS may enhance neural processing related to executive function (EF) in specific conditions. While tVNS demonstrated a positive impact on cognitive performance, this effect was observed only in individuals without burnout. Additionally, we introduce N2-P3 IPL, measured during a cognitive control task, as a potential marker of higher cognitive control efficiency. Although further research is necessary, IPL may serve as a sensitive, physiology-based measure for intervention studies aimed at optimizing brain health and EF.

To advance understanding, future studies should focus on larger, more homogeneous samples to clarify the effects of tVNS on affective and EF processes in clinical populations with alterations in these functions, including those experiencing burnout. Given the complexity and variability of burnout-related cognitive and affective changes, refined EEG methodologies and improved control over arousal levels will be crucial for detecting subtle neuromodulation effects. Furthermore, in addition to IPL, investigating frontal alpha asymmetry (FAA) during EF tasks and affective probes (Sun et al., 2017a) could provide valuable insight into the neural mechanisms underlying tVNS effects on cognitive and affective processes, as well as how they are altered in burnout.

In conclusion, tVNS holds promise as a potential cognitive enhancer in select contexts. We encourage further research to validate these preliminary findings, optimize stimulation parameters and targets, and explore EEG-derived indices such as N2-P3 IPL and FAA to gain a more comprehensive understanding of the impact of tVNS and other neuromodulation techniques on EF.

CRediT authorship contribution statement

Mia Pihlaja: Writing – original draft, Visualization, Formal analysis, Data curation. Kaisa M. Hartikainen: Writing – review & editing, Supervision, Resources, Project administration, Methodology, Investigation, Funding acquisition, Conceptualization.

Key points

• -TVNS accelerated neural processing in the context of response inhibition

• -TVNS enhanced executive function performance in healthy subjects

• -N2-P3 IPL might serve as a biomarker of cognitive control for neuromodulation studies

• -Preliminary findings suggest that TVNS shows promise as a cognitive enhancer

Data availability statement

The datasets presented in this article are not readily available because of hospital privacy policy and laws, which prohibit sharing the datasets outside European Union (EU) and European Economic Area (EEA) countries. Requests to access the datasets should be directed to KMH, kaisa.hartikainen@helsinki.fi.

Declaration of generative AI and AI-assisted technologies in the writing process

During the preparation of this work the authors used Microsoft Copilot and Grammarly in order to check and improve the English language. After using this tool/service, the author(s) reviewed and edited the content as needed and take full responsibility for the content of the publication.

Funding

This research was funded by the Finnish Ministry of Social Affairs and Health from the European Social Fund's (ESF) Programme for Sustainable Growth and Jobs 2014–2020 Finnish structural fund (Sustainable Brain Health, S21966) and financially partly supported by the Competitive State Research Financing of the Expert Responsibility area of Tampere University Hospital.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability

The data that has been used is confidential.