Abstract
Adapting to the ever-changing demands of the environment requires a
complex interplay between cognitive-affective, neuronal, and autonomic
processes. Vagally-mediated heart rate variability (vmHRV) is positively
associated with both cognitive-affective functioning and prefrontal
cortex (PFC) activity. Accordingly, the Neurovisceral Integration
Model has posited a shared role of the PFC in the regulation of
cognitive-affective processes and autonomic nervous system (ANS)
activity. While there are numerous correlational findings in this
regard, no study so far has investigated whether the manipulation of PFC
activity induces changes in vmHRV and cognitive-affective processing in
an inter-dependent manner. In this study, we examined the effects of
continuous (cTBS) and intermittent theta-burst stimulation (iTBS) over
the left dorsolateral PFC (dlPFC) on vmHRV and cognitive-affective
processing within an emotional stop-signal task (ESST) in 66
participants. Our results revealed that both resting vmHRV and
reactivity, at least partly, predicted cognitive-affective processing.
Furthermore, we found a dampening effect of cTBS on resting vmHRV, as
well as an enhancing effect of iTBS on ESST performance. Our results
show no direct association between vmHRV changes and ESST performance
alterations following stimulation. We interpret our results in the light
of a hierarchical model of neurovisceral integration, suggesting a
dynamical situation-dependent recruitment of higher-order cortical areas
like the dlPFC in the regulation of the ANS. In conclusion, our results
highlight the complex interplay between PFC activity, autonomic
regulation, and cognitive-affective processing, emphasizing the need for
further research to understand the causal dynamics of the underlying
neural mechanisms.
Introduction
To successfully adapt to the ever-changing demands of the environment,
individuals need to process environmental stimuli, both cognitively and
affectively. It has been suggested that the same neuronal circuits and
mechanisms involved in cognitive-affective processing contribute to the
regulation of the autonomic nervous system (ANS), thus allowing an
efficient and coordinated adaption on different levels (Smith et al.,
2017; Thayer et al., 2009). Consistent with these assumptions,
vagally-mediated heart rate variability (vmHRV), a measure of vagal
regulation of the heart or so called cardiac vagal activity, has been
found to correlate with cognitive and affective functioning (Appelhans
& Luecken, 2006; Balzarotti et al., 2017; Magnon et al., 2022).
Furthermore, higher vmHRV has been associated with higher activity of
the prefrontal cortex (PFC) which displays a key node involved in
cognitive, attentional, and affective processes (Thayer et al., 2012).
Although these results suggest a shared prefrontal regulation of
cognitive, affective, and autonomic functions, they do not allow causal
inferences about potential common underlying mechanisms given their
correlative nature. In this study, we used repetitive transcranial
magnetic stimulation (rTMS) to clarify whether common neuronal processes
within the PFC directly contribute to cognitive-affective and autonomic
processing alike.
The PFC is well known for its contribution to cognitive and affective
control (see e.g., Etkin et al., 2011; Miller & Cohen, 2001; Ochsner &
Gross, 2005). Besides that, it has been suspected to be integrated into
a neuronal network consisting of various cortical and subcortical
structures regulating the ANS, termed the central autonomic network
(CAN, Benarroch, 1993). The Neurovisceral Integration Modelsuggests that activity in different areas of the PFC regulates the
heartbeat via GABAergic inhibitory projections to subcortical
cardioacceleratory circuits of the CAN which are eventually transmitted
to the heart via the vagus nerve (Thayer & Lane, 2000). Accordingly,
these inhibitory projections are also involved in processes such as
cognitive control and emotion regulation, leading to cognitive-affective
and autonomic regulation being closely linked in adaptation to
environmental demands. Supporting this suggested link, higher levels of
vmHRV have been widely demonstrated to be associated with better
cognitive functioning, emotion regulation and mental health (Appelhans
& Luecken, 2006; Holzman & Bridgett, 2017; Magnon et al., 2022.
Further support is provided by results from brain imaging studies in
which higher vmHRV levels have been related to greater activity in
various prefrontal areas such as the dorsolateral PFC (dlPFC) and
ventromedial PFC (vmPFC) (McIntosh et al., 2020; Thayer et al., 2012).
While activity in the vmPFC has been found to be positively correlated
with vmHRV across situations, activity in the dlPFC appears to be
correlated with vmHRV only in certain circumstances, such as when high
cognitive or affective control is required (Lane et al., 2009). While
these results suggest a close relationship between vmHRV and prefrontal
activity, as well as associated cognitive and affective processes, they
do not allow any inferences regarding causality or direction of effects
given their correlational nature. Thus, it remains to be clarified
whether the link between prefrontal activity and vmHRV reflects top-down
regulation of both cardiac and cognitive-affective outcomes or bottom-up
processing of cardiac signals which in turn facilitates
cognitive-affective processes, or even both.
Repetitive TMS has proven to be a powerful non-invasive tool for
establishing causal relationships between activity in specific brain
regions and behavioral and physiological outcomes by inducing long
lasting changes in the stimulated cortical area and areas with
functional connectivity. Theta burst stimulation (TBS) is a relatively
new rTMS protocol that produces long-lasting aftereffects while having a
short duration of implementation. Whereas continuous TBS (cTBS) over the
primary motor cortex has been shown to suppress motor evoked potentials,
intermittent TBS (iTBS) enhances motor evoked potentials (Huang et al.,
2005). Thus, cTBS is assumed to inhibit the targeted cortical area by
producing long-term depression-like effects, whereas iTBS is thought to
produce long-term potentiation-like effects (Kirkovski et al., 2023).
Previous findings suggest that the application of TBS over prefrontal
areas, particularly over the left dlPFC, affects both cognitive and
affective processing, with iTBS having enhancing effects (Pabst et al.,
2022, Dumitru et al., 2020, Moulier et al., 2021) and cTBS having
disrupting effects (see Ngetich et al., 2020, e.g., Keuper, Dellert,
Junghoefer et al., 2019, Lowe, Staines, Mannochio et al., 2018, Maier,
Rosenbaum, Haeussinger et al., 2018, Perach-Barzilay, Tauber et al.,
2013, Rosemann, Dellert, Junghoefer et al., 2019). In addition, there is
recent meta-analytic evidence that rTMS targeting the left dlPFC exerts
modulatory effects on CVC (Makovac et al., 2017; Schmaußer et al.,
2022). In specific, inhibition of the left dlPFC by cTBS has been found
to decrease vmHRV (Era et al., 2021) whereas iTBS increases levels of
vmHRV (Iseger et al., 2020). While these results demonstrate that
stimulation of the left dlPFC influences cognitive and affective
processing as well as vmHRV, it remains unclear whether the induced
changes are directly related to each other, as posited by theNeurovisceral Integration Model .
As a result, this study was designed to investigate whether cTBS and
iTBS applied to the left dlPFC will induce alterations in vmHRV during
both resting state and cognitive-affective processing. Furthermore, the
study aimed to examine TBS-induced changes in cognitive-affective
processing and explore whether these changes are directly related to
vmHRV alterations. Cognitive-affective processing was operationalized as
the performance in an emotional Stop-Signal Task (ESST). Recently, we
found that resting vmHRV , vmHRV reactivity, and the interaction between
both measures significantly predicted stop-signal reaction times
(SSRTs), a measure of response inhibition, in an ESST (Schmaußer &
Laborde, 2023). More specifically, our results indicated higher levels
of resting vmHRV to be associated with faster SSRTs, whereas on-task
decreases in vmHRV predicted slower SSRTs. Interestingly, these effects
of vmHRV reactivity were more pronounced in individuals with low levels
of resting vmHRV, suggesting that high levels of vmHRV protect from
adverse effects of vagal withdrawal on cognitive-affective processing.
In this study, we aim to replicate these findings and extend them by
investigating whether TBS induced changes in vmHRV will be related to
changes in cognitive-affective processing. Consequently, we hypothesized
that individuals with higher resting vmHRV levels will produce faster
SSRTs than individuals with low resting vmHRV levels. In addition, we
expected that vagal withdrawal during performance of the ESST would
predict slower SSRTs, with this effect being more pronounced in
participants with low resting vmHRV levels. As we have outlined above,
cTBS and iTBS have been found to exert differential effects on both
cognition and vagal activity. Accordingly, we further hypothesized that
the excitatory effect of iTBS over the left dlPFC would lead to both
improved cognitive-affective processing (i.e., faster SSRTs) and
increased levels of vmHRV during rest and during the performance of the
ESST. Opposite effects were expected for cTBS given its proposed
inhibitory effects. Lastly, based on the assumptions of theNeurovisceral Integration Model proposing a shared role of the
PFC in the regulation of cognitive-affective as well as autonomic
processing, we hypothesized that TMS-induced changes in vmHRV and SSRT
would be directly linked to each other. In specific, we expected
increases in resting vmHRV to predict fast SSRTs in the ESST and vice
versa.
Methods
Participants
A required sample size of 57 was determined using G*Power . To account
for the potential loss of ECG data and the possibility of a smaller TBS
effect reported in the published research literature, a total of 78
participants was recruited in this study. According to the TMS safety
guidelines (Rossi et al., 2021) and HRV study recommendations (Laborde
et al., 2017), exclusion criteria included a history or current
diagnosis of a psychiatric, neurologic or cardiovascular disorder,
family history of epilepsy or hearing loss, use of psychopharmacological
or cardioactive medication, pregnancy, inner ear prosthesis, recent
neurosurgical procedures, pacemaker or other electronic implants, metal
objects or magnetic object in the brain or around the head (only
removable earrings and piercing were allowed), and skin disorders at the
level of the head. Further, participants were asked to follow a normal
sleep routine and to have no intense physical training the day before
the experiment, to have no meal or any caffeinated drinks 2 hours before
the experiment and to abstain from alcohol 24 hours before the
experiment. 12 participants were excluded following the stop signal task
quality control (see section 2.5.1). Two more participants were excluded
after outlier detection (see section 2.5.1). The exclusion of 12
participants resulted in a final sample of 66 participants.
Experimental design and procedure
Upon arrival, all participants provided written informed consent. The
local ethics committee approved the study.
At the beginning of the session, the participants were attached to ECG
electrodes and completed sociodemographic questionnaires. Afterwards,
participants were asked to sit and rest quietly in the sound-isolated
experimental room for 10 minutes to habituate to the situation and to
collect their resting baseline HRV. Participants then performed the
ESST. Each participant received standardized written instructions. The
investigator was available to answer any remaining questions. To become
familiar with the ESST, each participant performed a practice run
including 5 stop trials. After completion of the first ESST, the
respective TMS protocol was applied, including determination of the
resting motor threshold (RMT), determination of the left dlPFC by
neuronavigation, and the specific stimulation protocol (iTBS vs. cTBS.
vs. sham). Participants were assigned to the specific stimulation
protocol in a randomized manner. Specifically, participants were
assigned to the protocols by using an excel table, where the order of
the respective protocols to apply were shuffled by a randomization
algorithm. The TMS protocol was followed by a 10-minute resting period
and the completion of the second ESST. Finally, the participants were
debriefed about the purpose of the study and received 20 euros for their
participation. See Figure 1 for an overview of the experimental
procedure.
ECG recordings and analyses
VmHRV was indexed using the root mean square of successive differences
(RMSSD). VmHRV was measured via electrocardiography (ECG; Faros 180°,
Bittium, Kuopio, Finland), at a sampling rate of 500 Hz. Two disposable
pregelled ECG electrodes (Ambu L-00-S/25, Ambu GmbH, Bad Nauheim,
Germany) were used. The negative electrode was placed on the right
infraclavicular fossa (just below the right clavicle). The positive
electrode was placed on the left side of the chest, below the pectoral
muscle in the left anterior axillary line. To extract RMSSD, we usedEDFbrowser software (van Beelen, 2019) and the RHRVpackage in R, software version 4.0.2 (R Core Team, 2020). To remove
artifacts in the ECG signals, the default settings of theFilterNIHR() function of RHRV were used. Subsequently, the
ECG signals were visually inspected, and remaining abnormal beats were
manually removed. The ECG signal cleared of artifacts was interpolated
using the InterpolateNIHR() function from the RHRVpackage. To determine vmHRV during rest periods, RMSSD of the last 5 min
of the 10-min habituation period and the 10-min post-TMS rest period was
used. For the vmHRV reactivity scores, RMSSD was taken from the course
of the respective ESST, and this value was subtracted from the baseline
RMSSD. To determine the average RMSSD throughout the ESST, the RMSSD
recording of the entire ESST was divided into five 5-minute periods, for
each of which an RMSSD value was calculated. The average of these five
RMSSD values was then calculated. A positive reactivity score indicates
a decrease in vmHRV, whereas a negative reactivity score indicates an
increase in vmHRV.
Theta Burst Stimulation
Both continuous and intermittent theta burst stimulation (TBS) were
administered using a D70 Alpha figure-of-8 coil connected to a Magstim2
stimulator (Magstim Company Limited, Minneapolis, USA). For sham
stimulation, a D70 Alpha figure-of-8 sham coil was used, which mimics
the acoustic and somatosensory effects of active TBS. The Visor 2.0
Neuronavigation system (ANT Neuro, Enschede, The Netherlands) was
employed to identify the left dorsolateral prefrontal cortex (dlPFC) and
the corresponding scalp site where the coil was positioned during
MRI-guided rTMS. Instead of utilizing individual MRI scans, a template
MNI-152 scan was transformed to match the participant’s head for
neuronavigated rTMS. Previous research has demonstrated that this
approach accurately determines the location of the left dlPFC with
minimal deviation compared to the use of individual scans (Caulfield et
al., 2022). The coordinates reported in Rusjan et al. (Rusjan et al.,
2010) were employed to determine the position of the left dlPFC.
In the cTBS protocol, a theta burst stimulation pattern consisting of 3
pulses at 50 Hz every 200ms was administered continuously within a
40-second train, resulting in a total of 600 pulses. In the intermittent
iTBS protocol, the same stimulation pattern was delivered for 2 seconds
every 10 seconds for a total of 600 pulses (Huang et al., 2005). Both
cTBS and iTBS were administered at an intensity of 80% of each
participant’s resting motor threshold (RMT). RMT was determined prior to
the respective stimulation protocol and defined as the minimum intensity
of TMS output required to elicit a visible motor response in the right
abductor brevis muscle in 5 out of 10 consecutive attempts.
Emotional Stop Signal Task
The ESST is a modified version of the traditional stop-signal task,
initially developed by Logan and Cowan (1984) to examine response
inhibition by estimating so-called stop-signal reaction times (SSRT). In
the ESST, participants are asked to categorize via keypress the valence
of serially presented stimuli from the International Affective Picture
System (IAPS, Lang et al., 2008), randomized by valence within-block, as
either ‘pleasant/positive’ or ‘unpleasant/negative’ as quickly and
accurately as possible (i.e., go response). Given that in our
experimental design, the ESST had to be performed before and after
stimulation, two equivalent versions of the ESST were designed, each
containing 64 images (32 per IAPS category, i.e., negative, positive)
from the IAPS. In both versions, negative and positive pictures were
matched regarding arousal and intensity. Furthermore, to reach
equivalence between both ESST versions, we aimed to use thematic similar
images with comparable arousal and intensity levels in both versions
(see supplementary files). Both versions consisted of 4 blocks with 110
trials each leading to a total of 440 trials. On a minority of the
trials, a stop signal (i.e., auditory beep tone) was presented after a
variable delay, instructing the participants to suppress the ongoing go
response. As recommended , the task was designed so that approximately
25% of the trials (120 out of 440 trials) were stop trials, i.e.,
contained an auditory stop-signal. In our version of the ESST, the
temporal delay of the stop signal was continuously adjusted in 50ms
steps based on individual performance using a single staircase tracking
algorithm. This adaptive component is essential to estimate SSRTs using
the integration method (see Verbruggen et al., 2019), which requires the
accuracy or the total commission errors (false alarms) to remain around
50%. The order in which the two versions of the ESST were presented to
the single participants was randomized.
Data analysis
Data cleaning
As mentioned above, a total of 12 participants were excluded prior to
the final data analysis. Nine participants were excluded due to poor or
non-compliant performance on the ESST, following current recommendations
(Verbruggen et al., 2019). Of these: (a) One participant was excluded
due to a false alarm rate deviating strongly from 50% (i.e., 40%); (b)
Seven participants with more than 10% go omissions in either the first
or the second ESST or in both, which may lead to biased SSRT estimates
given the number of trials used in this study; (c) One participant
reported after the experimental procedure to have employed a strategy
wherein they awaited the occurrence of a stop signal before executing
the respective key presses.
Three additional participants were excluded from data analysis due to
extreme values in pre-TMS SSRTs (9.39 ms; z = 4.04), pre-TMS on-task
RMSSD (282.14 ms; z = 6.33), and in the change of RMSSD reactivity
(74.82 ms; z = 4.11), respectively.
Statistical analysis