![[Left] The Pipeline Illustrates Primarily Five Stages. [Right] Four Main Regions Are Shown (LF, RF, LP, RP).](https://www.researchgate.net/profile/Pankaj-Pandey-12/publication/364308991/figure/fig2/AS:11431281089208875@1665488220294/Left-The-Pipeline-Illustrates-Primarily-Five-Stages-Right-Four-Main-Regions-Are_Q320.jpg)
Content uploaded by Pankaj Pandey
Author content
All content in this area was uploaded by Pankaj Pandey on Oct 11, 2022
Content may be subject to copyright.
NeuroRegulation http://www.isnr.org
113 | www.neuroregulation.org Vol. 9(3):113–126 2022 doi:10.15540/nr.9.3.113
Exploration of Brain Network Measures Across Three
Meditation Traditions
Pankaj Pandey1, Pragati Gupta2, and Krishna Prasad Miyapuram1,3
1Computer Science and Engineering, Indian Institute of Technology Gandhinagar, Palaj, Gujarat, India
2Institute of Behavioral Sciences, National Forensic Sciences University, Gandhinagar, Gujarat, India
3Centre for Cognitive and Brain Sciences, Indian Institute of Technology Gandhinagar, Palaj, Gujarat, India
Abstract
Research into the similarities and differences between various forms of meditation practice is still in its early
stages. Here, utilizing functional connectivity and graph measures, we present our work examining three
meditation traditions: Himalayan Yoga (HT), Isha Shoonya (SNY), and Vipassana (VIP). EEG activity of the
meditative block is used to build functional brain connections to exploit the resulting networks between various
meditation traditions and a control group. Support vector machine is employed for binary classification, and
models are built with features generated via graph theory measures. We obtain maximum accuracy of 84.76%
with gamma1, 90% with alpha, and 84.76% with theta in HT, SNY, and VIP, respectively. Our key findings involve
(a) higher delta connectivity in Vipassana meditators, (b) synchronization of theta networks in the left hemisphere
inspected to be stronger in the anterior frontal area across meditators, (c) greater involvement of gamma2
processing observed among Himalayan and Vipassana meditators, (d) increased left frontal activity contribution
for all meditators in theta and gamma bands, and (e) modularity engaged extensively in gamma processing
across all meditation traditions. Furthermore, we discuss the implication of this research for neurotechnology
products to enable guided meditation among naive practitioners.
Keywords: EEG signals; meditation; functional connectivity; graph measures; support vector machine;
machine learning; brainwaves; Himalayan Yoga; Isha Shoonya; Vipassana
Citation: Pandey, P., Gupta, P., & Miyapuram, K. P. (2022). Exploration of brain network measures across three meditation traditions.
NeuroRegulation, 9(3), 113–126. https://doi.org/10.15540/nr.9.3.113
*Address correspondence to: Krishna Prasad Miyapuram, Indian
Institute of Technology Gandhinagar, Palaj, Gujarat, 382355, India.
Email: kprasad@iitgn.ac.in
Copyright: © 2022. Pandey et al. This is an Open Access article
distributed under the terms of the Creative Commons Attribution
License (CC-BY).
Edited by:
Rex L. Cannon, PhD, Currents, Knoxville, Tennessee, USA
Reviewed by:
Rex L. Cannon, PhD, Currents, Knoxville, Tennessee, USA
Estate M. Sokhadze, PhD, University of South Carolina, School
of Medicine Greenville, Greenville, SC, USA
Introduction
In recent years, neuroscientific research has
focused on meditation as a mental practice. This is
due to the large-scale benefit it offers, observed in
numerous studies, such as improved attentional
states, metacognitive awareness, cognitive control,
compassion, self-regulation, decreased states of
mind wandering, and so on (Brandmeyer & Delorme,
2018). Multiple studies determine how long- and
short-term meditation practice (measured in hours of
experience) impact the brain (in terms of neural
oscillation and executive functioning tasks such as
working memory). This approach is designed to
integrate mindfulness-based practices like
meditation in a clinical context to treat anxiety,
depression, chronic pain, and stress (Yordanova et
al., 2020). But most of the study misses out on the
significance of each meditation type on distinct brain
circuitry, frequency bands, and cognitive functions
that is unique in itself and cannot be generalized
fully to other types of meditation practices. As each
meditation type can uniquely influence the person
(both psychologically and physiologically) different
meditation practices require careful observation and
rigorous examination before making a causal
interpretation and generalization. With
neurotechnological advancements, meditation
researchers are using electroencephalogram (EEG),
functional magnetic resonance imaging (fMRI),
magnetic resonance imaging (MRI), and single-
photon emission computerized tomography
(SPECT). EEG and fMRI techniques are commonly
employed in meditation research. There are many
Pandey et al. NeuroRegulation
114 | www.neuroregulation.org Vol. 9(3):113–126 2022 doi:10.15540/nr.9.3.113
types of meditation traditions practiced worldwide,
for example, Himalayan Yoga (HT; focused
attention), Vipassana (VIP; open monitoring) and
Isha Shoonya (SNY; open awareness meditation),
and Loving Kindness meditation.
Spectral analysis used in earlier studies on VIP
revealed enhanced gamma activity over the parieto-
occipital electrodes (Braboszcz et al., 2017; Cahn et
al., 2010; van Lutterveld et al., 2017). Gamma band
has been associated with cognitive processes such
as attention, working memory, learning,
consciousness, microsaccades, and visual imagery
(Fries, 2009; Fries et al., 2007), and long-range
neural communication (Nikolić et al., 2013). The
sample entropy (SE) of VIP meditators was higher in
the study by Vivot and colleagues (Vivot et al.,
2020), especially in the alpha and low/high gamma
bands. The alpha band (7–11 Hz) was identified to
have a trait influence as observed in both the
conditions of mind wandering and meditation in a
recent study by (Braboszcz et al., 2017). According
to studies on HT practitioners, their brainwaves are
reported to have sensorimotor alpha, frontal-midline
theta, and parieto-occipital gamma (Braboszcz et al.,
2017; Brandmeyer & Delorme, 2018; Vivot et al.,
2020). Working memory has linkages with alpha
rhythms which are thought to be prevalent in HT
meditation, since it emphasizes the mental repetition
of the mantra and the breath (Braboszcz et al.,
2017). SNY was linked to gamma frequency in the
parieto-occipital, central, and frontal electrodes,
according to a study by Braboszcz et al. (2017).
Since the explicit focus is on "nothingness," it is
unclear what kind of object is sent to the attentional
system for SNY practitioners. Since higher gamma
power over the frontal and parieto-occipital
electrodes is demonstrated as a trait effect, this may
indicate that SNY meditation engages attentional
processes differently than VIP and HT meditation.
According to the research by van Lutterveld et al.
(2017), SNY meditators had greater separations in
their thought charts observed using Hausdorff
distance under the breath awareness condition. In
the current study, the brain states connected to
three crucial and distinctive types of meditation—HT,
VIP, and SNY—are examined. This study's goal is to
leverage functional network measurements to
examine variations between control subjects and
meditators on (a) frequency bands, (b) brain regions,
(c) network measures, and (d) commonalities and
discrepancies among mediators.
Complex network theory has recently gained
prominence (Li & Yang, 2016). Research has
demonstrated that EEG may be utilized to create
brain networks that retain several crucial topological
characteristics (Sun et al., 2019). The temporal
correlation between distant neurophysiological
events is often used to describe a functional
connection in the brain (Friston, 1994). In recent
decades, various neural coupling techniques have
been put forth. Coherence (coh), a linear
dependency measure between two nodes, has been
used to assess functional connectivity (Jalili, 2016).
However, coh is affected by volume conduction.
Since volume conduction is likely to detect brain
activity from the same sources, even if unrelated, it
may result in incorrect correlations between nearby
electrodes. To lessen the consequences of volume
conduction, additional metrics have been
incorporated. It has been shown that imaginary
coherence (imcoh) can eliminate any instantaneous
interactions that are caused by volume conduction
(Nolte et al., 2004). The phase lag index (pli), a
phase synchronization technique, is insensitive to
the volume conduction effect and reveals the
genuine coupling strength between pairs of channels
by excluding interactions produced by zero phase
differences (Stam et al., 2007). In the weighted
phase lag index (wpli), a modification of the pli
wherein observable phase leads and lags are
weighted by the amplitude of the imaginary
component of the cross-spectrum (Vinck et al.,
2011). Corrected imaginary phase locking value
(ciPLV), a metric for assessing synchronization in
the presence of volume conduction or source
leakage effects, was proposed by Bruña and
colleagues (Bruña et al., 2018). We have used all
five connectedness metrics because they can
distinguish between functional networks that are
similar and those that are distinct. Regardless of the
coupling method, our main goal is to find the
functional networks that discriminate between two
groups.
Examining functional connectivity with various graph
theoretical measures shows key topological
characteristics of brain networks (Rubinov & Sporns,
2010). EEG/MEG, functional MRI, diffusion MRI, and
structural MRI are just a few imaging modalities
using graph theory analyses of human brain
networks (He & Evans, 2010). Modularity, node
betweenness, centrality, clustering coefficient, and
the occurrence of highly connected hub regions are
a few network features that have been addressed
often (He & Evans, 2010; Sun et al., 2019; Wang et
al., 2010). Additionally, it has been found that these
network characteristics change over time under
various conditions, such as normal development,
aging, and pathological circumstances. Recent work
by Hiroyasu and colleagues and a dearth of works
Pandey et al. NeuroRegulation
115 | www.neuroregulation.org Vol. 9(3):113–126 2022 doi:10.15540/nr.9.3.113
on network modeling in meditation have shown how
to categorize resting and meditative states using the
centrality measure (Hiroyasu & Hiwa, 2017). In
earlier research on long-term meditators, trait effects
on meditators were examined (Braboszcz et al.,
2017). These effects may indicate a change in the
functional architecture of the human brain compared
to controls. Our research examines the four network
features between long-term practitioners of three
meditation traditions and the control group. Machine
learning classifiers are trained as the most practical
method for spotting differences due to their strong
pattern learning capabilities. Moreover, there has
been a surge in studies using machine learning to
categorize meditation states in recent years
(Chaudhary et al., 2022; Pandey et al., 2022;
Pandey & Miyapuram, 2020; Pandey & Miyapuram,
2021a, 2021c).
Figure 1. Measuring Four Scales of Improvement: Attention, Mind-Wandering, Drowsiness, and Functional Connectivity.
Note. A person is wearing an EEG headset while practicing meditation. Four scales of improvement can be examined on the
mobile screen, and the last one indicates Functional Connectivity. Recording of Day 1 shows the initial connectivity. With
progress in meditation, the application displays the change in connectivity after a few days and even explains the relation with
the connectivity of expert meditators. This is an illustration generated by us to display the potential idea for neurotechnology.
Research and Technology Relevance
Cognitive Relevance
Years of neuroscience research have shown several
advantages to meditation practice (Brandmeyer et
al., 2019). A recent article offered a possible course
of action with a unified framework (Dahl et al., 2020).
The framework suggests awareness, connection,
insight, and purpose as the four fundamental
characteristics of well-being. The only way to acquire
these qualities that offer direct access to one's well-
being is through intentional mental training. A
particular dimension denotes a specific method. A
practitioner can, for instance, utilize concentrated
attention to become aware of mind-wandering
occurrences, maintain focus, and use loving
kindness to cultivate fruitful relationships with others.
Neurotechnology
Many meditation applications are available to
improve awareness and train attention (Migala,
2021). Since no feedback is provided, a novice
practitioner feels pushed and gradually reduces their
practice to the minimal effort until stopping
altogether. Due to the availability of wearable EEG
technology, the market has been able to create
goods that can assist novice practitioners in learning
meditation through real-time feedback and
monitoring their progress over time, as
demonstrated by the use of Muse and Neuphony
meditation products. Here, we suggest a functional
connectivity module enabling practitioners to see
their incremental development and the changes in
connectivity patterns that go along with it.
In Figure 1, three of the four modules can assess
attention, daydreaming, and tiredness in real time.
When the mind wanders, or a person feels sleepy,
practitioners can receive immediate feedback so
they can refocus on the meditation object. After
some practice, people can evaluate their level of
attention, the amount of time their minds wander,
and whether they are awake or asleep while
meditating. In the final module, users can compare
their functional connectivity after a few sessions to
that of potential specialists in various types of
meditation. With the aid of neurotechnology,
cognitive scientists, computer scientists, and signal
processing experts can collaborate to identify the
brain correlates of various stages of meditation.
Therefore, it is conceivable to develop neural
markers for various levels of meditation using signal
processing and machine learning approaches.
Pandey et al. NeuroRegulation
116 | www.neuroregulation.org Vol. 9(3):113–126 2022 doi:10.15540/nr.9.3.113
Learning Representation
The most important step in separating the neural
signals of experts and beginners so that
neurofeedback can be implemented is through
feature engineering. Robust feature extraction
strategies are presented by deep learning and
machine learning to categorize the various stages.
Numerous articles from previous years have
described the brain correlates of meditation. Pandey
and Miyapuram describe a wavelet-based encoding
of the oscillatory signature of meditators (Pandey &
Miyapuram, 2020). In recent investigations,
functional connectivity networks were examined to
predict brain activity in meditators (Pandey et al.,
2021). Convolutional neural networks are used to
create a model that categorizes control and
meditators' cognitive states (Pandey & Miyapuram,
2021b). The SHAP (Shapley Additive Explanations)
explainable model, which employed three nonlinear
dynamics to extract the significance of the scalp
area, was used to analyze EEG data collected
before and after mindfulness-based stress reduction
(MBSR) training to determine the relevance of the
data (Pandey & Miyapuram, 2021c). A recent study
discusses and further categorizes various mental
states associated with meditation using different
machine learning approaches (Kora et al., 2021).
Cognitive science and machine learning researchers
have great potential to identify patterns and leverage
them to create systems that can guide novice
practitioners.
Data Description
Participants and Experimental Design
We used online open-access EEG data (Braboszcz
et al., 2017). Data were collected at the Meditation
Research Institute in Rishikesh, India, from 32
healthy control individuals and 20 meditators from
the VIP school, and 27 meditators from the HT
school, and 20 meditators from the SNY school. All
meditators were chosen for the study based on their
age, gender, and years of meditation practice.
Control subjects were also selected for the study
based on age, gender, and lack of meditation
practice. Researchers wanted to investigate uniform
groups of individuals for this study. Therefore, they
constructed groups based on age and gender to
match the individuals. As a result, there were four
groups of 16 subjects in each meditation group: 16
controls (45 ± 10 years, five females), 16 HT
meditators (43 ± 12 years, two females), 16 SNY
meditators (40 ± 10 years, two females), and 16 VIP
meditators (47 ± 15 years, five females). A single set
of individuals was a control group for all three
meditation traditions.
The experiment was divided into two 20-min
sessions, one titled "Meditation" and the other
"Instructed Mind Wandering." In the first 10 min of
the Meditation block, subjects were instructed to
focus on their breathing (breath focus or inhalation
and exhalation) to prepare for their meditation
practice. This task was used as a primitive practice
period in all three meditation traditions to help
people relax and deepen the depth of their
meditation practice. After 10 min, they were notified
to practice their specific meditation for the next 10
min. Both in the first and second half of the
Meditation block, control participants were instructed
to keep their focus on breath or inhalation and
exhalation. In the Instructed Mind Wandering block,
for the first 10 min, subjects were instructed to
perform mind-wandering tasks, wherein they were
asked to recall autobiographical events which were
emotionally neutral such as routine childhood life,
travels, etc. After the initial 10 min, they were
directed to continue their instructed mind-wandering
task for the next 10 min to preserve consistency with
the Meditation condition. To avoid any order effects,
the task sequence was counterbalanced; that is, in
each of the meditation groups and control group,
eight of the subjects either performed the mind-
wandering task first or the meditation task first. In
our study, we focused on comparing the second part
of the Meditation block between controls (i.e., breath
focus) and meditators (i.e., specific HT, VIP, SNY).
We used preprocessed open access data, and
preprocessing steps are mentioned in this article
(Braboszcz et al., 2017). Participants all signed
informed consent forms before participating. The
Meditation Research Institute Indian ethical
committee and University of California San Diego
ethical committee approved the project (IRB project
# 090731). Interested readers may refer to
Braboszcz et al. (2017) for complete details.
Methods
Functional Connectivity
To create the functional connectivity matrix, we
employed five coupling methods: coherence (coh),
imaginary coherence (imcoh), phase lag index (pli),
weighted phase lag index (wpli), and corrected
imaginary phase-locking value (ciplv). We started
from coh, the earliest measure of functional
connectivity, to ciPLV, the latest measure, as every
coupling method illustrates some similarities and
differentiating synchronization patterns for the same
dataset. In this study, we focus on capturing all the
crucial connectivity relationships that can provide
significant discrimination between control and
meditator irrespective of the coupling method. Each
Pandey et al. NeuroRegulation
117 | www.neuroregulation.org Vol. 9(3):113–126 2022 doi:10.15540/nr.9.3.113
brain connectivity preserves some network topology
that can be scrutinized and reveal new insights into
the meditative state. The subsection on spectral
connectivity presents a brief description of five
coupling methods (MNE, 2022). All the functional
connectivity matrices were calculated every 5 s with
a 2.5-s overlapping window for delta (1–4 Hz), theta
(4–8 Hz), beta (8–12 Hz), alpha (12–20 Hz),
gamma1 (20–60 Hz) and gamma2 (60–100 Hz)
frequency bands along with regions described in
Figure 2. Primarily four areas are left frontal (LF),
right frontal (RF), left parietal (LP), and right parietal
(RP). Based on these regions, intra- and
interfunctional connectivity are computed. Bands are
decided based on the recent article published on the
same dataset (Vivot et al., 2020). Regions are
determined based on the study discussing different
meditation techniques (Yordanova et al., 2020).
Binarization of Brain Networks
The topology of functional networks is often
obscured by faulty and weak connections (Sun et
al., 2019). Thresholding, which involves removing a
portion of the weakest links from the network, is a
popular technique for maintaining a sparse network.
However, deciding this threshold objectively remains
inconclusive. In the recent work of De Vico Fallani
and colleagues (De Vico Fallani et al., 2017), they
introduce a criterion, the efficiency cost optimization
(ECO), to identify the density threshold which filters
the connections depending on the network size
according to a power law. This method accentuates
a network's intrinsic features while maintaining its
sparsity. Hence, we used the ECO binarization
method to remove the weak links. Obtained
networks from coupling methods were binarized and
quantitatively analyzed using graph theory
measures.
Graph Theory Network Metrics
Several graph measures can characterize brain
networks (Rubinov & Sporns, 2010). We computed
functional segregation and integration measures of
binary brain networks for each subject, including all
coupling methods. The capacity for specialized
processing to emerge within tightly interconnected
clusters of brain regions is referred to as functional
segregation. Functional integration in the brain
quickly incorporates specialized information from
various brain regions. We identified four widely
employed network metrics. Functional integration
metrics were node betweenness centrality (NB) and
edge betweenness centrality (EBC). Functional
segregation metrics were clustering coefficient (CC)
and modularity (MU). The proportion of all shortest
routes in a network that connects a particular vertex
is known as NB, whereas the proportion of all
shortest routes in the network that involves a
particular edge is called EBC. Because the concept
of betweenness centrality readily extends to
linkages, it could be utilized to detect essential
anatomical or functional connections. The CC is the
number of triangles surrounding a node and is equal
to the number of neighbors who are neighbors of
each other. MU is a metric that measures how
efficiently a network may be separated into distinct
clusters. Mathematical equations and detailed
explanations can be accessed in this paper
(Rubinov & Sporns, 2010). Network measures were
computed in a Matlab environment.
Machine Learning
Features generated from different network measures
were used for classification. This study trained
binary classifiers between the control and meditator
groups. Support vector machine was selected for
this research due to its well-established theory and
more eloquent quality of easy interpretability. We
trained models by tuning hyperparameters, and
validation was performed using the 10-fold stratified
technique. The classifier’s performance was
evaluated using accuracy, precision, recall, and F1
score. In line with this, we further assessed the
statistical significance of the classifier using the
permutation test with 10,000 rounds. Several articles
have used this test (Ojala & Garriga, 2009) and
discussed the effectiveness of the results via
permutation tests. There were 1,200 models trained
for each group encompassing connectivity methods,
frequency bands, brain regions, and network
measures. A total of 3,600 models were developed,
of which only 154 models were selected based on
the significance of p < .05, and division is provided in
Figure 2. Since there was no class imbalance
present in our data, we found accuracy and the p-
value were sufficient for the presentation. Models
were developed using scikit-learn python
(Pedregosa et al., 2011). The outcome of classifiers
between the control and meditators resulted from
differences in network features and furthermore
explained the differences in connectivity and
synchronization patterns.
Results
The results presented in our study were based on
154 significant models (p < .05), selected using
permutation tests as illustrated in Figure 2. Each
model exhibited a unique combination of coupling
methods, bands, regions, and network metrics.
These values emphasized the discrimination
between control and meditation traditions. We
Pandey et al. NeuroRegulation
118 | www.neuroregulation.org Vol. 9(3):113–126 2022 doi:10.15540/nr.9.3.113
Figure 2. [Left] The Pipeline Illustrates Primarily Five Stages. [Right] Four Main Regions Are Shown (LF, RF, LP, RP).
Note 1 [Left]. The coupling method is selected, followed by the frequency band and region for constructing the brain network
from EEG recordings. The topology of the connectivity graph is explored using graph theory network metrics. Binary classifiers
are built based on the property of a graph. Permutation tests are performed to obtain the significant models (p < .05) for
analysis.
Note 2 [Right]. A combination of 10 electrodes forms each region. All four intrahemispheric regions are further used to form
two more intrahemispheric regions (LP-LF and RP-RF) and four interhemispheric regions (LF-RF, LP-RP, LF-RP, RF-LP),
overall making a total of 10 regions.
focused our study on bands, regions, and network
metrics. Hence, these 154 unique combinations
were segregated and discussed.
Role of Frequency Bands
For each meditation tradition, we have shown the
spread of 154 significant values across all frequency
bands in Figure 3. Each meditation type has some
consistency and some degree of variability in the
role played by a particular frequency band. Broadly,
theta frequency band was found to be uniform
across all meditators. For both HT and VIP
meditators, gamma2 was more dominant. VIP
meditators were found to have a greater amount of
slow frequency delta waves than other meditators.
For the CTR-HT group, all the accuracy was above
70% for most of the frequency bands except in the
gamma1 band, whose accuracy was found to be
around 85% (p < .01), as shown in Table 1. All
bands appear to function well in distinguishing HT
from control, but gamma1 played a more prominent
role than the others. For the CTR-SNY group, most
of the frequency bands, accuracy prediction was
within 70% to discriminate controls from SNY
meditators. The accuracy of alpha-band prediction,
on the other hand, was found to be 90% (p < .01),
with substantially higher efficiency. For the CTR-VIP
group, accuracy predictions were within the range of
70–80%, except for the theta band, which had an
accuracy of 85% (p < .01).
Participation of Regions
As shown in Figure 4, the synchronization of delta
networks was significant for a few clusters in a
variable fashion among all three meditation
traditions (HT, SNY, VIP). For VIP, synchronization
was found intrahemispheric in the LF, RF, and LP
regions and interhemispheric between LF-RF, LF-
RP, and LP-RP regions. During the synchronization
of theta networks, among all the meditation
traditions (HT, VIP, SNY), a stronger anterior-
posterior connectivity in the left hemisphere (LF-LP),
and anterior frontal connectivity (left to the right
hemisphere; i.e., LF-RF regions were found to be
Pandey et al. NeuroRegulation
119 | www.neuroregulation.org Vol. 9(3):113–126 2022 doi:10.15540/nr.9.3.113
Figure 3. A Significant Interaction of Bands With Meditation Traditions and Controls.
Note. The significant counts (p < .05) were obtained by performing permutation tests. It represents how 154 significant models
were distributed across each of the frequency bands, observed among meditators while they performed distinct meditation
types (HT, VIP, SNY).
Table 1
Representation of Accuracy to Correctly Distinguish Controls With Distinct Meditative States in Frequency Bands.
CTR-HT
CTR-SNY
CTR-VIP
Band
Accuracy (%)
p-value
Accuracy (%)
p-value
Accuracy (%)
p-value
delta
71.90
0.02
71.90
0.02
78.57
0.001
theta
78.57
0.003
75.71
0.01
84.76
0.0004
alpha
81.42
0.002
0.90
0.0001
79.04
0.001
beta
74.28
0.01
78.57
0.001
80.95
0.002
gamma1
84.76
0.00009
71.90
0.02
76.19
0.007
gamma2
80.95
0.001
75.71
0.008
80.95
0.003
All the p-values shown in the table are p < .05. Blue highlighted values suggest maximum accuracy in a particular column.
consistent). RF-LP regions were observed to have
interhemispheric connections only in the VIP
meditators. In the HT and SNY meditation, LF-RP
interhemispheric connections were observed. During
the synchronization of alpha networks,
interhemispheric connections between LF-LP were
common among HT and SNY meditators. A stronger
connectivity in the LF-RP region among the HT and
VIP meditators was observed. Moderate
connections in the SNY and VIP groups were
present in the LP-RP region. Overall, across all the
meditators, synchronization in LF, LP, and RP
clusters was indicative of its consistency.
Intrahemispheric connectivity in the RF and LF were
more robust in the HT and SNY meditators,
respectively. Higher intra- and interhemispheric
connections were present in the HT meditators
compared to the other two groups. During the
synchronization of beta networks, a stronger LF
Pandey et al. NeuroRegulation
120 | www.neuroregulation.org Vol. 9(3):113–126 2022 doi:10.15540/nr.9.3.113
Figure 4. Diagrammatic Representation of Statistically
Significant Differences (p < .05), Based on the Allocation
of 154 Values on Frequency Bands and Regions Across
Three Meditation Traditions.
Note. Circles indicate within-cluster (LF, LP, RF, RP)
significance; lines designate intraconnectivity (LF-LP, RF-
RP) and interconnectivity relationship (LF-RF, LP-RP, LF-
RP, RF-LP). Stronger links are shown by denser circles
and lines, based on the number of values obtained after
the permutation test. Different colors represent frequency
bands.
connectivity is observed for all meditators.
Interhemispheric connectivity is observed in the LF-
RF region and is seen across all meditators. A
greater interhemispheric connection can be viewed
in the HT and VIP meditators such as LF-LP, LP-RP,
etc. During the synchronization of gamma1
networks, both intra- and interhemispheric
connections were seen across meditators.
Intrahemispheric LF-LP connectivity was common
for all meditators. LP region had stronger
connections among SNY meditators. Higher intra-
and interhemispheric connections were found
among HT meditators. RF-RP synchronization was
found among VIP and HT meditators, whereas LP-
RP connectivity was only observed in HT meditators.
Interhemispheric connections in the LF-RF and LF-
RP regions were consistent only in the SNY and VIP
meditators. During the synchronization of gamma2
networks, consistent and higher intra- and
interhemispheric connections are observed in HYT
and VIP meditators (i.e., stronger connectivity in the
LF-LP, LF-RP, LP-RP, RF-RP regions). Connectivity
in the SNY meditators is not so dense both in the
intra- and interhemispheric regions.
In Table 2, across all meditation traditions, the
accuracy of most brain regions with frequency bands
is greater than 70%. Across all intrahemispheric
regions, the LF region was revealed to have the
highest accuracy, especially for the HT (in gamma1)
and VIP (in theta) meditation groups. For HT and
SNY meditators, the alpha band was shown to play
a role in the RF region, with an accuracy of 81% and
75%, respectively. In the RF-RP region, the SNY
meditator's maximum accuracy was obtained in the
LP region in the alpha band. In the LF-LP region,
gamma2 was expressed in both HT and VIP
meditator groups. Gamma2 bands can be seen for
the mediators, notably for the HT and SNY groups.
The beta band was observed only in RF and RF-RP
regions for VIP and in the LF region for SNY
meditators, but not for HT meditators. Broadly, most
brain regions were found to have an accuracy within
70–80%, distinguishing frequency bands across all
the meditation traditions, as shown in Table 3. In the
anterior LF-RF region, a beta band is present across
HT and SNY meditators, with 70% and 79%,
respectively. In LP-RP region, maximum accuracy of
78% is obtained for HT group in gamma1 band and
90% in SNY group in alpha band. Maximum
accuracy is obtained in the RF-LP regions for VIP
meditators in the theta band, but RF-LP regions for
HT and SNY meditators did not have significant
accuracy.
Significance of Network Metrics
We observed the maximum number of allocations in
modularity followed by NB as shown in Figure 5.
EBC showed the maximum involvement between
interconnectivity of the left and right frontal areas
(LF-RF). The CC primarily engaged in the left and
right frontal regions. In Figure 6 (regions), the
interconnectivity of the left frontal and left parietal of
SNY and VIP were observed in MU, NB, and EBC.
In contrast, HT was engaged in MU and. NB was
attributed across all regions in VIP, whereas MU was
Pandey et al. NeuroRegulation
121 | www.neuroregulation.org Vol. 9(3):113–126 2022 doi:10.15540/nr.9.3.113
Table 2
Maximum Classification Accuracy of Intrahemispheric Brain Regions Along With Frequency Bands.
CTR-HT
CTR-SNY
CTR-VIP
Region
Accuracy
(%)
Band
p-value
Accuracy
(%)
Band
p-value
Accuracy
(%)
Band
p-value
LF
84.76
gamma1
0.0009
75.71
beta
0.005
84.76
theta
0.0004
RF
81.42
alpha
0.002
75.23
alpha
0.009
80.95
beta
0.002
LP
80.95
gamma2
0.001
78.57
alpha
0.005
78.57
delta
0.001
RP
70.47
gamma1
0.02
71.90
delta
0.02
69.04
gamma2
0.04
LF-LP
77.61
gamma2
0.005
75.71
theta
0.01
80.95
gamma2
0.003
RF-RP
76.19
gamma2
0.004
74.76
gamma2
0.02
78.57
beta
0.003
Blue highlighted accuracy values suggest maximum accuracy in a particular column. All the p-values shown in the table are p
< .05.
Table 3
Maximum Accuracy Obtained to Distinguish Specific Meditation Traditions Based on Interhemispheric Regions
and Frequency Bands.
CTR-HT
CTR-SNY
CTR-VIP
Region
Accuracy
(%)
Band
p-value
Accuracy
(%)
Band
p-value
Accuracy
(%)
Band
p-value
LF-RF
70.47
beta
0.02
78.57
beta
0.001
76.19
gamma1
0.007
LP-RP
78.09
gamma1
0.007
90.46
alpha
0.0001
78.09
beta
0.007
LF-RP
77.61
alpha
0.005
72.38
gamma2
0.01
78.09
delta
0.003
RF-LP
-
-
79.04
theta
0.003
Blue highlighted accuracy values suggest maximum accuracy in a particular column. All the p-values shown in the table are p
< 0.05.
involved in HT and SNY. VIP showed a greater
number of connections in interconnectivity between
left and right parietal, including EBC and NB. The
right frontal of SNY was less involved than other
groups, and network properties were captured with
modularity and CC. Only VIP exhibited
interconnectivity of the right frontal and left parietal
with MU and NB. In Figure 6 (bands), VIP involved
all network metrics in the delta and gamma2,
whereas theta, alpha, and gamma2 in HT and SNY
were involved in alpha and beta. NB and EBC were
contributed across all bands in VIP, whereas MU in
SNY. The similarity between all meditators was
observed in frequency bands: (a) theta band
engaged in left and right frontal interconnectivity via
EBC, (b) more cross-connections involved in gamma
processing using MU, and (c) beta waves in left
frontal and interconnection with right frontal reflected
connections with NB and EBC.
Discussion
Our findings show that (a) VIP practitioners have
higher delta connectivity; (b) theta network
synchronization in the left hemisphere is observed to
be greater and more constant across meditators in
the LF-LP region and in the anterior frontal area; (c)
high levels of gamma2 processing in HT and VIP
practitioners favorably correlated with the number of
hours spent meditating in these two meditation
traditions; (d) the left frontal activity contributes to
theta and gamma bands for all meditators; (e) in
contrast to EBC and CC, MU and NB are heavily
weighted in graph measurements; and (f) MU is
engaged extensively in gamma processing across
all meditation traditions. Furthermore, left-right intra-
inter hemisphere networks are engaged in varied
ways, with each meditation state having unique
synchronization patterns.
We observed that gamma2 was more noticeable in
both HT and VIP meditators. This might result from
Pandey et al. NeuroRegulation
122 | www.neuroregulation.org Vol. 9(3):113–126 2022 doi:10.15540/nr.9.3.113
Figure 5. This Image Illustrates an Overall Distribution of Network Metrics Across Regions and Bands, Including All Traditions.
Figure 6. Detailed Representation of Network Metrics Concerning Regions and Bands Across Meditator Traditions.
Pandey et al. NeuroRegulation
123 | www.neuroregulation.org Vol. 9(3):113–126 2022 doi:10.15540/nr.9.3.113
more hours of meditation practice (Braboszcz et al.,
2017), which can be presented as a trait effect
exhibiting gamma2. Previous studies had observed
high-frequency gamma band activity during
meditation when participants had an increased hour
of meditation experience (Ferrarelli et al., 2013;
Hauswald et al., 2015).
Theta band activity was observed in the current
study across all meditation practices. This can be
linked with the cultivation of long-term meditation
practice exhibiting theta activity over the frontal
cortex, which is associated with sustained and
internally directed attention states (Brandmeyer &
Delorme, 2018). Theta activity is related to executive
functioning tasks such as working memory and
others that require cognitive control (Cavanagh &
Frank, 2014; Cavanagh & Shackman, 2015). Theta
rhythm was observed among all meditation traditions
with stronger left anterior-posterior (LF-LP) and
anterior frontal connectivity (LF-RF). Theta band's
importance in meditation has been mostly related to
top-down control mechanisms, such as heightened
conflict monitoring and neural communication over
long and broad networks related to cognitive
processing (Cavanagh & Frank, 2014). In work by
(Manna et al., 2010; Marzetti et al., 2014; Yordanova
et al., 2020), similar observations of theta coupling
across the left hemisphere anterior posterior (LF-LP
areas) have been reported throughout three
meditation traditions (focused attention, open
monitoring, and loving kindness). The engagement
of leftward asymmetry (Cahn & Polich, 2009),
anterior frontal (Banquet, 1973), and frontal midline
(Brandmeyer & Delorme, 2018) in the theta band
has been observed consistently among meditators.
VIP practitioners were shown to have an increase in
delta power. Past findings were found to support our
results (Cahn et al., 2010; Cahn & Polich, 2009) and
found that decreased frontal delta power in long-
term VIP practitioners, while increased frontal delta
in long-term meditators has been reported in zen
(Faber et al., 2008) and qi-gong (Tei et al., 2009).
VIP meditators may reflect a functional inhibition of
brain appraisal systems in keeping detached from
analysis, judgment, and expectation. For VIP
meditators, delta power synchronizes intra- (LF, RF,
LP) and interhemispheric (LF-RF, LF-RP, LP-RP).
Prior research on meditation has shown that this
increased frontal delta activity manifests as a
baseline relative suppression of cognitive attention
and a more vital detachment from current daily
experiences (Faber et al., 2008; Tei et al., 2009).
The LF, LP, and RP clusters for alpha
synchronization were seen among all meditation
techniques. The LF, LP, and RP clusters of alpha
synchronization were observed for all meditation
practices. Alpha power is essential for processing
and integrating somatosensory information, working
memory, and cognitive entrainment during
meditation (Brandmeyer & Delorme, 2018).
According to studies, different meditation types may
affect alpha power changes (Amihai & Kozhevnikov,
2015). This can be inferred to some extent from our
study's observations of regional variability (inter- and
intrahemisphere) due to different meditation
practices, such as the increased power of alpha LF-
LP frequently observed among HT and SNY
meditators but not VIP practitioners.
The study by Yordanova et al. (2020) specifically for
open monitoring meditation found left frontal
coupling in beta bands, documented for all
meditation traditions. We identified that lateralized
increase in intra- and interhemispheric beta
synchronization distinguished particular stages of
meditation with shared involvement in the related
clusters. The most frequently associated tasks with
beta oscillations are endogenous, top-down
regulated processing, and conscious processing,
which promotes long-range re-entrant connections
between cortical areas and greater communication
through coherence. Lateralized beta connection may
represent the amount of selected information (little
vs. large) or the type of attentional process of
selection (narrow/focused vs. wide/monitoring;
Yordanova et al., 2020).
During the gamma1 synchronization, LF-LP
connectivity was common for all meditation types.
While LP-RP connectivity was only noticed in HT
meditators, VIP and HT meditators displayed RF-RP
synchronization. This is due to the function of
gamma in the overall attentive state, working
memory activation, information integration, and
neuronal transmission underlying conscious
awareness (Braboszcz et al., 2017; Cahn et al.,
2013; Vivot et al., 2020). Neural coupling of gamma
2 frequency is primarily seen with higher inter- and
intrahemispheric interaction between brain regions
in HT and VIP meditators. It indicates the trait effect
with increased hours of meditation practice, leading
to neuroplastic change with the increase in neural
connections.
Our research showed that modularity makes a
considerable contribution. Modules are crucial for
breaking more extensive networks into basic
"building blocks," like internally highly connected
Pandey et al. NeuroRegulation
124 | www.neuroregulation.org Vol. 9(3):113–126 2022 doi:10.15540/nr.9.3.113
clusters with weaker linkages. In neurobiology,
modular divisions are significant because they
distinguish brain parts with similar functions (Sporns,
2022). There appears to be ample room for future
research to comprehend the underlying phenomena
of modularity between two groups (meditators vs.
controls).
Our results have been presented using a data-driven
methodology, making them more interpretable and
subject to further investigation for graph measures.
However, this work offers a viable concept for
consumer wearable headsets that can show how
functional connectivity evolves as meditation
practice progresses. The naive practitioner can
comprehend the relationship between their
functional connectivity patterns with different types
of experienced meditators.
Conclusion
In this study, we compared three meditation
traditions to a control group to find differences in
frequency bands, regions, and network topological
organization. Five coupling methods—including coh,
imcoh, pli, wpli, and ciplv—were used to construct
functional brain networks from the earliest to the
most recent. Four separate graph theory network
metrics (NBC, EBC, CC, MU), including functional
segregation and integration, were used to examine
six frequency bands, six intrahemispheric, and four
interhemispheric connections. The 3600 models
were reduced to 154 for examination using
permutation tests, which provided diverse insights
into the meditator groups. Left hemisphere theta
synchronization (LF-LP) and anterior frontal (LF-RF)
areas were visible for all meditation practitioners.
Here, the presence of the gamma2 band (strong
connections between the intra-interhemispheres) is
consistently observed across HT and VIP
meditators, indicating a characteristic influence (due
to the increased hours of meditation practice). The
research done in earlier literature on a comparable
dataset supports this. Additional data showed the
importance of various frequency bands and brain
regions in differentiating between different styles of
meditation, such as elevated delta power in VIP and
improved left parietal (LP) connectivity in SNY
practitioners. These neural connections among
meditators are still in the early stages of research as
to how and why they develop. Using brain
connectivity and graph measurements, this study
generally sheds light on the interaction effect of
neural oscillations with intra- and interhemispheric
brain areas during a particular meditative state, both
globally and specifically. Future research can focus
on the biomarkers found in graph measures for the
various meditation traditions.
Author Acknowledgements
We are grateful for the support provided by the
Science and Engineering Research Board (SERB)
and PlayPower Labs to Pankaj Pandey for the Prime
Minister's Research Fellowship (PMRF). We thank
the Federation of Indian Chambers of Commerce &
Industry (FICCI) for facilitating this PMRF.
Author Disclosure
The author declares no conflicts of interest.
References
Amihai, I., & Kozhevnikov, M. (2015). The influence of Buddhist
meditation traditions on the autonomic system and attention.
BioMed Research International, 2015. https://doi.org/10.1155
/2015/731579
Banquet, J. P. (1973). Spectral analysis of the EEG in meditation.
Electroencephalography and Clinical Neurophysiology, 35(2),
143–151. https://doi.org/10.1016/0013-4694(73)90170-3
Braboszcz, C., Cahn, B. R., Levy, J., Fernandez, M., & Delorme,
A. (2017). Increased gamma brainwave amplitude compared
to control in three different meditation traditions. PLoS ONE,
12(1), Article e0170647. https://doi.org/10.1371
/journal.pone.0170647
Brandmeyer, T., & Delorme, A. (2018). Reduced mind wandering
in experienced meditators and associated EEG correlates.
Experimental Brain Research, 236(9), 2519–2528.
https://doi.org/10.1007/s00221-016-4811-5
Brandmeyer, T., Delorme, A., & Wahbeh, H. (2019). The
neuroscience of meditation: Classification, phenomenology,
correlates, and mechanisms. Progress in Brain Research,
244, 1–29. https://doi.org/10.1016/bs.pbr.2018.10.020
Bruña, R., Maestú, F., & Pereda, E. (2018). Phase locking value
revisited: Teaching new tricks to an old dog. Journal of Neural
Engineering, 15(5), Article 056011. https://doi.org/10.1088
/1741-2552/aacfe4
Cahn, B. R., Delorme, A., & Polich, J. (2010). Occipital gamma
activation during Vipassana meditation. Cognitive Processing,
11(1), 39–56. https://doi.org/10.1007/s10339-009-0352-1
Cahn, B. R., Delorme, A., & Polich, J. (2013). Event-related delta,
theta, alpha and gamma correlates to auditory oddball
processing during Vipassana meditation. Social Cognitive and
Affective Neuroscience, 8(1), 100–111. https://doi.org
/10.1093/scan/nss060
Cahn, B. R., & Polich, J. (2009). Meditation (Vipassana) and the
P3a event-related brain potential. International Journal of
Psychophysiology, 72(1), 51–60. https://doi.org/10.1016
/j.ijpsycho.2008.03.013
Cavanagh, J. F., & Frank, M. J. (2014). Frontal theta as a
mechanism for cognitive control. Trends in Cognitive
Sciences, 18(8), 414–421. https://doi.org/10.1016
/j.tics.2014.04.012
Cavanagh, J. F., & Shackman, A. J. (2015). Frontal midline theta
reflects anxiety and cognitive control: Meta-analytic evidence.
Journal of Physiology-Paris, 109(1–3), 3–15. https://doi.org
/10.1016/j.jphysparis.2014.04.003
Chaudhary, S., Pandey, P., Miyapuram, K. P., & Lomas, D.
(2022). Classifying EEG signals of
mind-wandering across different styles of meditation. In M.
Mahmud, J. He, S. Vassanelli, A. van Zundert, & N. Zhong
(Eds.), Brain Informatics. BI 2022. Lecture Notes in Computer
Pandey et al. NeuroRegulation
125 | www.neuroregulation.org Vol. 9(3):113–126 2022 doi:10.15540/nr.9.3.113
Science (vol. 13406, pp. 152–163). Springer, Cham.
https://doi.org/10.1007/978-3-031-15037-1_13
Dahl, C. J., Wilson-Mendenhall, C. D., & Davidson, R. J. (2020).
The plasticity of well-being: A training-based framework for
the cultivation of human flourishing. Proceedings of the
National Academy of Sciences of the United States of
America, 117(51), 32197–32206. https://doi.org/10.1073
/pnas.2014859117
De Vico Fallani, F., Latora, V., & Chavez, M. (2017). A topological
criterion for filtering information in complex brain networks.
PLoS Computational Biology, 13(1), Article e1005305.
https://doi.org/10.1371/journal.pcbi.1005305
Faber, P. L., Steiner, M. E., Lehmann, D., Pascual-Marqui, R. D.,
Jäncke, L., Esslen, M., & Gianotti, L. R. R. (2008).
Deactivation of the medial prefrontal cortex in experienced
Zen meditators. Brain Topography, 20, 172.
Ferrarelli, F., Smith, R., Dentico, D., Riedner, B. A., Zennig, C.,
Benca, R. M., Lutz, A., Davidson, R. J., & Tononi, G. (2013).
Experienced mindfulness meditators exhibit higher parietal-
occipital EEG gamma activity during NREM sleep. PLoS
ONE, 8(8), Article e73417. https://doi.org/10.1371
/journal.pone.0073417
Fries, P. (2009). Neuronal gamma-band synchronization as a
fundamental process in cortical computation. Annual Review
of Neuroscience, 32(1), 209–224. https://doi.org/10.1146
/annurev.neuro.051508.135603
Fries, P., Nikolić, D., & Singer, W. (2007). The gamma cycle.
Trends in Neurosciences, 30(7), 309–316. https://doi.org
/10.1016/j.tins.2007.05.005
Friston, K. J. (1994). Functional and effective connectivity in
neuroimaging: A synthesis. Human Brain Mapping, 2(1–2),
56–78. https://doi.org/10.1002/hbm.460020107
Hauswald, A., Übelacker, T., Leske, S., & Weisz, N. (2015). What
it means to be Zen: Marked modulations of local and
interareal synchronization during open monitoring meditation.
NeuroImage, 108, 265–273. https://doi.org/10.1016
/j.neuroimage.2014.12.065
He, Y., & Evans, A. (2010). Graph theoretical modeling of brain
connectivity. Current Opinion in Neurology, 23(4), 341–350.
https://doi.org/10.1097/wco.0b013e32833aa567
Hiroyasu, T., & Hiwa, S. (2017, March). Brain functional state
analysis of mindfulness using graph theory and functional
connectivity. In 2017 AAAI Spring Symposium on Wellbeing
AI: From Machine Learning to Subjectivity Oriented
Computing Technical Report SS-17-08. https://www.aaai.org
/ocs/index.php/SSS /SSS17/paper/view/15333/14622
Jalili, M. (2016). Functional brain networks: Does the choice of
dependency estimator and binarization method matter?
Scientific Reports, 6(1), Article 29780. https://doi.org/10.1038
/srep29780
Kora, P., Meenakshi, K., Swaraja, K., Rajani, A., & Raju, M. S.
(2021). EEG based interpretation of human brain activity
during yoga and meditation using machine learning: A
systematic review. Complementary Therapies in Clinical
Practice, 43, Article 101329. https://doi.org/10.1016
/j.ctcp.2021.101329
Li, X.-J., & Yang, G.-H. (2016). Graph theory-based pinning
synchronization of stochastic complex dynamical networks. In
IEEE Transactions on Neural Networks and Learning
Systems, 28(2), 427–437. https://doi.org/10.1109
/tnnls.2016.2515080
Manna, A., Raffone, A., Perrucci, M. G., Nardo, D., Ferretti, A.,
Tartaro, A., Londei, A., Del Grattta, C., Belardinelli, M. O., &
Romani, G. L. (2010). Neural correlates of focused attention
and cognitive monitoring in meditation. Brain Research
Bulletin, 82(1–2), 46–56. https://doi.org/10.1016
/j.brainresbull.2010.03.001
Marzetti, L., Di Lanzo, C., Zappasodi, F., Chella, F., Raffone, A.,
& Pizzella, V. (2014). Magnetoencephalographic alpha band
connectivity reveals differential default mode network
interactions during focused attention and open monitoring
meditation. Frontiers in Human Neuroscience, 8, Article 832.
https://doi.org/10.3389/fnhum.2014.00832
Migala, J. (2021, May). These 7 apps will deepen your meditation
practice. Very Well Mind. https://www.verywellmind.com/best-
meditation-apps-4767322
MNE. (n.d.). mne.connectivity.spectralconnectivity.
https://mne.tools
Muse. (n.d.) Meditation made easy. https://choosemuse.com
Neuphony. (n.d.) https://neuphony.com
Nikolić, D., Fries, P., & Singer, W. (2013). Gamma oscillations:
Precise temporal coordination without a metronome. Trends
in Cognitive Sciences, 17(2), 54–55. https://doi.org/10.1016
/j.tics.2012.12.003
Nolte, G., Bai, O., Wheaton, L., Mari, Z., Vorbach, S., & Hallett, M.
(2004). Identifying true brain interaction from EEG data using
the imaginary part of coherency. Clinical Neurophysiology,
115(10), 2292–2307. https://doi.org/10.1016
/j.clinph.2004.04.029
Ojala, M., & Garriga, G. C. (2009). Permutation tests for studying
classifier performance. 2009 Ninth IEEE International
Conference on Data Mining, 908–913. https://doi.org
/10.1109/icdm.2009.108
Pandey, P., Gupta, P., Chaudhary, S., Miyapuram, K. P., &
Lomas, D. (2022, July). Real-time sensing and
neurofeedback for practicing meditation using simultaneous
EEG and eye tracking. In 2022 IEEE
Region 10 Symposium (TENSYMP; pp. 1–6). IEEE.
Pandey, P., Gupta, P., & Miyapuram, K. P. (2021, September).
Brain connectivity based classification of meditation expertise.
In M. Mahmud, M. S. Kaiser, S. Vassanelli, Q. Dai, and N.
Zhong (Eds.), Brain Informatics. BI 2021. Lecture Notes in
Computer Science (vol. 12960, pp. 89–98). Springer, Cham.
https://doi.org/10.1007/978-3-030-86993-9_9
Pandey, P., & Miyapuram, K. P. (2020, July). Classifying
oscillatory signatures of expert vs nonexpert meditators. 2020
International Joint Conference on Neural Networks (IJCNN),
1–7. https://doi.org/10.1109/ijcnn48605.2020.9207340
Pandey, P., & Miyapuram, K. P. (2021a, April). Non-linear
analysis of expert and non-expert meditators using machine
learning. https://doi.org/10.13140/RG.2.2.18323.60968
Pandey, P., & Miyapuram, K. P. (2021b, July). BRAIN2DEPTH:
Lightweight CNN model for classification of cognitive states
from EEG recordings. In Annual Conference on Medical
Image Understanding and Analysis (pp. 394–407). Springer,
Cham. https://doi.org/10.1007/978-3-030-80432-9_30
Pandey, P., & Miyapuram, K. P. (2021c, December). Nonlinear
EEG analysis of mindfulness training using interpretable
machine learning. 2021 IEEE International Conference on
Bioinformatics and Biomedicine (BIBM), 3051–3057.
https://doi.org/10.1109/bibm52615.2021.9669457
Pedregosa, F., Varoquaux, G., Gramfort, A., Michel, V., Thirion,
B., Grisel, O., Blondel, M., Prettenhofer, P., Weiss, R.,
Dubourg, V., Vanderplas, J., Passos, A., Cournapeau, D.,
Brucher, M., Perrot, M., & Duchesnay, E. (2011). Scikit-learn:
Machine learning in Python. Journal of Machine Learning
Research, 12, 2825–2830.
Rubinov, M., & Sporns, O. (2010). Complex network measures of
brain connectivity: Uses and interpretations. NeuroImage,
52(3), 1059–1069. https://doi.org/10.1016
/j.neuroimage.2009.10.003
Sporns, O. (2022). Graph theory methods: Applications in brain
networks. Dialogues in Clinical Neuroscience, 20(2), 111–
121. https://doi.org/10.31887/dcns.2018.20.2/osporns
Stam, C. J., Nolte, G., & Daffertshofer, A. (2007). Phase lag
index: Assessment of functional connectivity from multi
channel EEG and MEG with diminished bias from common
sources. Human Brain Mapping, 28(11), 1178–1193.
https://doi.org/10.1002/hbm.20346
Pandey et al. NeuroRegulation
126 | www.neuroregulation.org Vol. 9(3):113–126 2022 doi:10.15540/nr.9.3.113
Sun, S., Li, X., Zhu, J., Wang, Y., La, R., Zhang, X., Wei, L., & Hu,
B. (2019). Graph theory analysis of functional connectivity in
major depression disorder with high-density resting state EEG
data. IEEE Transactions on Neural Systems and
Rehabilitation Engineering, 27(3), 429–439. https://doi.org
/10.1109/tnsre.2019.2894423
Tei, S., Faber, P. L., Lehmann, D., Tsujiuchi, T., Kumano, H.,
Pascual-Marqui, R. D., Gianotti, L. R. R., & Kochi, K. (2009).
Meditators and non-meditators: EEG source imaging during
resting. Brain Topography, 22(3), 158–165. https://doi.org
/10.1007/s10548-009-0107-4
van Lutterveld, R., van Dellen, E., Pal, P., Yang, H., Stam, C. J.,
& Brewer, J. (2017). Meditation is associated with increased
brain network integration. NeuroImage, 158, 18–25.
https://doi.org/10.1016/j.neuroimage.2017.06.071
Vinck, M., Oostenveld, R., van Wingerden, M., Battaglia, F., &
Pennartz, C. M. A. (2011). An improved index of phase-
synchronization for electrophysiological data in the presence
of volume-conduction, noise and sample-size bias.
NeuroImage, 55(4), 1548–1565. https://doi.org/10.1016
/j.neuroimage.2011.01.055
Vivot, R. M., Pallavicini, C., Zamberlan, F., Vigo, D., &
Tagliazucchi, E. (2020). Meditation increases the entropy of
brain oscillatory activity. Neuroscience, 431, 40–51.
https://doi.org/10.1016/j.neuroscience.2020.01.033
Wang, J., Zuo, X., & He, Y. (2010). Graph-based network analysis
of resting-state functional MRI. Frontiers in Systems
Neuroscience, 4, Article 16. https://doi.org/10.3389
/fnsys.2010.00016
Yordanova, J., Kolev, V., Mauro, F., Nicolardi, V., Simione, L.,
Calabrese, L., Malinowski, P., & Raffone, A. (2020). Common
and distinct lateralised patterns of neural coupling during
focused attention, open monitoring and loving kindness
meditation. Scientific Reports, 10(1), Article 7430.
https://doi.org /10.1038/s41598-020-64324-6
Received: June 19, 2022
Accepted: July 31, 2022
Published: September 30, 2022
