Dissociating meditation proficiency and experience dependent EEG changes during traditional Vipassana meditation practice

Accepted Manuscript Title: Dissociating meditation proficiency and experience dependent EEG changes during traditional Vipassana meditation practice Authors: Ratna Jyothi, Ajay Kumar Nair, Rahul Venugopal, Arun Sasidharan, Prasanta Kumar Ghosh, John P. John, Seema Mehrotra, Ravindra Panth, Bindu M. Kutty PII: DOI: Reference:

S0301-0511(18)30172-8 https://doi.org/10.1016/j.biopsycho.2018.03.004 BIOPSY 7515

To appear in: Received date: Revised date: Accepted date:

10-9-2017 5-3-2018 5-3-2018

Please cite this article as: Jyothi, Ratna, Nair, Ajay Kumar, Venugopal, Rahul, Sasidharan, Arun, Ghosh, Prasanta Kumar, John, John P., Mehrotra, Seema, Panth, Ravindra, Kutty, Bindu M., Dissociating meditation proficiency and experience dependent EEG changes during traditional Vipassana meditation practice.Biological Psychology https://doi.org/10.1016/j.biopsycho.2018.03.004 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Dissociating meditation proficiency and experience dependent EEG changes during traditional Vipassana meditation practice

Ratna Jyothia1, Ajay Kumar Naira1, Rahul Venugopala, Arun Sasidharana, Prasanta Kumar Ghoshb, John P Johnc, Seema Mehrotrad, Ravindra Panthe, Bindu M Kuttya*

Department of Neurophysiology, cDepartment of Psychiatry, dDepartment of Clinical Psychology, National Institute of Mental Health and Neuro Sciences (NIMHANS), Bengaluru-560029, India. b Department of Electrical Engineering, Indian Institute of Science (IISc), Bengaluru-560012, India e Department of Buddhist Philosophy, Nava Nalanda Mahavihara, Nalanda, Bihar, India.

SC R

IP T

a

1

These authors contributed equally to this work.

A

N

U

*Corresponding author, Bindu M. Kutty, Professor and Head, Department of Neurophysiology, National Institute of Mental Health and Neuro Sciences (NIMHANS), P.B. No. 2900, Dharmaram P.O, Hosur Main Road, Bengaluru-560029, Karnataka, India. Tel: +91 80 2699 5170. Fax: +91 80 2656 2121 Email: [email protected]

ED

M

Email addresses of all the authors: [email protected], [email protected], [email protected], [email protected], [email protected], [email protected], [email protected], [email protected], [email protected]

Different EEG signatures of attention, mindfulness and loving-kindness meditation Ecologically sound design using traditional Vipassana meditation module Dissociation between proficiency and duration of practice Converging evidence from power spectra, permutation entropy and fractal dimensions

CC E

   

PT

Highlights

Abstract

A

Meditation, as taught by most schools of practice, consists of a set of heterogeneous techniques. We wanted to assess if EEG profiles varied across different meditation techniques, proficiency levels and experience of the practitioners. We examined EEG dynamics in Vipassana meditators (Novice, Senior meditators and Teachers) while they engaged in their traditional meditation practice (concentration, mindfulness and loving kindness in a structured manner) as taught by S.N. Goenka. Seniors and Teachers (vs Novices) showed trait increases in delta (1-4 Hz), theta-alpha (6-10 Hz) and low-gamma power (30-40 Hz) at baseline rest; state-trait increases in low-alpha (8-10 Hz) and low1

gamma power during concentrative and mindfulness meditation; and theta-alpha and low-gamma power during loving-kindness meditation. Permutation entropy and Higuchi fractal dimension measures further dissociated high proficiency from duration of experience as only Teachers showed consistent increase in network complexity from baseline rest and state transitions between the different meditation states. Keywords: Vipassana meditation; EEG; Permutation entropy; Fractal dimensions; Neural plasticity

IP T

1. Introduction

U

SC R

Meditation is an umbrella term for a set of heterogeneous techniques that engage several different neurocognitive processes and typically induce beneficial effects on brain and behavior (Boccia, Piccardi, & Guariglia, 2015; Fox et al., 2016; Nair, Sasidharan, John, Mehrotra, & Kutty, 2017). A recent attempt at classifying meditative practices based on cognitive mechanisms acknowledged that many meditative practices might span multiple categories (Dahl, Lutz, & Davidson, 2015). As an example, mindfulness based meditation has aspects of concentration, mindfulness and loving kindness (Manuello, Vercelli, Nani, Costa, & Cauda, 2016) and thus straddles the attentional, constructive and deconstructive families (Dahl et al., 2015).

PT

ED

M

A

N

We had three considerations while undertaking the present study. Since mindfulness might mean several different things (Davidson & Kaszniak, 2015), the first consideration was to specify the context under which a study is carried out. In particular, the traditional practice and philosophical position underlying the meditative practice needs to be articulated (Awasthi, 2013). EEG studies examining the neurophysiology of mindfulness based meditation techniques have found consistent changes in theta and alpha power (Cahn & Polich, 2006; Lomas, Ivtzan, & Fu, 2015). This might imply that there are at least some common mechanisms underlying these meditation techniques as they finally involve some aspect of mindfulness (Manuello et al., 2016). On the other hand, it is possible that a context based study of mindfulness meditation might reveal a more nuanced understanding of the neurophysiological underpinnings of the different meditative techniques.

A

CC E

The second consideration was that long term practice has a trait influence on meditation state (Davidson & Kaszniak, 2015). Several studies examine the influence of duration of practice on meditation state by comparing EEG changes in long term meditators in comparison to novice meditators. The challenge is that there is wide variability in the operational definition of novice and non-novice meditators (Lomas et al., 2015) and sometimes the duration of practice that is categorized as ‘experienced’ by one study could be categorized as ‘short term’ by another study (Lomas et al., 2015; Nair et al., 2017). A further complication is that proficiency in meditation is often conflated with duration of practice (as a reasonable approximation), but there can be systematic differences in practice in those who teach meditation (as compared to those who just practice for a long time) that can lead to different meditation proficiency levels. The third consideration was that most EEG studies examine power spectral decompositions using Fourier transforms. These techniques assume statistical stationarity of the EEG time series over short intervals. While this might be a suitable approximation, it is known that EEG time series is nonlinear, noisy and non-stationary in nature (Stam, 2005). Therefore, it might be valuable to examine these 2

dynamical systems using non-linear measures of complexity such as permutation entropy and fractal dimensions. These key concepts are briefly introduced in the next two paragraphs.

SC R

IP T

Permutation entropy (PE) is a natural complexity measure of any time series that is calculated by assigning symbols to patterns of neighboring time point sequences and then examining the information entropy (a measure derived from probability) of the symbol dynamics over the course of the full time series (Bandt & Pompe, 2002). The two defining parameters of PE are order and delay (Riedl, Muller, & Wessel, 2013). Order is the number of data points taken in one set to generate the symbol. If the order is n, there can be n! symbols. Delay is the extent of shift from the first data point after which the next symbol is generated. If the time series is well sampled, a time delay of 1 is an appropriate default. Thus, in a time series, an order of 3 will group data points 1, 2 and 3 into one set which will be represented as one symbol. A delay of 1 implies that data points 2, 3 and 4 will form the next set. For a tutorial on computing the ordinal patterns from time series, please see (Unakafova & Keller, 2013).

U

Higuchi’s fractal dimension (HFD) is another measure of the complexity of a time series that captures the space filling ability of the data points and reflects the self-similarity property of a fractal time series (Higuchi, 1988). For a detailed review of HFD analysis, please see (Kesić & Spasić, 2016).

ED

M

A

N

We have previously examined the influence of Vipassana meditation practice on sleep in a series of whole night polysomnography studies (Maruthai et al., 2016; R. Nagendra et al., 2017; R. P. Nagendra, Maruthai, & Kutty, 2012; Pattanashetty et al., 2010; Sulekha, Thennarasu, Vedamurthachar, Raju, & Kutty, 2006) finding that senior meditators had enhanced slow wave sleep and REM states, increased REM activity, evidence of better sleep architecture preservation even with aging, and enhanced parasympathetic activity during sleep. Overall, these studies point out to the neuroplastic changes due to long term practice of meditation. In the present study, we address the aforementioned three considerations: context of traditional complex meditation practice, relevance of proficiency and duration of experience, and examination of linear and non-linear EEG measures.

A

CC E

PT

Vipassana meditation is an ancient Buddhist mindfulness based practice that comprises of several meditation techniques. A popular school of Vipassana meditation (as taught by S.N. Goenka in the tradition of Sayagyi U Ba Khin) uses a structured module with three components of guided meditation (Fig. 1) – focused attention on breath (Anapana), focused attention with awareness of the impermanence of sensations (Vipassana, often called ‘mindfulness’) and radiating good will to oneself and others (Metta, often called ‘loving-kindness’ meditation). At an introductory level, the basic method of Vipassana meditation is taught as a ten-day residential course during which a ‘code of discipline’ is prescribed. Attendance of several such ten-day courses is required before a novice practitioner becomes eligible for advanced training that is imparted during ‘long retreats’, which are intense meditative sessions lasting between 20 and 90 days. The long retreats provide senior practitioners with exposure to the philosophical basis of these practices. Senior practitioners can then volunteer to become full time teachers who then get frequent and detailed exposures to the theoretical underpinnings of the technique and become eligible to conduct meditation courses for others. Teachers can thus have similar years of meditation experience as senior practitioners but have higher proficiency due to their focused training and practice.

3

We recorded 128 channel EEG from three groups of Vipassana meditators (novices, senior meditators and teachers) while they engaged in their traditional structured meditation practice. We then examined power spectral changes as well as complexity based measures across the different meditation components.

SC R

IP T

2. Methods 2.1. Participants Participants were recruited with permission from Vipassana Research Institute (VRI), Igatpuri, India. Notices about the study were put up in the long course conducting centers across India. Study details were published in the VRI newsletter. Participation criteria (age range 30-65) were as follows: Novice practitioners (2 or 3 ten-day courses with less than 3 years of practice); Senior practitioners (at least one long retreat with daily practice of more than 7 years); Teachers (instructors of Vipassana courses at meditation centers with a daily practice of more than 7 years and have undergone several long retreats). Participants with neurological or psychological disorders, history of substance abuse, on psychiatric or neurological medication, or practicing any other forms of meditation other than Vipassana were excluded from the study.

CC E

PT

ED

M

A

N

U

We recruited three groups of age, gender, education and socio-economic status matched participants (n = 68) of which EEG data from one novice participant could not be analysed: Novice practitioners (Nov; n = 24; 12 females; age 48.4 ± 10.1; meditation years: 2.2 ± 0.9 with max 3 years; meditation hours: 1,080.1 ± 599.9); Senior practitioners (Sen; n = 22; 11 females; age 54.2 ± 12.6; meditation years: 13.0 ± 4.4; meditation hours: 10,364 ± 5,229); and Teachers (Tea; n = 21; 10 females; age 51.8 ± 12.2; meditation years: 16.3 ± 5.3; meditation hours: 15,349 ± 9,307). There were no significant differences in terms of age, gender, education or income (as expected) and there was a significant difference between groups in terms of years of practice (F(2,64) = 79.39; p = 0.000) and hours of meditation experience (F(2,64) = 32.93; p = 0.000). The mean difference and 95% confidence intervals (CI) for the between group differences for years of meditation experience (Sen vs Nov: 10.8 (8.0, 13.6) p = 0.000, Tea vs Nov: 14.1( 11.2, 16.9) p = 0.000 and Tea vs Sen: 3.2 (0.3, 6.1) p = 0.025) and hours of meditation experience (Sen vs Nov: -9,283.7 (5,024, 13,543) p = 0.000, Tea vs Nov: 14,269 (9,957, 18,581) p = 0.000 and Tea vs Sen: 4985.6 (583, 9,388) p = 0.023) showed that these groups were clearly different in terms of overall meditation experience. Seniors and Teachers had overlaps in terms of years of experience but Teachers had more hours of meditation experience as well as higher proficiency due to their focused training in the theory and formal roles of teaching meditation.

A

Participants had at least high school education, were fluent in English, and the majority was from the middle income category as per Indian standards. Participants were all healthy, right handed, nonsmokers and refrained from any caffeinated beverages on the day of the study. They were recruited from all over India. Food, accommodation and travel expenses were offered with no other kind of financial incentives. All participants provided written informed consent as approved by NIMHANS Institute Human Ethics Committee. 2.2. EEG acquisition All EEG recordings were carried out in the forenoon (starting 9 AM), in a sound attenuated chamber of the Human Cognitive Research laboratory in the Department of Neurophysiology, NIMHANS. The 4

ambient temperature was maintained at 25C. The participants sat comfortably in a cushioned chair with armrest and E-prime 2.0 stimulus presentation software (Psychology Software Tools, Inc., Sharpsburg, PA, USA) was used for presenting audio instructions.

IP T

EEG data were acquired using a Geodesic EEG System 300 (Electrical Geodesics, Inc., USA) with Net Amps 300 amplifier and Net Station software version 4.5.7. Appropriate sizes of HydroCel Geodesic Sensor Nets with 128 channels were used. Preparatory procedures as recommended by the vendor were carefully carried out. EEG was digitized with a resolution of 24 bits at a sampling rate of 1 KHz. No notch filters were used. Impedance for all electrodes was maintained at less than 50 KΩ as recommended by the vendor.

A

N

U

SC R

The Meditation EEG protocol (Fig. 2) lasted just over one hour and consisted of the following structure: Pre-Rest Eyes Open (RO) and Eyes Closed (RC) for one minute each, alternating twice – 4 minutes; Anapana (Ana) – 3 minutes; Vipassana (Vipa) – 40 minutes; Metta – 6 minutes; Post-Rest (RO and RC) for one minute each, alternating twice – 4 minutes. All meditative states were in eyes closed condition. At the start of each meditative condition, time locked audio instructions were played in the voice of S.N. Goenka, with permissions from VRI. The participants were explicitly instructed to be relaxed and to avoid meditating during the rest sessions. The transcript of the instructions is provided in the Supplementary Material (SM). At the end of the protocol, there was a debriefing session and all participants mentioned that they were able to comply with the instructions.

CC E

PT

ED

M

2.3. Data pre-processing and Analysis The EEG data were preprocessed with a 0.1 Hz first order high pass filter using Net Station 4.5.7 and then exported as Net Station simple binary files. Further EEG preprocessing and analysis was done with custom scripts using EEGLAB v13.4.4b (Delorme & Makeig, 2004) – an open source toolbox using MATLAB version R2013a (Math Works Inc., Natick, MA, USA). Channel locations were set as per the 129 channel file supplied by the vendor. Following a low pass FIR filter of 40 Hz, artifact correction and removal was done using the artifact subspace reconstruction (ASR) method implemented in the ‘clean_rawdata’ plugin ver 1.2 of EEGLAB. As meditation data often contains high amplitude alpha and theta bursts (see Fig. 3), a high threshold of 20 standard deviations was set for detecting artifacts in the EEG. The bad channels that were removed were interpolated, and the cleaned data down sampled to 250 Hz and re-referenced to average.

A

For analyses, the last 2-minute data from Anapana and middle 2-minute portion of Metta were taken. The 40-minute-long Vipassana data was epoched into four equal ‘10’ minute portions (Vipa1 to Vipa4) out of which the middle 2-minutes data from each epoch was taken for analyses. Data from the 2-minute eyes closed pre-rest (R1C) condition were taken as baseline for comparisons. The data was z-scored to minimize confounds due to individual variability. Power spectral density was computed and the results of the topographical distribution of delta (14Hz), low-theta (4-6Hz), high-theta (6-8Hz), low-alpha (8-10Hz), high-alpha (10-12Hz), low-beta (1215Hz), high-beta (15-30Hz) and low-gamma (30-40Hz) were used for statistical analyses. We could not evaluate high-gamma power due to the 40 Hz low-pass filter that was applied to achieve good 5

quality artifact rejection with our data. In the present paper, we focus on high-theta and low-alpha power bands. For the complexity measures, each epoched dataset (120 seconds duration) was divided into sub epochs of 0.5 seconds yielding 240 bins per state for each participant. PE and HFD were calculated for each bin and averaged across 240 bins for each of the 129 electrodes for each subject. For PE, the order was set as 3 and the delay as 1. For HFD, the scale value of 5 was chosen. Custom scripts were written in MATLAB for the non-linear analyses and for generating scalp topography plots.

IP T

Surrogate data based statistical tests were carried out with non-parametric permutation based twoway ANOVA using 2000 random partitions and False Discovery Rate (FDR) correction for multiple comparisons. Statistical significance level was set at p < 0.05.

SC R

The median values of all electrodes for each power band for each state were used for creating the median plots and for correlating with meditation experience (in hours), HFD and PE.

A

N

U

3. Results We present the salient results in this section. Since we had four time points in the Vipassana condition, we use Vipa2 as a representative condition in this section and present the results of all time points in the Vipassana condition in the SM figures. The summary of all the power spectral changes (including post hoc analyses) can be found in SM Table 1.

M

3.1. State changes due to meditation

PT

ED

As compared to eyes-closed pre-rest (R1C), each group (Nov, Sen and Tea) showed global increases in each power band (delta, theta, alpha and low-beta) during each meditation condition (Ana, Vipa and Metta). These changes demonstrate that the meditation protocol elicited robust state (statetrait in the case of long term meditators) changes in each group. The post-rest condition (R2C) also showed the after-effect of the hour long meditation session with global power increases, even though the participants had explicit instructions not to meditate during the rest condition. Fig. 3 shows representative Raw EEG traces of Nov and Tea.

A

CC E

Fig.4 (and SM Fig. 1) shows median power spectral values for the three groups as they went through the time course of the meditation protocol. All the groups showed higher power across the different bands during Post-Rest than in Pre-Rest but these values were lower than the preceding meditative conditions. SM Fig. 2 and SM Fig. 3 show the comparison between Pre-Rest and Post-Rest for both eyes open and eyes closed states. 3.2. Trait changes due to long term meditation experience Comparison of pre-rest (R1C) between the groups revealed trait differences between the groups in the delta, high-theta, low-alpha and low-gamma bands (Fig. 5, first row in every sub-panel). Post hoc analyses revealed that there were no differences between Sen and Tea but they both had higher low-alpha and low-gamma power as compared to Nov (SM Table 1). Thus, the differences in resting state power at baseline are attributable to the effects of long term meditation experience.

6

3.3. State-Trait changes due to long term meditation experience State-trait interactions (trait differences attributable to long term practice as well as state changes during the meditation session) were visible in low-alpha band for Ana, Vipa2, Metta and R2C conditions (Fig. 5, 2nd row in each plot). The power spectra for the Vipa condition (SM Fig. 4) showed differing state-trait changes across the four time points. SM Fig. 5 shows reduction in power in R2C as compared to the meditative states as well as state-trait effects leading to condition differences even in the R2C condition.

M

A

N

U

SC R

IP T

Post hoc between-group analyses revealed that there were no differences between Sen and Tea in any power band for any of the meditation conditions and that Nov group had lower low-alpha power and higher low-gamma power as compared to both Sen and Tea in all the conditions. It is important to note that the higher low-gamma power in Nov had a bitemporal distribution in all the conditions, suggestive of possible high frequency artifacts from the temporalis muscle (Goncharova, McFarland, Vaughan, & Wolpaw, 2003). Since the Nov group had power increases during meditation, these analyses highlight that long term practice enabled further low-alpha power increases and lesser beta and gamma power increases during meditation. It is noteworthy that there were widespread hightheta differences between Nov and Tea in all the conditions, but not between Nov and Sen or Sen and Tea. Similarly Nov had higher low-gamma as compared to Tea in the R1O and R2O whereas there were no differences in low-gamma between Nov and Sen or Sen and Tea. These results indicate that the Sen group was intermediate group between Nov and Tea in terms of their EEG profile of meditation proficiency and/or experience.

3.4. Higuchi fractal dimension changes during meditation

ED

Analysis of HFD data showed that Tea and Nov had increased HFD complexity (as compared to R1C) during Vipassana meditation as compared to rest (Fig. 6A). Sen did not show any HFD differences in any meditation state.

CC E

PT

There were between-group differences for all the meditation conditions and post hoc analyses revealed that there were no differences between Sen and Tea in terms of HFD while Nov had significantly higher HFD values than both Sen and Tea for each state. This result suggests a nonlinear ‘U’ shaped relationship in terms of complexity increases during meditation with meditation proficiency as Sen was an intermediate group with no significant differences in complexity.

A

3.5. Permutation entropy changes during meditation Interestingly, Tea showed increased PE complexity (as compared to R1C) during the various meditation conditions (Ana, Vipa and Metta) as compared to rest but there were no differences between R1C and R2C (Fig. 6B and Fig. 7). Nov showed increased PE complexity from baseline only during Vipassana meditation. Sen did not show any complexity differences in any meditation condition.

7

There were between-group differences for all the meditation conditions. Post hoc analyses revealed that there were no differences between Sen and Tea in terms of PE in any condition while Nov had significantly larger PE values than both Sen and Tea in every condition. The group differences are more apparent when examining the median PE and HFD values across all conditions (Fig. 8) which also shows the increased processing by all groups during Metta. 3.6. Correlations between power spectra, meditation experience and complexity values

SC R

IP T

Median power-spectra and hours of meditation experience were positively correlated in the thetaalpha band (especially pre-rest and post-rest conditions) and negatively correlated with delta, beta and gamma bands. On the other hand, power spectra and both complexity measures were negatively correlated in the theta-alpha band and positively correlated with delta, beta and gamma bands. However, the number of significant correlations was more for the complexity measures that for the meditation experience. Fig.9 summarizes all the significant correlations. Discussion

PT

ED

M

A

N

U

Our study results reveal several distinctive state-trait changes among Vipassana mindfulness meditators. Long term practice (as in the case of Sen and Tea) yields trait differences at baseline rest conditions seen via both linear and non-linear methods. Complexity based measures reveal a further nuance that there are proficiency based trait-state differences as Teachers showed heightened information processing and fractal characteristics during meditative states (as compared to rest) whereas Senior meditators did not show these effects. The value of studying the Vipassana mindfulness based meditation as an entire module was revealed as (compared to rest) Anapana and Vipassana were marked by state-trait differences in delta, low-alpha, high-beta and low-gamma power but no group differences in high-theta power. Metta, on the other hand, was marked by differences in high-theta, low-alpha and low-gamma power but showed no group differences in high-beta power. Metta also required the highest amount of information processing and complexity among all the states. We discuss the possible implications of these changes in this section, but before that, we examine our results in the light of the three considerations we had for undertaking this study.

A

CC E

Our first consideration was to examine the context in which a meditation study is carried out. Vipassana (as taught by S.N. Goenka) has a structure in which multiple meditation techniques are practiced in a specific progressive sequence – starting with concentration on breath, moving on to mindful attention to sensations (potentially yielding insight into impermanence of things) and ending with a loving kindness meditation in which goodwill is generated towards the self and others. This structure is designed for specific outcomes (see Fig. 1) at each step. It is a useful exercise to study a particular meditation technique in isolation as is common practice. Several authors have emphasized the need to place these meditation studies in the traditional context (Dahl et al., 2015; Josipovic & Baars, 2015). We have previously examined the ability of meditators to rapidly shift between rest and meditation states (Nair et al., 2017) based on the traditional practice of Rajayoga meditators of performing one minute meditations. In this context a couple of new studies have attempted to examine the cumulative effects of doing multiple meditative techniques in a designed sequence (DeLosAngeles et al., 2016; Schoenberg, Ruf, Churchill, Brown, & Brewer, 2018) as- studying different techniques in isolation does not provide this value. Our approach of following an exact traditional sequence in the protocol contributes in this direction and is ecologically a more valid 8

study than examining any meditation component in isolation. On the other hand, there have been attempts to classify meditation groups into higher abstractions – such as considering all Buddhismrelated groups together as distinct from Hinduism-related groups (Tomasino, Chiesa, & Fabbro, 2014) or indeed all meditation types together (Hinterberger, Schmidt, Kamei, & Walach, 2014). These approaches are valuable as they provide insight into what is common between the varieties of techniques in terms of EEG profiles or outcomes (Cahn & Polich, 2006; Sedlmeier et al., 2012) but they do not provide insight into how any particular meditation approach is effective in its traditional context. Our study provides a novel contribution in this latter direction.

N

U

SC R

IP T

Our second consideration was to examine trait influences on state, in particular due to amount of meditation experience as well as their proficiency profiles. In our study, we used only experienced meditators. Even the Nov group had considerable experience and indeed other studies have classified people with about 1000 hours of meditation as highly experienced – see for example: (Hinterberger et al., 2014). The Sen and Tea groups had over 10 years of regular meditation practice and more than 10,000 hours of meditation experience. These groups would have typically been clubbed as one expert meditation group. However, from the traditional Vipassana viewpoint, these groups are different as the Tea group has exposure to more theory and focused practice as well as a role specific effect as they are formally involved in teaching meditation. We therefore examined these groups separately and found them to have similarities and differences. We suggest that future studies could consider examining proficiency differences in addition to duration of practice while determining groups of expert meditators.

CC E

PT

ED

M

A

Our third consideration was the use of both linear and non-linear approaches to study EEG profiles. Linear methods (power spectra) have been the dominant mode of EEG analysis and have yielded a lot of insight into similarities and differences between meditation techniques (Cahn & Polich, 2006; Lomas et al., 2015). In our study too, power spectral changes could help distinguish between different meditation states as well as show some differences between groups. However, non-linear methods provide additional insight by examining different aspects such as complex information processing or fractal geometry or dimensional complexity. These methods have found applications in widely varying domains such as patterns in seismic activity, stock market fluctuations, heart rate changes and EEG analysis (Aftanas & Golocheikine, 2002, 1998; Gao et al., 2016; Lutzenberger, Elbert, Birbaumer, Ray, & Schupp, 1992). In our study, we found that employing two different nonlinear methods (fractal dimensions and permutation entropy) provided different insights and so we suggest that future EEG meditation studies could also benefit from using complexity measures. We now discuss the EEG profiles of the three groups under study.

A

Firstly, all the groups were able to make shift from rest to a meditation state as evidenced by global power increases in each meditation condition as compared to rest and these increases were different for the various conditions and the three groups. Several meditation studies have found changes in these different power bands (see (Lomas, Ivtzan, & Fu, 2015) for a review) but most studies have focused on one or two bands. The most common reports have been for changes in theta-alpha and gamma bands and our study provides further evidence in that direction. The robust global enhancements across all these power bands show that these meditation techniques induce distinct states of consciousness that are vividly different from different stages of sleep, rest and relaxed wakefulness or active task performance. Further, these states don’t show a monotonic increase in power with passage of time. Instead, there are differences in how the power changes 9

across the various conditions and for the different groups showing an interaction between technique and proficiency levels. Additionally, the power changes were also observed in Nov group (although not at the same level as the long-term groups), indicating that this group was also experienced enough in each meditation technique. While there is a strong likelihood of spillover effects of each meditation condition on the next, there were different spectral and complexity profiles in the different meditation conditions that support the notion that these techniques have clear distinctions in terms of neural processing.

ED

M

A

N

U

SC R

IP T

Tea and Sen (vs Nov) showed enhanced low-alpha power in all meditative states (Ana, Vipa and Metta). Cortical alpha activity is inhibitory and routes information by functionally blocking off the task-irrelevant pathways (Brefczynski-lewis, Lutz, Schaefer, Levinson, & Davidson, 2007; Jensen & Mazaheri, 2010; Klimesch, Sauseng, & Hanslmayr, 2007). Thus, enhanced low-alpha power during meditation supports the state of sustained attention on the selected object by blocking off irrelevant information. Ana and Vipa are cognitively intense meditative states (Gross & Thompson, 2007) as they have a specific attentional component whereas Metta has both a cognitive and affective component as it is emotionally intense (Goenka, 1987) and the enhanced high-theta changes only in this condition supports its affective role (Aftanas & Golocheikine, 2001). While it is possible that the differences in theta power could be attributed to differences in duration of the practice (Ana was 3 min, Vipa was 40 min and Metta was 6 min) or that it is because Metta was towards the end of the practice, it is unlikely to be so. As is evident from Fig.4, Tea had a very different profile of power spectral changes in high-theta and low-alpha power as compared to the other two groups suggesting proficiency linked differences rather than differences in duration of the technique being practiced. Nov had enhanced delta and low-gamma power as compared to the long term groups. It has been suggested that increased delta is associated with enhanced internal processing (Lomas et al., 2015) and increased low-gamma with enhanced active state (Cahn, Delorme, Polich, Diego, & Jolla, 2013). It is thus plausible that Nov had a more internally engaged effortful state of consciousness as compared to the long term groups.

A

CC E

PT

The enhanced theta-alpha power and reduced low-gamma power at rest in long term meditators indicate the trait differences that are likely due to neuroplastic changes although as mentioned in the limitations below, we cannot rule out the possibility that the differences in low-gamma power could be due to muscle artifacts. Several studies have documented a variety of structural and functional changes in the brain following long term practice of mindfulness meditation (Tang, Holzel, & Posner, 2015). It has been suggested that Nov might employ ‘top-down’ emotion regulation strategies whereas experts might use a ‘bottom-up’ strategy as a consequence of long-term practice (Chiesa, Serretti, & Christian, 2013). Since normal controls have self-referential processing by default (Gusnard, Akbudak, Shulman, & Raichle, 2001), these trait changes could suggest decreased self-referential processing and enhanced objective stance towards oneself and others. It has been suggested that anterior theta and alpha in Sahaja yoga meditators could reflect positive emotions and internalized attention (Aftanas & Golocheikine, 2001). In the present study, the long-term groups showed enhanced central and posterior theta-alpha power across the three different meditation techniques (Ana, Vipa and Metta). While Sahaja yoga has strong focus on internalized attention and bliss, the three techniques in the Vipassana meditation module in our study (Goenka tradition) focus on non-judgmental observation or awareness while paying attention (as a starting point) on breath (Ana), being mindful of bodily sensations (Vipa) or radiating goodwill (Metta). It is noteworthy that there are other traditions of Vipassana meditation where the focus is open 10

awareness without focus on any specific object. Thus, the differing results between Aftanas et al and our study are likely to be due to differences in the meditation technique being studied. While positive results from studies on Vipassana meditators have usually been attributed to mindfulness alone, it is important to reiterate that the Vipassana meditative program has several distinct components including attention, mindfulness and loving-kindness (Ivanovski & Malhi, 2007).

SC R

IP T

We used two different complexity measures in our study. Neuronal assemblies inside the brain are coupled at varying levels and they oscillate at varying frequencies based on the task and nature of functional coupling. The inherent nature and interactions of the dynamical system is the source of all the complex patterns and behaviors that emerge from it. Multiple systems doing totally different information processing can produce similar patterns and same system can generate a plethora of patterns by a small change in the initial conditions (Wolfram, 1983). Each nonlinear parameter reveals one aspect of the dynamical system. Permutation entropy represents the Shannon information of the distribution of order patterns. It gives a measure of how ordered and predictable the system is from the time series data. Fractal dimensions represent the geometrical pattern in a time series at multiple scales.

A

CC E

PT

ED

M

A

N

U

Both the complexity measures clearly discriminated between novice and long term meditators. Nov had significantly higher information processing as reflected by PE measures. Novice meditators are in a more effortful stage as they make efforts to reach and maintain the meditative states while the long term groups are in a relatively effortless state of meditation due to the trait benefits of long term practice (Tang, Rothbart, & Posner, 2012). While both Sen and Tea were similar and clearly proficient in practice (there were no significant differences between these groups), only Tea showed complexity increases across the meditation states while Sen did not. The correlational summary (Fig. 9) indicated that there was an inverse relationship between duration of meditation experience and complexity changes. Specifically, high complexity was positively correlated with delta, beta and lowgamma and negatively correlated with theta-alpha and conversely, long term experience was negatively correlated with delta, beta and low-gamma and positively correlated with theta-alpha. However, there were some power spectral differences (SM Table 1) that were only between Nov and Tea (such as high-theta during Metta, high-beta during R1C, low-gamma during R1O and R2O) and some other differences that were only between Nov and Sen (such as high-beta during Ana) that show that there were minor differences between Sen and Tea that did not reach statistical significance. Qualitatively too (Fig. 4) Sen and Tea showed some differences. Overall, these suggest that both the long term groups had considerably experience and thus were similar in their overall profile, but there were differences that can be attributed to the conscious attention to detail and accuracy in practice due to Tea’s role in teaching meditation. On the other hand, Nov and Tea were very different from each other but showed similar increased information processing (unlike Sen) during the various meditation conditions. This ‘U’ shaped relationship between information processing and proficiency levels is not unprecedented as it was found that there was an inverted ‘U’ shaped relationship between regional brain activation during sustained attention and meditation experience (Brefczynski-lewis et al., 2007). We found that all the groups had highest global average HFD and PE values during Metta which relates with others and has an affective component that distinguishes it from Ana and Vipa conditions that have a more internally focused attention component. Remarkably, the complexity measures between post-rest and pre-rest were not different for any group even though the power spectra showed a huge influence of the intervening hour long meditation state. This demonstrates that the meditators were actually at rest and not 11

meditating (as instructed) during post-rest but that the functional changes due to the earlier meditative state continued to exert an influence on the overall brain state.

A

CC E

PT

ED

M

A

N

U

SC R

IP T

While the study makes several contributions to the literature as discussed above, there are several limitations that need to be taken into account. Firstly, we distinguished the three groups in terms of duration and proficiency based on their years of experience (Nov vs others) and formal roles (Tea vs Sen). It would be valuable to formally examine proficiency using a questionnaire that is designed to consider the traditional theoretical framework of these practitioners. Such studies are lacking in the literature. While the EEG profiles in our study supported the categorization into three groups, having a formal proficiency score would help correlate these values. Secondly, we used scalp topographical differences with 129 electrodes while examining power spectral changes. In our view, this is better than averaging a few regional electrodes and suggesting frontal or parietal activation etc. However, it would be useful to carry out source localization analysis (Canuet et al., 2011) and examine activation in different brain regions during the various conditions. Such an approach can also be achieved using fMRI studies (Fox et al., 2016) but EEG approaches are less intrusive and better suited for mimicking a traditional meditative practice. Thirdly, the gamma band has prominently featured in many meditation studies (Cahn, Delorme, & Polich, 2010; Fell, Axmacher, & Haupt, 2010), but we had noise limitations due to which we had to filter out this band to a large extent. We could only focus on the low-gamma band in this study and which yielded some valuable information but we cannot rule out confounds due to muscle artifacts especially for Nov. It might be possible to use different artifact removal approaches that still allow examination of changes across the full extent of the gamma band. Fourthly, since our study employed a fixed order and differing durations of meditation techniques to closely follow the traditional practice, order effects are to be expected. This does not allow us to interpret our findings related to Vipassana and Metta portions as being specific to these states. Also, the rest eyes open and closed states were of one minute duration each and were combined to get two minute epochs that were used for comparison with two minute epochs during the various meditative states. A possible confound is that EEG power in the low frequency bands increase with eyes closed state. This however, did not seem to be the case in our study as we observed power increases from pre-rest state to the meditative states, differing power changes across the three meditative states and finally power decreases (as compared to meditative states) in the post-rest state (Fig. 4 and SM Fig. 5). Future studies could consider having the participants to undergo a control protocol (on a separate day) of the same overall duration but with instructions for suitable thought engagement. Fifthly, we used global averages of the two different complexity measures when we provided results of the median differences for HFD and PE between the groups. The scalp topographies of the complexity measures indicate regional differences which would have been lost when using global averages. Finally, since this was a cross-sectional study, we cannot attribute all the observed group differences to the duration and proficiency of meditation practice or even rule out the possibility that there may be other pre-disposing factors that lead practitioners to engage in intensive meditation practice and thereby become more proficient. Nevertheless, these limitations do not detract much from our overall finding that it is possible to use EEG profiles using linear and non-linear methods to dissociate between extent of meditation experience and proficiency of practice. 4. Conclusion

12

Our study demonstrates the value of studying meditation within the traditional context of the practitioners. The study shows that both duration of practice and role based proficiency of practice influence the brain states that can be seen in the EEG profiles. Finally, our study demonstrates the value of using both linear and non-linear methods that can complement and supplement the results to provide a fuller understanding of brain changes during meditation. We suggest that future EEG studies on meditators could incorporate these considerations to examine the brain mechanisms underlying these practices in the quest for using such practices for therapeutic or performance enhancement benefits.

IP T

Acknowledgements and Funding:

A

CC E

PT

ED

M

A

N

U

SC R

This study has been funded by Cognitive Science Research Initiative, (DST-CSI) Department of Science & Technology, Government of India, New Delhi (Grant: SR/CSI/63/2011 to B.M.K). We are grateful to VRI (Vipassana Research Institute, Global Pagoda, Mumbai, India) for giving us permission and logistical support for recruitment of meditators; and NIMHANS for providing facilities support. We thank the two anonymous reviewers for their valuable inputs. Finally, we thank the Vipassana meditators who were generous with their time and sincere participation.

13

References Aftanas, L. I., & Golocheikine, S. a. (2002). Non-linear dynamic complexity of the human EEG during meditation. Neuroscience Letters, 330(2), 143–6. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/12231432 Aftanas, L. I., & Golocheikine, S. A. (1998). Non-linear dynamic complexity of the human EEG during meditation. Neuroscience Letters, 330(2), 143–146. http://doi.org/10.1016/S03043940(02)00745-0

IP T

Aftanas, L. I., & Golocheikine, S. A. (2001). Human anterior and frontal midline theta and lower alpha reflect emotionally positive state and internalized attention : high-resolution EEG investigation of meditation. Neuroscience Letters, 310, 57–60.

SC R

Awasthi, B. (2013). Issues and perspectives in meditation research: In search for a definition. Frontiers in Psychology, 3(JAN), 1–9. http://doi.org/10.3389/fpsyg.2012.00613

Bandt, C., & Pompe, B. (2002). Permutation Entropy: A Natural Complexity Measure for Time Series. Physical Review Letters, 88(17), 174102. http://doi.org/10.1103/PhysRevLett.88.174102

U

Boccia, M., Piccardi, L., & Guariglia, P. (2015). The Meditative Mind: A Comprehensive Meta-Analysis of MRI Studies. BioMed Research International, 2015, 419808. http://doi.org/10.1155/2015/419808

A

N

Brefczynski-lewis, J. A., Lutz, A., Schaefer, H. S., Levinson, D. B., & Davidson, R. J. (2007). Neural correlates of attentional expertise in long-term meditation practitioners. PNAS, 104(27), 11483–11488.

M

Cahn, B. R., Delorme, A., & Polich, J. (2010). Occipital gamma activation during Vipassana meditation. Cognitive Processing, 11(1), 39–56. http://doi.org/10.1007/s10339-009-0352-1

ED

Cahn, B. R., & Polich, J. (2006). Meditation states and traits: EEG, ERP, and neuroimaging studies. Psychological Bulletin, 132(2), 180–211. http://doi.org/10.1037/0033-2909.132.2.180

PT

Canuet, L., Ishii, R., Pascual-Marqui, R. D., Iwase, M., Kurimoto, R., Aoki, Y., … Takeda, M. (2011). Resting-state EEG source localization and functional connectivity in schizophrenia-like psychosis of epilepsy. PloS One, 6(11), e27863. http://doi.org/10.1371/journal.pone.0027863

CC E

Chiesa, A., Serretti, A., & Christian, J. (2013). Mindfulness : Top – down or bottom – up emotion regulation strategy ? Clinical Psychology Review, 33(1), 82–96. http://doi.org/10.1016/j.cpr.2012.10.006

A

Dahl, C. J., Lutz, A., & Davidson, R. J. (2015). Reconstructing and deconstructing the self: cognitive mechanisms in meditation practice. Trends in Cognitive Sciences, 1–9. http://doi.org/10.1016/j.tics.2015.07.001 Davidson, R. J., & Kaszniak, A. W. (2015). Conceptual and Methodological Issues in Research on Mindfulness and Meditation. American Psychologist, 70(7), 581–592. http://doi.org/10.1037/a0039512 Delorme, A., & Makeig, S. (2004). EEGLAB: an open source toolbox for analysis of single-trial EEG dynamics including independent component analysis. Journal of Neuroscience Methods, 134(1), 9–21. http://doi.org/10.1016/j.jneumeth.2003.10.009 DeLosAngeles, D., Williams, G., Burston, J., Fitzgibbon, S. P., Lewis, T. W., Grummett, T. S., … 14

Willoughby, J. O. (2016). Electroencephalographic correlates of states of concentrative meditation. International Journal of Psychophysiology, 110, 27–39. http://doi.org/10.1016/j.ijpsycho.2016.09.020 Fell, J., Axmacher, N., & Haupt, S. (2010). From alpha to gamma: electrophysiological correlates of meditation-related states of consciousness. Medical Hypotheses, 75(2), 218–24. http://doi.org/10.1016/j.mehy.2010.02.025

IP T

Fox, K. C. R., Dixon, M. L., Nijeboer, S., Girn, M., Floman, J. L., Lifshitz, M., … Christoff, K. (2016). Functional neuroanatomy of meditation: A review and meta-analysis of 78 functional neuroimaging investigations. Neuroscience & Biobehavioral Reviews, 65, 208–228. http://doi.org/10.1016/j.neubiorev.2016.03.021

SC R

Gao, J., Fan, J., Wu, B., Zhang, Z., Chang, C., Hung, Y., … Sik, H. (2016). Entrainment of chaotic activities in brain and heart during MBSR meditation. Neuroscience Letters. http://doi.org/10.1016/j.neulet.2016.01.001 Goenka, S. N. (1987). The Discourse Summaries. Dhammagiri, Igatpuri (Nasik): Vipassana Research Institute.

U

Goncharova, I. I., McFarland, D. J., Vaughan, T. M., & Wolpaw, J. R. (2003). EMG contamination of EEG: Spectral and topographical characteristics. Clinical Neurophysiology, 114(9), 1580–1593. http://doi.org/10.1016/S1388-2457(03)00093-2

A

N

Gross, J. J., & Thompson, R. A. (2007). Emotion regulation: Conceptual foundations. In Handbook of emotion regulation (pp. 3–24). Guilford Press. http://doi.org/http://doi.org/10.1080/00140130600971135

ED

M

Gusnard, D. A., Akbudak, E., Shulman, G. L., & Raichle, M. E. (2001). Medial prefrontal cortex and self-referential mental activity: Relation to a default mode of brain function. Proceedings of the National Academy of Sciences, 98(7), 4259–4264. http://doi.org/10.1073/pnas.071043098 Higuchi, T. (1988). Approach to an irregular time series on the basis of the fractal theory. Physica D, 31, 277–283.

PT

Hinterberger, T., Schmidt, S., Kamei, T., & Walach, H. (2014). Decreased electrophysiological activity represents the conscious state of emptiness in meditation. Frontiers in Psychology, 5(FEB), 99. http://doi.org/10.3389/fpsyg.2014.00099

CC E

Ivanovski, B., & Malhi, G. S. (2007). The psychological and neurophysiological concomitants of mindfulness forms of meditation. Acta Neuropsychiatrica, 19(2), 76–91. http://doi.org/10.1111/j.1601-5215.2007.00175.x

A

Jensen, O., & Mazaheri, A. (2010). Shaping functional architecture by oscillatory alpha activity: gating by inhibition. Frontiers in Human Neuroscience, 4(November), 186. http://doi.org/10.3389/fnhum.2010.00186 Josipovic, Z., & Baars, B. J. (2015). Editorial: What can Neuroscience Learn from Contemplative Practices? Frontiers in Psychology, 6(NOV), 1–3. http://doi.org/10.3389/fpsyg.2015.01731 Kesić, S., & Spasić, S. Z. (2016). Application of Higuchi’s fractal dimension from basic to clinical neurophysiology: A review. Computer Methods and Programs in Biomedicine, 133, 55–70. http://doi.org/10.1016/j.cmpb.2016.05.014 Klimesch, W., Sauseng, P., & Hanslmayr, S. (2007). EEG alpha oscillations: the inhibition-timing 15

hypothesis. Brain Research http://doi.org/10.1016/j.brainresrev.2006.06.003

Reviews,

53(1),

63–88.

Lomas, T., Ivtzan, I., & Fu, C. H. Y. (2015). A systematic review of the neurophysiology of mindfulness on EEG oscillations. Neuroscience & Biobehavioral Reviews. http://doi.org/10.1016/j.neubiorev.2015.09.018 Lutzenberger, W., Elbert, T., Birbaumer, N., Ray, W., & Schupp, H. (1992). The Scalp Distribution of the Fractal Dimension of the EEG and Its Variation with Mental Tasks. Brain Topography, 5, 27– 34.

IP T

Manuello, J., Vercelli, U., Nani, A., Costa, T., & Cauda, F. (2016). Mindfulness meditation and consciousness: An integrative neuroscientific perspective. Consciousness and Cognition, 40, 67– 78. http://doi.org/10.1016/j.concog.2015.12.005

SC R

Maruthai, N., Nagendra, R. P., Sasidharan, A., Srikumar, S., Datta, K., Uchida, S., & Kutty, B. M. (2016). Senior Vipassana Meditation practitioners exhibit distinct REM sleep organization from that of novice meditators and healthy controls. International Review of Psychiatry, 28(April). http://doi.org/10.3109/09540261.2016.1159949

U

Nagendra, R. P., Maruthai, N., & Kutty, B. M. (2012). Meditation and its regulatory role on sleep. Frontiers in Neurology, APR(April), 54. http://doi.org/10.3389/fneur.2012.00054

A

N

Nagendra, R., Sulekha, S., Sasidharan, A., Sathyaprabha, T., Pradhan, N., Raju, T. R., & Kutty, B. M. (2017). Vipassana Meditation Practices Enhance the Parasympathetic Activity during Sleep : a Case-Control Study of Heart Rate Variability across Sleep Cycles. International Journal of Complementary & Alternative Medicine, 5(2). http://doi.org/10.15406/ijcam.2017.05.00145

ED

M

Nair, A. K., Sasidharan, A., John, J. P., Mehrotra, S., & Kutty, B. M. (2017). Just a minute meditation: Rapid voluntary conscious state shifts in long term meditators. Consciousness and Cognition, 53(June), 176–184. http://doi.org/10.1016/j.concog.2017.06.002

PT

Pattanashetty, R., Sathiamma, S., Talakkad, S., Nityananda, P., Trichur, R., & Kutty, B. M. (2010). Practitioners of vipassana meditation exhibit enhanced slow wave sleep and REM sleep states across different age groups. Sleep and Biological Rhythms, 8(1), 34–41. http://doi.org/10.1111/j.1479-8425.2009.00416.x

CC E

Riedl, M., Muller, A., & Wessel, N. (2013). Practical considerations of permutation entropy: A tutorial review. European Physical Journal: Special Topics, 222(2), 249–262. http://doi.org/10.1140/epjst/e2013-01862-7

A

Schoenberg, P. L. A., Ruf, A., Churchill, J., Brown, D. P., & Brewer, J. A. (2018). Mapping complex mind states: EEG neural substrates of meditative unified compassionate awareness. Consciousness and Cognition, 57(June 2017), 41–53. http://doi.org/10.1016/j.concog.2017.11.003 Sedlmeier, P., Eberth, J., Schwarz, M., Zimmermann, D., Haarig, F., Jaeger, S., & Kunze, S. (2012). The psychological effects of meditation: a meta-analysis. Psychological Bulletin, 138(6), 1139–71. http://doi.org/10.1037/a0028168 Stam, C. J. (2005). Nonlinear dynamical analysis of EEG and MEG : Review of an emerging field. Clinical Neurophysiology, 116, 2266–2301. http://doi.org/10.1016/j.clinph.2005.06.011 Sulekha, S., Thennarasu, K., Vedamurthachar, A., Raju, T. R., & Kutty, B. M. (2006). Evaluation of sleep architecture in practitioners of Sudarshan Kriya yoga and Vipassana meditation. Sleep 16

and Biological Rhythms, 4(3), 207–214. http://doi.org/10.1111/j.1479-8425.2006.00233.x Tang, Y.-Y., Holzel, B., & Posner, M. I. (2015). The neuroscience of mindfulness. Nat Rev Neurosci, 16(4), 213–225. http://doi.org/10.1038/nrn3916 Tang, Y.-Y., Rothbart, M. K., & Posner, M. I. (2012). Neural correlates of establishing, maintaining, and switching brain states. Trends in Cognitive Sciences, 16(6), 330–337. http://doi.org/10.1016/j.tics.2012.05.001.Neural

IP T

Tomasino, B., Chiesa, A., & Fabbro, F. (2014). Disentangling the neural mechanisms involved in Hinduism-and Buddhism-related meditations. Brain and Cognition, 90, 32–40. Unakafova, V., & Keller, K. (2013). Efficiently Measuring Complexity on the Basis of Real-World Data. Entropy, 15(10), 4392–4415. http://doi.org/10.3390/e15104392

A

CC E

PT

ED

M

A

N

U

SC R

Wolfram, S. (1983). Statistical mechanics of cellular automata. Rev. Mod. Phys., 55(3), 602–644. http://doi.org/10.1016/j.cell.2013.07.024

17

IP T

A

N

U

SC R

Fig. 1: Outline of the traditional Vipassana meditation module. This is the sequence of meditative practices followed in the tradition of Sayagyi U Ba Khin. The salient features of each practice and the expected outcomes are represented. The notation used in this paper is: Anapana, Vipassana and Metta.

A

CC E

PT

ED

M

Fig. 2: Meditation EEG Protocol. RO = Rest Eyes Open, RC = Rest Eyes Closed. The Pre-Rest and Post-Rest conditions lasted four minutes each. Arrows indicate the time points when audio instructions were provided for the next condition. The durations indicated for each condition do not include the time for instructions. Anapana had instructions for focused attention on breath. Vipassana had instructions for being mindfully aware of bodily sensations and Metta had instructions for radiating goodwill to be in a state of loving-kindness.

Fig. 3: Representative EEG traces of Novice (A) and Teachers (B) during meditation.

18

IP T SC R

A

CC E

PT

ED

M

A

N

U

Fig. 4. Median power spectral changes across the meditation protocol. R1C = Pre-Rest with eyes closed. Ana = Anapana meditation. Vipa2 = Vipassana meditation 2nd time point. R2C = Post-Rest with eyes closed. Nov = Novice meditators. Sen = Senior meditators. Tea = Teachers of meditation.

19

IP T SC R U N A M ED PT CC E A Fig. 5: EEG Topography Comparisons: Pre-Rest vs Anapana, Vipa2 and Metta. Average topography plot using z-score normalized values for each subject. Row and column labelled ‘Diff’ show statistically significant differences across conditions - red dots indicate electrode locations (p < 0.05 using permutation based two-way ANOVA with 2000 random partitions, FDR corrected). R1C = Pre-Rest with eyes closed. Ana = Anapana meditation. Vipa2 = Vipassana meditation 2nd time point. Nov = Novice meditators. Sen = Senior meditators. Tea = Teachers of meditation. The various panels show the plots for corresponding frequency bands (1-4Hz, 6-8Hz, 8-10Hz and 30-40Hz).

20

IP T SC R

A

CC E

PT

ED

M

A

N

U

Fig. 6: Changes in Higuchi Fractal Dimensions and Permutation Entropy during Vipassana meditation. Average topography plot using z-score normalized values for each subject. Row and column labelled ‘Diff’ show statistically significant differences across conditions - red color indicates areas with significant changes (p < 0.05 using permutation based two way ANOVA with 2000 random partitions, FDR corrected). R1C = Pre-Rest with eyes closed. Vipa = Vipassana meditation. Nov = Novice meditators. Sen = Senior meditators. Tea = Teachers of meditation. Panel A: Changes in Higuchi Fractal Dimensions (HFD). Panel B: Changes in Permutation Entropy (PE).

21

M

A

N

U

SC R

IP T

Fig. 7: Changes in Permutation Entropy during Anapana and Metta meditation. Average topography plot using z-score normalized values for each subject. Row and column labelled ‘Diff’ show statistically significant differences across conditions - red color indicates areas with significant changes (p < 0.05 using permutation based two way ANOVA with 2000 random partitions, FDR corrected). R1C = Pre-Rest with eyes closed. Ana = Anapana meditation. PE: Permutation Entropy. Nov = Novice meditators. Sen = Senior meditators. Tea = Teachers of meditation. Panel A: Changes in PE during Anapana meditation. Panel B: Changes in PE during Metta meditation.

CC E

PT

ED

Fig. 8: HFD and PE values for all rest and meditation states. Median HFD and PE values for each group. R1C = Pre-Rest with eyes closed. Ana = Anapana. Vipa = Vipassana. R2C = Post-Rest with eyes closed. HFD: Higuchi Fractal Dimensions. PE: Permutation Entropy. Nov = Novice meditators. Sen = Senior meditators. Tea = Teachers of meditation. Panel A: Median HFD values across rest and meditation states. Panel B: Median PE values across rest and meditation states.

A

Fig. 9: Correlation values for all rest and meditation conditions with meditation experience and complexity measures. Median power spectral values across all electrodes for each power band and each condition were used for calculating the correlations. Significant negative correlations are shown in blue and to the left of the dashed lines whereas positive correlations are shown in red to the right of the dashed lines. R1C = Pre-Rest with eyes closed. Ana = Anapana. Vipa1 to Vipa4 = Vipassana for the 4 time points. R2C = Post-Rest with eyes closed. The Greek alphabets from left to right represent the frequency bands delta (1-4 Hz), low-theta (4-6 Hz), high-theta (6-8Hz), low-alpha (8-10Hz), high-alpha (10-12Hz), lowbeta(12-15Hz), high-beta(15-30Hz) and low-gamma(30-40Hz).

22