Taiwanese Married Women's Lived Experience of Zen Meditation A DISSERTATION SUBMITTED
Mindfulness meditation experience is associated with ... · 04.02.2020 · 21 1 Introduction 22...
Transcript of Mindfulness meditation experience is associated with ... · 04.02.2020 · 21 1 Introduction 22...
Mindfulness meditation experience is associated with increased
ability to monitor covert somatosensory attention
Stephen Whitmarsh1, Ole Jensen2, and Henk Barendregt1
1Foundations of mathematics and computer science, Faculty of Science, Radboud
UNiversity, Nijmegen, The Netherlands
2Department of Psychology, University of Birmingham, Edgbaston, Birmingham, B15 2TT,
United Kingdom
Abstract1
The distinguishing practice of mindfulness meditation is the intentional regulation of attention to-2
wards the present moment. Mindfulness meditation therefore emphasizes metacognitive functions, in3
particular the ability to monitor the attentional focus on a moment-by-moment basis. In this study4
we set out to test whether mindfulness meditation experience is associated with an increased ability5
to monitor moment-by-moment fluctuation in the attentional state. In response to auditory cues, par-6
ticipants maintained somatosensory attention to either their left or right hand. At random moments,7
trials were terminated by a probe sound to which participants reported their level of attention at that8
moment. MEG was recorded during the attention interval preceding probe onset. Using a beamformer9
approach, alpha activity in contralateral primary somatosensory regions was quantified. Alpha activity10
for self-reported high versus low attention trials was compared both within and between groups of either11
highly experienced experienced mindfulness meditators, novice meditators or meditation-naive partic-12
ipants (controls). As predicted, generally contralateral alpha power was associated with self-reported13
attention. Novice meditators (< 1000 h of meditation) showed temporal profiles similar to controls,14
displaying a correspondence between self-report and alpha power preceding probe onset. Expert med-15
itators (� 1000 h) showed a strikingly different pattern, however. Their self-reported attentional state16
corresponded with alpha power during a more extended time interval preceding those of controls and17
novice meditators. In addition, self-reported low attention trials showed a distinctive alpha suppression18
preceding probe onset, suggesting that the ability for moment-by-moment monitoring of the attentional19
state permitted greater attentional control.20
1
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2020. ; https://doi.org/10.1101/2020.02.04.933606doi: bioRxiv preprint
1 Introduction21
The goal of this study is to investigate whether extensive meditation experience can be associated with22
changes in meta-metacognitive monitoring of attention. While meditation practices are generally considered23
to incorporate the practice of attentional control, mindfulness meditation distinguishes itself by an emphasis24
on continuous metacognitive monitoring of mental contents and processes (Lutz, Slagter, Dunne, & David-25
son, 2008; Lutz, Dunne, & Davidson, 2007). Mindfulness based interventions have been proven successful26
in treating a wide range of clinical syndromes ranging from depression (Chiesa & Serretti, 2011; Piet &27
Hougaard, 2011; van Aalderen et al., 2012) to anxiety (Roemer & Orsillo, 2006; Wurtzen et al., 2013) and28
ADHD (Zylowska et al., 2008). Several clinical interventions have integrated mindfulness, e.g. in Mind-29
fulness Based Stress Reduction (MBCT, (Kabat-Zinn, 2014, 2013), Mindfulness Based Cognitive Therapy30
((Segal, Williams, & Teasdale, 2002; Teasdale, Segal, & Williams, 1995) and Acceptance and Commitment31
Therapy (Hayes, 2004). The mechanisms by which mindfulness meditation might benefit the mental health32
of the practitioner remain unclear, however. Initial proposals have been put forward from cognitive psycho-33
logical (Baer, 2003; Brown, Ryan, & Creswell, 2007) or neuroscientific perspectives (e.g. (Chiesa & Serretti,34
2011; Holzel et al., 2011; Tang & Posner, 2013). in which mindfulness has been defined in terms of “pay-35
ing attention in a particular way: on purpose, in the present moment, and nonjudgmentally” (Kabat-Zinn,36
2014) or “the state of being attentive to and aware of what is taking place in the present” (Brown & Ryan,37
2003) and similarly in (Bishop et al., 2004). The practice of mindfulness meditation is therefore understood38
beyond the ability to maintain attention upon a specific object or to safeguard oneself from (especially39
negative) distraction. Rather, mindfulness meditation emphases metacognitive awareness of experience re-40
gardless of its desirability or relevance. In fact, it is this explicit awareness of ongoing mental content, called41
meta-awareness (J. W. Schooler et al., 2011) or metacognition (Fleming, Dolan, & Frith, 2012), that allows42
distracting mental content to be noticed so that attention can subsequently be reoriented (J. W. Schooler43
et al., 2011; Smallwood, Beach, Schooler, & Handy, 2007; Smallwood & Schooler, 2006). Mindfulness, or44
“attending to the present moment”, can therefore be operationalized as the metacognitive monitoring of the45
focus of attention.46
It is well known that mind-wandering often occurs without the awareness that one’s mind has drifted47
(Giambra, 1995; J. Schooler & Schreiber, 2004). Importantly, recent work has show that the tendency to48
mind-wander correlates negatively with a mindful disposition (Mrazek, Smallwood, & Schooler, 2012). Fur-49
thermore, successful outcome after MBCT has been associated with increased availability of metacognitive50
reflection on negative experiences (Teasdale et al., 2002). Although metacognition is understood to be an51
important aspect of mindfulness, experimental evidence for improved metacognitive functioning after mind-52
2
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2020. ; https://doi.org/10.1101/2020.02.04.933606doi: bioRxiv preprint
fulness meditation has been only inferred indirectly from studies showing increased performance in tasks53
requiring cognitive control and response inhibition. Jha and colleagues (2007) found that mindfulness med-54
itation improved conflict monitoring in the Attention Network task. Allen and colleagues (2012) showed a55
reduction in affective Stroop conflict and in Heeren and colleagues (2009), mindfulness training was associated56
with increased cognitive inhibition in go/nogo task. Metacognition has been a traditional topic of interest57
in memory research for decades (see: (Metcalfe & Shimamura, 1996) and (Dunlosky & Bjork, 2014) for an58
overview). Recently, metacognition of action and perceptual processes have become a topic of interest as well59
(see (Fleming et al., 2012) for an overview). Since mindfulness meditation is first and foremost an attentional60
practice, recent work on metacognition of attention, or conversely of mind-wandering, could be a promising61
direction of neuroscientific research. Previous experiments on metacognition of attention or mind-wandering62
have used control tasks or external stimulation (Braboszcz & Delorme, 2011; Christoff, Gordon, Smallwood,63
Smith, & Schooler, 2009; Macdonald, Mathan, & Yeung, 2011; Wilimzig, Tsuchiya, Fahle, Einhauser, &64
Koch, 2008). When studying neural correlates of attention during meditation, however, the performance of65
a concomitant task can be expected to interfere with attentional monitoring, as well as preventing ecologi-66
cally valid conclusions regarding meditation. Furthermore, task performance and stimulus processing might67
provide information about the attentional state during stimulus processing or task performance, preventing68
conclusions in terms of metacognition of attention per se (Fleming et al., 2012). Importantly however, the69
attentional state can be determined without a control task or external stimulus. Cortical alpha activity has70
been shown to reflect both the degree and the location of covert somatosensory attention (Haegens, Handel, &71
Jensen, 2011; Haegens, Luther, & Jensen, 2012; van Ede, de Lange, Jensen, & Maris, 2011; van Ede, Jensen,72
& Maris, 2010). Furthermore, occipital alpha power was shown to be strongly correlated with self-reported73
attention during a visual detection task (Macdonald et al., 2011). We have reported that contralateral alpha74
power corresponds to self-reported attentional focus as well (Whitmarsh, Barendregt, Schoffelen, & Jensen,75
2014; Whitmarsh, Oostenveld, Almeida, & Lundqvist, 2017). In this study we used an identical paradigm76
as in Whitmarsh et al. (2014) to investigate whether metacognitive awareness of attention is improved after77
extensive mindfulness meditation. Participants were presented with auditory cues to attend either to their78
left or right hand while their brain activity was measured using MEG. After a random time interval trials79
were terminated with a probe sound to which they reported the degree of attentional focus. Self-reported80
attentional focus was compared with contralateral somatosensory alpha preceding probe onset, in eleven one81
second intervals. We compared the degree of correspondence over time between novice and expert mindful-82
ness meditators as well as with that previously reported for meditation-naive participants (Whitmarsh et83
al., 2014). We predicted that meditators would show a stronger and more sustained correspondence between84
self-reported attentional focus and contralateral alpha power over time. Furthermore, we expected these85
3
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2020. ; https://doi.org/10.1101/2020.02.04.933606doi: bioRxiv preprint
differences to be more pronounced for expert meditators compared to novices.86
2 Method87
2.1 Participants88
The control group consisted of fifteen healthy participants (9 female, mean = 29.4 years, SD = 10.4). One89
control participant was excluded from the analysis due to excessive movement artifacts. The meditation90
group consisted of sixteen healthy experienced mindfulness meditators (5 female, mean = 48.0 years, SD91
= 16.1). They were recruited via personal network of mindfulness teachers or contacted directly. To be92
included in our study they had to be familiar with mindfulness meditation through a meditation retreat and93
had to practice mindfulness mediation regularly for at least a year at the time of the study. Participants94
were enrolled after providing written informed consent and were paid in accordance with guidelines of the95
local ethics committee (CMO Committee on Research Involving Humans subjects, region Arnhem-Nijmegen,96
the Netherlands). The experiment was in compliance with national legislation as well as the code of ethical97
principles (Declaration of Helsinki).98
2.2 Procedure99
Participants were instructed to attend continuously to the cued hand while remaining aware of their attentive100
state until a probe sound (2000 Hz tone) was presented (Figure 1A). Cues consisted of two sequential tones101
of 400 ms each, 200 ms apart, either ascending in pitch for the right hand (2000 Hz followed by 2500 Hz)102
or descending for the left hand (2000 Hz followed by 1500 Hz). Cue side was determined pseudo-randomly.103
The subjective evaluation of attention was done by button press, selecting one out of four options: (1)104
not at all, (2) little, (3) much, (4) fully/maximally attentive. To minimize head movements and provide105
more comfort participants were measured in supine position. To minimize eye movements and blinks and106
increase the chance of fluctuations in attentional focus, participants were instructed to remain with their eyes107
closed throughout the experiment. The experiment started with a training session until participants were108
familiar with the paradigm, followed by three continuous experimental blocks of 125 trials (≈ 25 min each)109
separated by self-paced breaks. Non-evaluation trials (10% ad random) were indicated by the presentation of110
an extra third cue of similar pitch to the second. Participants were instructed that no evaluation of attention111
was needed at these trials, but to respond as quickly as possible with their index finger. On these trials112
only, participants received a short electrical stimulation of the cued hand after probe offset, to provide an113
additional functional localizer (which was not used in the analysis). Non-evaluation trials were discarded114
4
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2020. ; https://doi.org/10.1101/2020.02.04.933606doi: bioRxiv preprint
from all further analysis. Cue-probe intervals were generated according to an exponential distribution with115
mean of 3 s and a cut-off time of 27 s seconds, providing a flat hazard rate. In other words, the chance of the116
probe occurring after trial onset was held constant. A minimal cue-probe interval of 5 s was added, resulting117
in an average cue-probe interval of 8 s seconds, and maximal of 32 s.118
Cue
5 seconds
max. 32 seconds
Prob
e
Eval
uatio
n
2 seconds
Cue
294−2
−1
0
1
2
trials
left highleft low
right highright low
trialnr.block1block2
response timetrial length
coil1 xcoil1 ycoil1 zcoil2 xcoil2 ycoil2 zcoil3 xcoil3 ycoil3 z
1
example design matrixparadigmA B
Figure 1: Schematic of paradigm and example of the design matrix used for the source level General LinearModel. A) Schematic depiction of paradigm showing timing parameters of trial. B) Example design matrixshowing regressors for conditions (0’s and 1’s) and confound regressors (normalized).
2.3 Data Preprocessing119
Continuous MEG data was recorded using a 275-sensor axial gradiometer system (CTF MEG TM Systems120
Inc., Port Coquitlam, BC, Canada) placed in a magnetically shielded room. The ongoing MEG signals were121
low-pass filtered at 300 Hz, digitized at 1200 Hz, and stored for off-line analysis. The subjects’ head position122
was continuously recorded relative to the gradiometer array using coils positioned at the subject’s nasion123
and at the left and right ear canals. High-resolution anatomical images (1 mm isotropic voxel size) were124
acquired using a 1.5 T Siemens Magnetom Sonata system (Erlangen, Germany). The same earplugs, using125
vitamin E instead of the coils, were used for co-registration with the MEG data.126
2.4 Data Analysis127
MEG data was analyzed using the Matlab-based Fieldtrip toolbox, developed at the Donders Institute for128
Brain, Cognition and Behavior (Oostenveld, Fries, Maris, & Schoffelen, 2011). Trials containing movement,129
muscle, and superconducting quantum interference device (SQUID) jumps were discarded by visual inspec-130
tion. Independent component analysis (ICA) was used to remove eye and heart artifacts.131
5
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2020. ; https://doi.org/10.1101/2020.02.04.933606doi: bioRxiv preprint
2.5 Source Reconstruction132
Source reconstruction was performed using a frequency-domain beamformer approach (Dynamic Imaging of133
Coherent Sources) which uses adaptive spatial filters to localize power in the entire brain (Gross et al., 2001;134
Liljestrom, Kujala, Jensen, & Salmelin, 2005). The brain volume of each individual subject was discretized135
to a grid with a 0.8 cm resolution. For every grid point a spatial filter was constructed from the cross-spectral136
density matrix and the lead field. The lead fields were calculated from a subject specific realistic single-shell137
model of the brain (Nolte, 2003), based on the individual anatomical MRIs. We calculated the cross-spectral138
density matrix based upon the full interval between cue offset and probe onset to obtain the most accurate139
estimation of the alpha sources. Individual alpha frequencies were used, determined by the maximum log140
power 7 Hz to 15 Hz on all trials and sensors.141
Separate analyses were performed for one second segments preceding probe onset, to a maximum of 11142
seconds before probe onset. A sufficient number of trials (≈ 100) had trial lengths of at least 11 seconds143
to enable source statistics at those intervals. After calculating the spatial filter for each grid point and144
one-second time segment, alpha activity was estimated using a (Slepian) multitaper approach to accomplish145
accurate frequency smoothing for the alpha band (±2 Hz) around the subject-specific alpha peaks. To enable146
valid voxel-by-voxel comparisons in the face of the beamformer depth bias, alpha estimates were standardized147
over trials.148
A voxel-by-voxel first-level GLM approach was then used for every subject and time segment. Figure149
1B shows an example design matrix, including cue side (left/right) and self-report (high/low), dichotomized150
according to a median split per subject and time-point. In addition, trial number as well as the mean X, Y151
and Z position of the three fiducial coils were entered as separate regressors. The locations of the fiducial152
coils indicate the position of anatomical landmarks of the subject’s head (nasion and pre-auricular points) in153
the MEG helmet. By including the average fiducial positions during each trial, variance that was caused by154
differences in head position over trials was also reduced. To further reduce variance that could be explained155
by response preparation, regressors for evaluation response times and cue-probe duration were added to the156
GLM, together with separate regressors for the response hand, which was switched between each block and157
randomized over subjects. The cue side and self-report predictors consisted of 0’s and 1’s, thereby yielding158
mean standardized alpha power after multiplication with the standardized data. By standardizing the159
remaining covariates (response time, trial length and fiducial position) multiplication with the standardized160
data resulted in correlation values (ρ). Prior to averaging and group statistics, the resulting beta-values161
and correlations values were spatially normalized using SPM2 to the International Consortium for Brain162
Mapping template (Montreal Neurological Institute, MNI, Montreal, QC, Canada).163
6
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2020. ; https://doi.org/10.1101/2020.02.04.933606doi: bioRxiv preprint
2.6 Functional localization of somatosensory regions164
After the reconstruction of alpha power for each voxel and time-point, somatosensory regions of interest165
(ROIs) were determined based on alpha power during the last second preceding probe onset. A voxel-166
by-voxel comparison was made between left and right attention trials. A cluster-based permutation test167
(Maris & Oostenveld, 2007) was then used to identify significant spatial clusters. This resulted in a distinct168
somatosensory alpha-ROI for each hemisphere. Each ROI therefore depended on cue condition (left versus169
right), but remained independent of the self-reported evaluation of attention.170
2.7 Region of Interest analysis171
Alpha power values within the left and right ROI voxels were averaged according to cue condition (ipsi versus172
contra), evaluation (high versus low) and time-point (eleven one-second intervals preceding probe onset). The173
effects of cue condition and evaluation on mean alpha power were tested over time using repeated measures174
ANOVA. Differences in these effects over time were tested using post-hoc t-tests per time-point.175
3 Results176
Attentional focus fluctuated over trials, as was reflected by the use of the full range of responses (Figure177
2A). Participants were generally confident about their ability to perform the task, as was shown by the large178
fraction of high evaluations. A repeated measures ANOVA with one within-subject factor with four levels179
(evaluation) and one group factor (meditators vs. controls) showed that the number of trials per level of180
attention were significantly different (F (3) = 30.931, p < 0.001), and assumed a negative linear relationship181
(F (1) = 73.889, p =< 0.001). No effect of group on evaluation was found (F (3) = 1.451, p = .234).182
Evaluation times were also significantly different between levels of attention (F (3) = 55.390, p < 0.001) and183
showed a positive linear relationship (F (1) = 69.862, p < 0.001). No effect of group on evaluation time was184
found (F (3) = 2.038, p = .116). Furthermore, evaluation times correlated negatively with cue-probe duration185
for both controls (ρ = −0.130, t(13) = −5.505, p < 0.001) and meditators (ρ = −0.078, t(15) = −2.648, p =186
0.018), showing that longer cue-probe durations did not result in a loss of vigilance. The correlation between187
evaluation times and cue-probe duration did not differ between the groups (t(28) = −1.3450, p = 0.189).188
The behavioral analysis therefore showed no differences between meditators and controls, and both groups189
were confident in their performance of the task.190
7
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2020. ; https://doi.org/10.1101/2020.02.04.933606doi: bioRxiv preprint
controlsmeditators
0
100
0
1
2
high lowevaluation of attention
high lowevaluation of attention
eval
uatio
n tim
e (s
)
num
ber o
f tria
ls
A B
Figure 2: Behavioral differences between evaluations. A Distribution of number of trials per evaluation ofattention show that attention was generally rated high, showing a negative relationship with level of attention.B Evaluation times were reduced when attention was evaluated higher, showing a positive relationship withlevel of attention. No differences between groups were found.
3.1 Regions of interest analysis191
Somatosensory alpha regions of interest (ROIs) were determined for controls and meditators on the basis of192
the lateralization of alpha power (alpha cue left minus alpha cue right) during the last second before probe193
onset. A cluster-based permutation test (Maris & Oostenveld, 2007) was used to identify the significant194
clusters responsive to cue direction. In both groups the analysis resulted in two significant clusters, one in195
each hemisphere, including primary sensorimotor areas (see Figure 3). A cluster-based permutation test of196
alpha lateralization showed no significant differences between groups. Mean values of alpha power within197
the group-specific regions-of-interest were used in further analysis. We then computed contralateral alpha198
power preceding probe onset. Evaluations were dichotomized according to a median split per subject and199
time-point, resulting in time courses for high versus low attention trials (Figure 4).200
Differences between high and low attention trials were shown to be uncorrelated with age at any of the201
significant time points for either controls ([−3 : −2], ρ = −.31, p = .915; [−2 : −1], ρ = −.016, p = .956; [−1 :202
0], ρ = .313, p = .275) or meditators: ([−10 : −9], ρ = −.912, p = .476; [−6 : −5], ρ = .096, p = .724; [−5 :203
−4], ρ = −.191, p = .478; [−4 : −3], ρ = .188, p = .486).204
3.2 Controls versus meditators205
We performed a repeated measures ANOVA with time and self-report (high vs. low) as within-subject factors206
and group (meditators vs. controls) as a between-subjects factor. Self-report showed a significant effect on207
contralateral alpha power (F (1, 28) = 9.509, p = 0.005). Time showed to be a significant factor as well208
(F (10, 19) = 2.983, p = 0.019). No significant interaction with group or between time and self-report effects209
8
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2020. ; https://doi.org/10.1101/2020.02.04.933606doi: bioRxiv preprint
0.4
-0.4
0
mea
n st
anda
rdiz
ed a
lpha
cu
e le
ft -
cue
right
CONTROLS MEDITATORS
0.4
-0.4
0
Figure 3: Somatosensory alpha power lateralized in response to cue direction in both groups. Source recon-structed alpha activity during one second preceding probe-onset shows clear lateralization in areas includingprimary somatosensory regions. Significant voxels were thresholded based on cluster-based permutation test(p¡ 0.05) (Maris & Oostenveld, 2007)
were found. As expected, no effect of self-report was found for the ipsilateral ROI (F (1, 28) = 2.318, p =210
0.139). We took a closer look at the temporal profile of alpha in the high versus low evaluated trials. In211
controls, the last three second interval before probe onset showed significantly lower alpha power in high212
versus low attention trials (Figure 4A; paired t-tests, two-tailed: [−3 : −2], t(13) = −2.357, p = 0.036; [−2 :213
−1], t(13) = −2.746, 0.017; [−1 : 0], t(13) = −2.648, p = 0.020). Although the full model ANOVA did not214
show a significant interaction between time and group, differences in alpha power preceded those of controls215
(paired t-tests, two-tailed: [−10 : −9], t(15) = −2.806, p = 0.04; [−6 : −5], t(15) = −2.156, p = 0.048; [−5 :216
−4]t(15) = −4.413, p < 0.001; [−4 : −3], t(15) = −2.275, p = 0.036). As predicted, no time points showed217
a difference between high and low attention on the ipsilateral ROI, for either the control group or the218
meditation group.219
3.3 Novice versus expert meditators220
To explore differences between novices and expert meditators we applied a median split of the meditation221
group on the basis of total hours of meditation experience. Hours of meditation experience were calculated222
from the self-reported estimated time spend in meditation per week, multiplied by the total duration of223
consistent meditation practice. For meditation retreats an estimated ten hours per day were added. The224
9
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2020. ; https://doi.org/10.1101/2020.02.04.933606doi: bioRxiv preprint
******
−0.1
0
0.15
PROBE ONSET
-1-10 TIME (s)
−0.1
0
0.15HIGH ATTENTIONLOW ATTENTION
CONTROLS MEDITATORS
PROBE ONSET
STA
ND
ARD
IZED
AL
PHA
POW
ER
HIGH ATTENTIONLOW ATTENTION
-1-10 TIME (s)
STA
ND
ARD
IZED
AL
PHA
POW
ER
*
A B
Figure 4: Time-resolved analysis of contralateral alpha power for controls (A) and meditators (B). Asteriskdepict significant differences between high- and low-attention at individual time points. Shaded surfacerepresents standard error of mean.
median split resulted in two very separate groups for novices (n = 7, mean = 555 hrs, range 220-885 hrs)225
and expert meditators (n = 8, mean = 12962 hrs, range 3275-24819 hrs). One meditator did not reply226
to our follow-up questions and could not be included in the analysis. As can be seen in Figure 5A, the227
temporal profile of the novice meditators showed a remarkable similarity with those of controls, showing a228
significant effect of self-report at a distinct period directly preceding probe onset ([−2 : −1]t(6) = −4.057, p =229
.006; [−5 : −4]t(6) = −2.298, p = .062; [−6 : −5]t(6) = −3.865, p = 0.008). Expert meditators, however,230
showed a dramatically different profile (Figure 5B), with significant differences in alpha power preceding the231
differences found in novices ([−10 : −9]t(7) = −2.391, p = 0.048; [−8 : −7]t(7) = −2.201, p = 0.064; [−7 :232
−6]t(7) = −2.008, p = 0.085; [−5 : −6]t(7) = −2.580, p = 0.036; [−5 : −4]t(7) = −3.769, p = 0.008).233
Furthermore, the last interval in high meditators showed a significant difference in the opposite direction234
after what seems to be a gradual reduction of alpha power in low attention trials during the last seconds of235
the trial (t(7) = 3.593, p = 0.009). We therefore divided the temporal profile midway in time and averaged236
alpha power for high and low self-report within the late and early period preceding probe onset. An ANOVA237
with self-report (high vs. low) and time (early vs. late) as within-subjects factors and group (novice vs.238
expert) as between-subjects factor showed a significant main effect of self-report (F (1, 13) = 7.672, p = 0.016)239
as well as significant interaction between self-report, time and group (F (1, 13) = 8.908, p = 0.011).240
10
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2020. ; https://doi.org/10.1101/2020.02.04.933606doi: bioRxiv preprint
STA
ND
ARD
IZED
AL
PHA
POW
ER
STA
ND
ARD
IZED
AL
PHA
POW
ER
0
0.1
PROBE ONSET
-1-10 TIME (s)
PROBE ONSET
-1-10 TIME (s)
−0.1
0
0.1
−0.1
*******HIGH ATTENTION
LOW ATTENTION
HIGH ATTENTIONLOW ATTENTION
NOVICES EXPERTSA Bearly lateearly late
Figure 5: Time-resolved analysis of contralateral alpha power for novice meditators (A) and expert meditators(B). Asterisk depict significant differences between high- and low-attention at individual time points. Shadedsurface represents standard error of mean.
4 Discussion241
Compared to both meditation-naive controls and novice meditators, expert mindfulness meditators showed242
a distinctly different temporal profile of metacognitive awareness. While the self-reported attentional state243
of controls and novices were reflected by differences in alpha power preceding probe-onset, expert meditators244
showed a correspondence between self-report and alpha power during an extended period of time preceding245
that of both controls and novice meditators. Furthermore, expert meditators displayed a pronounced sup-246
pression of contralateral alpha power before probe onset, even surpassing those of high attention trials. Both247
these results are unexpected and warrant only tentative conclusions. Given these reservations, the results248
suggest shed light on previously unexplored territory and suggest new avenues of exploring neurocognitive249
mechanisms of mindfulness meditation. Firstly, the findings suggest that while novice meditators and con-250
trols were able to report on their attentional state retrospectively (i.e. after probe onset), expert meditators251
monitored their attention continuously and used it proactively rather than reactively. The ability maintain252
moment-by-moment evaluation of attention is in accordance with the proposed aim of mindfulness medi-253
tation. This might have enabled expert meditators to reestablish their attentional focus when distraction254
was noticed, as reflected by the suppression of alpha power in the latter half of low attention trials. How-255
ever, since these trials were still reported as low attention trials, expert meditators might have confounded256
metacognitive awareness of their attentional state with that of attentional control. Future research could257
test this hypothesis by using additional self-report probes of both attentional state as well as effort/control.258
11
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2020. ; https://doi.org/10.1101/2020.02.04.933606doi: bioRxiv preprint
Positive findings on both indexes would provide objective evidence for the claim that mindfulness meditation259
increases the ability to differentiate internal cognitive processes (Bishop et al., 2004).260
Alpha values for each voxel and time interval were standardized over all trials to remove the beamformer261
depth bias. Because we were interested in the correspondence between self-reports and alpha power over262
time, our data was standardized per time-interval as well. This made our paradigm particularly sensitive to263
metacognitive awareness of attention, but might have missed other effect of mindfulness meditation such as264
an increased ability to modulate somatosensory alpha (Kerr et al., 2011). Our findings are, however, in line265
with recent studies reporting improvements in attentional stability (Lutz et al., 2009) and improved conflict266
monitoring (Jha et al., 2007) as both are considered to be under metacognitive control (Fleming et al., 2012).267
The experiment required a minimum of behavioral responses, and these were not experienced as inter-268
fering with the usual practice of mindfulness meditation. In fact, the paradigm was well received by the269
expert meditators we measured. The benefit of a less intrusive paradigm in meditation research cannot be270
overstated. The absence of any external stimuli or behavioral measurements too often precludes unequivocal271
interpretation of psychophysical findings (see e.g. (Cahn & Polich, 2006)), while at the same time, exogenous272
perceptual and behavioral tasks might interfere with the introspective practice.273
A possible limitation to our study is the fact that the meditation and control groups were not matched in274
age, with the meditation group being significantly older. However, for both controls and meditators, age did275
not correlate with the differences in alpha power between high and low attention trials. It is therefore highly276
unlikely that results were dependent on age. As a between-subjects cross-sectional study our results might277
also suffer from a selection bias. The fact that our findings depended on the hours of meditation experience278
speaks against an interpretation in terms of differences preceding their meditation practice. Lastly, we cannot279
exclude any ‘third variable’ that might correlate both with meditation experience as well as metacognitive280
ability. For instance, it is shown that metacognition is impaired under conditions such as cigarette craving281
(Sayette, Schooler, & Reichle, 2010) or inebriation (Sayette, Reichle, & Schooler, 2009). Since meditation282
is often practiced in a social and religious context, it cannot be excluded that influences of life-style might283
have played a role.284
Mindfulness meditation has been theorized to distinguish itself from meditation practices that focus285
on sustained and selective attention through its emphasis on metacognitive monitoring (Lutz, Dunne, &286
Davidson, 2006; Lutz et al., 2008). While there is growing evidence that mindfulness meditation improves287
attentional abilities, convincing evidence for improved ability for metacognitive monitoring has been lacking288
(Chiesa & Serretti, 2011). This is unfortunate, since these metacognitive abilities are proposed to underlie289
its clinical efficacy (Baer, 2003; Teasdale et al., 2002). Our findings therefore provide important preliminary290
evidence for the involvement of metacognitive monitoring in mindfulness meditation. Furthermore, our291
12
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2020. ; https://doi.org/10.1101/2020.02.04.933606doi: bioRxiv preprint
results suggest that such proactive attentional monitoring abilities could improve the ability to manage292
attentional resources on a moment-by-moment basis. However, in line with the conclusion of Chiesa and293
colleagues (2011), our results also show that measurable effects might require extensive practice.294
5 Acknowledgements295
We would like to express our gratitude to David A. Holmes for his invaluable insights in understanding the296
practices and potential of mindfulness meditation, and for his help in proofreading the manuscript. The297
authors also gratefully acknowledge the support of the BrainGain Smart Mix Programme of the Netherlands298
Ministry of Economic Affairs and the Netherlands Ministry of Education, Culture and Science the Nether-299
lands Initiative Brain and Cognition, a part of the Organization for Scientific Research (NWO) under grant300
number SSM0611,301
References302
Allen, M., Dietz, M., Blair, K. S., van Beek, M., Rees, G., Vestergaard-Poulsen, P., . . . Roepstorff, A. (2012,303
October). Cognitive-Affective Neural Plasticity following Active-Controlled Mindfulness Intervention.304
Journal of Neuroscience, 32 (44), 15601–15610. doi: DOI:10.1523/JNEUROSCI.2957-12.2012305
Baer, R. A. (2003, May). Mindfulness Training as a Clinical Intervention: A Conceptual and Empirical Re-306
view. Clinical Psychology: Science and Practice, 10 (2), 125–143. doi: DOI:10.1093/clipsy.bpg015307
Bishop, S. R., Lau, M., Shapiro, S., Carlson, L., Anderson, N. D., Carmody, J., . . . Devins, G. (2004,308
September). Mindfulness: A Proposed Operational Definition. Clinical Psychology: Science and309
Practice, 11 (3), 230–241. doi: DOI:10.1093/clipsy.bph077310
Braboszcz, C., & Delorme, A. (2011, February). Lost in thoughts: Neural markers of low alert-311
ness during mind wandering. NeuroImage, 54 (4), 3040–3047. Retrieved 2015-12-18, from312
http://www.sciencedirect.com/science/article/pii/S1053811910013005 doi: DOI:10.1016/313
j.neuroimage.2010.10.008314
Brown, K. W., & Ryan, R. M. (2003). The benefits of being present: Mindfulness and its role in psychological315
well-being. Journal of Personality and Social Psychology , 84 (4), 822–848. doi: DOI:10.1037/0022316
-3514.84.4.822317
Brown, K. W., Ryan, R. M., & Creswell, J. D. (2007, October). Mindfulness: Theoretical Foundations318
and Evidence for its Salutary Effects. Psychological Inquiry , 18 (4), 211–237. doi: DOI:10.1080/319
10478400701598298320
13
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2020. ; https://doi.org/10.1101/2020.02.04.933606doi: bioRxiv preprint
Cahn, B. R., & Polich, J. (2006). Meditation states and traits: EEG, ERP, and neuroimaging studies.321
Psychological Bulletin, 132 (2), 180–211. doi: DOI:10.1037/0033-2909.132.2.180322
Chiesa, A., & Serretti, A. (2011, May). Mindfulness based cognitive therapy for psychiatric disorders:323
A systematic review and meta-analysis. Psychiatry Research, 187 (3), 441–453. doi: DOI:10.1016/324
j.psychres.2010.08.011325
Christoff, K., Gordon, A. M., Smallwood, J., Smith, R., & Schooler, J. W. (2009, May). Experience sampling326
during fMRI reveals default network and executive system contributions to mind wandering. Proceed-327
ings of the National Academy of Sciences, 106 (21), 8719–8724. doi: DOI:10.1073/pnas.0900234106328
Dunlosky, J., & Bjork, R. A. (2014). Handbook of Metamemory and Memory. Routledge. doi: DOI:10.4324/329
9780203805503330
Fleming, S. M., Dolan, R. J., & Frith, C. D. (2012, May). Metacognition: computation, biology and331
function. Philosophical Transactions of the Royal Society B: Biological Sciences, 367 (1594), 1280–332
1286. Retrieved 2015-10-09, from http://rstb.royalsocietypublishing.org/cgi/doi/10.1098/333
rstb.2012.0021 doi: DOI:10.1098/rstb.2012.0021334
Giambra, L. M. (1995, March). A Laboratory Method for Investigating Influences on Switching Attention to335
Task-Unrelated Imagery and Thought. Consciousness and Cognition, 4 (1), 1–21. doi: DOI:10.1006/336
ccog.1995.1001337
Gross, J., Kujala, J., Hamalainen, M., Timmermann, L., Schnitzler, A., & Salmelin, R. (2001, January).338
Dynamic imaging of coherent sources: Studying neural interactions in the human brain. Proceedings of339
the National Academy of Sciences, 98 (2), 694–699. Retrieved 2015-11-12, from http://www.pnas.org/340
cgi/doi/10.1073/pnas.98.2.694 doi: DOI:10.1073/pnas.98.2.694341
Haegens, S., Handel, B. F., & Jensen, O. (2011, April). Top-down controlled alpha band activity in342
somatosensory areas determines behavioral performance in a discrimination task. The Journal of343
Neuroscience: The Official Journal of the Society for Neuroscience, 31 (14), 5197–5204. doi: DOI:344
10.1523/JNEUROSCI.5199-10.2011345
Haegens, S., Luther, L., & Jensen, O. (2012, March). Somatosensory anticipatory alpha activity increases346
to suppress distracting input. Journal of Cognitive Neuroscience, 24 (3), 677–685. doi: DOI:10.1162/347
jocn a 00164348
Hayes, S. C. (2004). Acceptance and commitment therapy, relational frame theory, and the third wave349
of behavioral and cognitive therapies. Behavior Therapy , 35 (4), 639–665. doi: DOI:10.1016/S0005350
-7894(04)80013-3351
Heeren, A., Van Broeck, N., & Philippot, P. (2009, May). The effects of mindfulness on executive processes352
and autobiographical memory specificity. Behaviour Research and Therapy , 47 (5), 403–409. doi:353
14
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2020. ; https://doi.org/10.1101/2020.02.04.933606doi: bioRxiv preprint
DOI:10.1016/j.brat.2009.01.017354
Holzel, B. K., Lazar, S. W., Gard, T., Schuman-Olivier, Z., Vago, D. R., & Ott, U. (2011, November).355
How Does Mindfulness Meditation Work? Proposing Mechanisms of Action From a Conceptual356
and Neural Perspective. Perspectives on Psychological Science, 6 (6), 537–559. doi: DOI:10.1177/357
1745691611419671358
Jha, A. P., Krompinger, J., & Baime, M. J. (2007, June). Mindfulness training modifies subsystems359
of attention. Cognitive, Affective, & Behavioral Neuroscience, 7 (2), 109–119. doi: DOI:10.3758/360
CABN.7.2.109361
Kabat-Zinn, J. (2013). Full catastrophe living: using the wisdom of your body and mind to face stress, pain,362
and illness (Revised and updated edition ed.). New York: Bantam Books trade paperback.363
Kabat-Zinn, J. (2014). Wherever you go, there you are: mindfulness meditation in everyday life. New York,364
N.Y.: Hatchette Books. (OCLC: 1099747185)365
Kerr, C. E., Jones, S. R., Wan, Q., Pritchett, D. L., Wasserman, R. H., Wexler, A., . . . Moore, C. I.366
(2011, May). Effects of mindfulness meditation training on anticipatory alpha modulation in primary367
somatosensory cortex. Brain Research Bulletin, 85 (3-4), 96–103. doi: DOI:10.1016/j.brainresbull368
.2011.03.026369
Liljestrom, M., Kujala, J., Jensen, O., & Salmelin, R. (2005, April). Neuromagnetic localization of rhythmic370
activity in the human brain: a comparison of three methods. NeuroImage, 25 (3), 734–745. Retrieved371
2015-11-12, from http://linkinghub.elsevier.com/retrieve/pii/S1053811904007189 doi: DOI:372
10.1016/j.neuroimage.2004.11.034373
Lutz, A., Dunne, J. D., & Davidson, R. J. (2006). Meditation and the neuroscience of consciousness:374
an introduction. In P. Zelazo, M. Moscovitch, & E. Thompson (Eds.), he Cambridge Handbook of375
Consciousness. Cambridge: Cambridge University Press.376
Lutz, A., Dunne, J. D., & Davidson, R. J. (2007). Meditation and the Neuroscience of Consciousness: An377
Introduction. In The Cambridge Handbook of Consciousness. Cambridge University Press. Retrieved378
from http://dx.doi.org/10.1017/CBO9780511816789.020379
Lutz, A., Slagter, H. A., Dunne, J. D., & Davidson, R. J. (2008, April). Attention regulation and monitoring380
in meditation. Trends in Cognitive Sciences, 12 (4), 163–169. doi: DOI:10.1016/j.tics.2008.01.005381
Lutz, A., Slagter, H. A., Rawlings, N. B., Francis, A. D., Greischar, L. L., & Davidson, R. J. (2009,382
October). Mental Training Enhances Attentional Stability: Neural and Behavioral Evidence. Journal383
of Neuroscience, 29 (42), 13418–13427. doi: DOI:10.1523/JNEUROSCI.1614-09.2009384
Macdonald, J. S. P., Mathan, S., & Yeung, N. (2011). Trial-by-Trial Variations in Subjective Attentional385
State are Reflected in Ongoing Prestimulus EEG Alpha Oscillations. Frontiers in Psychology , 2 , 82.386
15
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2020. ; https://doi.org/10.1101/2020.02.04.933606doi: bioRxiv preprint
doi: DOI:10.3389/fpsyg.2011.00082387
Maris, E., & Oostenveld, R. (2007, August). Nonparametric statistical testing of EEG- and MEG-data.388
Journal of Neuroscience Methods, 164 (1), 177–190. Retrieved 2015-11-13, from http://linkinghub389
.elsevier.com/retrieve/pii/S0165027007001707 doi: DOI:10.1016/j.jneumeth.2007.03.024390
Metcalfe, J., & Shimamura, A. P. (1996). Metacognition: knowing about knowing (1st MIT Press paperback391
ed ed.). Cambridge, Mass: MIT Press. (OCLC: 41438436)392
Mrazek, M. D., Smallwood, J., & Schooler, J. W. (2012, June). Mindfulness and mind-wandering: Finding393
convergence through opposing constructs. Emotion, 12 (3), 442–448. doi: DOI:10.1037/a0026678394
Nolte, G. (2003, November). The magnetic lead field theorem in the quasi-static approximation and its use395
for magnetoencephalography forward calculation in realistic volume conductors. Physics in Medicine396
and Biology , 48 (22), 3637–3652.397
Oostenveld, R., Fries, P., Maris, E., & Schoffelen, J.-M. (2011). FieldTrip: Open Source Software for398
Advanced Analysis of MEG, EEG, and Invasive Electrophysiological Data. Computational Intelligence399
and Neuroscience, 2011 , 1–9. Retrieved 2015-11-12, from http://www.hindawi.com/journals/cin/400
2011/156869/ doi: DOI:10.1155/2011/156869401
Piet, J., & Hougaard, E. (2011, August). The effect of mindfulness-based cognitive therapy for prevention402
of relapse in recurrent major depressive disorder: A systematic review and meta-analysis. Clinical403
Psychology Review , 31 (6), 1032–1040. doi: DOI:10.1016/j.cpr.2011.05.002404
Roemer, L., & Orsillo, S. M. (2006, May). Expanding Our Conceptualization of and Treatment for405
Generalized Anxiety Disorder: Integrating Mindfulness/Acceptance-Based Approaches With Exist-406
ing Cognitive-Behavioral Models. Clinical Psychology: Science and Practice, 9 (1), 54–68. doi:407
DOI:10.1093/clipsy.9.1.54408
Sayette, M. A., Reichle, E. D., & Schooler, J. W. (2009, June). Lost in the sauce: the effects of alcohol on mind409
wandering. Psychological Science, 20 (6), 747–752. doi: DOI:10.1111/j.1467-9280.2009.02351.x410
Sayette, M. A., Schooler, J. W., & Reichle, E. D. (2010, January). Out for a Smoke: The Impact of411
Cigarette Craving on Zoning Out During Reading. Psychological Science, 21 (1), 26–30. Retrieved 2019-412
11-13, from http://journals.sagepub.com/doi/10.1177/0956797609354059 doi: DOI:10.1177/413
0956797609354059414
Schooler, J., & Schreiber, C. (2004). Experience, meta-consciousness, and the paradox of introspection.415
Journal of Consciousness Studies, 11 (7-8), 17–39.416
Schooler, J. W., Smallwood, J., Christoff, K., Handy, T. C., Reichle, E. D., & Sayette, M. A. (2011,417
June). Meta-awareness, perceptual decoupling and the wandering mind. Trends in Cognitive Sciences.418
Retrieved 2015-10-07, from http://linkinghub.elsevier.com/retrieve/pii/S1364661311000878419
16
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2020. ; https://doi.org/10.1101/2020.02.04.933606doi: bioRxiv preprint
doi: DOI:10.1016/j.tics.2011.05.006420
Segal, Z. V., Williams, J. M. G., & Teasdale, J. D. (2002). Mindfulness-based cognitive therapy for depression:421
a new approach to preventing relapse. New York: Guilford Press.422
Smallwood, J., Beach, E., Schooler, J. W., & Handy, T. C. (2007, November). Going AWOL in the Brain:423
Mind Wandering Reduces Cortical Analysis of External Events. Journal of Cognitive Neuroscience,424
20 (3), 458–469. Retrieved 2016-03-24, from http://dx.doi.org/10.1162/jocn.2008.20037 doi:425
DOI:10.1162/jocn.2008.20037426
Smallwood, J., & Schooler, J. W. (2006, November). The restless mind. Psychological Bulletin, 132 (6),427
946–958. doi: DOI:10.1037/0033-2909.132.6.946428
Tang, Y.-Y., & Posner, M. I. (2013, January). Special issue on mindfulness neuroscience. Social Cognitive429
and Affective Neuroscience, 8 (1), 1–3. doi: DOI:10.1093/scan/nss104430
Teasdale, J. D., Moore, R. G., Hayhurst, H., Pope, M., Williams, S., & Segal, Z. V. (2002). Metacognitive431
awareness and prevention of relapse in depression: Empirical evidence. Journal of Consulting and432
Clinical Psychology , 70 (2), 275–287. doi: DOI:10.1037//0022-006X.70.2.275433
Teasdale, J. D., Segal, Z., & Williams, J. G. (1995, January). How does cognitive therapy prevent depressive434
relapse and why should attentional control (mindfulness) training help? Behaviour Research and435
Therapy , 33 (1), 25–39. doi: DOI:10.1016/0005-7967(94)E0011-7436
van Aalderen, J. R., Donders, A. R. T., Giommi, F., Spinhoven, P., Barendregt, H. P., & Speckens, A. E. M.437
(2012, May). The efficacy of mindfulness-based cognitive therapy in recurrent depressed patients with438
and without a current depressive episode: a randomized controlled trial. Psychological Medicine, 42 (5),439
989–1001. doi: DOI:10.1017/S0033291711002054440
van Ede, F., de Lange, F., Jensen, O., & Maris, E. (2011, February). Orienting attention to an upcoming441
tactile event involves a spatially and temporally specific modulation of sensorimotor alpha- and beta-442
band oscillations. The Journal of Neuroscience: The Official Journal of the Society for Neuroscience,443
31 (6), 2016–2024. doi: DOI:10.1523/JNEUROSCI.5630-10.2011444
van Ede, F., Jensen, O., & Maris, E. (2010, June). Tactile expectation modulates pre-stimulus beta-445
band oscillations in human sensorimotor cortex. NeuroImage, 51 (2), 867–876. doi: DOI:10.1016/446
j.neuroimage.2010.02.053447
Whitmarsh, S., Barendregt, H., Schoffelen, J.-M., & Jensen, O. (2014, January). Metacognitive awareness of448
covert somatosensory attention corresponds to contralateral alpha power. NeuroImage, 85 , 803–809.449
doi: DOI:10.1016/j.neuroimage.2013.07.031450
Whitmarsh, S., Oostenveld, R., Almeida, R., & Lundqvist, D. (2017, February). Metacognition of attention451
during tactile discrimination. NeuroImage, 147 , 121–129. doi: DOI:10.1016/j.neuroimage.2016.11452
17
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2020. ; https://doi.org/10.1101/2020.02.04.933606doi: bioRxiv preprint
.070453
Wilimzig, C., Tsuchiya, N., Fahle, M., Einhauser, W., & Koch, C. (2008, May). Spatial attention increases454
performance but not subjective confidence in a discrimination task. Journal of Vision, 8 (5), 7. Re-455
trieved 2019-11-13, from http://jov.arvojournals.org/article.aspx?doi=10.1167/8.5.7 doi:456
DOI:10.1167/8.5.7457
Wurtzen, H., Dalton, S. O., Elsass, P., Sumbundu, A. D., Steding-Jensen, M., Karlsen, R. V., . . . Johansen,458
C. (2013, April). Mindfulness significantly reduces self-reported levels of anxiety and depression:459
Results of a randomised controlled trial among 336 Danish women treated for stage I–III breast cancer.460
European Journal of Cancer , 49 (6), 1365–1373. doi: DOI:10.1016/j.ejca.2012.10.030461
Zylowska, L., Ackerman, D. L., Yang, M. H., Futrell, J. L., Horton, N. L., Hale, T. S., . . . Smalley, S. L.462
(2008, May). Mindfulness Meditation Training in Adults and Adolescents With ADHD: A Feasibility463
Study. Journal of Attention Disorders, 11 (6), 737–746. doi: DOI:10.1177/1087054707308502464
18
(which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission. The copyright holder for this preprintthis version posted February 4, 2020. ; https://doi.org/10.1101/2020.02.04.933606doi: bioRxiv preprint