See Yang 2014, Review of neurobiology of mindfulness
2018 RCT of different mindfulness techniques. Takeaway: different types of meditation have different effects.
Mrazek 2013, be skeptical of such bold claims.
Why does nutrition program improve each TUT category slightly? Should the meditators be more or less aware of TUT’s? They seem to be less aware, i.e. they’re reporting less, but arguably they should be reporting more because they’re more aware of when these TUT’s spontaneously surface…
Then again, the improvements in OSPAN and vGRE speak for themselves...
Jain (2007), 23 dropouts Effect sizes for distress were large for both meditation and relaxation (Cohen's d = 1.36 and .91, respectively), whereas the meditation group showed a larger effect size for positive states of mind than relaxation (Cohen's d =. 71 and .25, respectively). The meditation group also demonstrated significant pre post decreases in both distractive and ruminative thoughts/behaviors compared with the control group (p < .04 in all cases; Cohen's d = .57 for rumination and .25 for distraction for the meditation group), with mediation models suggesting that mindfulness meditation's effects on reducing distress were partially mediated by reducing rumination.
DNMTL (purple): hippocampal formation (HF) and parahippocampal cortex (PHC), projections including retrosplenial cortex (RSC), the ventral mPFC (vmPFC) and the pIPL.
DNSUB3: dorsomedial PFC (dmPFC), the lateral temporal cortex (LTC) extending into the temporopolar cortex (TPC), and parts of the inferior frontal gyrus (IFG).
Salience network: AI and ACC
Fronto-parietal control network: dlPFC and anterior IPL
Cingulo-opercular control network (COCN) includes the dorsal ACC (dACC)–medial superior frontal cortex (msFC) and bilateral AI–frontal operculum
Does this generalize to general population? This sample has a high degree of psychopathology…
Salience network responsible for switching between DMN and CEN (see, for example, Goulden 2014).
“Using resting-state functional connectivity MRI, we found no associations between trait mindfulness in youth and measures of static connectivity between ‘core’ neurocognitive networks implicated in mindfulness in adults. Interestingly, associations with trait mindfulness...did emerge, however, when a sliding windows approach was employed to elucidate how network interaction varies over time (i.e. dynamic connectivity)”
Group level ICA identified 4 networks of interest.. Mediational model of mindfulness and anxiety via the number of state transitions → greater capacity for volitional switching, not getting “stuck’?
State 2: High DMN - left and right CEN connectivity,
Problem of reverse inference...
Taking a dynamical systems approach to neuroimaging vs. averaging over the whole time period
Problems: Small n No control
Hypothesis confirmed: activation of expected seed regions during each phase...
Seed 2 (PCC, vmPFC) → ACC decreased in experienced meditators; can switch from DMN to CEN with less SEN in-between.
Note: LTM’s did not enter a meditative state during the task
Concern: LTM’s reflect a priori personality effects, not results of meditation??
Psilocybin research, dissolution of self: coupling on PCC and mPFC, coupling of mPFC and dlPFC.
Left: 1a Right: 1b, attempted volitional control
(“efforting” vs. “not efforting”; “interpret” vs. “observe”; “trying to meditate” vs. “not trying”)
Problems: Small n Heterogeneity of meditation techniques Coding requires introspective ability
LTM’s vs novices and the dlPFC and ACC
mPFC and PCC are hubs of the DMN. During MTT, regions such as the hippocampus and anterior temporal lobe relay info to the DMN.
Do the FA and rsFC findings reflect state-related changes? Would sampling across more occasions provide better data?
Thalamus relays sensory info to cerebral cortex; insula for interoception. Corrected for false discoveries (family-wise error)
Heterogeneity of meditation practices and time accrued. Easy to tell a just-so story/hand-wave about the WM changes we see. Didn’t test for multiple comparisons because group-by-interaction effects didn’t survive multiple comparisons? Lol what? No bold claims on the basis of this one paper
Thalamus (left), insula (right)
Low degree of confidence on these claims….
Connection between psychological stress and neurogenic inflammation.
Think about within vs. between group analysis... HEP = health enhancement program (activity, learn about nutrition, music therapy)
Problems: 5 dropouts from MBSR, 1 dropout from waitlist. Selection effect?
Where’s the control group? HEP-like control program to account for social-interaction confounds?
Good control for behavioral confounds--sleep and exercise--didn’t seem to change among groups.
Again, small sample size.
Stressed adults, not necessarily representative of the population as a whole..
Problems: participants didn’t continue techniques after the intervention...
Meditators = red dots Concerns: Selection effect of meditation
Oligodendrocytes (a type of glia) are necessary to myelinate axons and, consequently, increase WM volume.
My understanding of this study is lacking.
Mindfulness: Behavioral Outcomes, Neural Substrates, and Inflammation
Behavioral benefits, neural substrates, and inflammation.
Changes in rsFC
What is mindfulness? Why should I care?
Monitoring, detecting, disengaging, and
Many different forms of meditation:
● FA (focused attention)
● OM (open monitoring)
MBSR (Kabat-Zinn) incorporates elements of FA
and OM into a secular, 8-week program.
Not about staying in one mental state.
Behavioral and Health Outcomes
Associated with Mindfulness
Improve working memory capacity and
(modified) GRE scores
N = 48 (26 mindfulness class, 22
2 weeks: 1 class/week + meditate 10
min./log food intake daily
OSPAN task and GRE verbal with task-
unrelated thought probes (self-report,
during, and post-test).
Average improvement analogous to 16
Moore et al 2012: Attentional Blink and
improved conflict monitoring
Longitudinal RCT, N = 40 meditation naive
individuals, randomized to meditation or waitlist
EEG and Stroop task at T1=0, T2= 8 weeks, T3=16
Look for differences in ERP’s (error monitoring,
Stroop interference effect).
● Possibly no behavioral
improvements on Stroop task
(longitudinal vs. cross sectional).
● Increased n200 irrespective of
congruence → vigilance (as
opposed to habituation)?
● Reduced p300 ERP for incongruent
stimuli → less resources
marshalled in conflict monitoring
because of increased n2?
Improved emotional regulation
n=83 undergrads reporting distress,
randomized to 1-month MBSR-like
program vs. somatic relaxation control
program vs. control
Dependent variables: GSI (Global Severity
Index), a measure of general
psychological distress, and Positive
States of Mind Scale (PSOM)
No difference between groups (MBSR v. relax): GSI: d=1.36, .91.
Medium difference: PSOM: d=.71, .25
Big difference: rumination (p <0.04, d=.57) and distraction (p<0.04, d=.25)
Cherkin 2017: RCT of MBSR vs. CBT for pain
N = 342 adults with chronic low-back pain; 3 groups, follow-up 4 wks, 8 wks...2-years
Takeaway: CBT and MBSR are equivalent for treating back pain.
Goyal 2014: Meta-analysis of effects sizes
● Only look at studies with a proper control group, 47 made the cut
● “Over the course of 2–6 months, mindfulness meditation program effect size
estimates ranged from 0.22–0.38 for anxiety symptoms and 0.23–0.30 for
depressive symptoms. These small effects are comparable with what would be
expected from the use of an antidepressant in a primary care population,
without the associated toxicities”
● Caution: publication bias, small N’s, etc.
Takeaway: mindfulness training has many
Changes in rsFC (DMN)
Neural Changes and the
Phenomenal Correlates of
For our purposes: DMN, SEN, & CEN
● Social thinking
● Mental time travel (MTT)
● Bottom-up awareness
of salient stimuli
Central Executive Network:
● Top-down control
● Task-switching and
Changes in rsFC (DMN)
Volitionally change activity
Reduction of DMN
Theta frequency (EEG)
Marusak 2018: dFC in children & adolescents
● N = 42 children (ages 6-17, mean =
● Complete measure of trait
mindfulness in children, and index of
anxiety and depression symptoms. →
62% exceeded thresholds for anxiety
● Dynamic rsFC approach (4 networks, 5
● More mindful kids spent less time in
state 2 (positive DMN-CEN coupling,
negative SEN-CEN, negative DMN-
SEN.) → FC pattern linked to mind-
wandering in adults.
● SEN-left CEN connectivity in state 4
→ negative association with
● The seductive allure of post-hoc,
reverse inferences and functional
Hasenkamp 2012: Neural correlates of mind-
wandering and attention
● n=14 meditators, some high practice
time, others low
● Focused attention (FA) meditation
stages/seeds: FOCUS -- MW -- AWARE --
● Button press when mind wanders.
● DMN: vmPFC, PCC.
● SEN: dACC, anterior insula
● CEN: right dlPFC.
● Practice time effects of vmPFC during
SHIFT phase (evaluative processing)
● Other practice time effects:
● Seed 1 (right dlPFC) → right insula, l
● Seed 2 (PCC) → increased connection with
left OFC; reduced connection with ACC →
processing viscero-motor states?
Berkovich-Ohana 2016: task-induced
● N = 18 LTM’s, 18 med-naive controls
● Visual recognition memory task
(basically a one-back memory
● In LTM’s, spontaneous fMRI
fluctuations (i.e. variance in
amplitude) in the DMN are
decreased; in the visual cortex
● LTM’s also had shorter reaction
Functional Connectivity of DMN & visual cortex
● Inter-network DMN-Vis functional
connectivity higher in LTM’s (more
extrinsic awareness), while intra-
network (e.g. DMN-DMN) functional
connectivity was higher in controls
● Strong negative correlation
between years of meditation
practice and DMN-DMN fc (r=-.340)
● Caution: small N’s, possible a priori
Young 2018: Meta-analysis of fMRI studies
● Regions listed in order
of robustness of
● Only 7 articles qualified
for the meta-analysis?
● Long term practitioners
(>10,000 hours) vs.
Garrison 2013a: rtfMRI, meditator v. controls
● PCC = DMN hub, implicated in self-related
processing, social cognition, response error on
● 1a: n = 22 meditators (Theravada), 22 controls
● FA task: focus on breath, graph in background
(displayed PCC activity for runs 1-4, randomized
[PPC or PCC] for runs 5-6).
● 1b: n=9 meditators, 11 controls
● Real-time feedback. Object of
attention → Volitional
manipulation of PCC graph.
Garrison 2013b: blinded neurofeedback
● N = 10 experienced meditators; meditation with
rtfMRI (PCC seed), then self-report.
● 4 runs: no feedback → offline feedback → real
time feedback → volitional change
● “Undistracted awareness” and “effortless
doing” elicit PCC deactivation
● Volitionally able to reduce PCC response
● Obviate the problem of reverse inference
Changes in rsFC (DMN)
Increased ACC activity (novice v. LTM)
Amygdala (reduced in novices)
Reduced PCC, mPFC
Reduced DMN-DMN rsFC
Increased SEN-CEN coupling
Increased insula, dlPFC activity
White Matter Changes Associated
Karapanagiotidis et al. 2017: MTT (FA and rs)
● n=86 (non-meditators), two cohorts (A &
B); 15 min. choice reaction time task
● “multidimensional experience-sampling”
(MDES) throughout the task
● Measured outcomes: structural
connectivity analysis (fractional
● FA increase in temporo-limbic white
matter region for those engaging in MTT
● Right hippocampus was the most well-
connected GM region with the temporo-
limbic WM pathway
● Right Hipp-dACC/mPFC functional
connectivity (i & ii) positively correlated
with MTT scores
Laneri 2016: White Matter Changes and Age
● n=33 meditators, n=31 non-meditators
● Fractional anisotropy (white matter imaging) of
ROI’s: ACC, insula, amygdala, thalamus, and
● In all ROI’s besides right Insula & left ACC, non-
meditators demonstrate age-related FA decline
● higher FA in left insula, right thalamus, right
amygdala (p <0.05) and left thalamus and right
hippocampus (p <0.1).
Issues with cross-sectional study design.
Changes in rsFC (DMN)
Changes in WM/GM
Reduced atrophying of:
Inflammation and the brain (Marsland 2015)
Inflammation is linked to cognitive
decline in midlife
N = 408 (ages 30-54), measure IL-6 and
CRP, cognitive function and MRI
Inflammation inversely associated with
total GM and WM volumes, and
Possible mediation effects but…
correlation != causation.
Rosenkranz 2016: Inflammatory response
● N = 37 healthy controls, 31 long-term
● Add topical capsaicin to forearm; then
take Trier Social Stress Test (speech +
● Measure cortisol (HPA) and alpha
amylase (SNS), flare size, self-report
● After controlling for age, sex and flare
peak, significant intergroup differences
in cortisol AUC (a summary indicator)
● Meditators had smaller flare
response peak than healthy
● Flare response highly correlated
with cortisol response to TSST but
not AA response.
● Meditators perceived level of
stress correlated with cortisol max
% change (r=.5, p=.009); HC’s
(r=.24, p=.21). Differences not
significant across groups.
So why do meditators:
a) present an attenuated inflammatory response?
b) subjectively rate this response more accurately
than healthy controls?
Are people who are more stress-resistant more
likely to meditate; is there a selection effect for
who sticks with meditation; or does meditation
have a salutary effect in and of itself?
Rosenkranz 2013: MBSR vs. active control
● 8 week RCT, n = 49 mindfulness-naive
● Randomized to MBSR or HEP; TSST and
capsaicin skin flare
● Results: MBSR and HEP comparable for
● Daily cortisol release negatively correlated
with MBSR and HEP practice times!
MBSR group showed smaller flare AUC post-
Creswell (2012): MBSR RCT for lonely old folks
● N = 40 individuals (M=65, SD=7) → 20
MBSR program, 20 waitlist
● NF-kB transcription factor: controls
DNA transcription and cytokine
● Mindfulness, loneliness, gene
expression (NF-kB target genes that
code for WBC’s) and pro-inflam.
protein (C-reactive protein and IL-6,
implicated in cardiovasc. disease)
● MBSR reduces loneliness as
compared to control
● Reduction in activity of NF-kB
target genes in MBSR group
relative to controls
● CRP (p = 0.075) condition x
time interaction was
marginally significant between
● IL-6 was not significant.
Creswell 2016: Randomized trial + inflammation
● IL-6: pro-inflammatory cytokine, important
● Experimental test of if mindfulness training
(MBSR) can alter DMN rsFC and IL-6
● N = 35 unemployed (i.e. stressed) → 18
meditation training, 17 relaxation retreat (3
● Measure IL-6 and rsFC at baseline and 4-
months post treatment.
● Mindfulness training reduced IL-6
from baseline to 4-month follow up
as compared to relaxation group
(not due to reemployment).
● Increased left dlPFC-PCC
connection in meditation group →
explains IL-6 changes?
Creswell (2014) Stress-Buffering Account of
Mindfulness: dlPFC increased top-down regulation
stress-related brain areas (HPA), prevent signal
cascade throughout the peripheral nervous system
(see Rosenkranz 2016).
Why is IL-6 significant in Creswell 2016 but
not in Creswell 2012?
Different samples (old people vs. distressed unemployed adults)?
Old people received a longer training program (8 weeks vs. 3 days)?
Creswell 2012 didn’t have an active control (would expect more pronounced effects
between meditation group and control, not less?), whereas Creswell 2016 did?
Kaliman 2014: HDAC and inflammation
● N = 19 LTM’s and 21 HC’s; day of mindfulness vs.
day of leisure (e.g. documentaries)
● Histone deacetylase: epigenetic modification
enzyme (remove acetyl) → reduce gene
transcription. Known to modulate inflammatory
● Post-intervention, meditation group showed
decreased expression of HDAC2, 3, and 9.
● Reduction correlated with HDAC
expression; downregulation of
● HDAC3 normalized using inverse
● RIPK2 implicated in NF-kB
signalling pathway (Yin, 2010)
● Lower post-intervention levels
of HDAC2 & RIPK2 gene
expression → better cortisol
recovery following TSST
Changes in rsFC (DMN)
(HDAC inhibition→ IL-6
Changes in /GMWM
Some conjecture: white matter
changes, inflammation, and age-
Wang 2015: HDAC inhibition prevent WM
injury in mice after TBI
● TBI is a pro-inflammatory stimulus; known
to cause WMI (white matter injury);
positive feedback loop of microglia (M1 =
exacerbate, M2 = neurorepair).
● Some HDAC inhibitors preferentially
promote transcription of neuroprotective
genes (e.g. Scriptaid).
● HDAC inhibition → long, convoluted
biological pathway (NF-kB) → protect
oligodendrocytes → preserve/grow white
Changes in rsFC (DMN)
Changes in WM/GM
(HDAC inhibition→ IL-6
Concerns with the State of Mindfulness
Small sample sizes and survivorship bias
Longitudinal vs. cross-sectional study
Control for confounds (MBSR vs. HEP, see RJ Davidson most-cited study).
Different levels of practice (experienced vs. novice).
Epistemological Concerns w/ Neuroimaging
Are our voxels too big, thus not allowing us to account for regional functional
heterogeneity? (e.g. Kanwisher and the FFA)
Hand-waving → Problems with forward and reverse inference (attributing a cognitive
function to a brain region/network; attributing a cognitive function to a task on the
basis of certain brain regions “lighting up”)
Small sample sizes
● Confounds: cross-sectional design (SES, survivorship bias, other selection
● Is meditation’s influence on pain perception operating through neural or
inflammatory changes? Both? Change the subjective appraisal?
● Critical period for mindfulness? Mindfulness and imagination mutually
● Linear relationship between WM changes and practice time (see Holzel 2011)?
● Dissociation between mindfulness in kids and adults?
● M.E.D. of mindfulness to induce anti-inflammatory effects? Personality effects?
● Decay of salutary effects over time?
● Clinical applications of mindfulness to inflammation-based diseases?