Yale University School of Medicine Study

Manuscripts submitted to Social Cognitive and Affective Neuroscience
Real-time fMRI links first and third person experience:

subjective correlation between meditation and PCC activity
Journal: Social Cognitive and Affective Neuroscience

Fo
Manuscript ID: SCAN-11-242
Manuscript Type: Original Manuscript

r
Date Submitted by the Author: 15-Nov-2011
Complete List of Authors: Scheinost, Dustin; Yale University School of Medicine, Diagnostic Radiology
Pe
Worhunsky, Patrick; Yale University School of Medicine, Psychiatry

Elwafi, Hani; Yale University School of Medicine, Psychiatry
Thornhill IV, Thomas; Yale University School of Medicine, Psychiatry
Kober, Hedy; Yale University School of Medicine, Psychiatry
er
Hampson, Michelle; Yale University School of Medicine, Diagnostic

Radiology
Gray, Jeremy; Yale University, Psychology
Constable, R.; Yale University School of Medicine, Diagnostic Radiology
Papademetris, Xenophon; Yale University School of Medicine, Diagnostic
Re
Radiology
Brewer, Judson; Yale University School of Medicine, Psychiatry
meditation, real-time neurofeedback, self-referential processing, posterior

Keywords:
vi
cingulate cortex, mind-wandering, default mode network

ew
http://mc.manuscriptcentral.com/scan
Page 1 of 32 Manuscripts submitted to Social Cognitive and Affective Neuroscience
Real-time neurofeedback of meditation
1
2 Title: Real-time fMRI links first and third person experience: subjective correlation between meditation
3
4 and PCC activity
5
6
7
8
9 Authors: Dustin Scheinost, Patrick D. Worhunsky, Hani M. Elwafi, Thomas A. Thornhill IV, Hedy Kober,
10
11 Michelle Hampson, Jeremy R. Gray, R. Todd Constable, Xenophon Papademetris, and Judson A. Brewer*
12
13
14
15
16 *To whom correspondence should be addressed. Department of Psychiatry, Yale University School of
17
18
Medicine, 300 George St. Suite 901, New Haven, CT, 06511, Ph: 203-937-4840, Fax: 203-937-3478
Fo
19
20
21 judson.brewer@yale.edu
22
r
23
24
Pe
25
26 Text: 7225 words
27
28
er
29
30
31
32
Re
33
34
35
36
vi
37
38
39
ew
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Manuscripts submitted to Social Cognitive and Affective Neuroscience Page 2 of 32
1
2 Abstract:
3
4 Establishing links between first-person experience and its underlying neuronal processes is important
5
6
7
yet challenging. Generating a contemporaneous introspective report will either modify the experience
8
9 itself or actually be obtained retrospectively, and hence be subject to memory bias. We assayed the
10
11 feasibility of using real-time fMRI neurofeedback to improve the correspondence between first-person
12
13
14 experience and specific brain activation patterns in a manner that minimally affects the experience itself.
15
16 Twenty-two experienced meditators and 22 case-matched controls meditated during fMRI scanning,
17
18
while viewing real-time feedback of activity in their own posterior cingulate cortex (PCC), an area that
Fo
19
20
21 has been conversely implicated in both mind-wandering and meditation. Subjects reported strong
22
r
23 subjective correlations between mind-wandering and an increase in PCC activity, and meditation and a
24
Pe
25
26 decrease in PCC activity. Importantly, they were also able to differentiate PCC activity from brain activity
27
28 in a related control region. Furthermore, meditators demonstrated significant control over their PCC
er
29
30
31 activation compared to novices. Our findings establish the feasibility of using fMRI neurofeedback to
32
Re
33 directly correlate subjective experience with neural activation, and consistent with previous reports,
34
35
further implicate the PCC in neural bases of mindfulness.
36
vi
37
38
39
ew
40 Keywords: meditation, real-time neurofeedback, self-referential processing, posterior cingulate cortex,

41
42
43 mind-wandering, default mode network
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1
2 Finding the connection between the mind and the brain has fascinated neuroscientists for centuries. A
3
4 rich and complex history has emerged around the study of first-person subjective reporting in pursuit of
5
6
7
understanding human experience. Recent technological advances have increasingly refined the objective
8
9 measurement of neuronal processes that are present during human experience. However, links between
10
11 these third-person measurements and first-person subjective reports have not been established due, at
12
13
14 least in part, to several methodological challenges that are inherent in these methodologies. One of the
15
16 main challenges in gathering subjective data is that self-reports can be inaccurate or biased (Nisbett
17
18
1977). Another difficulty is in capturing and characterizing subjective experiences that may regularly
Fo
19
20
21 reside outside of the conscious attention, such as the visual experience of seeing a color, or the auditory
22
r
23 experience of listening to music. Additionally, people vary greatly in their ability to observe and report
24
Pe
25
26 upon their experiences due to variability in the degree of their awareness of the contents of their own
27
28 thoughts (Christoff 2009) as well as fluctuations in their attention levels (Tononi 2008). Furthermore, the
er
29
30
31 act of generating a contemporaneous introspective report about an experience may serve to modify the
32
Re
33 experience itself (Lutz 2003). For example, during introspective states such as meditation, the act of self-
34
35
reflection pulls the individual out of the meditative state.
36
vi
37
38
39
ew
40 Relating subjective experiences and observed measurements or data seems simple enough in theory --
41
42
43 gather a subjective first-person report about an experience as contemporaneously with the experience as
44
45 possible, and gather third-person, objective data about behavior and brain processes simultaneously with
46
47
48
the experience, then formulate concepts of principles of mechanisms that might underlie the experience
49
50 based upon possible correlations between the two (Chalmers 2000). This study design has proven to be
51
52 more difficult in practice than it is in theory, as noted above. A number of recent studies have devised
53
54
55 strategies for solving this problem. For example, in their investigation of the neural basis of mind-
56
57 wandering, Christoff and colleagues used experience sampling during fMRI scanning, where, during a
58
59
60
sustained attention task, subjects were intermittently asked to report where their attention was focused
1
2 and whether they were on- or off-task preceding the query (Christoff 2009). In this study, when subjects
3
4 reported being off-task, they showed activation in the brain’s default mode network (DMN), an network
5
6
7
that reliably demonstrates involvement in both mind-wandering, and self-referential processing (Kelley
8
9 2002; Northoff 2006; Weissman 2006; Mason 2007). Importantly, DMN activation was strongest when
10
11 subjects were unaware of their own mind-wandering. Similarly, Hasenkamp and colleagues designed a
12
13
14 method in which subjective information was simultaneously collected alongside fMRI data (Hasenkamp
15
16 2011). They instructed meditators to meditate in the fMRI scanner and to press a button whenever they
17
18
realized their minds had wandered. They differentially analyzed periods before and after the button
Fo
19
20
21 press to determine brain activation patterns that were activated during presumptively different cognitive
22
r
23 states. They, too, found that DMN activation correlated with mind-wandering, and that salience/attention
24
Pe
25
26 network regions (e.g., dorsal anterior cingulate) were activated during awareness of mind-wandering.
27
28 The objectivity of these studies is much improved over the use of retrospective recall alone. Nevertheless.
er
29
30
31 a query or button press pulls individuals out of their current mind-state and the amount of subjective
32
Re
33 information that can be gathered during these types of experiments is somewhat limited, potentially
34
35
leading to reverse inference of the cognitive processes that may actually be active at the times of the
36
vi
37
38 probes (Poldrack 2006; Christoff 2009). Methods are now needed to refine the improved resolution
39
ew
40 afforded by using fMRI and self-report together, and to include more detailed subjective accounts in a less
41
42
43 disruptive manner.
44
45
46
47
48
Another recent advance in the field of neuroimaging has been the development of real-time functional
49
50 magnetic resonance imaging (rt-fMRI) neurofeedback. This technique retains the advantage of collecting
51
52 objective sampling data contemporaneously with the event, as in the studies highlighted above, but has
53
54
55 the theoretical advantage of reduced interference with the ongoing task or mind-state. Rt fMRI
56
57 neurofeedback has demonstrated preliminary success in several areas, including training the brain to
58
59
60
manipulate external computerized devices such as prostheses (deCharms 2004; Birbaumer 2008),
1
2 communicating with locked-in patients or those previously thought to be in vegetative states (Owen
3
4 2008; Monti 2010), lie detection (Spence 2004; Langleben 2005), controlling symptoms of chronic pain
5
6
7
(deCharms 2005) and modulating brain activation in regions associated with particular cognitive states
8
9 (Caria 2007; Caria 2010; Hamilton 2011). However, to our knowledge, it has not been used as a tool for
10
11 exploring the correlates between 1st person subjective experience and brain activity. This modality may
12
13
14 be especially useful in investigating neural correlates of introspective states that are conceptually difficult
15
16 to convey in the first-person. For example, during the practice of mindfulness meditation, individuals
17
18
practice dropping in to states of “bare awareness,” that, by definition, are free of all concepts including
Fo
19
20
21 the concept of ‘someone’ paying attention (Gunaratana 2002; Lutz 2008). Meditators can report being
22
r
23 mindfully aware just after, but not during, moments of mindfulness because the act of observing disrupts
24
Pe
25
26 the state itself. In extreme examples, such as absorptive concentration meditative states, awareness is
27
28 described as “one-pointed” in the sense that conscious experience is so focused that it becomes literally a
er
29
30
31 single point of focus (Buddhaghosa 1991). The use of rt-fMRI neurofeedback may help bridge the gap
32
Re
33 between subjective self-report, as individuals link their experience (including quality, such as depth) with
34
35
neural activation in a time-precise manner, but with minimal interruption of their state from an external
36
vi
37
38 probe or other interference.
39
ew
40
41
42
43 In this report, we describe the use of rt-fMRI neurofeedback to test the feasibility of linking subjective
44
45 self-report of a meditative state to neural activity in the posterior cingulate cortex (PCC). We chose the
46
47
48
PCC for several reasons. First, it has been implicated as a central node of the DMN. Second, recent data
49
50 have shown that the PCC is specifically and robustly deactivated during different types of meditation
51
52 (Brewer in press). In experiment 1, we hypothesized that individual reports of first-person experience of
53
54
55 mind-wandering or other types of self-referential activity would correlate with increased PCC activation
56
57 and that meditation would correlate with decreased activation in this brain region. Further, we
58
59
60
hypothesized that individuals would be able to discriminate between PCC activity and activity in the
1
2 posterior parietal cortex, a region in the DMN that is tightly temporally coupled to the PCC but has not
3
4 been strongly correlated with self-referential thoughts. In experiment 2, we hypothesized that
5
6
7
experienced meditators would be able to more easily volitionally decrease activation in the PCC
8
9 compared to novice controls. Together, these results support the feasibility of using rt-fMRI feedback for
10
11 linking first- and third-person data, and further establish the sensitivity and specificity of PCC activity for
12
13
14 assessing mind-wandering and meditation.
15
16
17
18
EXPERIMENT 1
Fo
19
20
21 METHODS.
22
r
23 Subjects. 22 right-handed experienced meditators and 22 matched novice controls were recruited to
24
Pe
25
26 participate. Meditators reported on average 13.9 ± 7.9 years and 9249 ± 6799 hours of mindfulness
27
28 meditation experience (one meditator practiced both mindfulness and non-dual meditation as primary
er
29
30
31 practices). Right-handed meditation-naïve controls were case-control matched for gender, age, race,
32
Re
33 education, and employment status (see Table 1). All participants gave informed consent in accordance
34
35
with the procedures of the Yale University Human Investigation Committee.
36
vi
37
38
39
ew
40 Task. Each study run consisted of collecting a 30-second active baseline in which individuals viewed
41
42
43 adjectives and mentally decided whether or not the words described them (Kelley 2002), followed by a 3-
44
45 minute meditation period in which a graph depicting BOLD percent signal change relative to the average
46
47
48
activation from the baseline period in either the PCC or the posterior parietal cortex (control region),
49
50 was displayed to the subject (see Figure 1a). An active baseline condition was used to provide a relatively
51
52 more standard baseline signal, as it has been argued and recently shown that experienced meditators
53
54
55 may adopt a more meditative stance during a passive baseline (Holzel 2007; Brewer in press). The PCC
56
57 was chosen as an a priori region of interest due to recent findings showing it’s activation during self-
58
59
60
referential processing (Kelley 2002; Northoff 2006; Weissman 2006; Mason 2007), and, importantly,
1
2 because it is a common area of deactivation during different types of meditation (Brewer in press). The
3
4 posterior parietal cortex was used as a control region as it is a major node of the DMN that has been
5
6
7
shown not to correlate as strongly with self-referential processing as the PCC (Northoff 2004; Northoff
8
9 2006; Andrews-Hanna 2010). Thus, in theory, the posterior parietal cortex should show similar patterns
10
11 of activity to the PCC, but its activity should not correlate as well with subjective reports.
12
13
14
15
16 Participants were instructed to meditate with their eyes open, as is standard in many meditative
17
18
traditions (instructions below), letting the graph of BOLD percent signal change stay in the background
Fo
19
20
21 or off to the side of their awareness, and, from time to time, to check the graph to see if it correlated with
22
r
23 their experience. Participants were instructed that they would be receiving feedback from a brain region
24
Pe
25
26 that was thought to be involved in self-referential processing and that increased (red) signal on the graph
27
28 reflected self-referential processing (examples of which were mind-wandering, trying to win something
er
29
30
31 or thinking about what they are going to do later), and decreased (blue) signal reflected meditation. They
32
Re
33 were educated to the fact that, due to the nature of fMRI signal (i.e. the hemodynamic response function
34
35
has been shown to have a time lag that peaks between 4-8 seconds after neuronal activation (Lee 2005);
36
vi
37
38 (Bandettini 1993)), the feedback that they were receiving would show a delay of up to eight seconds from
39
ew
40 what was happening in their brain. Thus, they were instructed to check the graph particularly after
41
42
43 periods of excessive mind-wandering or deep meditation, and then immediately return their attention to
44
45 meditation. Their aim was only to determine how well their experience correlated with the graph.
46
47
48
Participants were also informed that they may receive feedback from different parts of the brain during
49
50 different runs, and thus should consider each run individually. The graph began displaying values directly
51
52 after the active baseline period, with an additional value plotted roughly every two seconds, concomitant
53
54
55 with each new BOLD frame that was collected (TR = 2 seconds). Each subject performed six runs total.
56
57 Runs five and six were randomized to show feedback from either the PCC or posterior parietal cortex.
58
59
60
After each run, participants were asked to rate on a scale of 0-10 how well they were able to follow the
1
2 instructions and how well their subjective experience of mind-wandering and meditation correlated with
3
4 the graph. Additionally, they were instructed to briefly describe how they knew that their experience
5
6
7
lined up with meditation. Responses were recorded and transcribed. After the 6th run, subjects were
8
9 informed which of the feedback runs (#5 vs. #6) was from the control region (posterior parietal cortex)
10
11 of the brain.
12
13
14
15
16 Meditation instructions. All participants were instructed in standard mindfulness concentration
17
18
meditation, as follows: “Pay attention to the physical sensation of the breath wherever you feel it most
Fo
19
20
21 strongly in the body. Follow the natural and spontaneous movement of the breath, not trying to change it
22
r
23 in any way. Just pay attention to it. If you find that your attention has wandered to something else, gently
24
Pe
25
26 but firmly bring it back to the physical sensation of the breath”(Gunaratana 2002; Brewer in press).
27
28 Participants practiced meditation outside of the scanner and confirmed that they understood and could
er
29
30
31 follow the instructions before proceeding.
32
Re
33
34
35
Imaging data acquisition and real-time registration. Subjects were scanned in a Siemens 1.5 Tesla
36
vi
37
38 Sonata scanner. After a first localizing scan, a high-resolution sagittal scan was collected using a
39
ew
40 magnetization prepared rapid gradient echo (MPRAGE) sequence (TR = 2530 ms, TE = 3.34 ms, 160
41
42
43 contiguous sagittal slices, slice thickness 1.2 mm, matrix size 192×192, flip angle = 8°). Next, a T1-
44
45 weighted anatomical scan (TR = 500 ms, TE = 11 ms, FoV = 220 mm, thickness = 4 mm thick, gap = 1mm)
46
47
48
was collected with 25 AC-PC aligned axial-oblique slices. After these structural images, acquisition of
49
50 functional data began in the same slice locations as the axial-oblique T1-weighted data. Functional
51
52 images were acquired using a T2* sensitive gradient-recalled single shot echo-planar pulse sequence (TR
53
54
55 = 2000ms, TE = 35ms, flip angle = 90, Bandwidth = 1446 hz/pixel, matrix size = 64x64, FoV = 220 mm,
56
57 interleaved acquisition). Prior to feedback, a short functional series of 10 volumes (first 2 discarded) was
58
59
60
collected. This functional series was used as the single subject reference space for motion correction and
1
2 ROI analysis (see below). Feedback functional runs consisted of 113 volumes with the first two and last
3
4 volume(s) discarded.
5
6
7
8
9 Region of interest definition. Gray matter ROIs were defined on a standard template brain (Holmes
10
11 1998) using BioImage Suite (www.bioimagesuite.org). The first 26 ROIs were defined based on prior
12
13
14 analysis by Brewer, et al. Three of these ROIs (the left posterior insula, left PCC, and left posterior
15
16 parietal cortex) were used for pilot testing of rt-fMRI feedback. The left posterior insula was defined
17
18
anatomically (volume=2869 mm3) using the Yale Brodmann Atlas (Lacadie 2008). The PCC and posterior
Fo
19
20
21 parietal cortex ROIs were functionally defined from Brewer, et al. (Brewer in press). 9 mm diameter
22
r
23 spheres (volume=461 mm3) centered at MNI coordinates (-6, -60, 18) for the PCC (local maximum for
24
Pe
25
26 between-group differences for concentration meditation using GLM analysis) and (-55, -51, 19) for the
27
28 posterior parietal cortex (significant functional connectivity with PCC in meditators and controls, but no
er
29
30
31 significant activity from GLM analysis). The final ROI consisted of the remaining gray matter that was not
32
Re
33 included in the previous ROIs (960455 mm3), and was used to control for scanner drift. Feedback
34
35
displayed to the subjects was only determined from the PCC, posterior parietal, and gray matter drift
36
vi
37
38 control ROI. These ROIs are shown in Figure 1b.
39
ew
40
41
42
43 Real-time image processing. Prior to feedback, the ROIs were transformed from template space to
44
45 single subject reference space through a series of linear and non-linear registrations. Similar to previous
46
47
48
studies, a non-linear transformation was first applied to warp the template brain to the individual
49
50 MPRAGE (3D anatomical) image (Martuzzi 2010; Hampson 2011). Next, the individual T1 axial-oblique
51
52 (2D anatomical) image was linearly registered to 3D anatomical image. Finally, the short functional series
53
54
55 was linearly registered to the 2D anatomical image. All transformations were visually inspected for
56
57 accuracy and were estimated using the intensity-only component of the method implemented by
58
59
60
BioImage Suite as previously reported (Papademetris 2004).
1
2
3
4 Real-time fMRI neurofeedback. The moment-to-moment feedback signal shown to the subject was
5
6
7
determined through a series of preprocessing steps for each frame of the fMRI time-series using the rt-
8
9 fMRI system described previously (Hampson 2011; Scheinost 2011). Briefly, each slice of an fMRI volume
10
11 is reconstructed in real-time and each volume is analyzed immediately after acquisition. This analysis
12
13
14 includes motion correction, after which the mean activation of in each ROI was calculated for each frame.
15
16 To account for motion correct and partial volume effects near the edge of the brain, voxels with intensity
17
18
less than 25% of the overall brain mean were excluded from the calculations of the mean activation.
Fo
19
20
21 Second, any ROI measurement with greater than a 10% change from the previous frame was treated as
22
r
23 an outlier and was replaced by the previous measurement (0th order interpolation). Next, the ROI
24
Pe
25
26 measurements were temporally smoothed based on the last five values with a zero mean, unit variance
27
28 Gaussian kernel. Percent signal change in the ROI (either the PCC or posterior parietal cortex) as
er
29
30
31 compared to the ROI average value across the 30 second baseline was corrected for scanner drift by
32
Re
33 subtracting the percent signal change from the gray matter control region, as previously described
34
35
(deCharms 2005). This corrected ROI measurement was then graphically presented to each subject in
36
vi
37
38 real-time (e.g., see Figure 1a) using E-Prime v. 1.2 (Psychology Software Tools, www.pstnet.com). The
39
ew
40 entire processing stream from functional volume acquisition to feedback display required less than one
41
42
43 second of delay from data acquisition of each new functional brain volume.
44
45
46
47
Statistical analysis of self-report data. Statistical analyses were performed using SPSS/PASW 18 (SPSS,
48
49
50 Inc; Chicago, IL). All tests of significance are two-tailed and means are reported with ± standard deviation.
51
52
53
54
55 RESULTS
56
57 Correlations between subjective self-report and neural activation. As expected, all subjects reported
58
59
60
success in following the instructions (average scores across all six runs = 9.0 ± .01). Additionally,
1
2 individuals reported a high degree of subjective correlation between their first-person experience of self-
3
4 referential processing correlating with red and meditation correlating with blue (average scores across
5
6
7
first four runs = 7.5 ± .24). These were similar among groups (meditators = 7.4 ± .16, controls = 7.7 ± .29).
8
9
10
11 Discrimination between PCC and control region feedback. To determine whether individuals were
12
13
14 able to discriminate between feedback derived from signals from the PCC (which was hypothesized to
15
16 show increased activity during self-referential processing [red] and decreased activity during meditation
17
18
[blue]) as compared to the parietal cortex), we compared individuals’ subjective reports during runs five
Fo
19
20
21 and six, in which they received feedback from these two regions in a randomized order. As expected, due
22
r
23 to the high correlation shown previously between the PCC and posterior parietal cortex in both novices
24
Pe
25
26 and individuals with meditation experience (Andrews-Hanna 2010; Brewer in press), correlations of
27
28 subjective self-reports were relatively high in both regions (PCC = 8.1 ± 1.8, posterior parietal cortex =
er
29
30
31 6.9 ± 2.5); however, correlations between subjective experience and feedback from the PCC were
32
Re
33 significantly higher than those with the posterior parietal cortex (t42= 3.1, p = .004).
34
35
36
vi
37
38 Subjective self-reports. In order to determine the relative sensitivity of rt-fMRI feedback from the PCC
39
ew
40 in correlation with subjective reports, individuals reported on their experience directly after each run
41
42
43 and at the conclusion of their entire scanning session. Representative narratives from three meditators
44
45 and three controls are shown below where we present the subject’s reported correlation between self-
46
47
48
experience and the displayed feedback and the subject’s brief description of this correlation for feedback
49
50 runs 1-4. If subjects were asked any additional questions or clarifications, we present the question
51
52 (preceded by a Q:) and the subject’s response (preceded by an A:). The corresponding feedback
53
54
55 presented to each subject is shown in figure 2 for the meditators and figure 3 for the controls. Finally, we
56
57 present a fourth self-report with additional responses for runs 5 and 6 and the post feedback interview
58
59
60
from a meditator in figure 4.
1
2
3
4 SUBJECT A (meditator, Figure 2a). Run 1. Reported correlation: 7. How did I know? As I was getting used
5
6 to looking at the image there was a lot of self-referential thoughts a lot of worries around it and as I was able to
7
8
9 look away for a period of time and settle back into it. And the little red at the end, I don't know, I didn't feel so
10
11 lost in thought at that period. Run 2. Reported correlation: 7. The red spots were me checking in with the graph.
12
13
14
Yeah, the first one was definitely me checking in with the graph. The second one I don't know... But the little
15
16 ones were definitely. Run 3. Reported correlation: 6. The spike at the beginning was me thinking "I'm doing
17
18 good look at all the blue I did" And after that I was starting to feel some discomfort in my feeling and thinking
Fo
19
20
21 about how to alleviate it. And just getting into the patience of it and so there was a bit more red than I expected.
22
r
23 Run 4. Reported correlation: 7. The smaller spikes again were me thinking about the pain in my finger. And the
24
Pe
25 spikes at the end were me thinking "good job there is a lot of blue."
26
27
28
er
29
30 SUBJECT B (meditator, Figure 2b). Run 1. Reported correlation: 8. Generally when... there were certain
31
32
times when I registered thinking and the thoughts will come and those are the times that corresponded to the red
Re
33
34
35 spikes. Q: There are segments of deeper blue in the middle. Do you remember anything particular there? A:
36
vi
37 Yes, there was a slow inhale. I was very focused on the breath and it slowed down. Run 2. Reported correlation:
38
39
ew
40
10. Well it seemed to be just at that same spot where the breath changes when you inhale and exhale. And it
41
42 appears to roughly correspond with the red spikes. Q: And there were some deep blue in the middle. Did that
43
44 correspond to anything particular? A: There are moments when the concentration is deep and the breath is very
45
46
47 stable. Run 3. Reported correlation: 9. Well there were some thoughts. There were not many thoughts. There
48
49 were some subtle thoughts that I thought would show up but didn't show up. Concentration was fairly smooth.
50
51 Run 4. Reported correlation: 8. Same as before. I had a few more thoughts or concerns in the beginning, so that
52
53
54 seems to be recorded in the more thinking. Q: And in the end I see some deeper blue. What was going on there?
55
56 A: There was a kind of… it felt more stable, the breath and the focus.
57
58
59
60
1
2 SUBJECT C (meditator, Figure 2c). Run 1. Reported correlation: 8. I was aware during [the run] that mind
3
4 was having gaps and differently focusing and strayed from breath. There was a background of thinking. Run 2.
5
6 Reported correlation: 9. I was aware of being with the object -the breath. Run 3. Reported correlation: 9. I was
7
8
9 conscious of mind being focused on breath. A third of the way in I looked at the graph. In the middle -relaxation,
10
11 presence. [Then] it veered away. The red at the end was when I looked at the graph. Run 4. Reported
12
13
correlation: 10. There was a sense of flow, being with the breath -flow deepened in the middle.
14
15
16
17
18 SUBJECT D (control, Figure 3a). Run 1. Reported correlation: 3. At the beginning I felt it was correlating
Fo
19
20
21
well and although I was just breathing and focusing, I kept seeing red so maybe at the end I got anxious but
22
r
23 before I got anxious it stopped correlating. I was thinking did I move or something...I don't think so. Run 2.
24
Pe
25 Reported correlation: 9. At the beginning I wasn't checking in. And then like every time when I go and check in
26
27
28 with the graph it looks like it is going red with a delay but [when I] wasn't paying attention to the graph it
er
29
30 looked blue so it looked like it was correlating with the graph. And towards the end I was thinking this is really
31
32 [neat? garbled] and so I wasn't really paying attention to my breath and so it turned red. Run 3. Reported
Re
33
34
35 correlation: 9. At the beginning was left over from the words and then every time I check in with the graph, it
36
vi
37 keeps going red again…and when I don't pay attention to the graph and just concentrate on my breathing it
38
39
drops back into blue and when I didn't check for a long time it stayed blue for along time. And at the end it was
ew
40
41
42 me being really impressed with this. Run 4. Reported correlation: 6. I feel like I was focusing on the task, like
43
44 on my breathing and at some parts I did get anxious. But um yeah I felt like I was focusing on my breathing a
45
46
47
lot more than before. I kept looking up and it was red all the time.
48
49
50
51 SUBJECT E (control, Figure 3b). Run 1. Reported correlation: 10. I felt like that I was able to focus on
52
53
54 breathing, but occasionally I would have a flickering thought. And I think that what the red shows. Q: I see in
55
56 the beginning there is a little more sustained blue, did you notice anything different between that and any of the
57
58
later parts? A: Ah, I think that there was an eyelash that I was bothering my eye a little bit. And that created a
59
60
1
2 lot of wondering thoughts for a second. Run 2. Reported correlation: 10. I think that I was more wondering
3
4 thoughts, when I'm not doing it right. It kinda looks like that. Q: And I notice a big red spike in the beginning
5
6 and a big blue spike in the middle, anything that you noticed that correlated with those? A: I think in the
7
8
9 beginning I wasn't quite sure if I was supposed to start, and I saw the graph move, and I was like wow, and I
10
11 tried a lot harder to focus on my breathing and watch that. Run 3. Reported correlation: 4. I felt like that I was
12
13
on, really concentrating on my breathing, but it looks like on the graph that I had a lot of wondering thoughts.
14
15
16 Run 4. Reported correlation: 8. I was able to focus on my breathing, the physical sensation, and not thinking of
17
18 breathing. But I felt like that I had [two?] wandering thoughts. Q: I'm sorry, did you say that you weren't
Fo
19
20
21
thinking of breathing? A: Yeah, I was focused more on the physical sensation instead of thinking in and out.
22
r
23
24
Pe
25 SUBJECT F (control, Figure 3c). Run 1. Reported correlation: 5. Because I felt like that I was more focused
26
27
28 on my breath than the graph says. Run 2. Reported correlation: 7. I think I now realize that I thought that I was
er
29
30 more focused on my breath than I was at the last one, comparing my experience from the last one to this one. Q:
31
32 And that blue right there, what did that correlate with? A: Being really focused on my breath, that's the most
Re
33
34
35 intensely. Run 3. Reported correlation: 10. It just looks like that it tracks exactly… I had nothing in, like I
36
vi
37 experienced definite blank spaces, mind wandering. Run 4. Reported correlation: 10. The run looks like what I
38
39
exactly experienced feeling. Q: And what's the difference between this run and the previous run? A: I felt a lot
ew
40
41
42 more relaxed this time. Q: Anything else? A: It felt like less of a struggle to prevent my mind from wandering.
43
44
45
46
47
Subject G (meditator, Figure 4). Run 1. Reported correlation: 8. I don't tend to have a lot of self reference. I
48
49 tried to generate some at the end by saying my name over and over. Run 2. Reported correlation: 8. red peak in
50
51 the middle -I had a memory of swimming in a pool Run 3. Reported correlation: 7. The second red spike at the
52
53
54 beginning [I was] thinking about, evaluating the task. A couple of times in the middle [I was] thinking about
55
56 reporting on what I was noticing Run 4. Reported correlation: 6. Little things. Run 5 (control region). Reported
57
58 correlation: 6. I don't know what that big red part at the end was. Run 6. Reported correlation: 7. [At the]
59
60
1
2 beginning, memory was red, second red was me discussing it with myself.
3
4
5
6 This individual also reported the following after finishing the rt-fMRI session.
7
8
9 “Run 1 is interesting because after several minutes of blue, I wondered if this paradigm actually
10
11 did measure self-referential processing so I effortfully broke the period of resting in awareness
12
13
and generated a sense of self by saying "[my name], [my name], [my name]," while trying to
14
15
16 visualize my face and sense of myself as a subject in the scanner "doing something". This
17
18 produced a large red spike at the end of the run. OK... it works...interesting. Run 2 also shows an
Fo
19
20
21
oscillation that I have noticed while meditating. There are certain "memory-images" that
22
r
23 recurrently appear in practice, and there have been sets of them over the years. For example, for
24
Pe
25 maybe 2 or 3 years, there was one memory- image of trees and vines that were on a path near my
26
27
28 parents home that would appear recurrently in practice, especially at the beginning. In run 2, the
er
29
30 memory-image is one of swimming at my aunt’s swimming pool around age 8. This corresponds
31
32 with the largest red spike in run 2. In Run 3+4, these two types of oscillations appear in the
Re
33
34
35 larger background of awareness. I also tried to focus more on breath sensations but quickly
36
vi
37 remembered how strained that felt because it created a subject object split of the watcher and
38
39
breath. Between run 4 and run 6 (run 5 was the dummy), the real-time feedback was suggesting
ew
40
41
42 something that I had not considered: that these memory images, which I am not in (as an object),
43
44 show up as red or self-referential. They don't have the same jarring tension or contracted feel to
45
46
47
them as the other type of oscillation (when subject is separate from the object) so I have just
48
49 assumed that nothing much is happening with them and kind of disregarded them as relevant. On
50
51 Run 6, [Reported correlation: 7] I had a familiar memory image appear, one of a pond, willow
52
53
54 tree and fields of my parents farm. I noticed the strong red deflection in response to this,
55
56 although I don't appear in the image. I went back to the image to see if there was a sense of
57
58 watcher-subject and noticed that image has a sense of being seen through a child's eyes. The
59
60
1
2 somewhat desolate feeling landscape corresponds to that child's subjectivity. So there is a subject
3
4 there, even though I never noticed it before, the scanner feedback made me look for it. If you
5
6 look at run 6 you can see me exploring the image in a long run of red in the middle. Then I
7
8
9 remembered I wasn't doing the task so I let it go for a while. Then I started imagining myself in
10
11 the future, telling Jud about what I had discovered about childhood memories, which you can see
12
13
clearly in the second run of red at the end of run 6. I am sorry that I blew off the directions, but I
14
15
16 learned something new and very subtle about those recurring memory-images that I have had for
17
18 more than a decade. Something I may not have learned otherwise.”
Fo
19
20
21
22
r
23 Experiment 2:
24
Pe
25 Using rt-fMRI neurofeedback as feedback. Recent studies of meditation have suggested that the PCC is
26
27
28 a key region of deactivation during meditation (Taylor 2011; Brewer in press). As meditators and
er
29
30 controls appeared to be learning key elements of meditation from their feedback sessions in Experiment
31
32
1 (relaxed focus, paying attention to the physical sensations of the breath rather than thinking about
Re
33
34
35 them etc.), and to confirm these literature reports, we administered an additional task to half of the
36
vi
37
subjects who completed Experiment 1. These individuals were instructed to decrease PCC activation as
38
39
ew
40 much as possible. Given meditator’s previous experience at performing a meditation task in a

41
42 selfless/non-‘striving’ way and the fact that novices may not be as adept, we hypothesized that
43
44
45 meditators would be able to more easily intentionally cause a relative decrease in PCC activation
46
47 compared to controls. This is particularly important in that it has been hypothesized that the key to
48
49
50
meditation is “not doing”, suggesting the paradoxical notion that something is happening (attention is
51
52 being paid, albeit to the sensation of the breath), without someone behind that action, which would
53
54 theoretically increase PCC activation (Buddhaghosa 1991; Goldstein 1993; Gunaratana 2002).
55
56
57
58
59
60
1
2 METHODS. Subjects were the same as in Experiment 1 and were recruited sequentially throughout the
3
4 second-half of that study (meditators, n = 10; controls n = 11). After completing the 6th run of Experiment
5
6
7
1, and informing subjects whether run 5 or 6 was feedback from the control region, they were given the
8
9 following instructions: “Now you have gotten a chance to see how activation of this region (the PCC)
10
11 correlates with meditation. Given what you have learned from the previous runs, in the next run, please
12
13
14 see how much you can actively make it go blue.” Subjects then performed a single run and BOLD percent
15
16 signal change in the PCC was calculated as described in experiment 1.
17
18
Fo
19
20
21 Statistical analysis of PCC activation. Statistical analyses were performed on average BOLD percent
22
r
23 signal change values using independent samples t-tests (SPSS/PASW 18 Inc; Chicago, IL). All tests of
24
Pe
25
26 significance are two-tailed, and means are reported with ± standard deviation.
27
28
er
29
30
31 RESULTS.
32
Re
33 As hypothesized, meditators demonstrated a significant voluntary reduction in PCC activation during

34
35
meditation relative to controls (meditators = -.30 ± .32%; controls = .07 ± .29%; t19 = -2.73, p = .013,
36
vi
37
38 Figure 5).
39
ew
40
41
42
43 DISCUSSION.
44
45 Methods for improving the correlation between first-person reports of subjective experience and third-
46
47
48
person measurement of neuronal activation have improved over the past few years, though clear hurdles
49
50 remain. Rt-fMRI neurofeedback is a relatively new and promising technology that has demonstrated
51
52 utility in a number of areas. However, to our knowledge, it has not been tested for use in correlating first-
53
54
55 person subjective experience with third-person objective observation. In this report, we show that using
56
57 rt-fMRI neurofeedback for linking subjective reports to neuronal processes is feasible using the example
58
59
60
of mind-wandering and meditative experience linked to PCC activation (Experiment 1). Not only did
1
2 individuals report correlation between their mental state and the objective rt-fMRI data, but they also
3
4 were able to discriminate between two highly temporally linked brain regions (the PCC and the posterior
5
6
7
parietal cortex) in a significant manner, which provides validation to the accuracy and truthfulness of
8
9 their reports (Experiment 1). Finally, meditators significantly voluntarily decreased activation in the PCC
10
11 compared to controls, confirming previous reports of the PCC’s role in meditative states (Experiment 2)
12
13
14 (Taylor 2011; Brewer in press).
15
16
17
18
This study substantiates the proof-of-principal that using rt-fMRI neurofeedback in cognitive
Fo
19
20
21 neuroscience studies is feasible. In studies designed to explore the correlation between self-reported
22
r
23 experience of introspective brain states and their underlying neuro-mechanisms, such as this one, use of
24
Pe
25
26 rt-fMRI with self-reports can achieve clear and reproducible results that are easily accessible to the
27
28 subjects. Additionally, it obviates possible reverse inference of correlation between subject reports and
er
29
30
31 objective data, as individuals report their direct experience as it correlates with neuronal activation
32
Re
33 without interference from a probe etc. It may also be useful in confirming or further teasing apart
34
35
differential cognitive processes that may be hypothesized to activate certain brain regions. For example,
36
vi
37
38 Mason and colleagues linked mind-wandering to default mode network activation by training individuals
39
ew
40 ‘to boredom’ in a working memory task and measured average neuronal activity during blocks of a
41
42
43 learned versus a novel working memory task (Mason 2007). In our study, subjects reported their direct
44
45 subjective experience of mind-wandering and its relative correlation with PCC activation, adding
46
47
48
convergent validity and resolution to this theory that had not been demonstrated previously. However, it
49
50 should be noted that this methodology, when used solely to document neural correlates of certain mind
51
52 states (i.e., non-self-referential focus) is still limited to some degree by the interruption of the subject as
53
54
55 observer, as highlighted by reports of the graph becoming red when individuals turned their attention to
56
57 it.
58
59
60
1
2 Several questions that have been put forward in the literature about meditation can begin to be
3
4 addressed by this feasibility study. It sheds light on the question of whether default mode brain regions
5
6
7
are involved in self-referential processes (i.e., is meditation truly ‘selfless’?) by showing links between
8
9 thinking about ‘me’ (self-referential focus) and increased activity in the PCC and, conversely, between a
10
11 meditative (non-self-referential) focus and decreased activation of the PCC. Another question is, how
12
13
14 does subjective relaxation affect meditation (and related PCC activity)? Indeed, subjective experiences
15
16 both in novices and meditators from Experiment 1 give the first direct clues that a relaxed awareness of
17
18
phenomena (e.g. the breath) is necessary for deactivation of regions involved in self-referential
Fo
19
20
21 processing (e.g., the PCC), as has been previously hypothesized (Goldstein 1993; Gunaratana 2002;
22
r
23 Taylor 2011). Experiment 2 indirectly supports this as well, as novices demonstrated an increase in PCC
24
Pe
25
26 activation on average while ‘trying’ to make their graph go blue, while experienced meditators were
27
28 significantly more effective at causing their graphs to be blue by ‘allowing’ themselves to drop into a
er
29
30
31 meditative state. This may indicate the critical difference between ‘someone’ making something happen
32
Re
33 and ‘dropping into’ meditation, and/or “may reflect an adaptive process through which present-moment
34
35
awareness is enhanced in individuals with long-term meditation experience, and information in the
36
vi
37
38 environment is processed with reduced distractibility and interference from self-referent thought or
39
ew
40 ruminative processes”(Taylor 2011). However, future studies that are directly designed to test these
41
42
43 hypotheses are needed.
44
45
46
47
48
As brain activity is notoriously ‘noisy’, rt-fMRI neurofeedback may be able to add a degree of sensitivity in
49
50 linking subjective experience with neuronal activity that may otherwise be lost with block-design
51
52 averaged regional activity. For example, Subject C (Figure 2c) reported a deepening of the meditative
53
54
55 state that correlated with significantly reduced PCC activation in run 4 vs. run 3. Additionally, Subject G
56
57 reported nuanced experience that was linked to very minor changes in neuronal activity. Especially in
58
59
60
1
2 individuals who may be trained to proficiency in particular tasks, such as meditation, these techniques
3
4 may be remarkably suited for probing and refining first-person data (Lutz 2003).
5
6
7
8
9 Real-time fMRI neurofeedback has been used in studies in which individuals empirically learned to train
10
11 specific brain regions. These studies use trial-and-error to train subjects to literally gain control of their
12
13
14 brains. This can be very time consuming, results are often variable and some individuals never acquire
15
16 control (deCharms 2008). As observed by Hampson, et al., individuals trained in this manner will often
17
18
continue to try novel methods even after a successful method has been found (Hampson 2011). This
Fo
19
20
21 continued experimentation with different methods could confound these data sets. However, in instances
22
r
23 in which cognitive techniques have been honed over many years (or centuries, in the case of meditation),
24
Pe
25
26 this technology may be useful for confirming ‘correct’ techniques, especially those that are purely mental.
27
28 With meditation, instruction can be maddeningly simple (e.g., “when sitting, just sit”), and thus
er
29
30
31 conceptually difficult to convey. Surprisingly, in our study, we found several instances in which
32
Re
33 individuals spontaneously learned seemingly important distinctions between self-related processes, as

34
35
indicated at first by increased PCC activation and later confirmed by their own subjective experience. For
36
vi
37
38 example, Subject E reported paying attention to the breath in run #3, but then noticed the difference
39
ew
40 between feeling the physical sensation of the breath as compared to thinking about it in the next run,
41
42
43 which was highlighted by a complete reversal of PCC activation (Figure 2b). Also, Subject F noticed that
44
45 relaxation related to decreased PCC activation, while Subject G discovered a very subtle subjectivity
46
47
48
related to an affective tone (“desolate feel…”) during meditation that s/he hadn’t noticed before. These
49
50 suggest the intriguing possibility that rt-fMRI neurofeedback may be used to augment traditional
51
52 meditation instruction. As a yoga teacher provides feedback to guide one’s posture, so too might a
53
54
55 machine, one day, provide feedback on neuronal activation that correlates with ‘striving’, ‘relaxation’ and
56
57 possibly even ‘selfing.’
58
59
60
1
2 There are several noteworthy strengths and limitations to this study. Though a pilot feasibility study, we
3
4 included a relatively large sample size consisting of two different groups that reportedly differed in their
5
6
7
experience in attaining a meditative state. This sample is roughly twice the size of many previous
8
9 realtime fMRI neurofeedback studies. In general, previous neurofeedback studies also only tested
10
11 whether normal individuals can control a brain region and did not draw inferences across different
12
13
14 populations. We also included experienced meditators as they may be more adept at teasing apart subtle
15
16 mind-states and thus may not only improve the theoretical limits of subjective experiential report, but
17
18
also be able to directly test the relationship between a brain region and its purported function (in this
Fo
19
20
21 case PCC activation during meditation). Furthermore, we included a tightly correlated brain region as a
22
r
23 test condition to provide a rigorous control for our study. Though subjects were able to discriminate
24
Pe
25
26 between these regions to a significant degree, future studies might include several control regions of
27
28 varying degrees of difficulty. The current study was also limited to investigation of signal in the PCC and
er
29
30
31 the control region, and future studies could examine additional regions of interest. For example, in
32
Re
33 studies that measure subjective experiences of self over time, individuals might be instructed to perform
34
35
specific past and future related self-referential tasks (Andrews-Hanna 2010) while given feedback from
36
vi
37
38 various nodes of the default mode network (as well as control regions not linked to this network) in
39
ew
40 order to separate specific elements of the subjective experience that correlate with each node.
41
42
43
44
45 We tested the feasibility of using rt-fMRI neurofeedback in exploring first-person subjective experiences
46
47
48
by examining a region of the default mode network (specifically, the PCC) that, theoretically, is conversely
49
50 linked to both mind-wandering and meditation. This allowed testing of two related but distinct mind-
51
52 state experiences in a single study. We used open-ended questions when asking subjects to describe their
53
54
55 experiences, which reduces bias but can lead to ambiguous and/or non-specific reports. This can be
56
57 especially true in individuals who are not trained to observe their minds, as demonstrated by the
58
59
60
subjective reports in this study. As suggested by Lutz and Thompson, future studies using rt-fMRI
1
2 neurofeedback might employ iterative sampling methods in which individuals are systematically trained
3
4 to become reflectively attentive to experience in order to more precisely link brain mechanisms and mind
5
6
7
states (Lutz 2003). For example, when studying visual perception, Lutz and colleagues instructed
8
9 subjects to press a button when a 3D shape emerged from 2D random dot images and, immediately
10
11 afterwards, give a brief verbal report of their experience. In each successive round, the subjects drew
12
13
14 upon what they had noticed (or discovered) in the previous round of training (Lutz 2002). Over time, this
15
16 allows for stabilization of phenomenal invariants, thereby ‘removing noise’ from the system. This type of
17
18
experimentation may also help address the limitation of the delay in BOLD signal from neuronal
Fo
19
20
21 activation, which is secondary to the brains hemodynamic response function. Future studies might
22
r
23 partially overcome this delay between experience and signal feedback by using tasks that present specific
24
Pe
25
26 stimuli in an event-related rather than block design fashion. For example, to examine relationships
27
28 between the experience of craving and its regulation, individuals might be given tasks in which they
er
29
30
31 alternate between craving induction and volitional regulation, while viewing feedback from regions of
32
Re
33 the brain that are hypothesized to be important thereto (e.g. the ventral striatum and dorsolateral
34
35
prefrontal cortex) (Kober 2010).
36
vi
37
38
39
ew
40 To conclude, our results demonstrate the feasibility of using rt-fMRI neurofeedback for more tightly
41
42
43 linking 1st person subjective experience with 3rd person objective measurement without interrupting the
44
45 on-going experience. Our results also demonstrate that this type of feedback can be sensitive and
46
47
48
relatively specific, given a particular task(s) and brain region. Finally, our data also suggest the possibility
49
50 of using rt-fMRI neurofeedback for the augmentation and/or calibration of specific mental training
51
52 techniques such as meditation.
53
54
55
56
57 ACKNOWLEDGMENTS
58
59
60
1
2 We would like to thank the participants for their interest and willingness to partake in this study. We
3
4 would also like to thank Joseph Goldstein and Ginny Morgan for input regarding meditation instructions
5
6
7
and the task paradigm. We would also like to thank Jithendra Bhawnani, and Maolin Qiu for their
8
9 technical expertise in setting up the rt-fMRI feedback, Alice Brewer for help with editing the manuscript,
10
11 and Reza Farajian for his comments on the manuscript. Finally, we would like to thank Marc Potenza and
12
13
14 Kathleen Carroll, the staff of the Yale Therapeutic Neuroscience Clinic, and Hedy Sarofin and the team of
15
16 MRI technologists in the Yale Magnetic Resonance Research Center for their contributions to this
17
18
research. This study was funded by the following grants: NIDA K12-DA00167, 1R03DA029163-01A1, and
Fo
19
20
21 the U.S. Veterans Affairs New England Mental Illness Research, Education, and Clinical Center (MIRECC).
22
r
23
24
Pe
25
26
27
28
er
29
30
31
32
Re
33
34
35
36
vi
37
38
39
ew
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1
2 Tables and figure legends.
3
4 Table 1.
5 Table 1. Baseline characteristics of participants
6
CONT F or
7 2
MED (n=22) (n=22) Total (n=44) χ df p
8
9 Sex N (%) N (%) N (%) 0 1 1
10 Male 13 (59.1) 13 (59.1) 26 (59.1)
11 Female 9 (40.9) 9 (40.9) 18 (40.9)
12 Age 44.7 ± 12.6 42.9 ± 13.8 43.8 ± 13.1 0.185 1 0.67
13
Race
14
15 White 22 (100) 22 (100) 44 (100) N/A
16 Education level 5.451 3 0.142
17 Completed graduate/prof training 14 (63.6) 11 (50.0) 25 (56.8)
18 College grad 5 (22.7) 6 (27.3) 11 (25.0)
Fo
19
Partial college 3 (13.6) 1 (4.5) 4 (9.1)
20
21 High School 0 4 (18.2) 4 (9.1)
22 Years of Education 17.9 ± 3.1 16.6 ± 3.4 17.3 ± 3.3 1.852 1 0.181
r
23
24
Pe
25
26
Figure 1. Schematic of rt-fMRI neurofeedback. a) Individuals performed a 0.5 min baseline task in
27 which they viewed adjectives and decided if these described them, and then were asked to meditate for 3
28 minutes with their eyes open while their brain activation in the posterior cingulate cortex (PCC), or
er
29 posterior parietal cortex control region, was displayed in the background. During meditation, they were
30 instructed to check the graph periodically to determine how well it matched their subjective experience.
31
32
b) Regions of interest: PCC (red, MNI coordinates: -6, -60, 18), posterior parietal cortex (yellow, MNI
coordinates: -55, -51, 19), and remaining grey matter (light blue).
Re
33
34
35 Figure 2. Examples of rt-fMRI neurofeedback from meditators. Percent signal change (corrected for
36 whole brain signal drift) from the PCC during feedback runs 1-4 are shown for a) subject A, b) subject B
vi
37
38 and c) subject C.
39
ew
40 Figure 3. Examples of rt-fMRI neurofeedback from controls. Percent signal change (corrected for
41 whole brain signal drift) from the PCC during feedback runs 1-4 are shown for a) subject D, b) subject E
42 and c) subject F.
43
44
45 Figure 4. Examples of rt-fMRI neurofeedback from meditator (subject G). Percent signal change
46 (corrected for whole brain signal drift) from the PCC during feedback runs 1-6 are shown (run #5 shows
47 feedback from posterior parietal cortex).
48
49
50 Figure 5. Experienced meditators demonstrate relative PCC deactivation during meditation.
51 Average percent signal change (corrected for whole brain signal drift) from the PCC during feedback run
52 7 (Experiment 2) is shown for meditators (blue) and controls (red). p = .013.
53
54
55
56
57
58
59
60
1
2 Andrews-Hanna, J. R., J. S. Reidler, J. Sepulcre, R. Poulin and R. L. Buckner (2010). "Functional-Anatomic
3 Fractionation of the Brain's Default Network." Neuron 65(4): 550-562.
4 Bandettini, P. A., A. Jesmanowicz, E. C. Wong and J. S. Hyde (1993). "Processing strategies for time-course
5 data sets in functional mri of the human brain." Magnetic Resonance in Medicine 30(2): 161-173.
6
7
Birbaumer, N., A. R. Murguialday and L. Cohen (2008). "Brain-computer interface in paralysis." Current
8 Opinion in Neurology 21(6): 634-638 610.1097/WCO.1090b1013e328315ee328312d.
9 Brewer, J. A., P. D. Worhunsky, J. R. Gray, Y. Tang, J. Weber and H. Kober (in press). "Meditation experience
10 is associated with differences in default mode network activity and connectivity." Proceedings of
11 the National Academy of Sciences.
12
13
Buddhaghosa, A. (1991). The path of purification: Visuddhimagga, Buddhist Publication Society.
14 Caria, A., R. Sitaram, R. Veit, C. Begliomini and N. Birbaumer (2010). "Volitional Control of Anterior Insula
15 Activity Modulates the Response to Aversive Stimuli. A Real-Time Functional Magnetic Resonance
16 Imaging Study." Biological Psychiatry 68(5): 425-432.
17 Caria, A., R. Veit, R. Sitaram, M. Lotze, N. Weiskopf, W. Grodd, et al. (2007). "Regulation of anterior insular
18
cortex activity using real-time fMRI." Neuroimage 35(3): 1238-1246.
Fo
19
20 Chalmers, D. J. (2000). "What is a neural correlate of consciousness." Neural correlates of consciousness:
21 Empirical and conceptual questions: 17ñ40.
22 Christoff, K., A. M. Gordon, J. Smallwood, R. Smith and J. W. Schooler (2009). "Experience sampling during
r
23 fMRI reveals default network and executive system contributions to mind wandering." Proc Natl
24
Acad Sci U S A 106(21): 8719-8724.
Pe
25
26 deCharms, R. C. (2008). "Applications of real-time fMRI." Nat Rev Neurosci 9(9): 720-729.
27 deCharms, R. C., K. Christoff, G. H. Glover, J. M. Pauly, S. Whitfield and J. D. E. Gabrieli (2004). "Learned
28 regulation of spatially localized brain activation using real-time fMRI." Neuroimage 21(1): 436-
er
29 443.
30
31 deCharms, R. C., F. Maeda, G. H. Glover, D. Ludlow, J. M. Pauly, D. Soneji, et al. (2005). "Control over brain
32 activation and pain learned by using real-time functional MRI." Proc Natl Acad Sci U S A 102(51):
Re
33 18626-18631.
34 Goldstein, J. (1993). Insight meditation: the practice of freedom. Boston, Shambhala Publications.
35
Gunaratana, H. (2002). Mindfulness in Plain English. Somerville, MA, Wisdom Publications.
36
Hamilton, J. P., G. H. Glover, J.-J. Hsu, R. F. Johnson and I. H. Gotlib (2011). "Modulation of subgenual
vi
37
38 anterior cingulate cortex activity with real-time neurofeedback." Human brain mapping 32(1): 22-
39 31.
ew
40 Hampson, M., D. Scheinost, M. Qiu, J. Bhawnani, C. M. Lacadie, J. F. Leckman, et al. (2011). "Biofeedback of
41
Real-Time Functional Magnetic Resonance Imaging Data from the Supplementary Motor Area
42
43 Reduces Functional Connectivity to Subcortical Regions." Brain Connectivity 1(1): 91-98.
44 Hasenkamp, W., C. D. Wilson-Mendenhall, E. Duncan and L. W. Barsalou (2011). "Mind wandering and
45 attention during focused meditation: A fine-grained temporal analysis of fluctuating cognitive
46 states." Neuroimage.
47
48
Holmes, C. J., R. Hoge, L. Collins, R. Woods, A. W. Toga and A. C. Evans (1998). "Enhancement of MR
49 Images Using Registration for Signal Averaging." Journal of Computer Assisted Tomography 22(2):
50 324-333.
51 Holzel, B. K., U. Ott, H. Hempel, A. Hackl, K. Wolf, R. Stark, et al. (2007). "Differential engagement of
52 anterior cingulate and adjacent medial frontal cortex in adept meditators and non-meditators."
53
54
Neuroscience Letters 421(1): 16-21.
55 Kelley, W. M., C. N. Macrae, C. L. Wyland, S. Caglar, S. Inati and T. F. Heatherton (2002). "Finding the self?
56 An event-related fMRI study." J Cogn Neurosci 14(5): 785-794.
57 Kober, H., P. Mende-Siedlecki, E. F. Kross, J. Weber, W. Mischel, C. L. Hart, et al. (2010). "Prefrontal-striatal
58 pathway underlies cognitive regulation of craving." Proc Natl Acad Sci U S A 107(33): 14811-
59
60
14816.
1
2 Lacadie, C., R. K. Fulbright, J. Arora, R. T. Constable and X. Papademetris (2008). Brodmann Areas defined
3 in MNI space using a new Tracing Tool in BioImage Suite. Human Brain Mapping (abstract).
4 Langleben, D. D., J. W. Loughead, W. B. Bilker, K. Ruparel, A. R. Childress, S. I. Busch, et al. (2005). "Telling
5 truth from lie in individual subjects with fast event-related fMRI." Human Brain Mapping 26(4):
6
7
262-272.
8 Lee, J. H., Y. Lim, B. K. Wiederhold and S. J. Graham (2005). "A functional magnetic resonance imaging
9 (FMRI) study of cue-induced smoking craving in virtual environments." Appl Psychophysiol
10 Biofeedback 30(3): 195-204.
11 Lutz, A., J.-P. Lachaux, J. Martinerie and F. J. Varela (2002). "Guiding the study of brain dynamics by using
12
13
first-person data: Synchrony patterns correlate with ongoing conscious states during a simple
14 visual task." Proceedings of the National Academy of Sciences 99(3): 1586-1591.
15 Lutz, A., H. A. Slagter, J. D. Dunne and R. J. Davidson (2008). "Attention regulation and monitoring in
16 meditation." Trends in Cognitive Sciences 12(4): 163-169.
17 Lutz, A. and E. Thompson (2003). "Neurophenomenology integrating subjective experience and brain
18
dynamics in the neuroscience of consciousness." Journal of Consciousness Studies, 10 9(10): 31-
Fo
19
20 52.
21 Martuzzi, R., R. Ramani, M. Qiu, N. Rajeevan and R. T. Constable (2010). "Functional connectivity and
22 alterations in baseline brain state in humans." Neuroimage 49(1): 823-834.
r
23 Mason, M. F., M. I. Norton, J. D. Van Horn, D. M. Wegner, S. T. Graffon and C. N. Macrae (2007). "Wandering
24
Minds: The Default Network and Stimulus-independent Thought." Science 315(5810): 393.
Pe
25
26 Monti, M. M., A. Vanhaudenhuyse, M. R. Coleman, M. Boly, J. D. Pickard, L. Tshibanda, et al. (2010). "Willful
27 Modulation of Brain Activity in Disorders of Consciousness." New England Journal of Medicine
28 362(7): 579-589.
er
29 Nisbett, R. E. and T. D. Wilson (1977). "Telling more than we can know: Verbal reports on mental
30
31 processes." Psychological Review 84(3): 231.
32 Northoff, G. and F. Bermpohl (2004). "Cortical midline structures and the self." Trends in Cognitive
Re
33 Sciences 8(3): 102.

34 Northoff, G., A. Heinzel, M. de Greck, F. Bermpohl, H. Dobrowolny and J. Panksepp (2006). "Self-referential
35
processing in our brain--A meta-analysis of imaging studies on the self." NeuroImage 31(1): 440-
36
457.
vi
37
38 Owen, A. M. and M. R. Coleman (2008). "Functional neuroimaging of the vegetative state." Nat Rev
39 Neurosci 9(3): 235-243.
ew
40 Papademetris, X., A. Jackowski, R. Schultz, L. Staib and J. Duncan (2004). Integrated Intensity and Point-
41
Feature Nonrigid Registration
42
43 Medical Image Computing and Computer-Assisted Intervention ‚Äì MICCAI 2004. C. Barillot, D. Haynor
44 and P. Hellier, Springer Berlin / Heidelberg. 3216: 763-770.
45 Poldrack, R. A. (2006). "Can cognitive processes be inferred from neuroimaging data?" Trends in
46 Cognitive Sciences 10(2): 59-63.
47
48
Scheinost, D., M. Hampson, J. Bhawnani, M. Qiu, R. T. Constable and X. Papademetris (2011). A GPU
49 Accelerated Motion Correction Algorithm for Real-time fMRI. Human Brain Mapping. Quebec City,
50 Canada.
51 Spence, S. A., M. D. Hunter, T. F. D. Farrow, R. D. Green, D. H. Leung, C. J. Hughes, et al. (2004). "A cognitive
52 neurobiological account of deception: evidence from functional neuroimaging." Philosophical
53
54
Transactions of the Royal Society B: Biological Sciences 359(1451): 1755.
55 Taylor, V. r. A., J. Grant, V. r. Daneault, G. v. Scavone, E. Breton, S. b. Roffe-Vidal, et al. (2011). "Impact of
56 mindfulness on the neural responses to emotional pictures in experienced and beginner
57 meditators." Neuroimage 57(4): 1524-1533.
58 Tononi, G. and C. Koch (2008). "The Neural Correlates of Consciousness." Annals of the New York
59
60
Academy of Sciences 1124(1): 239-261.
1
2 Weissman, D. H., K. C. Roberts, K. M. Visscher and M. G. Woldorff (2006). "The neural bases of momentary
3 lapses in attention." Nat Neurosci 9(7): 971-978.
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Fo
19
20
21
22
r
23
24
Pe
25
26
27
28
er
29
30
31
32
Re
33
34
35
36
vi
37
38
39
ew
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Fo
19
20
21
22
r
23 Figure 1. Schematic of rt-fMRI neurofeedback. a) Individuals performed a 0.5 min baseline task in which
24 they viewed adjectives and decided if these described them, and then were asked to meditate for 3 minutes
Pe
25 with their eyes open while their brain activation in the posterior cingulate cortex (PCC), or posterior parietal
26 cortex control region, was displayed in the background. During meditation, they were instructed to check
the graph periodically to determine how well it matched their subjective experience. b) Regions of interest:
27
PCC (red, MNI coordinates: -6, -60, 18), posterior parietal cortex (yellow, MNI coordinates: -55, -51, 19),
28 and remaining grey matter (light blue).
er
29
30
31
32
Re
33
34
35
36
vi
37
38
39
ew
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Fo
19
20
21
22
r
23
24
Pe
25
26
27
28
er
29
30
31
32
Re
33
34
35
36
vi
37
38
39
ew
40
41
Figure 2. Examples of rt-fMRI neurofeedback from meditators. Percent signal change (corrected for whole
42
brain signal drift) from the PCC during feedback runs 1-4 are shown for a) subject A, b) subject B and c)
43 subject C.
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Fo
19
20
21
22
r
23
24
Pe
25
26
27
28
er
29
30
31
32
Re
33
34
35
36
vi
37
38 Figure 3. Examples of rt-fMRI neurofeedback from controls. Percent signal change (corrected for whole brain
signal drift) from the PCC during feedback runs 1-4 are shown for a) subject D, b) subject E and c) subject
39
ew
F.
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16 Figure 4. Examples of rt-fMRI neurofeedback from meditator (subject G). Percent signal change (corrected
for whole brain signal drift) from the PCC during feedback runs 1-6 are shown (run #5 shows feedback from
17
posterior parietal cortex).
18
Fo
19
20
21
22
r
23
24
Pe
25
26
27
28
er
29
30
31
32
Re
33
34
35
36
vi
37
38
39
ew
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
Fo
19
20
21
22
r
23
24
Pe
25
26
27
28
er
29
30
31
32
Re
33
34
35
36 Figure 5. Experienced meditators demonstrate relative PCC deactivation during meditation. Average percent
vi
37 signal change (corrected for whole brain signal drift) from the PCC during feedback run 7 (Experiment 2) is
38 shown for meditators (blue) and controls (red). p = .013.
39
ew
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60

Yale University School of Medicine Study

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Yale University School of Medicine Study

Uploaded by

Copyright:

Available Formats

Manuscripts submitted to Social Cognitive and Affective Neuroscience

Real-time fMRI links first and third person experience:

Journal: Social Cognitive and Affective Neuroscience

Manuscript Type: Original Manuscript

Worhunsky, Patrick; Yale University School of Medicine, Psychiatry

Hampson, Michelle; Yale University School of Medicine, Diagnostic

meditation, real-time neurofeedback, self-referential processing, posterior

cingulate cortex, mind-wandering, default mode network

40 Keywords: meditation, real-time neurofeedback, self-referential processing, posterior cingulate cortex,

40 much as possible. Given meditator’s previous experience at performing a meditation task in a

33 As hypothesized, meditators demonstrated a significant voluntary reduction in PCC activation during

33 individuals spontaneously learned seemingly important distinctions between self-related processes, as

33 Sciences 8(3): 102.

You might also like