One of the brain’s more incredible properties is its plasticity: The brain constantly rewires itself in an endless run of updates. In a very real sense, you can never use the same brain twice.
Some of these changes are self-evident. For instance, many experiences, especially rare or emotional or highly important ones, leave more or less permanent imprints on the brain—that’s what memories are.
Some types of plasticity may be less obvious. For instance, repeatedly performing a particular action leaves its mark on the brain structure that is needed to perform this action: It starts to grow.
One of the first studies to show this type of plasticity in humans, by Eleanor Maguire and colleagues, was done on London taxicab drivers.1 London is a very complicated city to navigate (not like most North American cities, which are built on a rectangular grid, often with streets that have numbers for names). In order to be licensed, cab drivers have to prove they know the shortest route between any two points in the city—they have to pass a mysterious-sounding test called The Knowledge. Both the intense study of maps necessary to pass The Knowledge as well as the actual driving around in the city would build navigation skills. Maguire et al. found that the part of the brain that is associated with spatial navigation, the back part of the hippocampus, was larger in cab drivers than in other Londoners. They also found that this growth came at a price—cab drivers had a smaller front part of the hippocampus (the part that handles memory).
Sometimes such changes occur very rapidly. In one famous study,2 12 undergraduate students learned to juggle; they were compared with other students who did not practice this skill. After only three months, two parts of the jugglers’ brains had grown larger, namely V5, which specializes in motion perception, and the left posterior intraparietal sulcus, which plans and executes movements that involve objects. After another three months in which the jugglers were asked to not practice their newly acquired skill, these two areas shrunk about halfway back to their original volume. Thus plasticity can (at least under some circumstances, maybe with a relatively simple, well-defined skill) turn on and off quite rapidly.
In this chapter, I look at differences in the structure (or “morphology”) of the brains of meditators compared with the brains of nonmeditators and at changes in structure within meditators over the course of their practice. Chapters 2 and 3 were all about the brain activity associated with meditation; this chapter investigates what lasting changes are wrought by this activity.
Note that it is unlikely that we will find any qualitative differences between the brains of meditators and nonmeditators. That is, there is no reason to suspect that the brains of meditators would have different parts, or would be wired differently, or would be arranged in different ways. More likely, we will find quantitative differences, that is, differences between meditators and nonmeditators in shape, mass, or volume of particular brain regions, as we can see them on MRI scans.
In practice, three types of measures have been used to examine such differences. The first is simply volume: how large a particular brain region or structure is. The second is called gray matter density or gray matter concentration, and it is a bit less intuitive. Scanners do not take a 3D picture of the whole brain. Rather, they build a 3D representation of the brain by taking multiple 2D pictures of a slice of brain tissue (rather like making a cut and examining both sides of the cut); then they stack these images. These slices have a thickness to them (often a millimeter or so). Within each slice, the picture actually consists of a set of small blocks, called voxels. You can think of voxels like pixels on a computer screen, only in 3D, hence the name, short for “volume pixel.” Voxel size differs with the strength of the scanner and the way scanning is done but is typically around 2 mm3—enough to contain over a million neurons. Density is defined as the amount of gray matter inside a voxel; the gray matter density of a particular brain structure is the amount of gray matter combined over all voxels that fall within that structure.
For both of these measures, volume and density, the actual reason behind any changes we might find is ambiguous: Gray matter is made up of the cell bodies of neurons, the end nodes of neurons, glial cells (cells that provide support and protection for neurons), and capillaries (small blood vessels). Thus if we find that meditation increases gray matter volume or density, this could mean a number of things: New neurons have formed, existing neurons have built more connections, new glial cells have popped up, or new capillaries have appeared (the technical term for the latter is angiogenesis), or any combination of these four.
Note that although volume and density are different measures, at least two studies3 have found that they tend to go hand in hand; that is, if you find a difference in one, you are also likely to find a difference in the other, so we could combine studies that looked at only one of the two in the same meta-analysis—exactly what was done in the one meta-analysis on brain structure and meditation.4
The third measure, tractography, is quite different from the other two. Tractography is a technique for mapping the white matter fibers that interconnect different parts of the brain; it also allows for an estimate of how strong or efficient those connections are. White matter is the stuff that connects neurons; white matter fibers, often called white matter tracts, thus send information from one part of the brain to another.
If we find differences between meditators and nonmeditators for one or more of these measures in one or more brain regions, the hope, of course, is that these differences are telltale traces left by meditation practice. This is not a given, however. The issue is that most (but not all) of the relevant studies are cross-sectional studies, that is, studies in which the two different groups—meditators and nonmeditators—are scanned exactly once. This offers only weak evidence for the hypothesis that meditation is the cause of the difference. It could also be the case that people with a particular brain structure are attracted to meditation, or more successful at it, or find it more enjoyable and are thus more likely to stick with it.
Consider, for instance, the finding that meditators have larger attention centers in the brain (an outcome that has actually been observed, as we will see). I hope I have convinced you by now that meditation is a form of attention training, and so it is tempting to conclude that this training translates into an increase in relevant brain tissue. But it might also be possible that it is the other way around: People who are better at paying attention (and so might have larger attention centers in their brain to begin with) might take to meditation more organically than people who are naturally more scatter-brained.
So how can we establish that differences between meditators and nonmeditators are due to the meditation history of the meditators and not to preexisting differences between meditators and nonmeditators or to chance?
The previous argument, namely that what we see makes logical sense—meditation trains attention, and so if we see differences in attention skills between those who meditate and those who do not, this must be due to attention—is a weak argument. The convergence argument sometimes used in research papers—that is, that some regions found to show structural changes in the brain are also found to be activated during meditation—is not much stronger: The causal link could still be reversed: Brain regions might activate more strongly because they contain more neurons, more connections, or more capillaries, and so these preexisting differences between meditators and nonmeditators might show up in brain activation as well.
The only type of study that can unambiguously establish the direction of the effect (i.e., brain → meditation, or meditation → brain) is a longitudinal study. In longitudinal research, two groups—nonmeditators (the “control” group) and beginning meditators (the “treatment” or “intervention” group)—are followed over a period of time so that we can actually look at change. Crucially, in these studies, everyone’s brain is scanned at at least two time points, once before the treatment group starts meditating and once after this group acquires some skill in meditation. Two types of comparisons can be made. First, we can compare “after” with “before” in both groups and see if meditators make more progress than nonmeditators. Second, we can compare meditators with nonmeditators after the intervention to see if differences between meditators and nonmeditators emerge after treatment. (At first blush, it may seem strange to have the second comparison when we have the first. One reason to look at differences after treatment is that this is the statistic that matters to policymakers: How well are people who went through a particular treatment doing compared to their peers who did not undergo treatment? A second reason to look at differences after treatment is that cross-sectional studies by definition only provide this second comparison, and so by using this “after” contrast, we can directly link longitudinal and cross-sectional studies.) In an ideal study, we would also scan at multiple time points along the way, and have a follow-up test as well, so that we can have a good look at the development of expertise and check if expertise is sustained.
Longitudinal studies into true meditation expertise are not very feasible, alas—as we have seen, the people whom researchers in this field consider to be meditation experts have spent many thousands of hours of practice on the cushion. There are, however, a few intrepid researchers that have been brave enough to investigate the effects of short-term interventions, notably eight-week MBSR programs, on brain structure with a longitudinal design.
To some extent, the longitudinal design can be simulated by having differences stand in for change. You can do this by comparing meditators of varying degrees of expertise and relating the level of expertise to one or more measures of brain morphology. Suppose we find, for instance, that in meditators with one year of expertise, brain region X has a volume of 10 mm3 and that this structure has a volume of 12 mm3 in meditators with three years of expertise and 14 mm3 in those with five years of expertise. In this example, each year of expertise translates into a 1 mm3 increase in brain volume in region X. Researchers in pharmacology call this a dose–response relationship. The dose is the amount of meditation expertise; the response is brain volume in region X. It is still technically possible that a dose–response relationship is due to chance or to preexisting differences between these three groups, but it is also quite unlikely.
I should note here that dose–response relationships have been found in plasticity studies outside the field of meditation research. In the London cab driver study, for instance, the increase in volume in the back of the hippocampus was directly related to the number of years on the job; in the juggling study, the students who were better at juggling (presumably because they practiced more) also had larger increases in volume in the two critical regions.
As in my review of the literature on brain activation, I take a recent meta-analysis of the literature as my starting point. There is only one at the time of writing, by Kieran Fox and colleagues.5
As with the Tomasino et al. meta-analysis on brain activation, it may be a good idea to first have a good look at what type of meditators we are talking about here. Fox et al. collected the results from 21 studies. The total tally across all studies is 503 meditators and 472 nonmeditators. Because some participants were tested in more than one study, the number of actual people tested is smaller still, namely about 3706 meditators. All but one of the 21 studies looked at cross-sectional differences between meditators and nonmeditators; five also included a longitudinal component. The longitudinal sample is extremely small, however: Only 114 people were tested before and after participating in a meditation program; they were compared to 90 control participants.
There are four noticeable differences between this set of studies and those on brain activation during meditation. One is that the amount of meditation expertise is lower here (4,664 lifetime hours of lifetime expertise, on average, versus 11,552 lifetime hours for the Tomasino et al. analysis; that is still a lot of experience). A second difference is that the morphology studies, as far as I can tell, all use laypeople—no monastics have been tested. The third is that fewer studies (12 out of 21) focused on participants who practiced meditation within an explicit Buddhist tradition. The fourth is that five of these studies (i.e., about one-quarter) recruited their participants from an MBSR or MBSR-like program,7 that is, a relatively short and decidedly nonreligious training program.
Again, this is a sample that is, on average, highly proficient and well trained in meditation, and likely not very representative of the average Western meditator. That said, the sample is also less extreme in their accomplishments than the sample in the activation studies I discussed in Chapters 2 and 3. This is probably because it is easier to recruit participants at your nearest Zen or Vipassanā center than to find a Buddhist monastery, and it is less critical for these kind of studies to have truly unflappable meditators. In morphology studies, what your mind actually does inside the scanner is unimportant as long as you keep your head still, so you could simply rest, or daydream, or make your shopping list.
There is one important caveat with this set of studies, and that is the possibility of publication bias. What is publication bias?
Here is the scientific publication process in a nutshell: You conduct your study, you analyze your results, you write them up, and you send it off to a journal. The editor of the journal (a fellow scientist) will send your manuscript to two or three other scientists to check on its merits and demerits. These two or three colleagues write a detailed, anonymous review. These reviews almost always contain suggestions for changes—new citations to be added, new analyses to be performed, questions for clarification, errors in reasoning or in the design of the study, and so on. On the basis of those reviews, the editor makes a decision about whether to publish it. When the paper looks good, the decision usually takes the form of a “revise and resubmit”; that is, you are invited to rewrite your manuscript to incorporate some or all of the changes suggested by the reviewers or to counter the reviewer’s points of criticism.
The publication process has many rejection moments built in, starting with the researcher: When you analyze your results, do you like them? Assume, for instance, that you conduct a study on brain morphology and meditation and you do not find any differences, or you find something unexpected that runs counter your ideas (e.g., you find that the attention centers of meditators’ brains are actually smaller). In that case, you might be tempted to shelve the results right away, because you think they do not make sense, or because you think they will never make it past the reviewers.8 If you do write up the results, the reviewers might lambast you for trying to publish “null results” (another technical term—results that say there is nothing there) or for showing results that you, or they, cannot explain, and so the editor rejects the paper and it never sees the light of day.9 This creates a publication bias in that studies reporting a positive result are more likely to be published and studies that found a null or negative result are less likely to be published.
How can we know that publication bias exists in a particular set of studies? The best way is to ask around and see if there’s anyone who has unpublished studies lying around—but who do you ask, and will they be truthful in their answer? One slightly sneakier way to get at publication bias is to probe for indirect evidence that a specific type of result is missing. This involves capitalizing on the law of large numbers: If you have a study with many participants, it is more likely to reveal the truth than a study with only a small group of participants. I gave the example of political polling earlier—asking a few thousand folks whom they will vote for is a better way to predict the outcome of an election than asking just 20 people.
So one way to look for publication bias is to see what the effect size is in the largest studies, and then see if smaller studies show effect sizes that are symmetrical around this large-sample effect. That is, we would expect that smaller scale studies show more randomness, but if random is really random, this should work in both directions: Half of the studies should show a larger effect and half a smaller effect. Also, the studies with the smallest number of people should be off by a larger amount. That is not what Fox et al. found in the brain morphology studies: In effect, all smaller scale studies showed effects that were larger than the effect in the large-scale studies, and all studies showed a positive effect. This is highly suspicious, and it makes it very likely that publication bias is operating—studies with null effects simply don’t make it into print.
This doesn’t mean that the results cannot be trusted. First, publication bias in and of itself is unlikely to lead to convergent results between studies, which is what the Fox et al. meta-analysis ultimately set out to determine. If the effects that have been published were all random, there should be very little consistency in the actual brain regions that are being found to differentiate between meditators and nonmeditators. If we find that there is consistency, the result of publication bias is that we will overestimate the size of the effect, but its location should still most likely be correct. Second, the effects in large-scale studies are not zero—the large-sample effect size is still about 0.5 SD.
In sum, what the presence of publication bias does mean is that the “true” size of the effect (i.e., its size if we were able to get those unpublished studies out of the file drawer) is likely much smaller than what Fox et al. reported in their meta-analysis.
As I mentioned in Chapter 1, and again when discussing the Tomasino et al. map, the aim of meta-analysis is to find convergence among studies. It does so by averaging results across studies. Recall that one consequence of this is that differences in brain structure between meditators and nonmeditators that are only found in a few studies are not likely to figure heavily in the final verdict. One example of this is an excellent study10 that looked at brainstem volume and found that areas in the brainstem that regulate breathing are enlarged in meditators. This result is exciting because it makes sense—after all, meditators are keenly attuned to their breathing—and it also has important implications—for instance, it might explain the effects of long-term meditation on stress reactivity. (And you might read such claims in the media reports on this study.) Much as I like this result, it has not been replicated—no other study has found it—and so it remains a lonely data point that did not become significant in the Fox et al. meta-analysis. This is the way it should be—after all, one of science’s mottos is (or should be) “replication or it didn’t happen.”
Given the modest number of studies, the meta-analysis also by necessity averages across aspects that we probably shouldn’t average across, such as the particular tradition the meditators practice in, different levels of expertise, the ultimate goal of the training (stress reduction vs. enlightenment), and the like. How much this truly matters is unknown. For instance, we might expect that traditions that emphasize focused attention might lead to different brain changes (maybe more thickening in executive control regions) than traditions that put open monitoring to the fore (which might yield larger modifications in the salience network). An early review found exactly these differences between Vipassanā and Zen,11 but—perhaps unsurprisingly, given the mixed results in activation studies, as shown in Chapter 3—later studies have not borne this out.
Looking at changes in gray matter first, Fox et al. found eight brain regions with more or less consistent increases in volume and/or density: (a) the anterior and mid insula, (b) the sensory and motor cortex (including the supramarginal gyrus), (c) the anterior precuneus, (d) the rostrolateral prefrontal cortex, (e) the anterior cingulate cortex and the medial cingulate cortex, (f) the orbitofrontal cortex, (g) the inferior temporal gyrus and the fusiform gyrus, and (h) the hippocampus. They also found two regions that show more or less consistent decreases in volume and/or density: (a) the posterior cingulate cortex and (b) the precuneus.
One way of looking at these findings is to check for convergence with the Tomasino et al. map. The assumption would be that brain areas that are activated during meditation would also show structural changes. Under this view, meditation is a skill that is continuously being honed, and this honing leaves its marks on the brain.12 You may have noticed that a few of Fox et al.’s regions sound familiar from Tomasino et al.’s map or from the discussion in Chapter 3 (anterior insula, anterior cingulate cortex, supramarginal gyrus, posterior cingulate cortex, and precuneus). You may also have noticed absences—regions that are on Tomasino et al.’s map but not on Fox et al.’s (e.g., superior parietal lobe, medial prefrontal cortex)—and regions that are on Fox et al.’s map but not Tomasino et al.’s (e.g., the sensory and motor cortices and the hippocampus).
It seems to me that what is important here is not the region-to-region correspondence between the two maps but the parts of the wider story that overlap. Recall that we viewed Tomasino et al.’s map as a map of four fields of activity: (a) control over attention, which results in a silencing of the default-mode network; (b) an increased focus on body sensations; (c) a decreased sense of global body awareness; and (d) a quieting down of the storytelling mind. Some of these themes are also present in Fox et al.’s map, suggesting, indeed, that repeated activation of these brain regions in meditation might result in structural changes. (Although, as noted earlier, it might also be the other way around—structural differences leading to different levels of activation.)
With regard to attention, Fox et al. find morphological changes in the salience network—most notably an increase in the anterior cingulate cortex and the anterior insula. We’ve encountered plenty of studies in the previous chapter that show that meditation tends to activate these structures.
The evidence for changes in executive control is maybe harder to detect—the major players in the frontoparietal control network do not figure in Fox et al.’s map. Fox et al. did, however, find increased volume and/or density in the rostrolateral prefrontal cortex (sometimes labeled the anterior prefrontal cortex). This area is not on Tomasino et al.’s map, but it is an area that was activated in Hasenkamp et al.’s study on mind-wandering during meditation, notably during the phase of becoming aware of being distracted and during the phase of shifting attention. One influential theory13 about this region is that it selects the set of tasks that the mind needs to perform. More specifically, the rostrolateral prefrontal cortex acts as a kind of gateway that enables switching between attending to environmental stimuli (in meditation: the breath, sensations in the body, or sounds) and internal stimuli (in meditation: thoughts, mind-wanderings, memories, and the like). This, then, would be considered flexible executive control. You could even argue that this is the type of control that the meditating mind is practicing all the time. (Or at least the part of the time when it isn’t dreaming away.)
From Fox et al.’s map, it seems like these types of control might be successful, that is, major parts of the default-mode network (the posterior parietal cortex and the precuneus) are lower in volume and/or density in meditators than nonmeditators. Tomasino et al.’s map showed that both of these areas deactivate during meditation.
Quite a number of the increases in volume and/or density on Fox et al.’s map point at more efficient direct awareness of specific body sensations: the insula (also part of the salience network), the anterior precuneus, and the sensory and motor cortex. The increase in volume and/or density in the supramarginal gyrus (parallel to activation in this structure on Tomasino et al.’s map) points to changes in global body awareness.
Finally, some of Fox et al.’s results are in line with Tomasino et al.’s findings concerning the quieting down of the narrative self. Notable here is the decrease in volume and/or density in the precuneus—echoing its deactivation on Tomasino et al.’s map. The precuneus is an area within the default-mode network structure that is keenly associated with the self-as-story.
Thus the tale I spun from Tomasino et al.’s map is largely repeated here, suggesting that changes in brain functioning during meditation (as measured by activations and deactivations) go hand in hand with changes in gray matter structure. Again, this is a suggestion, a story, not a hard fact.
Interestingly, the Fox et al. map also has three regions that do not show up in the Tomasino et al. analysis. The first of these is the right orbitofrontal cortex.14 This region is associated with a number of functions, including decision-making; that is, it quickly assesses and integrates information to predict what a specific situation could mean to you and what the outcome of a specific decision might be.15 Others have pointed out that the orbitofrontal cortex (like the anterior insula, also on Fox et al.’s map) is also richly connected to the limbic system, that is, the emotional part of the brain, and serves to down-regulate emotion and reappraise negative situations in a more positive light.16
Both of these functions might fit with results of meditation practice as meditators often experience them. Meditation might help, for instance, with shifting the off-the-cushion meditator away from more reflexive, customary reactions to the world around them toward the more dynamic, predictive—shall we say mindful?—kind of decisions that the orbitofrontal cortex performs.17 And it may be the case that the repeated practice of letting go, practiced over and over through the orbitofrontal cortex, may in turn lead to positive effects on negative emotional states, such as depression and anxiety, as we shall see in Chapter 6.
The second result that is unique to Fox et al.’s map is the increase in volume and/or density of the fusiform gyrus and inferior temporal gyrus. It should be noted that some fMRI studies have found activations in this region as well, so this may be a case of convergence between activation and morphology after all.18 Fox et al. point out that these findings “appear puzzling and have been little discussed.” The reason is that these areas are usually associated with visual processing, and given that meditation is typically done with the eyes either closed or half-open, this makes little sense. Fox et al. suggest, with James Austin,19 that this area may be associated with the visual hallucinations or imagery that sometimes pop up, especially during long sitting periods or during retreats. (For more on these hallucinations, see Chapter 3.)
Finally, Fox et al. finds consistent evidence for an enlarged hippocampus in meditators. He offers three explanations.20 First, the hippocampus is part of the default-mode network, and so this difference might signal the higher levels of attention that meditators (especially of the open-monitoring variety) offer to spontaneous thoughts as they arise during meditation. (This, of course, stands in contrast to the shrinkage in the precuneus.) Second, the hippocampus is also a memory structure, specializing in the transfer of short-term memories into more permanent storage. Fox et al. speculate that it is possible that meditators spend a lot of time—not necessarily on the cushion—reviewing, reexamining, and reintegrating memories, placing past experiences in a new light, all in the interest of finding new freedom in their actions and reactions. Third, the hippocampus, with its many connections to the limbic system, also plays a role in emotion regulation and stress resilience. For instance, it is smaller in people with posttraumatic stress disorder. In rats, growing up in an exciting environment (by which we mean not in a bare cage) increases the size of the hippocampus, and this in turn works to buffer against stress.21
One possible solution to the puzzle comes from a study by Eileen Luders and colleagues,22 which looked in more detail at exactly which part or parts of the hippocampus were enlarged in meditators. They found significant effects only for the subiculum, the lower part of the hippocampus. The role of the subiculum is still not completely mapped out, but we do know that one of its functions is to regulate stress via what is called the hypothalamic–pituitary–adrenal axis, or HPA axis.23 Remember the discussion earlier about the parasympathetic and sympathetic nervous system? The HPA axis is what sets the parasympathetic system in motion. Basically, when something stresses you out, the following sequence of events unfolds: The hypothalamus (part of the limbic system in the brain) starts releasing corticotropin-releasing hormone, which causes the pituitary gland to release adrenocorticotropic hormone (also known as corticotropin), which then signals the adrenal cortex to start producing glucocorticoid hormones (mainly cortisol), epinephrine (also known as adrenaline) and norepinephrine, and off you go on your fight-or-flight response. The subiculum is intricately connected to this system, acting as a brake—it stops or limits the response of the HPA system, more specifically the release of cortisol that sets the actual stress response in motion. Thus, even when a stressor comes along, the subiculum can still intervene and prevent your heart rate from going up, your lungs from gasping for air, your blood pressure from skyrocketing, your sweat glands from pumping, and your digestion from coming to a grinding halt. (Ever known a meditator who is preternaturally calm, even when poked? That’s possibly a more than fully grown subiculum in action.)
Interestingly, this may be a circular process. Stress hormones (particularly cortisol) slowly destroy hippocampal tissue.24 If meditation, or the habit of meditating, brings about a reduction of the stress response (through repeated return to the breath, an increased parasympathetic response, or the increasing habit of being able to let go of the small stuff), then this might have the added effect of preserving the hippocampus from the normal wear and tear of daily life.
The average effect size over all these structures in all these studies is quite large—about .75 of a standard deviation (0.77 SD, to be precise). As I mentioned, this is likely an overestimation. Fox et al. applied some fancy statistics to obtain a better estimate and concluded that the true effect is more likely about 0.44 SD. This means that 67% of people in the control group have lower volume/density in these structures than the average meditator (or larger, for the precuneus and the posterior cingulate cortex).
Another (maybe more surprising) way to look at these brain changes is to consider that meditation might counteract aging. In many of the studies that are represented in the Fox et al. map, there is a wide age range in the meditators and controls. Two studies25 have calculated correlations between volume of particular regions and age. One found a correlation between total gray matter and age of r = –.54 for controls but a zero correlation in meditators (r = .01); one looked at the right frontal region and found a negative correlation in controls (r = –.76) and an essentially zero correlation in meditators (r = –.05). These are only two studies, but the suggestion is that meditators’ brains (or parts thereof) show less age-related decline (i.e., in these two cases, no aging at all) compared to the brains of those who do not meditate (which show quite a bit of shrinkage with age).
What happens in white matter, the bundles of nerve fibers that connect different parts of the brain?
Fox et al. found two pathways that show higher efficiency in meditators. The first is the corpus callosum. You may know that the brain comes in two halves (“hemispheres”). The corpus callosum connects those two halves. Meditators have an enlarged corpus callosum, especially in the front of the brain (the genu and the forceps minor), suggesting that information transfer between the two brain halves proceeds more efficiently. This may be a byproduct of all the activation going on in different areas of the front of the brain during meditation (such as the insula, the anterior cingulate cortex, the rostrolateral prefrontal cortex, and the orbitofrontal cortex).
The second is the superior longitudinal fasciculus. This is one of the brain’s main front-to-back-to-front pathways; it connects parietal regions (like those associated with body awareness) with frontal regions (like those involved in attention). The different subcomponents of this tract are responsible for things like the sense of your body in space, the moment-to-moment understanding of the state of your body, spatial attention, and control over your focus of attention.26 The most likely story here is that this enhanced connection represents the fruit of repeatedly and persistently paying close attention to fleeting sensations in the body (i.e., the breath and/or bodily sensations).
Eileen Luders and colleagues27 have looked at how well these white matter tracts28 still work with advancing age. Over 20 different tracts, the correlation with age was r = –.69 for nonmeditators, compared with –.25 for meditators. This is only a single study, but the message is again that meditators’ brains may be less susceptible to the negative effects of aging than those of nonmeditators.
All the previous findings fall within the “weak” category of evidence for the existence of a causal connection—the results could also be due to preexisting differences between meditators and nonmeditators rather than to the practice itself.
Fortunately, a number of studies have recruited meditators that differ widely in meditation experience. This allows the researchers to correlate experience (typically the number of years the meditators have been meditating) with volume and/or density of particular brain regions.29 As already explained, the existence of a dose–response relationship (more meditation experience is associated with an increase or decrease in volume and/or density) would be stronger evidence for the position that meditation practice changes the brain.30
With regard to the salience network, two studies Grant et al. (2010) and Hölzel et al. (2008) report a correlation between meditation experience and volume/density in the anterior cingulate cortex; the average correlation is .21. The average correlation between experience and volume/density in the insula (if we consider the insula as part of the salience network) is .38,34 Hölzel et al. (2008), Luders et al. (2009), and Luders et al. (2012b).
With regard to executive control and the default network, the single study34 that analyzed the correlation with the rostrolateral prefrontal cortex failed to find one; one study Grant et al. (2013) found a high positive correlation between experience and volume in the precuneus (.50); nothing is known about the posterior cingulate cortex.
With regard to direct awareness of bodily sensations, the correlation between meditation experience and volume/density in the insula (if we consider the insula as part of the body-sensation regions) is about .38; the average correlation between meditation experience and volume/density in the somatosensory cortices (one study)34 is .71; the two studies Grant et al. (2013) and Kang et al. (2013) that looked at the supramarginal gyrus found no effects. Nothing is known about the anterior precuneus.
No data are available about global body awareness and the narrative self.
With regard to regions unique to the Fox et al. map, dose–response correlations involving the orbitofrontal cortex average to .04,34 Kang et al. (2013), inferior temporal lobe correlations to .41 Hölzel et al. (2008), Luders et al. (2009) and Luders et al. (2012b), and hippocampus correlations to .14,34 Leung et al. (2013), Luders et al. (2009) and Luders et al. (2013).
Summarized, it seems that there are some positive correlations between experience and volume/density in gray matter, suggesting that meditation practice indeed drives the changes in brain morphology. This link is found for the salience attention network, for direct awareness of bodily sensations, for the inferior temporal lobe and for the hippocampus, but not for the orbitofrontal cortex. The verdict is still out for the executive control attention network, global body awareness, and the narrative self, simply because there are no or very few data. (It is often a cliché to state that more data are needed—here we really need them.)
Overall, these results are grounds for cautious optimism. Note, on the one hand, that the correlations are modest: Meditation experience is clearly not the only factor at play in shaping the brain. On the other hand, experience tends to correlate with age (as you gain experience, you also get older), and—as we have seen—gray matter volume and density decline with age, so finding even modest positive correlations in the face of brain aging is a wonderful result.
The strongest evidence for the point of view that meditation causes brain changes would come from longitudinal studies, in which a group of nonmeditators (the “treatment group or “intervention group”) receive meditation instruction and another group of nonmeditators (the “control group”) does not, and both groups are followed over time.
To date, four such longitudinal studies have been conducted. Two of these31 have looked at MBSR programs, which involve on average around 30 hours of meditation; all measures were related to gray matter. The other two32 have used an even shorter program, devised by their author, called “integrative body-mind training” (IBMT), with a total of 11 hours of meditation. IBMT involves body relaxation, mental imagery, and mindfulness training with musical background as help. Thus the level of expertise developed through these interventions is only a fraction of the level reached by the meditators in the Fox et al. meta-analysis (which was more than 4,500 hours on average). The total number of participants involved is small—98 meditators and 90 nonmeditators, or on average less than 25 per group in each study.
The picture emerging from these studies is mixed.
With regard to gray matter, one study finds no effects; one finds effects in the dorsal anterior insula and the dorsomedial prefrontal cortex; the other finds effects in the left hippocampus, the posterior cingulate cortex (with an increase in density, contrary to the results of the Fox et al. meta-analysis, which found lower density in experienced meditators), the left temporoparietal junction, and two structures in the cerebellum. Thus there is no convergence among these studies, but most of the structures that show growth over the course of the intervention also showed up in the Fox et al. meta-analysis, suggesting that, under certain circumstances, even very short-term meditation experience might lead to growth in brain regions exercised by meditation. This also adds modest (maybe very modest) confidence to the assumption that it is meditation experience that leads to brain changes. But, clearly, the main message is that we need a lot more studies on this topic, and we would probably need to follow meditators over a longer period of time.
With regard to white matter, there is more convergence (but then both studies were conducted in the same lab, using the same meditation program)—both studies observed an increase in white matter in the corpus callosum and the superior longitudinal fasciculus, as Fox et al. also found in their meta-analysis. The two studies also found consistent changes in the corona radiata, a white matter tract associated with attention and control over attention.33 One of these studies also tested for gray matter changes and found none. What intrigues me in this finding—if we combine it with gray matter changes in programs that are a little longer—is the possibility that the change in white matter precedes the change in gray matter. This would also suggest that the changes in white matter might be more crucial than those in gray matter—maybe the impact of meditation on the brain lies more in the honing of pathways, that is, in increasingly efficient information transfer, than in changes in local processing of information. Given what pathways are being honed, it seems that this effect might be primarily about transferring information into the field of attention. Again, two studies is a very small number, and we need more research.
The finding of almost immediate growth in some of these studies underscores how quickly the brain is able to reorganize itself. There is another implication here—a flipside—that may be more unwelcome to you, and that is that this opens up the possibility that unlearning is equally possible. The brain changes in response to meditation experiences are just that—a response, a reaction, a way to help out with the task. If the task is no longer being performed, there is no reason to maintain the brain change. (Brain tissue is expensive; it hogs a lot of resources. Recall the juggling study: The brain regions that grew with training tended to shrink back again [at least halfway] once the skill was no longer used.) In other words: Meditate for a few months and the brain starts to rewire itself; quit meditation for a few months and the brain might well fall back to its baseline premeditation state. This assumption is easy to test; all we need is a scanner, a good amount of money, and two groups of people: Folks who persisted with MBSR (or a similar course) after the program is over and folks who are willing to confess that they did not. Right now, we don’t know.
Most—although possibly not all—of the differences in brain morphology that I have described so far make sense: We can understand them as the brain responding to the demands of meditation. Pay close attention to the sensations in your body and your brain will sharpen the pathway from the representations of the bodily sensations to the attention centers. Let go of the story you build around yourself and your precuneus will shrink a little. And so on. In other words, the changes may correspond to your increasing experience with meditation.
There are also, however, a few studies that show intriguing side effects of the brain changes that come with meditation—positive changes in aspects of behavior that have little to do with the stated goals or demands of meditation, at least as it is typically practiced in a Buddhist context. Here I concentrate on four examples.
The first example, also mentioned in the previous chapter, concerns the brain’s response to pain. As all of us who have ever meditated know, pain is part of the reality of sitting still on a cushion for a long time. But we don’t meditate, of course, to get rid of the pain that meditation causes (that would be a bit, well, circular and unproductive). Still, this seems to be what happens: In the previous chapter, I mentioned that meditation makes (literally) painful experiences less unpleasant. I discussed a number of mechanisms that could explain this effect. One of these is reexamination, which helps dampen the emotional response. We saw that there is some evidence that the salience network is involved in this endeavor. And, of course, the salience network changes with meditation experience, as we saw earlier—the anterior cingulate cortex and the insula grow in volume and/or density.
In one study, Joshua Grant and colleagues34 looked at this meditation–salience network–pain connection directly, in 19 Zen practitioners and 10 nonmeditators. They put a thermode—a device that generates heat—on the left calf of each participant, heated it up, and asked how painful the sensation was, on a scale from zero to 11. They wanted to measure the experience of moderate pain, which they defined as 6 to 7 on that scale. They found that meditators were less sensitive to pain—their threshold for moderate pain was at 50.1°C (122.2°F), that of nonmeditators at 48.1°C (118.6°F).
Looking at brain morphology, they found that the size of the dorsal anterior cingulate showed a correlation with meditation experience—people who practiced meditation longer had a thicker anterior cingulate cortex. Interestingly, they found that pain sensitivity was also related to volume in this brain region, as well as a few others—the dorsal anterior cingulate cortex, the hippocampus, the secondary sensory cortex, and the insula35: thicker cortex, lower sensitivity to pain. What is happening here, it seems, is that the experience of repeatedly checking in on your sensations, thoughts, and emotions during meditation builds a brain that is also well equipped to deal with pain.
The second example concerns personal well-being. You would imagine that quite a number of people meditate to feel better about themselves and their world, but it isn’t something that is typically deliberately practiced during meditation. Omar Singleton and colleagues36 found a direct meditation–brain–behavior relationship, much like the relationship Grant et al. found for pain. In this case, meditation led to higher gray matter density in a few particular regions of the brainstem,37 which in turn correlated with meditators’ self-reports of well-being.38 What is particularly important and exciting is that this study was an intervention study—an eight-week MBSR program—and that the key finding was that brain changes over this relatively brief period correlated with changes in well-being, just about as strong an indicator of causality as you can get with this sort of design.
The third example concerns self-perceived stress. Note that self-perceived stress is not about the daily hassles or important life events that actually happen to you (psychologists call these “stressors” rather than “stress”) but about how you handle these—your feelings, thoughts, and behaviors in relation to those experiences. It is possible to quibble about whether changes in self-perceived stress are a side effect or a direct effect of meditation. One could argue that the effect is direct, that is, intended, because meditation instructions often do emphasize the calming nature of meditation. On the other hand, rarely do they involve asking meditators to handle stressful situations then and there, on the cushion; in fact, most meditation traditions I know of would explicitly discourage analytical thinking about your life issues during meditation time. Either way, in one intervention study39 with a standard eight-week MBSR program, changes in gray matter density in the amygdala were associated with changes in self-perceived stress—larger decreases in volume in the amygdala were associated with larger decreases in perceived stress.
Finally, one study40 found meditation–brain–behavior effects on mood. Like personal well-being, an improvement in mood is something people might expect as a result of their practice, but mood enhancement isn’t explicitly part of the meditation training. After a four-week training in an MBSR-like program, changes in the left sagittal stratum and corona radiata (both are white matter tracts) correlated with changes in mood—people with newly strengthened neuronal connections described themselves as feeling less angry, less confused, less dejected or depressed, and less tired or lethargic. Both pathways have been implicated in depression,41 so it makes sense to expect that changes in their efficiency would lighten mood.
In sum, there are some indications that practicing meditation has an impact on specific brain structures, which in turn leads to positive side effects. In two cases, the side effect aspect seems clear: The practice of one skill (vigilant monitoring) likely leads to brain changes (growth in the anterior cingulate cortex and changes in the corona radiata), which then in turn lead to side effects (lowered sensitivity to pain, improved mood). In the two other cases (stress and well-being), the reason for the brain changes is less obvious, and the observed changes weren’t on Fox et al.’s map.
Again, I end with the cliché that we need more studies. One reason is that none of the studies cited here have been replicated. It would be good if they were, so we can be certain that the results are not some odd, one-time occurrence. (Randomness happens.) Another reason is that what are truly side effects from a strict point of view are often effects that for at least some meditators are the real deal. That is, people who start meditating often do so because they have heard it might help them with stress or might help improve their mood. If these side effects are a natural consequence of the way meditation is practiced, we don’t necessarily need to change anything in our curricula to accommodate meditators with such goals and desires. If, however, stress reduction or mood improvement are not a direct consequence of the type of attention training that is the focus of most meditation programs, it might make some sense to see if we can find a more direct route for those seeking this kind of relief.
The main conclusion from the studies I have reviewed here is that meditation indeed seems to shape the brain. Specifically, meditators show measurable differences in gray matter in areas associated with attention, global and specific body awareness, the quieting of the self, emotion regulation, and a better regulated stress response. They also show stronger interconnections within the brain, back to front and side to side. Meditation even seems to have some anti-aging effects. There are dose–response relationships, suggesting that at least some of these brain changes are due to meditation per se, and there is some preliminary evidence that some of these changes may become visible after only eight weeks of mindfulness training. These changes lead to some desirable side effects—lowered pain sensitivity, heightened well-being, lower stress, better mood.
Throughout this chapter I have indicated further avenues for research. There is a lot we do not know: There is the puzzling finding that differences between meditation orientations do not seem to matter much; we still have a limited grasp on changes in white matter tracts; we know nothing about the sequencing of effects (Do white matter changes precede gray matter changes? Do changes in attention networks precede other changes?); we know nothing about dose–response relationships in the executive control attention network, in the areas associated with global body awareness, and in those associated with the narrative self; and we have little direct and hard evidence that the desirable side effects are indeed direct and inevitable consequences of the attention-related aspects of meditation practice. The presence of publication bias in morphology studies remains troubling, especially given that there is no exact correspondence between the results from cross-sectional, longitudinal, and dose–response relationships. Finally, we know a lot less about how these brain changes inscribe themselves into behavior than we should.
In the next few chapters, I investigate the behavioral changes that come with mindfulness and meditation in more depth. Chapter 5 looks at the intended effect of most types of meditation, namely a sharpening of attention. Chapter 6 takes on the changes in stress, sleep, and well-being, and Chapter 7 investigates the usefulness of mindfulness as therapy or medicine.