Methodological Constraints on Neuroscientifically Investigating Meditation

Meditation is an altered state of consciousness. Its investigation is not only spiritual, but has recently been given a neuroscientific dimension in a number of papers. There are great methodological problems with previous research however. Here I outline a few of these and propose ways for future studies to address and overcome them.

Picture: Dr Richard Davidson of the University of Wisconsin with Buddhist monk Matthieu Ricard. Image link to the University of Wisconsin-Madison. 1 Defining Meditation

While “meditation” might intuitively seem esoteric and empirically impenetrable, the reasons why neuroscientists should care about it are twofold: Firstly, there is highly compelling evidence that it is a genuine, “robustly reproducible” (Lutz,, 2007: 519) phenomenon that causes tremendous changes in brain states. As neglect and abuse can bring about structural brain alterations (Glaser, 2000), more positively, even a two-month “Mindfulness-Based Stress Reduction” course reliably produces changes in prefrontal activation patterns, that are a significant predictor for positive mood and immune response (Rosenkranz, 2003; see Davidson,, 2003). In the attentional blink paradigm, where participants ordinarily fail to detect a second stimulus in a rapid succession of visual information (see Raymond,, 1992), three months of open monitoring meditation shortens the attentional blink (Salgter,, 2007; in Brefczynski-Lewis, 2007: 11483). Consider also binocular rivalry, a bistable visual stimulus where images presented to each eye compete for awareness, in which long-term practitioners (LTPs) of focused attention can achieve significantly longer dominance durations (Carter,, 2005). Unlike voluntary movement, the indirect metabolic modulation from meditation seems to affect a variety of functions typically thought of as outside wilful control. Meditation might hence help us in sketching the limits of self-induced cognitive change as well as the boundaries of metacognition. This means secondly that meditation can aid in generating novel data on neuroplasticity. Considering the itch and scratch, inducing brain change through volition is intuitive. Meditation however should compel us to furthermore look at cognitive functions as learnt skills (McCelland & Rogers, 2003; in Lutz,, 2007: 520), whose changes in the long-term result from experience (Davidson & Lutz, 2007).

How shall we characterise “meditation” for scientific purposes? Many practices intuitively fall under the umbrella term. If we set the criteria coarse enough to include all kinds of self-induced, temporarily altered states of consciousness, then most mystical experiences, ritual African dances (West, 1987), perchance even drug-abuse are meditative. Manipulating a common element in all these is experimentally unmanageable. Candidates for such elements are rare. Consider for instance “hyperarousal” proposed by Fisher (1971), which is grossly inconsistent with Buddhist meditations (Austin, 1998; in Lutz, 2007: 498). From Yoga to Zen, researchers must not assume that the phenomenon studied is always the same. It is extremely important to recognise that different meditative practices aim to manipulate distinct internal processes, hence seem bound to engage different cognitive circuits (Davidson & Lutz, 2007: 176). This essay shall focus mainly on developing Buddhist working definitions since the scriptures offer detailed theories and instructions for meditation that can be scientifically exploited (Lutz,, 2007: 501).

The main Buddhist objective is the elimination of suffering, which entails changing emotional and cognitive states. Two such states are are at the core of all Buddhist practice: 1) śamatha, or mindfulness, awareness and meta-awareness, and 2) vipaśyanā, or insight into the practitioners habits and assumptions (Gethin, 1998). These are understood as two characteristics of a single meditative state, concerning its stability and vividness or intensity (Thrangu & Johnson, 2004). For novices, these two faculties are a trade-off wherein greater attentional stability of the experience leads to “dullness”, while intensity might sacrifice its steadiness. Balancing clarity and stability is the main task for the practitioner (Wangchug Dorjé, 1989; in Lutz,, 2007: 504). The confusingly diverse Buddhist practices differ now in their approach to attaining this balance, e.g. through concentration on breath, releasing distraction, relaxation or tensioning of experience. In this essay we will encounter three Buddhist practices in particular: 1) Focused Attention or “one-pointed concentration” aims to focus clarity on a single object, 2) Non-referential Compassion attempts to induce an emotional state of lovingkindness that is not focused on a specific object (Lutz,, 2007: 509, 515), while 3) Open Monitoring aims to “non-reactively monitor” one’s surroundings and internal states (Davidson & Lutz, 2007: 176).

2 Findings and Limitations in Neuroimaging

This section shall explore some methodological limitations of previous neuroscientific investigation of this ambiguous concept, with particular emphasis on ERP gamma synchrony and fMRI imaging of attention meditation and non-referential compassion under the above framework.

Neural oscillatory synchrony, or common action potential firing patterns, can be used to “tag” shared activity across neural populations (Singer, 1999). Such integration on a large-scale gives us clues about the functional connections across the brain active during a given activity (Schnitzler & Gross, 2005; Lutz, 2004: 16372). While faster gamma rhythms are associated with “neural integration”, consciousness, sensory integration and perception (see Tononi & Edelman, 1998); slower alpha patterns are more linked to attention and working memory (Fries, 2001). Lutz (, 2004) studied LTPs of Tibetan training with experience between 10 000 and 50 000 hours (estimate based on practising years, months in retreats, ect.; mean age=49), engaged in non-referential compassion meditation against a control group of 10 volunteers with one week prior experience in meditation (mean age=21). ERPs were measured before, during and after meditation. There is a difference already in baseline gamma oscillations between groups. Practitioners oscillate more and statistically not because of the mean age gap. This oscillation difference widened during meditation and stayed significantly larger post-meditation in comparison to pre-meditation baseline. Looking at absolute gamma patterns, Lutz found a “common topographical pattern […] across LTPs” in midfrontal and parietotemporal electrodes (16371). These areas next served as regions of interest (ROIs) for a long-distance gamma synchrony analysis, which revealed up to a 30% increase in synchrony as result of meditation in LTPs to levels that according to Lutz have never been recorded in healthy populations. Moreover, the increase was gradual during meditation. They interpreted these results as being consistent with Buddhist phenomenal accounts of the temporal order of meditation sensations.

The list of methodological flaws in that paper is long and varied. There are the more obvious problems associated with EEG, such as “blur” in estimates of synchrony at only 1011 neurones (Lutz,, 2007: 531), as well as confirmed muscle signal pollution (Lutz,, 2004: 16370). Beyond those, there are two more fundamental conceptual limitations that are of importance to us: 1) The expected baseline difference in gamma oscillations between controls and LTPs makes good sampling impossible. Buddhist meditation aims to change the baseline of mental states for followers, which might well be reflected in persistently higher gamma activity even outside meditation. Still, it is just as possible that there is a sampling flaw here in that those individuals with naturally high gamma activity are more proficient meditators and more likely to be asked to participate in studies. Since we hypothesise a group difference pre-meditation, any between subjects design is inherently flawed. 2) The significant differences in activity over lateral frontoparietal electrodes does not warrant many of Lutz’s interpretations. Consider LTPs in Yoga who show increased alpha activation consistent with attribution of attention and inhibition of non-relevant stimuli (Corby,, 1978). This “consistency” claim is a logical fallacy that turns the research into a guessing game that tries to find some accepted cognate theory to link to the meditative evidence. Gamma synchrony increase may be somewhat consistent with Buddhist phenomenal accounts, but is certainly no evidence thereof.

Imaging studies, which at the moment are also more “exploratory than hypothesis-driven” (Lutz,, 2007: 537), have previously suffered from equally grave methodological limitations. The first such study by Newberg (, 2001) employing SPECT to investigate focused attention in eight Buddhist LTPs with no control, revealed increased activity in dorsolateral prefrontal cortex, orbital frontal cortex and thalamus, where there was a significant negative correlation in activity of the superior parietal lobe and dorsolateral frontal areas. This was interpreted as involving some alteration in the participants experience of space. The lack of controls, useful resolution or rigour in interpretation led to more sensible studies as for instance by Brefczynski-Lewis (, 2007). Here Buddhist LTPs (10 000 to 54 000 hrs) and a non-experienced control were asked to meditate attention towards a visual stimulus under fMRI. Both groups exhibited activation in similar areas, namely intraparietal sulci, thalamus, insulae, frontal eye fields, lateral occipital areas and basal ganglia (ibid: 11483). LTPs showed significantly more activation though, especially in the anterior cingulate gyrus. The authors interpret these results as entailing greater error susceptibility and monitoring in the control group, consistent with our framework of stability and clarity. Once more caution is required, especially since these areas are certainly not exclusively involved in those tasks.

Interestingly, LTPs with mean expertise of 19 000 hrs showed greater activation than novices, LTPs of 44 000 hrs showed less activation again, forming an inverse u-shaped function of activation across expertise. Such functions are associated with skill learning, for instance in language acquisition (Davidson & Lutz, 2007: 173). Suspicously, Brefczynski-Lewis do not report any statistics in support of this conclusion. In fact, they did not even bother drawing a graph to supplement this claim, but were content abstracting it from looking at the MR images. Again there are sampling issues and no behavioural task to check for the meditator’s proficiency during the experiment. Nonetheless, the important take-home message of these imaging studies that cannot be talked away methodologically is that activation amplitude varies with practice.

In another experiment, the same authors instructed the same eight Buddhist LPTs as well as age-matched controls to enter a lovingkindness meditation (Brefczynski-Lewis, 2004). In both groups, significant activation occurred at the stratium, anterior insula, somatosensory and anterior cingulate cortices, left-prefrontal areas as well as decreased activation of the right inferior parietal area. This reoccurring pattern again became stronger with expertise of the practitioners. Prima facie, this is consistent with theories about emotion involving structures that monitor homeostasis and internal states, such as somatosensory cortex, cingulate cortex or insula (see Damasio, 1999; in Lutz,, 2004: 541). However, on closer inspection we must question the author’s supposition that since love and compassion necessitate social cognition, brain region activation associated with your own emotional state reflects both observation and imagination of someone else’s emotional state. As with all imaging studies, we must critically evaluate whether this truly is the most supported conclusion given the data. Do LTPs simply have stronger activation in attentional and social cognition modules? The same “consistency” critique must be made, especially since co-activation of insula, cingulate and somatosensory cortices might be consistent with a variety of theories.

3 Methodological Constraints and Future Research

So far, the methodological issues discussed have mainly centred around the absence of good control groups, lack of analysis power and difficulty in transporting empirical evidence into a working model of meditation. Beyond these, there are also wider constrains worth noting. As outlined in section 1, meditation is an extremely diverse set of practices. Whether it aims to relax or arouse, direct attention or distract from sensation, manipulate emotion or indifference, may affect the brain dissimilarly. That in itself is not problematic. However, researcher must start reporting a richer description of what their participants do. To help neuroscientists untangle nuances within practice, Dunne (of Lutz, 2007) proposes a catalogue of questions based on the common elements of most techniques. In 24 queries they aim to categorise the technique with reference to 1) the degree of stability and clarity aimed for, 2) intentional modality or whether there is an object of focus, 3) routines, as for instance reciting mantra or breath control, and finally anticipated changes 4) during and 5) post-meditation (ibid: 517-519). Their account is necessarily unsophisticated in respect to the historical tendencies between traditions and countless nuances in Sanskrit, but gives an easy to report overview for publication.

Another very problematic aspect of studying meditation is the lack of controllability in the proficiency-related independent variable. Traditionally, there are nine levels to mastery of balancing clarity and stability (Wangchug Dorjé, 1989; in Lutz,, 2007: 510), which are however useless for novice and secular samples. While Brefczynski-Lewis (, 2007: 11487) were for some reason content that pupil dilation showed that all subjects “were engaged in the task”, most research uses the subjective phenomenal report of meditators to see if all subjects were in fact doing the same thing at similar levels of effectiveness. This means that we have no quantitative measure for meditation. While for novices this is very problematic, one can argue that since LTPs can better modulate attention and have a heightened sense of awareness, they also have more refined and enriched first-person reports. This essay will not judge whether this leap of faith is justified. Still, unfortunately self-report is an inherent limitation of studying states of consciousness, which at our stage of research cannot be overcome.

Nonetheless, let us finish with a proposal for possible future research. If we pair the proposition that training and enhanced cognitive function leads to lesser thickness in associated areas (Kanai & Rees, 2011) with studies that found differences in grey matter density and cortical thickness in LTPs (see Luders 2009 vs. Lazar 2005) as well as our findings of altered functional connectivity reflected in gamma synchrony, then we should expect also some alteration in structural white matter connections. This can be investigated with diffusion tensor imaging, which evaluates the structural integrity of white matter connections between ROIs. Given the previous discussion in this essay, ROIs should include ventromedial and lateral prefrontal cortex as well as parietal areas. Beyond proposing an alternate technique of measurement, the single greatest methodological improvement would come from a longitudinal design. That brain area x is more active in LTPs during meditation is not as interesting as the question of how this modulation occurs. To overcome our sampling problem, the control and experimental group should both be novices, some of which are trained over the course of a year. Dunne’s catalogue of questions should always be used to trace differences in technique between individuals. Alas, LTPs have the advantage of showing brain alterations over time that are structurally coarse enough to be picked up and it is questionable whether novices would be proficient enough to produce measurable brain modulations.

Quintessentially, meditation research might turn out to be very much like studying dreams. By now we have some altered EEG and MR patterns, but know little beyond that. We want to do more research on the borderline between meditation and normal wakefulness. For novices, we have no good account of what this altered state of consciousness is supposed to be. Cross-traditional meditation techniques cannot be distinguished by us at any high level of sophistication. Reliance on first-person reports and lack of independent variables make this venture exceedingly difficult. However, the compelling evidence that LTPs can potentially structurally alter their brain by sheer will power makes meditation an irresistible field for research on neuroplasticity and I hope to have argued that there exist relatively simple steps to be implemented by researchers that go in that direction.

References Austin, J. H. (1998). Zen and the brain: Toward an understanding of meditation and consciousness. (2nd ed.). Cambridge: MIT Press. Brefczynski-Lewis, J. A., Lutz, A., & Davidson, R. J. (2004). A neural correlate of attentional expertise in long-time Buddhist practitioners. (Report no. 715.8.). San Diego: Society for Neuroscience. Brefczynski-Lewis, J. A., Lutz, A., Schaefer, H. S., Levinson, D. B., & Davidson, R. J. (2007). Neural correlates of attentional expertise in long-term meditation practitioners. PNAS 104 (27): 11483–11488. Carter, O. L., Presti, D. E., Callistemon, C., Ungerer, Y., Liu, G. B., & Pettigrew, J. D. (2005). Meditation alters perceptual rivalry in Tibetan Buddhist monks. Current Biology 15(11): 412–413. Corby, J. C., Roth, W. T., Zarcone, V. P., Jr., & Kopell, B. S. (1978). Psychophysiological correlates of the practice of Tantric Yoga meditation. Archives of General Psychiatry 35: 571–577. Damasio, A. R. (1999). The feeling of what happens: Body and emotion in the making of consciousness. New York: Harcourt Brace. Davidson, J. R., & Lutz, A. (2007). Buddha’s brain: neuroplasticity and meditation. IEEE Signal Processing Magazine 10: 172–176. Fischer, R. (1971). A cartography of the ecstatic and meditative states. Science 174(12): 897–904. Fries, P., Reynolds, J. H., Rorie, A. E., & Desimone, R. (2001). Modulation of oscillatory neuronal synchronization by selective visual attention. Science 291(5508): 1560–1563. Gethin, R. (1998). The foundations of Buddhism. Oxford: Oxford University Press. Glaser, D. (2000). Child abuse and neglect and the brain – A review. Journal of Child Psychology, Psychiatry & Allied Disciplines 41(1): 97–116. Kanai, R., Rees, G. (2011). The structural basis of inter-individual differences in human behaviour and cognition. Nature Reviews Neuroscience 12(4): 231–242. Lazar, S. W., Kerr, C. E., Wasserman, R. H., Gray, J. R., Greve, D. N., Treadway, M. T., McGarvey, M., Quinn, B. T., Dusek, J. A., Benson, H., Rauch, S. L., Moore, C. L., & Fischl, B. (2005). Meditation experience is associated with increased cortical thickness. Neuroreport 16(17): 1893–1897. Luders, E., Toga, A. W., Lepore, N., & Gaser, C. (2009). The underlying anatomical correlates of long-term meditation: larger hippocampal and frontal volumes of gray matter. Neuroimage 45(3): 672–678. Lutz, A., Greischar, L. L., Rawlings, N. B., Ricard, M., & Davidson, R. J. (2004). Long-term meditators self-induce high amplitude gamma synchrony during mental practice. PNAS 101(46): 16369–16373. Lutz, A., Dunne, J. D., & Davidson, R. J. (2007). Meditation and the neuroscience of consciousness: An Introduction. In Zelazo, P. D., Moscovitch, M., & Thompson, E. (Eds.). The Cambridge Handbook of Consciousness. Cambridge: Cambridge University Press. McClelland, J. L., & Rogers, T. T. (2003). The parallel distributed processing approach to semantic cognition. Nature Reviews Neuroscience 4(4): 310–322. Newberg, A., Alavi, A., Baime, M., Pourdehnad, M., Santanna, J., & d’Aquili, E. (2001). The measurement of regional cerebral blood flow during the complex cognitive task of meditation: A preliminary SPECT study. Psychiatry Research 106(2): 113–122. Raymond, J. E., Shapiro, K.L., Arnell, K.M. (1992). Temporary suppression of visual processing in an RSVP task: An attentional blink?. Journal of Experimental Psychology: Human Perception and Performance 18: 849–860. Schnitzler, A., & Gross, J. (2005). Normal and pathological oscillatory communication in the brain. Nature Review Neuroscience 6(4): 285–296. Singer, W. (2001). Consciousness and the binding problem. Annals of the New York Academy of Sciences 929: 123–146. H.A. Slagter, H.A., Lutz, A., Greischar, L. L., Francis, A. D., Nieuwenhuis, S., Davis, J.M & Davidson, R. J. (2007). Mental training affects use of limited brain resources. PLOS Biology 5(6): 1380001–1380008. Thrangu, K. & Johnson, C. (2004). Essentials of Mahamudra: Looking directly at the mind. Boston: Wisdom Publications. Tononi, G., & Edelman, G. M. (1998). Consciousness and complexity. Science 282(5395): 1846–1851. Wangchug Dorjé. (1989). The Mahamudra eliminating the darkness of ignorance (3rd ed.). Dharamsala: Library of Tibetan Works and Archives. West, M. A. (Ed.). (1987). The psychology of meditation. New York: Clarendon Press.