Introduction

Meditation has been a spiritual and healing practice for more than 5000 years. The word ‘meditation’ derives from Latin ‘meditari’, which means ‘to engage in contemplation or reflection’ and can be defined as both a process and a state. According to the Yoga Sutras, meditation is the act of inward contemplation and the intermediate state between mere attention to an object and complete fusion within it Reference Taylor, Murphy, Donovan and Sausalito(1). The practice of meditation has been related to significant improvements of health Reference Ospina, Bond and Karkhaneh(2) and of mental abilities Reference Barinaga(3). Among the great variety of existing meditation practices, a growing attention has recently been given to a particular subgroup of meditation practices named ‘Mindfulness meditations' (MM) Reference Chiesa and Serretti(4).

Mindfulness is currently defined in psychological terms characterised by paying total attention to the present moment with a non-judgmental awareness of the inner and outer experience Reference Kabat-Zinn(5,Reference Marlatt, Kristeller and Miller6). Although the concept of Mindfulness has its origins in many contemplative and philosophical traditions such as Hinduism and Buddhism Reference Buddhaghosa(7), its practice does not imply to follow any specific philosophical or religious tradition Reference Kabat-Zinn(8). Yet, MM can be considered a component of such ancient practices as Vipassana Reference Ahir(9) and Zen meditation Reference Mizuno(10). In addition, clinically oriented programs designed to organise the original concept of Mindfulness into standardised courses for health, such as Mindfulness based Stress Reduction (MBSR) Reference Kabat-Zinn(11) and Mindfulness based Cognitive Therapy (MBCT) Reference Segal, Williams and Teasdale(12) can be considered examples of MM as well.

To date, MM were found to be useful for a great variety of physical and psychiatric disorders as well as for healthy people. In particular, MBSR could be useful for several physical and psychiatric disorders (Reference Bishop13–Reference Allen, Chambers and Knight15) as well as for healthy people Reference Chiesa and Serretti(16) and MBCT could be able to prevent depression relapses and reduce residual depressive symptoms Reference Coelho, Canter and Ernst(17). In addition, Vipassana meditation could be efficacious for the reduction of alcohol and substance abuse in prisoners Reference Chiesa(18) and Zen meditation could reduce blood pressure Reference Chiesa(19). It is noteworthy, however, that the majority of current findings does not allow to distinguish specific from non-specific effects of MM Reference Chiesa and Serretti(4), implying the necessity for further better designed studies.

Both Vipassana and Zen meditation from one hand and MBSR and MBCT from the other include MM as an active feature of their practice, although with several differences. In any case, the four meditations mentioned above can be grouped together in accordance with the current theoretical framework of meditation that considers such meditations as belonging to the pole of Mindfulness and distinguishes them from other types of meditation practices (Reference Ospina, Bond and Karkhaneh2,Reference Cahn and Polich20,Reference Goleman21).

Thanks to the improvements of neuroimaging technologies, there has been in the last decade a wide spread of scientific investigations about the neurobiological correlates of different types of meditations, including MM, by means of different techniques such as Magnetic Resonance Imaging (MRI), functional MRI (fMRI), 18-fluorodeoxyglucose Positron Emission Tomography (PET) and 99mtechnetium hexamethylpropyleneamineoxime Single Photon Emission Computed Tomography (SPECT). Such investigations are particularly important because they provide preliminary evidence about the neurobiological modifications related to meditation practice and they answer the call for more rigorous neurobiological approaches to meditation research.

Findings about the neural correlates of a broad range of meditation practices have already been reviewed elsewhere Reference Cahn and Polich(20,Reference Rubia22) and are not the focus of the present review. Rather, the aim of the present work is to review available findings about the functional neural correlates of MM and compare them with the ones of other types of treatments such as antidepressants, psychotherapy and placebo.

Critical issues related to methodology in neuroimaging studies

Before reviewing neuroimaging findings, there are a number of issues that should be carefully considered in the interpretation of data. First of all, the functional neuroimaging modalities that have been used in meditation, pharmacotherapy, psychotherapy and placebo research vary substantially. They include fMRI, PET and SPECT. These imaging modalities differ with respect to mechanism, image resolution and patient-related limitations Reference Dougherty, Rauch and Rosenbaum(23). Also, fMRI has traditionally been used to probe the brain activation patterns during perceptual or cognitive tasks, whereas PET and SPECT have been usually used to measure baseline brain metabolism. As a consequence, while some studies are performed in subjects during a resting state, others are performed during the execution of a specific task, hence investigating different correlates of psychopathology.

Second, these techniques provide indexes of brain activity by measuring glucose metabolism or cerebral blood flow (CBF). Note, however, that the relationship between glucose metabolism or CBF and neuronal activity remains controversial Reference Giove, Mangia and Bianciardi(24), although very often the term ‘brain activity’ is used interchangeably with ‘metabolic activity’ or ‘haemodynamic activity’.

A further issue that has to be taken into account into the interpretation of reviewed data is that there are many specific differences in rationale, technique and efficacy of the various modalities related to different treatments, particularly psychotherapies, in use today. Such issue is particularly important for placebo treatments as well because recent research showed that placebo effect is not a unitary phenomenon but that there are many types of placebo responses driven by different mechanisms depending on the particular context in which the placebo is given Reference Price, Finniss and Benedetti(25,Reference Faria, Fredrikson and Furmark26). Methodological inconsistencies created by the use of single versus multiple therapists, differing numbers of sessions, and further issues (such as, for instance, individual versus group therapy) should also be considered when comparing studies.

Finally, a wide range of patients' populations are included. While MM and placebo studies were mainly performed in healthy subjects, studies about psychotherapy and pharmacotherapy were exclusively performed in samples of patients suffering from mood and anxiety disorders.

Methods

A literature search was undertaken using MEDLINE, ISI web of knowledge, the Cochrane database and references of retrieved articles. The search included articles indexed in the above mentioned web-based electronic bibliographic databases, published up to August 2009. The search strategy considered only studies published in English. The main search terms were the following: Mindfulness meditation, Vipassana meditation, Zen meditation, MBSR, MBCT, psychotherapy, pharmacotherapy and placebo in combination with functional neuroimaging, fMRI, PET and SPECT.

Studies which focused on the functional neural correlates of MM, psychotherapy, pharmacotherapy and placebo were screened by the reviewers in order to be considered for the inclusion. To be included, studies had to specify the used functional neuroimaging technique, the type of practice under examination and which kind of task was used (or its absence). Pertaining to psychotherapy and pharmacotherapy treatments, they had to be well established interventions for the specific disease under investigation. Excluded were: structural neuroimaging studies, studies not reporting details mentioned above, and reviews of the literature. No restriction was used with respect to patient population (e.g. healthy vs. non-healthy subjects), study design (e.g. uncontrolled vs. controlled study) and type of practice/technique under investigation (e.g. psychoanalysis vs. cognitive behavioural therapy (CBT)).

The findings related to psychotherapy focused on approved and well established psychotherapeutic interventions such as CBT and BT for mood and anxiety disorders (Reference Rupke, Blecke and Renfrow27–Reference Choy, Fyer and Lipsitz32), interpersonal therapy (IPT) for major depression (MD) Reference De Mello, De Jesus Mari, Bacaltchuk, Verdeli and Neugebauer(33) and eye movement desensitisation therapy (EMDR) for post-traumatic stress disorder Reference Shepherd, Stein and Milne(34). The focus of the present review with respect to pharmacotherapy regards well established treatments options for MD and anxiety disorders including selective serotonine reuptake inhibitors and serotonine and noradrenaline reuptake inhibitors Reference Cipriani, Furukawa and Salanti(35,Reference Zohar and Westenberg36).

Finally, the choice of including neuroimaging findings about the placebo effect was motivated by the decision to review commonalities and differences between brain areas activated by non-specific psychological mechanisms related to placebo such as expectancy of a benefit and classical conditioning which are supposed to be common to all treatments Reference Price, Finniss and Benedetti(25,Reference Benedetti37) and those activated by the so-called ‘specific treatments' which are supposed to include some specific ‘active ingredient' in addition to non-specific mechanisms. This issue is particularly important on account of the recent advances in our understanding of the placebo effect which shifted over the last decades from the impossibility of an inert substance to produce a clinical benefit to the meaning that the patient gives to such substance which is able to produce a measurable effect Reference Pollo and Benedetti(38).

The exposure of available findings will mainly focus on three topic brain areas: the prefrontal cortex (PFC), the anterior cingulate cortex (ACC) and the amygdala. Other brain areas are well involved in all such treatments; however, we selected these three areas with the aim to suggest that all investigated treatments can affect brain activity by means of a regulation of these areas, although through different modalities.

Results

Mindfulness meditations

Functional neuroimaging studies about MM showed a consistent activation of the PFC and ACC during meditation (Reference Baerentsen, Hartvig, Stodkilde-Jorgensen and Mammen39–Reference Kozasa, Radvany, Barreiros, Leite and Amaro42), although with some differences across the studies (Table 1). In an early study, Baerentsen et al. Reference Baerentsen, Hartvig, Stodkilde-Jorgensen and Mammen(39) reported the activation of the PFC and ACC during the initiation of Zen meditation. Of note, their observations are in accordance with findings reviewed by Newberg and Iversen Reference Newberg and Iversen(81) which suggested that, during meditation, distracting external events as well as memories represent conflicting events to task goals and that such brain areas, particularly the ACC, may contribute to the maintenance of attention by alerting those systems directly implementing top-down regulation to resolve this conflict Reference Van Veen and Carter(82).

Table 1 Functional neuroimaging studies

L and R before other words mean left and right, respectively. dlPFC, dorsolateral prefrontal cortex; EF, experiential (mindfulness) focus; IBS, irritable bowel syndrome; mPFC, medial prefrontal cortex; NF, narrative focus; ofc, orbitofrontal prefrontal cortex; PAG, peri-acqueductal grey; PCC, posterior cingulate cortex; PD, panic disorder; PTSD, post-traumatic stress disorder; vlPFC, ventrolateral prefrontal cortex; vmPFC, ventromedial prefrontal cortex.

In a following study performed on Vipassana meditators during a period of mindful attention to their breath Reference Holzel, Ott and Hempel(40), the authors observed an increased activation of bilateral ACC, in particular, its rostral portion, and of dorsal medial PFC [medial PFC (mPFC)]. The rostral/ventral division of the ACC and the mPFC are activated in emotional processing Reference Bush, Luu and Posner(83,Reference Phan, Wager, Taylor and Liberzon84) and it could be hypothesised that MM training leads to more cortical processing of emotional conflicts and processes during meditation, possibly reflecting an improved ability for emotional regulation. Interestingly, such explanation is consistent with the observation that subjects with higher dispositional mindfulness show an increased activation of the mPFC during the execution of an affective but not of a cognitive task Reference Creswell, Way, Eisenberger and Lieberman(85).

The relationship existing between these findings can be better understood if one considers that mindfulness may be conceived as a trait like quality that varies among different individuals Reference Baer, Smith, Hopkins, Krietemeyer and Toney(86,Reference Brown and Ryan87) and that MM practice has been found to increase mindfulness levels (Reference Lau, Bishop and Segal88–Reference Carmody, Reed, Kristeller and Merriam91). As a consequence, it could be argued that both subjects with higher dispositional mindfulness levels and those who increased their mindfulness levels through the practice of a MM may more easily engage the mPFC and the ACC during tasks that require emotional regulation.

At the opposite of the previous findings, however, Farb et al. Reference Farb, Segal and Mayberg(41) observed a decrease rather than an increase of the activity of the mPFC during meditation. More in detail, the authors observed an increased activation of the mPFC as well as of left language areas during a narrative focus in which subjects had to figure out what some words meant to them, and a decrease of the same areas during mindful attention to present thoughts and stimuli. It is worthwhile, however, that significant differences between these two trials Reference Holzel, Ott and Hempel(40,Reference Farb, Segal and Mayberg41) including expert versus novice meditators, Vipassana versus MBSR meditators as well as the different focus during meditation (the breath vs. mindfulness of thoughts and experiences while observing words) could explain why different findings have been observed.

Despite this finding, significant activations in the PFC were observed both in meditators and in non-meditators during the experiential (mindfulness) focus, even though with specific differences between the two groups Reference Farb, Segal and Mayberg(41). During the experiential focus non-meditators showed left dorsolateral PFC (dlPFC) activation, whereas meditators showed an activation of the right dlPFC and inferolateral PFC (ilPFC). Consistent with a dual-mode hypothesis of self-awareness, these results suggested a fundamental neural dissociation in modes of self-representation that support distinct, but usually integrated, aspects of self-reference: the higher order self-reference characterised by neural processes supporting awareness of a self that extends across time (related to left lPFC) and more basic momentary self-reference characterised by neural changes supporting awareness of the psychological present (related to right lPFC) Reference Farb, Segal and Mayberg(41). Interestingly, many authors suggested that the latter could represent a return to the neural origins of identity, in which self-awareness in every moment arises from the integration of basic interoceptive and exteroceptive bodily sensory processes (Reference Damasio92–Reference Panksepp94).

Further, the activation of the right dlPFC, along with other areas, has also been implicated in the switching between first and third person perspectives Reference Ruby and Decety(95) and was associated to a more detached observation state of events characterised by lower emotional reactivity related to reduced amygdala activation Reference Ochsner, Bunge, Gross and Gabrieli(96,Reference Kalisch, Wiech and Critchley97). Consonant with such findings, decreased activity in the left amygdala was observed in meditators during mindfulness focus in the study performed by Farb et al. Reference Farb, Segal and Mayberg(41). Note also that a similar pattern of activation of the right dlPFC was observed by Kazosa et al. in a independent sample of Zen meditators as well Reference Kozasa, Radvany, Barreiros, Leite and Amaro(42).

A further important finding was reported by Pagnoni et al. Reference Pagnoni, Cekic and Guo(43) about the neural correlates of conceptual processing during Zen meditation. In such study the authors observed that, in comparison to a control group, Zen meditators displayed a reduced duration of the neural response linked to conceptual processing in regions of the default network Reference Mason, Norton, Van Horn, Wegner, Grafton and Macrae(98) such as the angular gyrus and the superior frontal gyrus, suggesting that meditative training could foster the ability to control the automatic cascade of semantic associations triggered by a stimulus and, by extension, to voluntarily regulate the flow of spontaneous mentation. This result is particularly important if one considers that Farb et al. Reference Farb, Segal and Mayberg(41) reported that non-meditators showed a significant activation of left language areas in the PFC, even when they tried to focus their attention on present experience, suggesting the impossibility to dampen their spontaneous mentation. On the other hand, such reduction could be particularly useful for many psychiatric conditions characterised by excessive ruminations Reference Brosschot, Gerin and Thayer(99) such as obsessive compulsive disorder Reference Clayton, Richards and Edwards(100), anxiety disorders Reference Hofmann(101), alcohol dependence Reference Borders, Barnwell and Earleywine(102) and MD Reference Teasdale, Segal and Williams(103,Reference Siegle, Steinhauer, Thase, Stenger and Carter104). Interestingly, several findings suggested that MM could be useful for all such disorders (Reference Chiesa and Serretti4,Reference Coelho, Canter and Ernst17,Reference Kuyken, Byford and Taylor105–Reference Fairfax109).

Overall, these findings suggest that, consistently with historical accounts of MM which underscore that mindfulness training allows individuals to treat affective states as ‘objects’ of attention Reference Analayo(110,Reference Kapleau111), MM practice could be related to a more flexible emotional regulation and to a higher ability to detach from negative states by engaging frontal cortical structures to dampen automatic amygdala activation.

The comparison with other specific and placebo treatments

Current evidence suggests that both anxiety and depressive states could be considered as the result of increased and unmodulated amygdala activity (Reference Siegle, Steinhauer, Thase, Stenger and Carter104,Reference Abercrombie, Schaefer and Larson112–Reference Eren, Tukel, Polat, Karaman and Unal115), of a deficit of inhibitory connections to the amygdala meditated by cortical regions such as the orbitofrontal PFC (OFC), mPFC, dlPFC and ACC (Reference Ochsner, Bunge, Gross and Gabrieli96,Reference Davidson116–Reference Davidson118) or of both (Reference Eren, Tukel, Polat, Karaman and Unal115,Reference Siegle, Thompson, Carter, Steinhauer and Thase119,Reference Bremner120). Consistent with this model of mood and anxiety disorders, several studies found that both psychotherapy and pharmacotherapy were able to reverse baseline patterns of activation into two distinct ways. They were found to induce either higher activity of prefrontal areas such as the inferior frontal gyrus Reference Paquette, Levesque and Mensour(57,Reference Lansing, Amen, Hanks and Rudy63), the medial OFC Reference Schienle, Schafer, Hermann, Rohrmann and Vaitl(60,Reference Pagani, Hogberg and Salmaso64), the ventral PFC Reference Brody, Saxena and Silverman(44,Reference Kennedy, Evans and Kruger46), the dlPFC (Reference Mayberg, Brannan and Tekell45,Reference Kennedy, Evans and Kruger46,Reference Goldapple, Segal and Garson51) and the ACC (Reference Goldapple, Segal and Garson51,Reference Fu, Williams and Cleare55,Reference Felmingham, Kemp and Williams65), or lower activation of the amygdala (Reference Fu, Williams and Cleare55,Reference Furmark, Tillfors and Marteinsdottir56,Reference Schienle, Schafer, Hermann, Rohrmann and Vaitl60,Reference Felmingham, Kemp and Williams65) (Table 1). The majority of these studies used a fMRI technique and probed the brain activation patterns during perceptual or cognitive/affective tasks, hence allowing a higher comparability to studies performed on mindfulness meditators. Of note, one can observe that a large overlaps exists between findings in MM studies and those reported in pharmacotherapy and psychotherapy studies.

On the other hand, different and often contrasting results were observed in studies focusing on a resting state in depressed patients and to a lesser extent in other populations of patients. In particular, apart from some studies where an increased activation in some areas of the PFC was observed at rest following a successful treatment with antidepressants (Reference Brody, Saxena and Silverman44–Reference Kennedy, Evans and Kruger46,Reference Goldapple, Segal and Garson51), the majority of studies, especially those focusing on psychotherapy, observed decreases in many areas of the PFC including the vlPFC, vmPFC and dlPFC, and of the ACC (Reference Brody, Saxena and Silverman44–Reference Brody, Saxena and Stoessel47,Reference Fu, Williams and Cleare50,Reference Goldapple, Segal and Garson51,Reference Kennedy, Konarski and Segal54,Reference Paquette, Levesque and Mensour57–Reference Schienle, Schafer, Hermann, Rohrmann and Vaitl60,Reference Lindauer, Gersons and Van Meijel62). The apparent contrast between these findings and those suggesting that psychotherapy could allow greater frontal activation during emotional tasks could be resolved if one considers that recovery could involve the lowering of tonic resting-state activity, to allow for greater reactivity when executive control is recruited. However this hypothesis has not been verified yet by rigorous neuroimaging studies comparing brain activity at rest and during the execution of a specific task within the context of a single experimental design.

In addition, there is preliminary evidence suggesting that, while psychotherapy could allow a resetting of tonic prefrontal activity and yield greater ability for ‘top-down’ emotional regulation when it is needed (e.g. when phobic patients are exposed to phobic stimuli), drugs could increase ACC metabolism tonically Reference Brody, Saxena, Mandelkern, Fairbanks, Ho and Baxter(48,Reference Goldapple, Segal and Garson51). This, in turn, could create a ‘bottom-up’ effect whereby limbic regions are tonically inhibited during antidepressants administration.

On the basis of reviewed findings, we could preliminary suggest that MM could be more similar to psychotherapy, i.e. they could regulate frontal brain areas to dampen amygdala hyperreactivity specifically when it is necessary, whereas drugs could target amygdala directly. Then, all such treatments could result in a final common state characterised by lower amygdala and prefrontal activity at rest with the latter being able to reverse after MM or psychotherapy when it is required (Fig. 1). Nevertheless, such hypothesis should be verified by neuroimaging studies comparing MM to other interventions for specific disorders and assessing both resting and task related activation at different time points.

Fig. 1 Red/clear=hypometabolic area; blue/dark=hypermetabolic area; PFC=prefontal cortex.

Pertaining to the placebo effect, it is noteworthy that almost all studies, independently from the underlying condition, showed a consistent activation of the dlPFC (Reference Wager, Rilling and Smith67–Reference Kong, Gollub and Rosman72,Reference Craggs, Price, Perlstein, Nicholas Verne and Robinson78,Reference Mayberg, Silva and Brannan79), the OFC (Reference Petrovic, Kalso, Petersson and Ingvar66–Reference Petrovic, Dietrich, Fransson, Andersson, Carlsson and Ingvar68,Reference Wager, Scott and Zubieta75) and the rostral/ventral ACC (Reference Petrovic, Kalso, Petersson and Ingvar66–Reference Bingel, Lorenz, Schoell, Weiller and Buchel73,Reference Wager, Scott and Zubieta75,Reference Mayberg, Silva and Brannan79) during the placebo effect and the anticipation of a negative stimulus. Given that placebo does not exert a therapeutic effect by itself, it could be argued that the expectation of a benefit is able to activate such areas when needed. Furthermore, the notion that similar findings were observed across a great variety of conditions including pain, vision of unpleasant pictures, MD and social phobia suggests that modulatory processes in placebo are not specific for any single condition but are rather a part of the mechanisms involved in the regulation of emotional processes in general Reference Faria, Fredrikson and Furmark(26). Consistent with this hypothesis is also the observation that amygdala reactivity can be dampened during the anticipation of a noxious stimulus and in phobic patients after the administration of a placebo Reference Wager, Scott and Zubieta(75,Reference Furmark, Appel and Henningsson80).

Overall, these findings suggest that placebo responses seem mediated by ‘top-down’ processes dependent on frontal cortical areas that generate and maintain cognitive expectancies, hence suggesting a stronger similarity with MM and psychotherapy rather than with antidepressants. Nevertheless, we should remember that placebo is a phenomenon common to all treatments Reference Price, Finniss and Benedetti(25). As a consequence, it is difficult to distinguish the neural correlates of the specific effects of active treatments from the non-specific placebo-like effects (such as the expectancy of a benefit) common to all treatments. This difficulty is particularly evident in the study performed by Mayberg et al. in MD patients where an almost complete overlap between areas activated during the administration of placebo and of fluoxetine was observed Reference Mayberg, Silva and Brannan(79).

However, while studies performed in meditators and in subjects who underwent a pharmacological or a psychotherapeutic intervention were referred to long-term meditators or to a long-term treatment, respectively (for instance, a 2-months antidepressant treatment), placebo studies often referred to a short period of time (e.g. minutes). As a consequence, while it could be argued that the former observations could be relatively enduring, there is not sufficient evidence to understand to what extent placebo treatments could result in long lasting brain activation patterns.

Discussion

The observations reviewed in the present work suggest that a great overlap could exist between cerebral areas activated during MM, psychotherapy, pharmacotherapy and those activated during the placebo effect, although with some differences. It could be preliminary suggested that while MM, psychotherapy and placebo act through a top-down regulation, antidepressants could act through a bottom-up process. In particular, MM, psychotherapy and placebo could target the PFC which in turn could exert an inhibitory effect on limbic regions such as the amygdala when it is needed (Reference Ochsner, Bunge, Gross and Gabrieli96,Reference Mayberg, Liotti and Brannan121,Reference Ochsner, Ray and Cooper122), possibly through inhibitory connections from the OFC and ventromedial regions (including the rostral and subgenual cingulate gyri) to the amygdala Reference Ray and Price(123,Reference Ghashghaei and Barbas124). On the other hand antidepressants could target amygdala directly modulating specific emotional information processing mediated by the amygdala such as repetitive negative thinking in depressed patients Reference Derubeis, Siegle and Hollon(28). Note also that structural neuroimaging studies in long-term Vipassana and Zen meditators showed that MM practice could be related to enduring macroscopic modifications of brain areas including the PFC Reference Lazar, Kerr and Wasserman(125), the anterior insula Reference Lazar, Kerr and Wasserman(125,Reference Holzel, Ott and Gard126) and the putamen Reference Pagnoni and Cekic(127). Further, there is a partial overlap between thicker brain areas in long-term meditators and brain areas activated in functional neuroimaging studies (Reference Farb, Segal and Mayberg41–Reference Pagnoni, Cekic and Guo43) and it could be argued that the repeated activation of specific brain areas could lead to long-term macroscopic brain modifications. Notably, such explanation is consistent with observations that when a task that requires attention is consistently directed towards a behaviourally relevant sensory stimulus, robust changes in sensory cortical maps are observed (Reference Recanzone, Merzenich, Jenkins, Grajski and Dinse128–Reference Bangert and Altenmuller130) and that the entity of changes in specific brain areas seems to be related to meditation experience Reference Lazar, Kerr and Wasserman(125).

Despite the preliminary nature of these findings, it is noteworthy that they could have an important clinical impact. First of all, they preliminary support a neurobiological basis for the usefulness of MM in several clinical conditions that are only partial responsive to current treatments options such as anxiety and mood disorders. Even though no specific neuroimaging study on MM has yet been performed in clinical samples, it is noteworthy that several clinical trials about MM have already provided preliminary evidence about their efficacy on all such conditions (Reference Bishop13,Reference Coelho, Canter and Ernst17,Reference Baer131).

In addition, our findings underscore that the mechanisms underlying MM could share some similarities with other treatments, in particular psychotherapy, in that they engage frontal cortical structures to dampen exaggerated limbic responses to negative stimuli when it is needed, although specific neuroimaging studies of MM addressing such issue in psychiatric populations of patients are still lacking. Finally, our findings point out to the necessity of a more deep investigation of the comparison between the neurobiological modifications induced by specific treatments and those induced by non-specific treatments or placebos. Such investigation could allow to more deeply dissect the overlapping specific and non-specific effects of the so-called ‘specific treatments' in general and could be helpful, along with clinical trials designed to compare MM to active non-specific treatments such as educational support groups, in distinguishing specific from non-specific effects of MM in particular.

Other limitations have to be considered in the interpretation of our findings. First of all, the fact that studies about MM have been performed in healthy subjects could raise concerns about their applicability to psychiatric populations of patients. However, as reported above, preliminary data suggest that MM could be useful for disorders such as MD and anxiety disorders, and neuroimaging data on these populations of patients are warranted. Second, a great number of interventions and of active drugs with different rationales and mechanisms of actions were included. However, on account of the preliminary nature of the present review, this could represent a strength rather than a limitation in that we tried to compare MM to a large number of treatments in order to open new directions for future comparative research.

Also, we mainly focused on three brain areas while it is known that further cerebral areas are involved in the response to meditation, psychotherapy, pharmacotherapy and placebo (Reference Cahn and Polich20,Reference Faria, Fredrikson and Furmark26,Reference Linden132). Nonetheless, the aim of the present work was only to suggest a possible model of the mechanisms through which MM could be effective and to review possible similarities of our model with other treatment options. As a consequence we also excluded those interventions or conditions which main targets were not the PFC, the ACC or the amygdala such as Parkinson's disease and schizophrenia Reference Faria, Fredrikson and Furmark(26,Reference Roffman, Marci, Glick, Dougherty and Rauch133). Further, while we generally referred to MM as a unitary group of meditative practices, it is worthwhile that specific differences exist across different MM Reference Ivanovski and Malhi(134). As a consequence, while available studies mainly focused on long-term Vipassana and Zen meditators, further research is needed to investigate the neural correlates of MBSR and MBCT practice.

A final concern could derive from the decision to use neuroimaging findings to explain the correlates of normal physiology or psychopathology of the individual. Note, however, that neuroimaging studies could suggest possible brain areas where future studies could assess potential changes in neurotransmitter metabolites to determine which system is likely to be involved. Concurrently, molecular imaging will be needed to determine where exactly these changes take place and further research should aim at integrating functional imaging with molecular techniques such as radioligand imaging or biochemical analysis of metabolites in order to elucidate the molecular mechanisms of MM and its commonalities and differences with psychotherapy, pharmacotherapy and placebo.

In conclusion, our findings suggest that MM practice could allow a more flexible emotional regulation and a higher ability to detach from negative states by engaging frontal cortical structures to dampen automatic amygdala activation. Also, a great overlap could exists between cerebral areas activated during MM, psychotherapy, pharmacotherapy and those activated by expectancy in placebo studies. However, it could be preliminary suggested that while MM, psychotherapy and placebo act through a top-down regulation, antidepressants could act through a bottom-up process. Accordingly, further research is needed to replicate findings observed in healthy subjects in clinical populations of patients, to compare MM with other specific and non-specific treatments and to better investigate the neurochemical correlates of MM.