Keywords

Introduction

Mindfulness training (MT) represents a set of contemplative practices aimed at restructuring one’s relationship with immediate experience to promote well-being. In recent years, MT has grown rapidly in popularity, buoyed by its increasingly validated clinical efficacy in a variety of contexts, including depression, anxiety, substance use, and chronic pain conditions (Bohlmeijer et al., 2010; Chiesa & Serretti, 2010; Goyal et al., 2014; Li & Bressington, 2019). Mindfulness also appears to have substantial benefits for stress reduction in the general population (Astin, 1997; Bowen & Marlatt, 2009; Chiesa & Serretti, 2009), and has been growing in popularity in a variety of applications ranging from more traditional meditation groups to an explosion of MT applications for smartphones and web platforms (Walsh et al., 2019). Despite the availability of MT practices for millennia, the rapid uptake and popularization of MT in contemporary Western society suggests that MT addresses a need not fully satisfied by contemporary healthcare models. What unique approaches to stress reduction does MT bring to a culture that already seems deeply invested in self-improvement and resilience? How does the neurobiology of mindfulness interventions inform our broader understanding of stress reactivity and resilience?

In this chapter, we will briefly review mindfulness theory covered in greater detail in other chapters of the book. We will discuss how multiple definitions of mindfulness correspond with types of neural systems models, and review evidence on how the most popular manualized MT interventions seem to impact brain structure and function. The chapter will conclude with a narrative synthesis of the findings reviewed, proposing a more general neurobiological model of mindfulness and its mechanisms for reducing stress reactivity.

How Does Mindfulness Work?

We begin by briefly reviewing current theory on how mindfulness works. The consensus research literature suggests that MT’s efficacy stems from its ability to change maladaptive appraisal and coping habits that drive psychiatric symptoms, such as perceiving negative thoughts and feelings as objective truths rather than as transitory mental events (Farb et al., 2012; Hölzel, Lazar, et al., 2011b; Teasdale et al., 2013). Central to this theory is the idea MT promotes awareness of reactive habits that are used to cope with stress, allowing the practitioner to adjudicate between adaptive response and habitual, maladaptive responses.

For example, faced with criticism from one’s boss, one might feel an urge to quit to avoid further criticism. Alternatively, one might consider reviewing the criticism as a chance to improve one’s work. If avoidance is the overlearned response, one might find oneself leaving a job that one cares deeply about to protect oneself against the negative feeling of criticism. If one can recognize the impulse to quit as a natural urge to avoid criticism, one can work with that feeling and consider how else to manage it rather than making a harmful decision. Mindfulness training is precisely the practice of cultivating awareness of one’s experience to limit habitual reactions and possibly replace one’s most pernicious habits with alternative strategies that better align with one’s personal values.

In technical terms, MT addresses the rigidity and power of overlearned perceptual and behavioral habits by leveraging awareness to de-automatize stress appraisals and their associated regulatory responses (Chong et al., 2015; Lea et al., 2015; Vago, 2014). Intentional, nonjudgmental exploration of present-moment experience serves to disrupt the habitual implementation of existing perceptual and behavioral habits. The benefits of mindfulness meditation are believed to arise from the cultivation of awareness as a buffer against “knee-jerk” reactions to stressful events, allowing for novel perceptions and appraisals to emerge. In the reflective space of mindful awareness, a practitioner is empowered to explore novel perceptions, appraisals, and responses. Practitioners are encouraged to use this freedom to explore novel perceptions and responses that align with deeply held personal values.

Given growing research establishing the efficacy of MT for addressing a range of clinical syndromes, a second wave of research has begun to identify potential mechanisms underlying MT’s therapeutic effects (Alsubaie et al., 2017). Mechanistic research is critical for refining treatments to better target maladaptive processes aggravating psychological functioning (Holmes et al., 2018). To this end, neuroimaging has become an invaluable tool for the detection of putative mechanisms of MT, empowering researchers to go beyond self-report to examine both transient and lasting effects (Roffman et al., 2005). Yet despite a wealth of meta-analytic studies describing MT efficacy, we have only begun to understand MT mechanisms through a scientific lens. To address the gap between emerging mechanistic research and its integration into theory, this chapter will employ a narrative review of the neurobiological correlates of MT and their clinical significance. We will begin by briefly describing three popular paradigms in MT research, exploring mindfulness as (i) a short-term state, (ii) the product of more comprehensive clinical interventions, and (iii) a disposition or “trait”. To guide discussion toward practical interpretation of function, this review will focus on large-scale, intrinsically connected brain networks rather than specific brain regions. We will review empirical studies exploring three levels of analysis: mindfulness as a transient state, the impact of MT interventions, and correlations with “trait” or dispositional mindfulness. Finally, we will discuss the implications of these findings for the study and application of contemplative practices within the clinical domain.

Conceptualizations of Mindfulness

The term “mindfulness” is multifaceted, with at least three distinct connotations and accompanying research paradigms. First, mindfulness can be thought of as a transient mental state, in which a person is induced to intentionally engage in nonjudgmental attention to the present moment (Kabat-Zinn, 1990). Experimentally, transient mindfulness can be studied as a state induction, examining the temporally local effects of mindfulness prompts on cognition and mood. Such research can occur within an individual, but also to evaluate whether the nature of the state changes with expertise. Second, mindfulness can be thought of as a trajectory of personal development in the capacities such as awareness or equanimity, growth that is engendered via MT, often through clinically manualized interventions (Baer, 2003). As such, MT can be studied as a form of contemplative training , producing lasting changes to attentional networks and well-being. This sort of research is most in line with clinical trials, where an intervention is compared against a control group, but cross-sectional research that uses training history as a covariate is also possible. Finally, a combination of natural predispositions and training could yield chronic individual differences in the tendency to be mindful (Tomlinson et al., 2018). As such, MT can be studied as a disposition or trait. This is most commonly the domain of cross-sectional research, correlating measures of mindfulness with other variables of interest. We will describe each of these conceptualizations to better establish the context for neuroscientific models of MT.

Mindfulness as a State

Perhaps the simplest method for studying mindfulness is through state inductions: brief tasks or meditation exercises designed to induce a state of mindfulness in participants. This is accomplished by having participants practice a specific meditation or other contemplative practice such as breath meditation and the body scan, which would normally comprise a mindfulness training or intervention program. The advantage of such an approach is the ability to explore the momentary “building blocks” of mindfulness training – to understand the immediate impact of a brief practice and to determine which of the many possible effects are most consistent and replicable. Inductions can also be used between different cohorts with varying expertise in mindfulness techniques, to examine the interplay between brief practice and more dispositional factors that will be discussed below.

While the effects of brief practice may be transient, the potential for within-study replication of effects provides an opportunity for great scientific rigor. In addition, studies that employ induction techniques may be closest to directly probing the capacity to engage in sustained mindful states, assessing a person’s ability to adopt mindful awareness of the present moment given the situation at hand. Moreover, unlike more intensive manualized trainings, mindfulness induction paradigms can be adjusted in dosage (e.g., 10-min meditation vs. 40-min meditation), focus (e.g., sensations, thoughts), attitudinal stances (e.g., cognitive distancing, self-compassion), and mode of delivery (e.g., in-person vs. online).

A substantial number of induction studies now populate the research literature (Leyland et al., 2019). While the exact implementation of inductions varies substantially across studies, most studies link brief MT to improved mood, often introducing mental calmness and emotional stability (Keng et al., 2011), as well as improvements in cognitive processes such as attentional control (Gallant, 2016). Moreover, inductions also appear to improve central psychological mechanisms of MT including decentering, the ability to watch one’s experience with a sense of objectivity and psychological distance (Feldman et al., 2010; Lebois et al., 2015), and experiential acceptance, the ability to tolerate and explore intense or stressful experiences (Keng et al., 2011).

However, there are some limitations to mindfulness inductions. Specifically, the preponderance of studies using mindfulness inductions rely on post-meditation self-reports to capture features of the participant’s mindful state. As such, findings could be influenced by known drawbacks of self-report measures, such as social desirability bias and biases related to demand characteristics. A further limitation is that some skills may require real-life stressors, and time for reflection and repeated exposure to such stressors, before MT produces reliable alterations to emotion regulation or well-being (Garland et al., 2015). Thus, while induction techniques are attractive due to the low investment of time and resources, they may be limited in revealing the most meaningful changes associated with MT. Nevertheless, induction paradigms figure significantly in the neurobiological research literature due to their compatibility with standard experimental designs often employed in cognitive neuroscientific exploration.

Mindfulness as Contemplative Training

Ideally, inductions serve as experimental “micro-interventions” that could in theory aggregate over time to produce more impactful, lasting changes. Through practice in sustaining states of mindful awareness, interventions promote mindfulness as a regulatory skill that can be marshalled in response to life’s challenges (Kabat-Zinn, 2003). In mindfulness-based interventions (MBIs), the intention is for participants to move beyond experimenting with state inductions in formal meditation to employ the mindful state in response to a stressor. The project is therefore one that extends beyond immediate symptom resolution, encouraging participants to continue their contemplative practice as a lifelong skill; indeed continued practice following intervention completion is associated with the maintenance of therapeutic gains (Crane et al., 2014). Mindfulness training is now customized for a variety of ailments through a growing cadre of manualized clinical interventions, perhaps most famously through mindfulness-based stress reduction (MBSR) (Kabat-Zinn, 1990), the first and most general form of manualized training, and mindfulness-based cognitive therapy (MBCT) (Segal et al., 2012), which was developed to address relapse and recurrence in MDD. However, MT has also been integrated as a core element of emerging cognitive-behavioral therapies such as acceptance and commitment therapy (ACT) (Hayes et al., 1999) and in dialectical behavior therapy (DBT) (Linehan et al., 1999), whose mindfulness application actually inform and precede MBCT.

Both MBSR and MBCT are group interventions that provide participants a structured 8-week protocol for practicing various mindfulness meditations and learning about their underlying principles to strengthen emotion regulation capacities and improve quality of life. These interventions are designed to stimulate metacognitive awareness and experiential acceptance of internal experiences, as it is postulated that the willingness to initiate and remain in contact with uncomfortable internal states will allow practitioners to disengage from prepotent responses (e.g., worrying, rumination) and instead deliberately engage in more adaptive regulatory strategies (e.g., problem solving, behavioral activation) in the face of stressful life events. Although MBSR and MBCT were originally tailored to target chronic pain and the prevention of depressive relapse, respectively, the growing interest in MBIs and the mounting evidence supporting their clinical efficacy have encouraged others to adapt these treatment protocols for other stress-related medical and clinical disorders.

Currently, the empirical evidence for MBSR has supported its clinical efficacy for the general reduction of stress and negative mood states (e.g., depressed mood, anxiety) and maladaptive regulatory strategies (e.g., rumination, worry) in healthy individuals (Chiesa & Serretti, 2009); patients struggling with physical conditions such as chronic pain, breast cancer, multiple sclerosis, and fibromyalgia (Fjorback et al., 2011); and primary caregivers (Lengacher et al., 2012). In contrast, MBCT has been evaluated for its prophylactic outcomes for major depressive disorder, with randomized controlled trials demonstrating that its clinical effects extend beyond treatment as usual (Ma & Teasdale, 2004; Teasdale et al., 2000), providing protection equivalent to continued antidepressant treatment (Kuyken et al., 2016). With modifications, MBIs also show for addressing even acute phases of MDD, comparable to the effects of more conventional interventions such as cognitive therapy (Bockting et al., 2015).

Mechanistic research into MBIs has revealed that intervention effects operate on both the psychological and neurobiological aspects of stress in tandem, working in partnership to reduce stress reactivity associated with affective symptom burden (Brown et al., 2007; Chiesa & Serretti, 2009; Creswell et al., 2014). However, there are a number of limitations to the randomized controlled trials of MBSR and MBCT, as they have relied predominantly on research designs built to examine pre-post intervention effects within group or relative to waitlist controls, with relatively few studies that employ active controls (Goyal et al., 2014). As such, there is a need for more rigorous designs such as dismantling studies to better specify the MT’s central mechanisms of action.

While reviewing the emergence of recent dismantling studies on psychological constructs such as decentering, acceptance, and equanimity is beyond the scope of the current chapter, neurobiological accounts provide an alternative source of data for understanding intervention mechanisms. While applying neurobiological paradigms in brief induction research helps reveal transient changes in biological processes such as brain metabolism, same paradigms hold much greater promise when applied to intervention designs. Specifically, intervention designs allow comparisons between patients and healthy controls to track the biological covariates of symptom burden and its resolution (e.g., Farb et al., 2010). When participants are instead evaluated in terms of future disorder vulnerability (e.g., Farb et al., 2011), we can learn about biomarkers of vulnerability that may not be obviously expressed as symptoms. We can therefore use contemplative training designs to learn more about how MT impacts well-being, either by normalizing brain and physiological function or by introducing novel regulatory capacities that offset enduring markers of disease vulnerability.

Mindfulness as a Disposition or Trait

Dispositional or trait mindfulness is a construct intended to capture individual differences in the tendency to engage in mindful states, representing both naturally occurring variation across the population and also the internalized “product” of efficacious MT. Measured primarily through self-report instruments, the literature contains numerous proposed structures and subprocesses comprising mindful awareness (Chiesa, 2013). Self-report measures of state mindfulness are the easiest and most popular approaches for capturing stable characteristics of mindfulness. There is disagreement regarding exactly how many components are best for defining dispositional, self-reported mindfulness, with complexity growing over time. Indeed, the most commonly used dispositional mindfulness scale 15 years ago was likely the single-factor Mindful Attention Awareness Scale (Brown & Ryan, 2003), which has now been supplanted by the current “gold standard” Five Facet Mindfulness Scale (FFMQ) (Baer et al., 2008), which incorporates much of the content of the MAAS into a more complete canvassing of mindful qualities. The FFMQ features five interrelated subfactors that contribute to a global dispositional mindfulness score: observing and describing shifting internal states, acting with awarenes s in the present moment, and approaching internal experiences with a nonjudgmental and nonreactive stance. While debate is ongoing as to the optimal number of factors by which to conduct self-assessment (Gu et al., 2016; Rudkin et al., 2018), it seems clear that dispositional mindfulness is an easily reportable and relatively stable measure of individuals’ differences.

Despite their face validity, self-reports are susceptible to inaccuracies given the common methodological pitfalls of surveys and biases in retrospective recall. In addition, the subcomponents of mindfulness states measured in such questionnaires are also driven by the expectations of the questionnaire author that may miss out on canvassing central but unexpected consequences of MT. Finally, self-report may be limited in its access to subtle cognitive or physiological changes. As such, self-report questionnaires only provide limited utility in furthering our current understanding of mindfulness states, their underlying structure, and how they impact other cognitive states (e.g., awareness, attention, attitudes).

However, when integrated with neuroimaging techniques to form a multimethod approach, studying mindfulness using self-report allows researchers to investigate changes in neural processes and structures as a function of the time and effort invested in mindfulness practice. Although there is debate as to whether these questionnaires capture true changes in mindfulness levels or are driven by other nonspecific effects (Chiesa, 2013; Grossman, 2011), evidence does suggest that mindfulness measured in this fashion can differentiate between those who are practicing mindfulness and those who are not engaged in any training (Baer et al., 2008; Hölzel, Lazar, et al., 2011b). Neurobiological accounts of dispositional mindfulness therefore provide tantalizing clues around how to relate theories of stress resilience to information processing systems in the brain.

Psychological Mechanisms of Mindfulness

Currently, most research on the mechanisms of mindfulness has emphasized proposed mediators derived from the underlying theories of MBIs such as MBSR and MBCT. For this reason, investigations have been limited to psychological constructs such as decentering, ruminative thinking, and metacognition (Gu et al., 2015), which are mostly bolstered by self-report measures. Beyond these psychological constructs, other theoretical models have distinguished themselves by focusing primarily on the cognitive and affective elements associated with mindfulness meditation (Baer, 2003; Brown et al., 2007; (Baer et al., 2006; Brown et al., 2007; Hölzel, Lazar, et al., 2011b; Shapiro et al., 2006; Vago & Silbersweig, 2012). These models overlap with one another regarding the proposed subcomponents driving the effects of mindfulness and related interventions, positing self-regulation, attentional reorienting, and experiential exposure as prospective drivers of mindful awareness. However, these theoretical models have yet to explicitly establish core mediators linking mindfulness and stress reactivity and stress-related disorders.

To date, most of the research investigating the connection between mindfulness meditation and stress reactivity has been conducted with self-report measures representing the potential mechanisms of action. In addition, these studies seldom employ a strong a priori theoretical model to inform research design and the interpretation of findings, which impedes the process of building, refining, and contrasting theoretical accounts regarding intermediary factors. Though the influence of mindfulness meditation on stress outcomes has been well documented, the overreliance on self-report measures and the lack of a guiding theoretical framework have resulted in current models that are without a firm empirical foundation and only capture the subjective experience of mindfulness. Moreover, although the identification of potential psychological mediators has been invaluable for understanding the widespread effects of mindfulness meditation, these findings rely predominantly on self-report measures that presume self-awareness of such psychological states.

The complexities of mindfulness and its impact on several distinct aspects of human functioning (e.g., cognitive, biological) underscore the need for multimethod research approaches and more comprehensive theoretical accounts. For instance, assessing both phenomenological and neurobiological dimensions of stress-related conditions has potential to expand our understanding of how mindfulness operates at deeper levels of functioning to alter stress-related outcomes. Moreover, such an approach is crucial for marshalling empirical evidence at deeper levels of analysis that support the construct validity of a given psychological mediator.

Unfortunately, the mediating pathways underlying the relationship between mindfulness meditation and stress reactivity have not been fully explored. These limitations are also evident even in the popular target of relating MT to improved stress management. Although there are many studies reporting on the efficacy of MT for a variety of stress-related disorders, few studies have investigated and firmly established mechanisms of action, with even fewer adjudicating between competing mechanistic accounts in search of those that provide the greatest explanatory value (Gu et al., 2015). As the number of proposed mechanisms increases, such initiatives will become pivotal for establishing factors through which mindfulness meditation and related interventions function, especially for improving the design and delivery of such interventions. For instance, the identification of the major mediating players would allow for the augmentation of therapeutic mechanisms likely to produce the greatest amount of symptom alleviation while also differentiating between the unique contributions of MBIs and those that are attributable to common treatment features (Kazdin, 2007).

Given the limitations of relying on self-report measures and focusing primarily on the phenomenological experience of mindfulness, neuroscience methods have burgeoned in interest and appeal due to their methodological features that overcome such pitfalls. Neuroimaging and other biological measures are capable of noninvasively isolating and monitoring psychological processes that characterize pathological conditions and the influence of mindfulness at a deeper level of analysis, without requiring participant self-report and insight into their subjective experience. This is especially helpful when studying treatment effects in psychiatric disorders that are associated with lower levels of insight and psychological mindedness than that which is observed in the general population (e.g., schizophrenia). Ultimately, applying neuroscientific methods measuring the biological underpinnings of stress reactivity and related conditions following a stressor allows researchers to formulate more realistic models of vulnerability and evaluate predicted therapeutic processes underlying mindfulness meditation. When integrated into a single methodological approach, such studies can effectively identify the basic mechanisms of action and how they interact to produce the phenomenological experience of mindful awareness.

Recent investigations of the neural correlates of mindfulness meditation and MBIs have greatly expanded our current understanding of the major brain networks operating during a mindful state. These investigations have largely focused on examining the effects of mindfulness meditation using two approaches of neurological analysis, functional and structural analysis , which together have the potential to produce a more comprehensive neural account of mindfulness. Functional analyses are designed to detect differences in brain activation between experimental conditions (e.g., mindfulness meditation vs. control) or time points (e.g., pre- vs. post-treatment), as well as correlations of activation patterns between brain regions. For instance, in regional functional analysis , participants perform an experimental task (e.g., focusing on the breath) during brain scanning to detect task-specific neural activations that underlie a specified mental state. In addition, given its capacity to reveal activation patterns unique to an experimental condition (e.g., reward task, emotion provocation), researchers have also implemented functional analyses to measure the impact of contemplative training and therapeutic interventions on cognitive and affective processes (e.g., reward processing, emotion regulation). Similarly, resting-state functional analysis is also applied to uncover activation patterns across brain regions or networks while the participant is at rest and unengaged in an explicit task. During the imposed resting states, participants are often automatically engaged in internally focused processes such as memory retrieval and perspective taking, allowing researchers to further measure the generalizability of mindfulness meditation and MBIs to self-referential processes. In contrast, structural analyses detect changes in the anatomy of neural structures (e.g., amygdala) or pathways (e.g., corpus callosum) across time points. Within the context of mindfulness meditation, structural analyses are generally applied to ascertain the depth of the impact of contemplative training and related interventions on brain structures implicated in well-being (e.g., amygdala) or putative cognitive processes related to mindfulness (e.g., attentional neural hubs).

Proposed Neural Networks

Early on, neuroscientists traditionally studied the neural correlates of cognitive and affective states via the detection of activation rates and structural changes in isolated brain regions. The findings from these analyses catalyzed the production of modular accounts of neural functioning wherein the interconnectivity between brain regions was surmised, expanding the rudimentary understanding of various neural structures. However, over the past decade, research objectives have gradually transitioned to modeling circuits within and between brain networks rather than relying predominantly on theoretical accounts, as more methodological advancements were made available to empirically validate such claims. This shift in methodological approach was accompanied by a growing acknowledgment for the synergistic interactions among neural networks for producing complex cognitions and behaviors.

In the case of mindfulness meditation, several distinct advantages were obtained from network analyses of brain function. For instance, whereas the modular accounts of mindfulness meditation provide a snapshot of the neural structures involved, network accounts are better suited for modeling the sequential processing of information as it travels and changes from one network to the next. In addition, researchers can model bidirectional connectivity among brain networks and the mechanisms that determine the flow of processing across such networks given their appropriateness for the task at hand. Altogether, network accounts move beyond modular accounts by capturing the dynamic interactions of neural networks in the brain.

In the subsequent section, we will provide a brief introduction of the major neural networks due to their relevance to the current empirical literature and findings on mindfulness meditation and related interventions. We will describe three crucial networks: the default mode network (DMN), the salience network (SLN), and the central executive network (CEN).

Default Mode Network

The default mode network (DMN) was first discovered in a neuroimaging study by Raichle et al. (2001), wherein participants underwent brain scanning during task-positive periods – periods wherein the participant is engaged in goal-directed behaviors – and task-negative periods, also known as a rest period. After examining brain activity during task-negative periods, it was revealed that some neural structures were activated during these rest periods in comparison to task-positive periods (Raichle et al., 2001). These findings led Raichle and colleagues to conclude that the synchronous activations during task-negative periods are indicative of a non-goal-oriented intrinsic brain network comprised of functionally interconnected set of regions.

Since then, our understanding of the DMN has grown exponentially. Subsequent studies have extended the aforementioned findings and led to newer theoretical accounts that have refined original conceptualizations and linked the DMN to the cognitive processing of internal stimuli, including information about the self, mind-wandering, and cognitive elaborations in general (Buckner & DiNicola, 2019). In addition, researchers have identified core brain regions implicated in the DMN, three of which include the posterior cingulate cortex (PCC) and the precuneus, the medial prefrontal cortex (MPFC), and the angular gyrus. Individually, these brain regions have been implicated in numerous cognitive processes that contribute to the overall function of the DMN. For instance, the PCC is implicated in integrating incoming information with memory and perceptions to allow for such cognitive processes as memory recall, future-oriented thinking, and mentalization , which refers to understanding the mental states of others (Leech & Sharp, 2014). The precuneus and angular gyrus are also implicated in memory recall in addition to visuospatial imagery and navigation and attentional deployment (Cavanna & Trimble, 2006), whereas the MPFC represents self-referential schemas and supports decision-making affecting oneself in the future or in situations concerning social partners (Lieberman et al., 2019). In combination with other neural regions such as the temporoparietal junction (TPJ) and temporal poles, the DMN is capable of producing a continuous sense of self, as well as the capacity to understand and predict others (Qin & Northoff, 2011).

Within the context of mindful awareness, researchers have also found associations between DMN activations and mind-wandering, which typically occurs during task-negative periods, involves thinking about the self or others, and is generally categorized as a non-mindful state (Fox et al., 2015). Given that mindfulness meditation has also been shown to influence self-reported mind-wandering and self-focused cognitions (Mrazek et al., 2012; Rahl et al., 2017), it has been suggested that contemplative practices fostering mindful awareness could be targeting these cognitive processes by tuning DMN connectivity (Doll et al., 2015; Taylor et al., 2013). Moreover, it is widely believed that mindfulness meditation also has implications for the treatment of stress-related conditions that are purportedly driven by dysfunctions in the DMN (Marchetti et al., 2016). According to review (Buckner et al., 2008), DMN dysfunction is related to numerous clinical disorders, including major depressive disorder (MDD), posttraumatic stress disorder (PTSD), and schizophrenia. For instance, individuals diagnosed with MDD show greater functional connectivity among neural hubs in the DMN (Mulders et al., 2015), and the stronger interconnectivity within the DMN is also associated with ruminative thinking (Zhou et al., 2020), which is linked to the onset and maintenance of depressive symptoms. Altogether, if mindfulness meditation is capable of shifting the neural activation of the DMN, it could potentially correct maladaptive activation patterns in the DMN of individuals susceptible to psychiatric disorders, in turn predicting less self-referential thinking and a more consistent sense of self.

Central Executive Network

Whereas the DMN is associated with task-negative periods, the central executive network (CEN) is most activated during task-positive periods (Fox & Raichle, 2007). The CEN represents the neural underpinnings’ executive functions including attentional control, working memory, and planning (Collette & Van der Linden, 2002; Salmon et al., 1996). These hubs include the dorsolateral prefrontal cortex (DLPFC), the dorsal medial prefrontal cortex (DMPFC), the inferior parietal lobule (IPL), and the orbitofrontal cortex (OFC). Each of these brain regions are lynchpins of the network, as they integrate incoming information from other brain structures to perform functions such as planning and problem solving, impulse control, self-monitoring, and social behaviors (Baumgartner et al., 2011). As such, the CEN is seen as critical for cognitive forms of emotion regulation (Goldin et al., 2008). Generally, the CEN is activated when planning a response to novel situations that cannot be managed with learned, habitual responses associated with the DMN (Ballard et al., 2011). This is accomplished primarily by overriding DMN-driven prepotent responses that would typically be prompted by external cues and instead engaging in novel goal-directed actions and intentions to optimally handle the situation at hand (Koshino, 2017).

In addition to general executive functions, the CEN has also been implicated in mindfulness meditation and related practices, especially since mindful awareness is necessarily about orienting and sustaining attention toward internal cues, inhibiting prepotent responses that might follow from such cues, and adopting actions that best meet the individual’s goals in that moment (Farb et al., 2012). The CEN is most associated with controlled behavioral process, which puts it into direct opposition to the DMN, and indeed a CEN hub region appears to inhibit DMN activity when the CEN is activated (Chen et al., 2013). However, while the DMN is generally inhibited as the CEN works to overcome habitual responses, the two networks may become positively connected when the CEN is applied to monitoring internal processes and prepotent urges (Christoff et al., 2009).

Given the apparent association between the CEN and mindful awareness, mindfulness meditation could also have implications for the treatment of stress-related conditions associated with CEN dysfunction (Alfonso et al., 2011; Crowe & McKay, 2016). For instance, the CEN is found to have fewer connections in children diagnosed with PTSD, while the DMN is hyper-connected, possibly reflecting their inability to disentangle themselves from ruminative thinking processes (Suo et al., 2015). Similarly, individuals with anxiety disorders tend to exhibit hypo-connectivity within the CEN but increased connectivity between the CEN and orbitofrontal cortex, potentially reflecting the influence of worrying on their behavioral actions for coping with environmental stressors (Geiger et al., 2016). Improving attentional control through MT could therefore be expressed as a reconfiguration of CEN connectivity, increasing local connections with the CEN while re-tuning its connections to the DMN other brain areas.

Salience Network

To adjudicate between effortful processing and reliance on habit, the salience network (SLN) is purported to detect events that prompt switching between the DMN and CEN, or in other words, from task-negative to task-positive functioning (Menon & Uddin, 2010). Ultimately, the SLN represents the motivational relevance of internal and external cues by filtering incoming information and sending forward the most pertinent for goal-directed behavior (Craig, 2009). The primary hubs of the SLN are the anterior insula (AI) and the dorsal anterior cingulate cortex (DACC), both of which are linked via a specialized tract of Von Economo neurons (Allman et al., 2011). The SLN is associated with the detection and integration of emotional and sensory stimuli, as well as in modulating the switch between the internally directed cognition of the default mode network and the externally directed cognition of the central executive network (Seeley et al., 2007). More specifically, the AI has been primarily implicated in sensory monitoring, whereas the DACC has been implicated in motor monitoring (Medford & Critchley, 2010).

In addition to these core hubs, the SLN also includes brain regions such as the amygdala, putamen, and ventral striatum, which are broadly involved in arousal and emotional responses (Reynolds, 2005). As such, the SLN is the most “emotional” of the three brain networks described here, and it is also associated with visceral feelings rather than “conceptual thought” (Craig, 2002). Given its role in determining the motivational relevance of situational cues, the SLN is heavily recruited by stressors (Hermans et al., 2014), serving to allocate necessary resources to meet the current needs of the situation at hand, by tapping into the DMN and CEN forwarding information for processing as deemed necessary (Sridharan et al., 2008). As such, the SLN may support transitions out of self-referential thinking to support feelings of agency by integrating momentary sensation and available motor responses.

As for its relation to mindfulness meditation, the SLN might be implicated in mindfulness given that mindfulness meditation has been found to increase noticing of the somatic and visceral sensations, and catching oneself when the mind wanders (Price & Hooven, 2018). As such, it could very well be that mindfulness results in lower DMN activation through the SLN switching between the DMN and CEN (Sridharan et al., 2008). By operating on these networks, MT and related interventions could also have implications for the treatment of stress-related conditions driven by SLN dysfunction (Farb et al., 2012). For instance, the AI node of the salience network has been observed to be hyperactive in anxiety disorders, which is thought to reflect predictions of aversive bodily states leading to worrisome thoughts and anxious behaviors (Pannekoek et al., 2013). Increasing awareness of worrisome thoughts may allow the CEN to offload the activity of the SLN, reducing anxious feelings in the face of greater perceived agency and intentionality of responding.

Narrative Review of Neuroscientific Findings

This section will review neuroimaging findings around MT, focusing on regional activations, resting-state activations, and structural changes. The summary of findings will canvas the three levels of analysis discussed above: transient states, training effects, and trait-like or dispositional individual differences. However, before proceeding, it should be noted that the described findings are to be interpreted with caution. A common methodological drawback among most brain imaging studies is the small number of participants recruited and assigned to each experimental condition, and the neural investigation of mindfulness is no different. Although there are numerous practical explanations for the limited sample sizes, we nonetheless recommend that all findings and interpretations be treated as preliminary rather than conclusive.

Mindfulness as a State

Emotion Processing

Neuroscientific investigations of mindfulness customarily seek to address two questions: what neural networks support the mindful state, and are mindful brain states impacted by MT? To address these questions, researchers commonly recruit both novice and experienced meditators and have them engage in mindfulness to discern the impact of mindfulness practice on brain function. By comparing MT to a non-meditative control condition, researchers can detect more immediate neural activations associated with the mindful state, whereas the comparison of novice and experienced meditators would speak to the role of expertise in entering this state. To date, studies in this vein have identified significant widespread changes in neural hubs that together constitute large-scale brain networks such as the DMN and SLN.

For example, one study instructed novice and experienced meditators to view positive and negative images designed to elicit emotional reactivity while either mindfully observing the presented stimuli or viewing them passively without any attentional modifications (Taylor et al., 2011). Compared to the passive-viewing condition, mindful viewing of both positive and negative images attenuated the subjective emotional intensity reported by novice and experienced meditators. However, although MT reduced emotional reactivity equally across both experienced and novice meditators, the neural structures underlying increased emotional stability varied as a function of meditation experience. Compared to experienced meditators, novice meditators lower activation of the amygdala, an SLN hub region associated with emotional processing and reactivity. In contrast, compared to novice meditators, experienced meditators exhibited reduced activation in neural hubs comprising the DMN, including the MPFC and the posterior cingulate cortex (PCC). In this way, the mindful state in beginners may differ from experts: while the average person may experience mindfulness as being less emotionally reactive to events, more experienced meditators may experience the state as one more broadly free from habitual ways of responding to events rather than blunting emotional salience. A second recent study replicated the finding that task-related deactivations of DMN cortical midline structures were more pronounced in experienced meditators when compared to novice meditators (Lutz et al., 2016).

A third study corroborates the finding of DMN deactivation during mindful states, even in novice meditators with only brief exposure to MT (Doll et al., 2016). Researchers recruited novice meditators for a 2-week attention-to-breath mindfulness training. During brain scanning, participants were instructed to attend to aversive images either by i) mindfully attending to the breath or ii) passively viewing the images. Following 2 weeks of MT, the novice meditators increased deactivation of the amygdala during mindful viewing versus passive viewing of aversive stimuli, supported by greater connectivity between the CEN and SLN. Furthermore, these novice meditators increased deactivation of the dorsal MPFC, a DMN hub, during mindful viewing regardless of whether the aversive cue was present or not, potentially signaling a greater capacity for disengaging from cognitive elaborations in general. A fourth study by yet another independent group supports characterizing the mindful state as low DMN activation, but high CEN and SLN recruitment (Tomasino & Fabbro, 2016). After an 8-week mindfulness training program, novice meditators showed increased activation in CEN and SLN regions involved in attention regulation (i.e., dorsolateral PFC, caudate/anterior insula) and decreased activation of DMN hubs (e.g., rostral PFC) at post-training.

Thus, after even short-term MT, practitioners begin disengaging from cognitive elaborations by downregulating DMN processing and relying more on modulating attention. However, expertise seems to dictate to what extent DMN deactivation is also coupled with SLN deactivation, blunting the affective salience of events. Specifically, high-expertise meditators seem able to deactivate the DMN without also blunting activation of the SLN, including the amygdala, which acts as a detector of emotionally salient events (Cunningham et al., 2008). In addition, repeated finding of increased amygdala-PFC connectivity may suggest that, with increased trait mindfulness through continued training, meditators are better able to regulate their emotions by integrating CEN-driven cognitive control with SLN-driven momentary emotional experience.

The inclusion of both novice and expert samples adds further nuance to the neural depiction of the mindful state. The reviewed studies suggest that early MT may reduce the emotional impact of events by exerting control over affective processing, which may also indirectly reduce habitual patterns of emotional reactivity to evocative cues, as indexed by attenuated DMN recruitment. However, with more meditation experience, emotional stability might then be sustained by primarily acting upon these habitual ways of responding via a focus on reducing DMN recruitment rather than SLN inhibition, disengaging from cognitive elaboration (e.g., self-referential thinking) that heightens emotional reactivity without needing to control affective processing directly. In effect, with more experience, MT practitioners may be able to experience greater emotional clarity without responding, whereas early practitioners may need to inhibit emotional responses themselves to get a handle on reactive habits.

Pain

Although neuroscientific research on MT has largely focused on experimental conditions incorporating emotionally evocative cues, a growing number of studies suggest that the benefits of mindfulness could also extend to pain perception and regulation (Hilton et al., 2017). The neural mechanisms underlying these benefits are now becoming clearer: reductions in pain intensity are often associated with increased activation of SLN hubs (i.e., ACC, anterior insula), reflecting changes in saliency processing and attentional monitoring of painful somatosensory cues. Across a variety of experimental paradigms, similar results were also found for pain processing and anxiety in naïve and experienced meditators, with increases in SLN regions relating to reductions in pain anticipation and unpleasantness and improved anxiety relief (Gard, 2014; Lutz et al., 2013; Zeidan & Vago, 2016). Intriguingly, the shift from DMN- to SLN-oriented processing during mindfulness of painful stimulation elevates activity of sensory regions involved in processing pain. For instance, reductions in pain unpleasantness were associated with increased activation of the orbitofrontal cortex (OFC), a region involved in the contextual evaluation of aversive cues, and decreased activation of the thalamus, the sensorimotor switchboard of the brain (Zeidan et al., 2011). As such, although both experienced and naïve meditators demonstrate higher activity in the somatosensory cortex while mindfully attending to emotions and body sensations (e.g., Lutz et al., 2016), experienced meditators shift more of their cognitive resources to interoceptive awareness and basic sensory functioning while downregulating executive processes and cognitive elaboration in self-regulation.

Reward

Beyond impacting the processing of evocative emotional or painful stimuli, MT also appears to produce meaningful effects on the reward processing of positive cues. Although behavioral differences were not detected between participant groups, experienced meditators showed greater activation of reward-related hubs (e.g., caudate, putamen) and sensory hubs (e.g., posterior insula) and an attenuated connection between the caudate and anterior insula compared to non-meditators during reward anticipation (Kirk et al., 2015; Kirk & Montague, 2015). Compared to novice meditators, experienced meditators showed greater deactivation of cortical hubs of the DMN (e.g., MPFC) during the receipt of reward. The regional activations underlying formal mindfulness practices might therefore produce downstream neural changes to non-meditative states and mindful states relying on different sensory anchors (e.g., breath, body sensation), which supports the notion that mindfulness is generalizable and not strictly task specific. Moreover, experienced meditators may downregulate the saliency and valence of aversive stimuli and rewarding events, becoming less reactive to both intrinsic and extrinsic cues as they instead attend to incoming sensory information in a nonreactive manner.

Structural Changes

Numerous studies have investigated the structural brain changes associated with various meditation traditions, but to the authors, it is implausible that these structural changes could occur quickly enough to characterize a transient mindful states opposed to more intensive changes over longer periods of time. To our knowledge, the shortest training interval required to detect changes in brain morphology (as opposed to activity) is only an hour in the context of learning a complex balancing task (Taubert et al., 2016). However, no mindfulness have employed a comparable design to our knowledge. Structural changes will therefore be covered in more detail in the following sections on longitudinal training and mindfulness as an enduring disposition or trait.

Mindfulness as Contemplative Training

Enhanced Sensory Processing

The neuroscientific evidence for MBIs has grown rapidly over the past decade, with most studies investigating the neural correlates of mindful states and other cognitive processes ameliorated by such treatments. As these interventions were designed to promote psychological well-being, such studies have aimed to uncover the neural changes underlying MBIs and the active treatment mechanisms that lead to improved symptom burden and quality of life in both healthy and psychiatric populations. To date, studies have connected MBIs to a range of cognitive processes for a range of psychiatric conditions, most of which emphasize shifting away from self-referential processing and toward heightened body awareness. For instance, our group compared MBSR participants and untrained novices on a brain-imaging task, wherein they were instructed to focus on their moment-to-moment experience (experiential focus) or on self-referential thinking (narrative focus) while reading evocative trait descriptive words (Farb et al., 2007). Compared to untrained novices, MBSR participants showed a more pronounced reduction in activation of the cortical midline structures of the DMN (e.g., MPFC) while adopting an experiential focus. Furthermore, MT completers showed additional activation of neural structures involved in somatic and interoceptive awareness (e.g., insula, secondary somatosensory cortex, IPL), in accordance with characterization of the mindful state as one of increased sensory representation. Finally, MT completers showed greater connectivity between sensory regions and the CEN, and reduced connectivity with more ventral PFC hubs of the DMN, in keeping with less habitual and evaluative attention toward sensation. As MBSR participants become better able to disengage from self-referential thinking supported by the DMN, they were able to increase direct representation of momentary body sensations, shifting the criteria for self-reference.

Our program of research joins other studies in showing that MT has powerful effects on not only the upregulation of sensory representation but also altered communication between sensory regions and the prefrontal cortex, moving from ventral hubs that are part of the DMN to more dorsal hubs that are typical of CEN activity. For example, we recently showed that, relative to visual attention, attention to the breath resulted in greater activation of the body representation regions in the posterior insula and somatosensory cortex (Farb et al., 2013a). When MBSR completers were compared to a waitlisted control group using this paradigm, MT was associated with greater local connectivity within the posterior insula and middle insula, and increased connectivity between the posterior insula and the dorsal PFC associated with CEN processing (Farb et al., 2013b). Despite increased posterior insula connectivity, the anterior insula, the hub of the SLN, was predominantly deactivated during internal attention relative to external attention. Thus, greater SLN activation is not a general end-point of MT, even if greater SLN activity connectivity is often shown during mindful states.

This finding is corroborated by an independent study (Ives-Deliperi et al., 2011), which reported on training-related changes in brain activity following MBSR-based MT while participants meditated in the brain scanner. Although MBSR resulted in deactivated midline cortical structures belonging to the DMN, MBSR was also associated with greater deactivations of SLN hubs (e.g., anterior insula, ventral ACC). The findings suggest that MBIs do not unilaterally enhance emotional salience, but rather such activation may be context dependent – when exposed to external stimulation, greater SLN activation may represent a sense of engagement with momentary experience and be enhanced by MT. Conversely, when engaged in internal reflection, greater expertise is linked with more pronounced SLN deactivation, as participants aim to approach internal experience with equanimity and nonattachment. From this perspective, MT is more about tuning sensory salience (SLN), cognitive elaboration (CEN), and habitual perception and appraisal (DMN) than it is about promoting dominance of any one brain system over another. Such flexibility is particularly important when one considers that many affective disorders such as major depressive disorder are characterized by abnormally elevated recruitment of all three of these brain networks simultaneously (Kaiser et al., 2015; Sheline et al., 2010), belying the notion that “more activation means better mental health.” Instead, more selective recruitment of networks, and better communication particularly with the CEN and SLN and/or sensory cortices, seem to suggest the type of increased clarity and focus qualitatively associated with MT.

Emotion Processing

Beyond the neural changes observed in mindful states, MBIs are also expected to cause downstream effects on the neural profiles of other cognitive-affective states such as emotion processing and regulation. A large (N = 158) study of emotion processing compared experienced practitioners, MBSR participants, and participants enrolled in a health enhancement program (HEP) on an emotional face processing task (Kral et al., 2018). No group differences were reported when processing negative faces, though experienced practitioners showed greater deactivation of the amygdala when viewing positive pictures. However, this pattern was not observed in MBSR participants, who instead exhibited increased functional connectivity from pre- to post-treatment between the amygdala and ventromedial prefrontal cortex, SLN, and DMN regions respectively implicated in emotion processing and regulation. These findings suggest that, like novice meditators in the early stages of mindfulness practice, MBIs might be promoting efforts directed at self-regulation rather than the nonjudgmental, present-moment openness that is observed in more experienced practitioners.

Encouragingly, studies have found comparable effects of MT on emotion processing for individuals diagnosed with psychiatric conditions. For instance, individuals diagnosed with social anxiety disorder (SAD) who underwent MBSR were asked to complete a regulation task during brain scanning (Goldin & Gross, 2010). Participants were asked to regulate their emotions in response to presented negative self-beliefs by either attending to the breath or distracting themselves. Compared to distraction, attuning to the breath resulted in deactivations of SLN regions and greater activation of visual attention hubs in the occipital and parietal cortices. This finding parallels studies of advanced practitioners discussed above (Farb et al., 2013b; Kilpatrick et al., 2011), in which greater engagement sensory representation, and less engagement with affective salience, might be a relief for participants in conditions where the impact of emotional information is itself a contributor to symptom burden.

However, increasing sensory processing is not always a boon; we have discussed above how depressive disorders are characterized not by hyperactivity in the SLN, but rather a conflation of prefrontal network activity, to the exclusion of nonevaluative forms of processing (Sheline et al., 2010). Our group examined depressive symptom burden in a community sample, comparing MBSR completers to a waitlist control group (Farb et al., 2010). We observed that it was the suppression of middle and posterior insula activity rather than prefrontal hyperactivity that best predicted symptom burden, as though the prefrontal activation was starving processing of momentary sensation that might interfere with overarching negative schema. Completion of MBSR was associated with a restoration of insula activity, linked to processing of the body’s internal state (Craig, 2002), and a commensurate reduction of activation in the posterior cingulate posterior hub of the DMN, associated with conceptual self-reference (Farb et al., 2007; Yang et al., 2019). Indeed, learning to better recruit even visual attention regions may serve as a sign of protection against depression in addition to reduction of prefrontal reactivity (Farb et al., 2011). Thus, in the context of psychopathology, simply increasing affective salience may be insufficient to reduce symptom burden – instead, a return to sensory processing may be needed to disentangle maladaptive neurobiological responses to stress. Even though a shift from conceptual (DMN) to affective salience (SLN) may be sufficient in healthy individuals, the conflation of affective salience with habitual, maladaptive appraisals in disorders like depression may require a more drastic move away from DMN/SLN processing altogether, supported primary though direct sensory access via relatively non“emotional” or “selfish” dorsal CEN structures.

Yet as more clinical intervention studies emerge, it becomes increasingly important to remember that some mechanisms of treatment are likely disorder-specific, and mindfulness’ benefits may involve a skillful application of CEN integration into other forms of representation that extends beyond any “one-size-fits-all” pattern of connectivity. Unlike social anxiety disorder, in which treatment response led to reduced SLN activity, and depression, in which pathology treatment was associated with recovered sensory activation in the insula and other sensory cortices, posttraumatic stress disorder (PTSD) research suggests a third distinct mechanism of action. A study of combat veterans diagnosed with PTSD who underwent Mindfulness-Based Exposure Therapy (MBET) or Person-Centered Group Therapy found evidence consistent with this CEN-sensory connectivity account of symptom resolution (King et al., 2016). Both treatment groups showed decreased PTSD symptoms from pre- to post-treatment, and these improvements were associated with increased connectivity between the DMN and both the SLN and CEN. Thus, greater CEN engagement in both momentary salience attribution and cognitive elaborative habits was a sign of increased capacity to engage in the volitional shifting of attention, similar to the literature on MT states in health participants. Finally, a study of participants with bipolar disorder undergoing MBCT found yet another profile of results (Ives-Deliperi et al., 2013), in which training was associated with increased activation in both DMN and CEN hubs, commensurate with a need to better integrate self-knowledge into momentary decision-making and self-regulation.

Pain

Extending research on pain processing in mindful states, it seems that even brief training can alter pain processing habits. In a seminal study by Zeidan et al., novice meditators underwent 4 days of MT training and complete a pain stimulation task during brain imaging before and after training (Zeidan et al., 2011). During the task, the participants were asked to attend to their breath while receiving noxious thermal stimuli to the skin. While engaged in mindful breathing during pain stimulation, participants at post-training compared to pre-training reported increased reductions in pain unpleasantness and intensity, as well as a dampening of pain-related activation of the somatosensory cortex, which processes somatic sensations. Moreover, reductions in pain intensity were associated with increased activation of SLN hubs (i.e., ACC, anterior insula), reflecting changes in saliency processing and attentional monitoring of painful somatosensory cues.

Reward

Mindfulness may also allow practitioners to improve the feeling of pleasure or meaning associated with relatively mundane or normal events. The goal of many substance use recovery programs, such as Mindfulness-Oriented Recovery Enhancement (MORE), is to restructure reward processing away from the conditioned substance and more toward regular life events, thus rebalancing the award system away from focused drug craving and toward savoring of other potentially meaningful or pleasurable aspects of daily life (Garland et al., 2014). Accordingly, pilot work on participants enrolled in MORE for smoking cessation demonstrates reduced activation to substance-related cues within the ventral striatum, the putative neural hub of reward processing. Concurrently, the same participants demonstrated increased ventral striatum responses to nondrug emotionally evocative cues, supporting the stated intentions of the program (Froeliger et al., 2017). Thus, the same principle of increasing sensory representations may generalize to reward processing to help expand and normalize the brain’s selection of rewarding experiences in the face of addiction.

Reviewing examples of research across distinct disorders, it seems as though the implication of greater cognitive control by the CEN seems to be ubiquitous in the MT literature, but the application of such control seems to vary in observing the skillful resolution of different mental health conditions. Additional research may clarify the reliability of these disorder-specific response patterns, as in many conditions research is being driven only by a handful of labs, making it hard to determine how replicable and generalizable these finds truly are – the authors’ lab included! Nevertheless, increased CEN connectivity seems to be a more reliable, transdiagnostic marker of intervention-related MT effects on brain networks.

Structural Changes

With the use of more comprehensive interventions, the possible changes to brain structure seem more feasible than examinations of transient state changes. Although limited, a few studies have investigated structural changes linked to MBIs from pre- to post-treatment. Participants completing MBSR demonstrated increased gray matter density in the right insula and somatosensory cortex, with greater increases predicting decreased alexithymia symptoms (Santarnecchi et al., 2014). Hölzel et al. (2010) recruited participants who self-reported high levels of stress and underwent MBSR training and showed that compared to pre-treatment, MBSR participants at post-treatment reported lower levels of stress, and the stress reduction was associated with decreased gray matter density in the amygdala (Hölzel et al., 2010), suggesting that short-term mindfulness practice can result in neuroplastic changes to network hubs underlying emotion processing that underlie attenuated stress responses. These effects may play out in white matter tracts, both in connecting the two hemispheres of the brain (Luders et al., 2012), and also in connecting memory circuits in the hippocampus with the anterior insula, the sensory hub of the SLN (Britta K. Hölzel et al., 2016). Subsequent connectivity analyses suggested that this uncinate tract may help support extinction learning of fear-inducing stimuli, improving stress resilience (Sevinc et al., 2019).

Mindfulness as a Disposition or Trait

Resting-State Imaging

The discovery of intrinsically connected brain networks has led to a rapid expansion of “resting-state” fMRI, which examines brain connectivity in the absence of task demands. In this way, resting-state studies may be close analogues to psychometric studies of dispositional mindfulness, examining the “default” state of a participant rather than focusing on a transiently induced mindfulness state. Focusing on the connectivity between brain regions in the absence of an explicit task, resting-state activations can be used to ascertain whether MT influences habitual patterns of neural representation and computation. Accordingly, investigations linking MT with resting-state activity seek to determine the extent to which habitual modes of mind can be altered by mindfulness interventions.

Researchers investigating the resting-state effects of MT aim to uncover the degree to which mindfulness can alter connectivity within and between brain networks while the individual is at rest and not focused on a given task. In pursuit of this aim, researchers commonly recruit both novice and experienced meditators and compare their resting-state activations to evaluate these questions, although comparisons of participants before and after training, and in theory even before and after state inductions, could occur.

Compared to the number of neuroscientific investigations into regional activations, fewer studies have examined the impact of MT on resting-state activations. Nevertheless, these studies correspond with previous findings for regional activations, such that the DMN is downregulated while other networks such as the SLN are upregulated. In one pioneering study, experienced meditators reported on the spontaneous noticing of mind-wandering in a “task-free” neuroimaging paradigm (Hasenkamp & Barsalou, 2012). From this single key press, four cognitive processes related to focused-attention meditation, including mind-wandering (which preceded the button press), awareness of mind-wandering (the key press itself), refocusing of attention (immediately after the key press), and sustained attention (the period leading up to the next mind-wandering period).

Ultimately, more experienced meditators exhibited greater trait-like connectivity between the CEN (i.e., dorsolateral PFC) and SLN (i.e., right insula) and decreased functional connectivity between the DMN and SLN while sustaining their attention to the breath (Hasenkamp & Barsalou, 2012). The fact that Hasenkamp and Barsalou (2012) found increased connectivity between the CEN and SLN suggests that experienced meditators could be relying on less habitual processing as they increasingly cultivate present-moment awareness and acceptance of their internal experiences. These findings are supported by an independent study (Berkovich-Ohana et al., 2016), which reports that experienced meditators (relative to novice meditators) showed lower spontaneous fluctuations of neural responses in DMN hubs while exhibiting greater fluctuations in the visual cortex.

Given the putative role of the right anterior insula in switching between brain networks, the resting-state findings suggest that MT engenders a greater capacity for neural and cognitive shifts from the DMN and states of cognitive elaboration to the CEN and states of focused attention. As the neural hubs of the SLN and CEN become functionally connected, meditators become more apt at de-automatization, that is, breaking free from habitual, automatic processing for other cognitive states that might be of more utility in the moment. Finally, local prefrontal effects may further be reinforced by greater functional connectivity with other regions of the brain involved in sensory integration, emotion processing, and other important cognitive-affective functions (Kilpatrick et al., 2011; Tang et al., 2017).

Nevertheless, some studies have suggested that the current understanding of the resting-state effects of mindfulness practice might not be capturing more nuanced neural changes within brain networks. One study (Taylor et al., 2013) examined experienced meditators and novice practitioners who underwent a weeklong meditation training, later comparing them on resting-state functional connectivity. Compared to the novice meditators, the experienced meditators exhibited weaker functional connectivity between cortical midline hubs within the DMN, areas associated with conceptual self-elaboration. However, relative to novice meditators, experienced meditators also showed greater connectivity between other clusters of DMN that are associated with perspective taking and theory of mind. A complementary set of findings demonstrated that untrained participants who engaged in 2 weeks of daily mindfulness meditations showed a link between trait mindfulness and reduced functional connectivity between the DMN and SLN, indicating perhaps less habitual reactivity to emotionally salient events (Doll et al., 2015). This finding was replicated in a study of university students who had undergone an MBSR-inspired treatment, in which the meditative state was associated with weaker DMN and SLN connectivity, and better connectivity within the SLN (Yang et al., 2016). Together, these results emphasize that mindfulness mechanisms should not be simplified to a general weakening or strengthening of a neural network; instead, the DMN, which supports cognitive habits, can be tuned to shape habits through training rather than seeking to eliminate reliance on habits. These findings support an account where habits turn less toward reactive self-evaluation and more toward perspective taking. Thus, we must be careful not to demonize any one brain network in understanding MT, but instead ask how that network is being tuned by training to afford more adaptive perceptions and responses.

Accordingly, it is possible that the beneficial effects of MT stem not from weakening the DMN, but rather from rewiring in connectivity allowing practitioners to engage in more adaptive cognitive elaborations and self-referential processes. For instance, participants enrolled in a 4-day meditation intervention and underwent brain imaging at pre- and post-training to measure the impact of MT on resilience compared to relaxation training (Kwak et al., 2019). Participants from the MT and relaxation training exhibited increased self-reported mindfulness and resiliency from pre- to post-training, and the meditation group sustained these gains at a 3-month follow-up. Interestingly, the meditation group also exhibited strengthened resting-state connectivity between neural hubs of the DMN (i.e., rostral ACC) and CEN (e.g., dMPFC) compared to the relaxation group at post-training. Moreover, the strength of the DMN-CEN connectivity across time points mediated the association between self-reported mindfulness and resilience at post-training and positively predicted resilience at follow-up. These findings suggest that through MT, practitioners learned to bring awareness to their self-referential cognitions and other internal experiences, which in turn enhanced their resilience in the face of adversity.

Together, these findings support the notion that mindfulness practices can cultivate an orientation of open awareness toward internal experiences that can also translate to nonspecific states. Transfer may extend beyond simple reductions in self-referential processing and other forms of cognitive elaboration, increasing awareness of mental habits while reducing reactive responses to salient events; in neural terms, transfer may reduce reliance on the SLN in favor of attentional monitoring supported by the CEN.

The role of the SLN also appears to be modulated by MT; rather than broadly reducing SLN activity and blunting emotional responses to motivationally salient events, MT may upregulate exploratory, sensory responses to such events in place of cognitive judgments supported by the DMN. For example, when compared to untrained participants, participants completing MBSR showed increased connectivity within the auditory and visual networks and between these networks and the SLN (Kilpatrick et al., 2011). The auditory and visual networks also showed greater anticorrelations more strongly negatively associated with MBSR participants, suggesting a better capability of inhibiting irrelevant sensory information for mindfulness practitioners. Training has also been linked to greater interconnectivity between sensory and motor hubs of the SLN, as participants who suffered from moderate to severe pain showed greater coherence between the anterior insula and dorsal anterior cingulate cortex following MBSR training compared to healthy controls (Su et al., 2016). Intriguingly, this greater connectivity was linked to reduced pain scores, indicating that stronger communication with the SLN may afford greater regulation rather than just serving to alert an individual to a painful or distressing sensation. A recent systematic review bolstered such interpretations (Gotink et al., 2016), showing that on average neuroimaging reduced activity but greater connectivity between the amygdala, an SLN hub, and the prefrontal cortex, indicating better communication but less reactivity to emotional events. While it may be too early to speak definitively, the “dorsal shift” in connectivity between the SLN and the prefrontal cortex may be a characteristic of mindful states, leveraging CEN rather than DMN processing to response to motivationally relevant information (Farb et al., 2007).

These results may extend beyond just SLN-CEN connectivity to also show DMN activity becoming coupled with CEN control. Patients with trauma-related symptoms who underwent mindfulness-based exposure therapy (MBET) showed increased connectivity between the CEN and both the DMN and SLN (King et al., 2016), and effect that seems to be supported by recent meta-analysis (Boyd et al., 2018). Similarly, unemployed participants who underwent a 3-day intensive residential MT program showed increased connectivity between the DMN and CEN (Creswell et al., 2016). These findings were further supported by a large (n = 14) study aimed at replicating the functional enhanced connectivity findings between the DMN and CEN (Kral et al., 2019). Untrained participants were recruited and assigned to MBSR or an active or waitlist control condition, and each participant underwent brain scanning at three time points: baseline, post-treatment, and follow-up approximately 5 months post-treatment. MBSR participants exhibited increased PCC-dlPFC connectivity from pre- to post-treatment relative to controls, but these differences were not sustained at follow-up. Nevertheless, the increased PCC-dlPFC connectivity was more strongly related to the number of days spent practicing therapy skills in the MBSR group compared to the active control group, as well as strengthened white matter connectivity between these same regions. Again, these findings support the notion that the connection between the CEN and DMN may be strengthened through mindfulness practice and may be related to an improved capacity for attuning to sensory cues while inhibiting or containing cognitive elaborations.

Together, these findings suggest that dispositional mindfulness may be indicated by a normative profile of brain connectivity which features higher-than-normal connectivity between the CEN and the SLN/DMN, and perhaps greater connectivity within the DMN and SLN as well. Typically, high dispositional mindfulness is to greater network connectivity between prefrontal networks and sensory representation regions, and increased DMN or SLN connectivity strength may be a sign of successful MT to the extent that these networks more richly correspond to activation in sensory regions.

Structural Changes

Perhaps the most obvious level of analysis in which to detect brain structure changes is at the level of stable differences in mindfulness as a disposition or trait. Accordingly, one of the first papers leverages lifetime MT experience as a predictor of gray matter volume, demonstrating that meditation practice seemed to reduce the rate of cortical atrophy that naturally occurs with human aging (Lazar et al., 2005). In a cross-sectional analysis of cortical thickness, meditators showed preserved gray matter density in a variety of brain regions, including all three of the major networks (SLN/CEN/DMN) discussed in this chapter. The implication of this first study was that meditation practice helps preserve brain health over the life span, as MT-related differences were particularly pronounced in older participants included in the sample. These general protective effects were recently replicated in a large (n = 100) cross-sectional study (Luders et al., 2015).

Since the Lazar et al. (2005) paper, many other investigators have explored the effects of MT on brain structure, employing both cross-sectional and intervention studies. Generally, results have been inconsistent and appear biased toward findings of increased gray matter, as to our knowledge almost no gray matter reductions have been linked to MT. The lone exception is a study relating dispositional mindfulness to brain structure, in which higher levels of trait mindfulness were linked to less gray matter in the amygdala and caudate, elements of the SLN (Taren et al., 2013), putatively indicating less “practice” in evaluating stimuli for their emotional importance, although the cross-sectional nature of the study leaves the possibility that those with smaller amygdala are also more likely to meditate.

On the other hand, many studies have linked greater levels of gray matter with MT, and somewhat surprisingly these reports tend to include aspects of the DMN. One cross-sectional investigation found greater gray matter in meditators than controls in the left inferior temporal gyrus and right hippocampus in long-term meditators, aspects of the DMN memory system (Holzel et al., 2008). A within-participant MT intervention study by the same group found increased gray matter in DMN hubs such as the cingulate cortex and temporoparietal junction (Hölzel, Carmody, et al., 2011a). Another small (n = 6) within-participant longitudinal study of MT over 6 weeks found precuneus (a posterior DMN hub) increases in older adults (Kurth et al., 2014).

What can we make of preserved DMN and (possibly) reduced SLN gray matter associated with mindfulness? The interpretation of gray matter density is not so straightforward as greater gray matter indicating greater use or function of a region, because pruning (reduction) of brain structures is a critical aspect of healthy brain development (Casey et al., 2008; Sowell et al., 2001). Greater DMN volume may therefore not necessarily indicate a greater reliance on habit, but rather a restoration/preserving of plasticity in these regions, whereas lower gray matter regions may already have been “pruned” to a support more efficient but more static population of neurons.

Effects within the DMN aside, other investigations have focused more on increases in gray matter in brainstem regions associated with the production of neurotransmitters and supporting the cranial nerves. Studies have found increased matter in these regions (Vestergaard-Poulsen et al., 2009), and subsequent research has linked this increased gray matter to greater levels of personal well-being (Singleton et al., 2014).

Evidently, more research is needed to test the consistency of such effects and the relationship between MT and function. As a broad principle, MT appears to help preserve gray and white matter from age-related atrophy, and greater gray matter in the DMN specifically may reflect greater plasticity in habitual cognition rather than a greater reliance on any one particular set of cognitive habits.

Neural Models of Mindfulness

From this review, it is evident that the prefrontal brain networks (CEN/DMN/SLN) can help to explain mindful awareness and are potentially linked to clinical outcomes in the therapeutic context. Beyond these empirical investigations, the sheer range of findings across various dimensions of brain structure and functioning prompted the development of theoretical models to integrate said findings into a cohesive and comprehensive account of mindfulness.

One of the most notable models comes from Holzel et al. (2011b), who proposed that the effects of mindfulness meditation are primarily driven by four internal processes. These processes include attention regulation , the capacity to sustain attention on a given sensory cue and return one’s attention to said cue when distracted; body awareness, the capacity to attend and bring awareness to sensory and visceral sensations; emotion regulation , particularly the ability to adopt a nonjudgmental attitude toward emotional experiences and contact unpleasant internal cues in a nonreactive manner; and change in self-perspective, the ability to refrain from overidentifying with inner narratives regarding self-identity (Hölzel, Lazar, et al., 2011b). Moreover, Holzel et al. (2011b) also proposed that the four internal processes were associated with neural activation of the ACC (attention regulation), the insular cortex (body awareness), the dorsolateral and ventromedial PFC and amygdala (emotion regulation), and the PCC and cortical midline structures (self-perspective). In essence, it is argued that mindfulness meditation and related interventions exert its therapy effects by correcting maladaptive changes in brain function and structure associated with a given clinical disorder (e.g., attention regulation deficits in bipolar disorder, elevated emotion reactivity in major depression and anxiety disorders) (Holzel et al., 2011b), which could explain the divergent findings regarding the effects of mindfulness training when studying different psychiatric populations.

Another noteworthy model comes from Vago and Silbersweig (2012), who proposed that mindfulness practice influences self-processing through the strengthening of three internal processes: self-awareness , the ability to focus and maintain awareness on internal states; self-regulation , the ability to inhibit and modulate prepotent responses; and self-transcendence , the ability to transcend self-focused needs and adopt more prosocial characteristics. Similar to Hozel et al. (2011b), this model proposes that changes in these internal processes reflect functional and structural alterations in brain networks broadly associated with attention and emotion regulation, as well as motivation and prosociality (Vago & Silbersweig, 2012). The authors propose that clinical disorders stem from negative self-beliefs that are reinforced by attentional biases toward emotionally evocative cues, which subsequently aggravate and are reinforced by psychiatric symptoms. Through this continuous cycle, negative self-beliefs can become deeply entrenched into internal schemas of the self, others, and the world, leading one to habitually engage in maladaptive cognitive processes (e.g., rumination) and behaviors (e.g., situational avoidance). Moreover, these maladaptive cognitions and behaviors can also interfere with learning from present-moment contingencies and engaging in novel behaviors or adopting more functional interpretations of the self and evocative situations. However, through the practice mindfulness meditation, psychiatric populations can minimize the deleterious influence of negative self-beliefs by assuming an experiential and transcendent sense of self. More specifically, mindfulness training modulates attention regulation and executive monitoring via alterations in brain activation and structural changes to integrate the evaluative self with experiential and transcendent senses of self and in turn correcting affect-biased attention. Similar to Holzel et al. (2011b), many of the neural structures proposed to underlie these varying senses of self are associated with the DMN (e.g., ventromedial PFC, PCC), SLN (e.g., ACC, ventral striatum), and CEN (e.g., dMPFC). Interestingly, many of the neural structures described by Holzel et al. (2011a,b) and Vago and Silbersweig (2012) serve as neural hubs within the DMN, SLN, and CEN, suggesting that apparent deficits in one of the aforementioned internal processes and senses of self could reflect broader maladaptive functioning and structural changes in the network subsuming that process.

Concluding Remarks

The literature reviewed suggests an emerging theory of mindfulness-based stress reduction, enhaning attentional control while reducing habitual evaluation. We must consider such mechanistic inferences cautiously, as few studies have been replicated in the research literature. At the same time, several themes have become apparent in our narrative review.

First, no single brain network can explain or track mindfulness, as mindfulness is a multifaceted construct with many components, and likely requires the coordination of multiple brain systems to alter existing mental habits and eventually replace them with more adaptive cognitive patterns for perceiving and responding to life events. The central elements of mindfulness are themselves under investigation, and a new generation of dismantling studies that seek to hone in on the minimum essential ingredients of mindfulness are currently underway, for example, examining the role of homework practices in MBCT (Williams et al., 2013), or comparing training in focused attention against training in open monitoring (Chin et al., 2019). As these dismantling studies mature, they will inform neurobiological paradigms by allowing for more specific forms of induction and training. Over time, this may allow for a closer correspondence between specific mechanisms of MT and the underlying neural dynamics that best support efficacious contemplative training.

Second, there is room for optimism in that mindfulness seems associated with multiple levels of analysis with a “dorsal” shift, i.e., the more dorsal CEN seems to become increasingly empowered to activate and communicate with other brain regions through MT, and offers an alternative to evaluation and self-referencing in the face of stressful experience. Learning to “decenter” or see things from a wider, more objective perspective is a central goal of MT, and is very consistent with a shift from DMN/SLN connectivity with sensory cortices to greater connectivity with the CEN.

Third, there is evidence from multiple sources that MT involves not just enhanced CEN connectivity, but also greater representation and integration of sensory information to the prefrontal cortex. This is often reported via CEN connectivity, but occasionally through SLN or DMN connectivity, especially in clinical studies. The integration of sensory information represents a marked shift away from the brain’s tendency to prune out sensory signals in favor of more rarefied conceptual and response distinctions in the prefrontal cortex. As such, mindfulness may represent a radical shift toward neural plasticity as sensory information is prioritized, likely provoking downstream changes in how we represent and respond to perceived events. This move to become “sensory learners” may also account for the widespread findings of cortical growth or reduced atrophy across the life span in meditation practitioners.

Fourth, the cultivation of mindfulness may be expressed differently depending on the expertise and mental health of the practitioner. This means that no single neural configuration is going to accurately represent an ideal mindful state or proof of training across diverse populations that include novices, experts, and those with clinical conditions. For example, MT in healthy individuals seems to involve reducing DMN activity and increasing sensory connectivity, but expert meditators and people recovering from depression both demonstrate patterns of sensory activation through a combination of SLN and CEN connectivity to sensory cortices. So, we must urge caution in interpreting any one study as providing normative data for all practitioners.

In conclusion, neurobiological accounts of mindfulness are growing rapidly as mechanistic research becomes increasingly justified in the face of a growing clinical and public consensus that MT provides benefits in a variety of populations and conditions. The neuroimaging literature largely supports accounts of nonjudgmental attention that integrates sensory representations in an exploratory way, which may eventually be reintegrated into a set of perceptual and behavioral habits. The coming decades of research will hopefully clarify the critical components of efficacious MT and their underlying neurobiological mechanisms, to the benefit of all those interested in the emerging field of contemplative science.