The Neuroscience of Mindfulness: Introduction to Large-Scale Brain Networks written by Richik Neogi

Our brains are complex organs composed of multiple cell types (neurons are not the only cells one can find in the brain). The brain is part of the central nervous system, which also includes the spinal cord. The brain consists of many different distinct anatomical regions. Each region serves a different purpose, but consciousness as we know it is formed by the collaboration between different regions. This article will elaborate on the function of networks that are central to a neural understanding of mindfulness; the next article will elaborate on how mindfulness training regulates the activity of these large-scale brain networks and how this regulation of activity can benefit individuals who practice mindfulness. These large-scale brain networks consist of several nodes distributed across the brain; the nodes communicate with one another to form a network. It is important to realize that each node in a network may be composed of multiple different cell types. The most relevant general cell type in the brain for our discussion is the neuron, but there are a great many different types of neurons in any given brain region. Neuroscientists still routinely discover new types of neurons. For this article, we will largely ignore cellular and molecular neuroscience, but it is critical to an understanding of neuroscience that one appreciates the cellular diversity within the brain. Instead, we will focus on the activity of a few distinct networks (which consist of individual nodes or regions) as they pertain to the practice of mindfulness. Before we dive into these large-scale brain networks, it is beneficial to understand the general structure of the brain.

              The part of our brain most directly responsible for our conscious experience as humans is the cerebral cortex. This comprises the outermost region of the brain. The word cortex derives from the Latin word for “bark.” The cortex is thus like the bark that envelops most of the other regions of the brain. The cerebral cortex is divided into several regions, each with a particular function. The frontmost part of the cerebral cortex is known as the frontal cortex – it is responsible for higher-level processing that enables the many cognitive functions that make us humans as well as our executive functioning. The part of the cortex that is at the very back of your head is called the occipital cortex. This region is responsible for high-level processing of visual information. The cerebral cortex does not operate in isolation. It requires input from and provides input to several subcortical structures. For example, all our senses (with the exception of smell) are relayed through a structure that lies near the center of the brain called the thalamus. The thalamus provides extensive input to regions of the cerebral cortex and receives feedback from cells in the cerebral cortex. Another important subcortical structure, particularly in the context of the benefits of mindfulness, is the amygdala. The amygdala is responsible for processing emotional memories (the amygdala serves other functions as well, but emotional memory is perhaps its most well-known function). When we develop fear of a stimulus, the emotional memory of this fear is stored in the amygdala. Every time we are exposed to that fear-evoking stimulus, the amygdala increases in activity. The amygdala works closely with the hippocampus, a region of the brain responsible for the formation of memories (among other functions) and the contextualization of memories (i.e. framing memories in the context of environmental cues). The most important takeaway from this very brief discussion of cortical and subcortical structures is that all regions of the brain work together to create our perception of reality. It is important to note that there are structures in the brain that are considered neither cortical nor subcortical (even though these structures may lie below the cerebral cortex) such as the medulla oblongata, which is critical for the control of breathing and reflexes. However, most of these structures are less relevant to our discussion regarding mindfulness. If you are interested in learning more about the anatomy of the brain, I encourage you to consult a neuroanatomy textbook – during my undergraduate education, we spent an entire term studying neuroanatomy – it is not possible for me to provide a detailed overview of the neuroanatomy of the human brain in this article, but I hope this single paragraph provides some background.

            Now that we are aware of the general structure of the brain, we can begin discussing large-scale brain networks as they pertain to mindfulness. There are three networks relevant to our discussion: the default mode network (DMN), the task-positive network (TPN), and the salience network (SN). We will briefly go through the constituent parts of these large-scale networks, their functions, and how they work together. Before we get into describing the three networks, it is important to discuss how neuroscientists investigating large scale networks measure brain activity. 

            There are two general means of measuring brain activity in humans: imaging and electrophysiology. Imaging comprises many different techniques. The most common technique for imaging technique that you will see in the literature today is functional magnetic resonance imaging (fMRI) with blood oxygenation level-dependent (BOLD) contrast. This technique takes advantage of the fact that oxygenated blood has different magnetic properties than deoxygenated blood. It essentially is an indirect measure of brain activity because more active regions of the brain will have higher oxygen demands (it does not directly measure the electrical activity of neurons in the brain). Another method of imaging that is sometimes used in the literature is positron emission tomography (PET). PET takes advantage of radioactive decay of certain compounds. These compounds, called tracers, are injected into the bloodstream. Different properties in the brain can be measured using different tracers. Binding to specific receptors can be measured using radioactive compounds that bind to the receptor of interest. In a landmark paper by Shulman and coworkers (as well as another publication by Raichle and colleagues in 2001), PET using a tracer that permitted measurement of cerebral blood flow allowed the authors to detect key regions of the default mode network (Shulman et al., 1997; Raichle et al., 2001). The other major technique used to measure brain activity in humans is electrophysiology. Neurons are electrically active cells. It is possible to measure the electrical activity of several neurons either by directly implanting electrodes in the brain (this is usually not done because it is extremely invasive) or by measuring the electrical activity of large populations of neurons through the skull. Due to the invasiveness of the former approach, the latter approach is usually preferred – this latter technique is called electroencephalography (EEG) though a more spatially sensitive variant exists called magnetoencephalography (MEG). Neuroscientists have developed a number of techniques to monitor brain activity, but imaging and electrophysiology remain the two dominant means of doing so in human subjects.

            The most well-known large-scale brain network is probably the DMN. A common notion in public perception is that most of our brain is not active when we are not actively engaged in a task. This is actually not true: for starters, our brains must coordinate functions that keep us alive (functions such as breathing). Moreover, there exists a distributed network within the brain known as the DMN that is responsible for constructing our internal state. This network is most active when we are at rest (not engaged in tasks or conscious perception), though Vinod Menon’s group posits that it is also active during passive sensory processing and other authors claim that the DMN is active during internally and externally oriented tasks (Kelly et al., 2008; Elton and Gao, 2015). Thus, a more appropriate interpretation (in light of findings by Elton and Gao in 2015) of the literature is that the DMN alters its connectivity with other networks depending on the task and its activity is the greatest in isolation while we are at rest. The DMN is responsible for multiple functions within the brain: self-referential cognition, social cognition, personally relevant episodic memory, language/semantic memory, and mind wandering (Menon, 2023). The DMN has nodes across the entire brain, but the most important nodes are the medial prefrontal cortex (mPFC), the posterior cingulate cortex (PCC), the precuneus, the entorhinal cortex (part of the medial temporal lobe), and the lateral parietal cortex (Raichle, 2015). The PCC, precuneus, and the mPFC are core components of the DMN. As in all large-scale brain networks, each of these nodes serves a different purpose. In the discussion section of their landmark 2001 study, Raichle and coworkers proposed that the PCC and the precuneus may be associated with the general monitoring of streams of consciousness, whereas the mPFC may be associated with evaluating the relevance of these streams (Raichle et al., 2001). When the DMN is active, our brains are busy monitoring and interpreting our internal state (not to be confused with the notion of interoception). The lateral parietal areas of the cortex as well as the PCC are associated with recollection of memories – these regions are heavily interconnected with the hippocampus (the hippocampus is not a node of the DMN, but it interacts heavily with it). It should be noted that the entorhinal cortex, which Raichle states is often considered part of the DMN (Raichle, 2015), is a key source of input to the hippocampus. Menon states that the PCC is associated with autobiographical recall while the angular gyrus (part of the lateral parietal cortex) is associated with elaboration of the contents of memories (Menon, 2023). Another key function of the DMN is mind-wandering. Mind-wandering refers to thought processes unrelated to active tasks. If you are ever engaged in a task and then you suddenly find yourself “stuck in your head” and not focused on the task at hand, that is called mind wandering: it is a state of introspection that is often associated with rumination. The mPFC along with other nodes in the DMN (the lateral parietal cortex, precuneus, and PCC) are strongly associated with mind-wandering. What then, is the general purpose of the DMN? The DMN is responsible for integrating the five functions described by Menon in a manner that generates an internal narrative. It is responsible for our internal dialog, and it interprets our internal experience in the context of the past as well as our predictions for the future. If the DMN in general mediates our internal narrative at rest, then what network is most active when we are actively engaged in tasks that require our interaction with the external world?

               The answer to this question is the task-positive network (TPN). The TPN is often spoken of with reference to the DMN, so much so that the term “task-negative network” is sometimes used interchangeably with the DMN. The TPN does not consist of a single network, but rather several different networks that work together when we are engaged in an externally oriented task. Elements of the TPN include the dorsal attention network (DAN) and the frontoparietal control network (FPCN). Often, in the literature, a common theme is the presence of an anticorrelation between activity in the DMN and activity in networks that comprise the TPN. Indeed, Menon cites evidence that disengagement of the DMN is required during externally oriented tasks (Menon, 2023). However, Raichle cites evidence that the anticorrelation between the DAN (as we mentioned previously, this is a component of the TPN) and the DMN is only present a certain percentage of the time in the cat homologs of these networks when electrophysiology is used to measure activity. He goes on to discuss methodological differences in the analysis of fMRI data may account for why there has been demonstrated a consistent anticorrelation between the DAN and the DMN – in the analysis of fMRI data, signal that is always present in the background is often removed or subtracted. This “hides” the potential relationship between the DAN and the DMN (Raichle, 2015). This data processing was absent from the studies he cites conducted in the cat. Other researchers have gone on to demonstrate that elements of the TPN (the FPCN specifically) can mediate the anticorrelation between the DAN and the DMN during a movie watching task (Gao and Lin, 2012), although this data falls victim to the same issues that Raichle mentioned. Thus, in studies conducted in humans using fMRI, there is consistent evidence suggesting that the DMN is most active when certain elements within the TPN are not active and vice versa. Before moving on to the final large-scale brain network in our discussion, let us cover some key nodes within the TPN, as we did for the DMN.

         Given that the TPN consists of a few different subnetworks within the brain, it is challenging to isolate which nodes belong to which subnetworks during the tasks in which the TPN is observed to be active. However, an early study by Fox and coworkers was instructive in determining the general components of the TPN. In this study, the authors determined the regions in the brain in which activity was anticorrelated with activity in the DMN. Two components of the DAN were identified as part of the TPN: the intraparietal sulcus and the frontal eye fields (Fox et al., 2005). The intraparietal sulcus is a groove in the parietal lobe of the cortex (the term sulcus, plural sulci, refers to the grooves on the wrinkled surface of the brain versus the ridges which are called gyri). It is located towards the back of the brain on the top portion. The frontal eye field is an area in the frontal cortex that is part of the precentral sulcus. Components of the FPCN were also identified as part of the TPN such as the dorsolateral prefrontal cortex (dlPFC) and parts of the inferior parietal lobule. The dlPFC is a region at the very front part of the frontal cortex, sitting on the side that points away from the midline. It plays an important role in working memory functions during a variety of tasks. The inferior parietal lobule is located directly below the intraparietal sulcus. The full list of nodes within the TPN as described by Fox and coworkers in their seminal 2005 paper is as follows: the intraparietal sulcus, inferior parietal lobule, precentral sulcus (the frontal eye fields), the dlPFC, the middle temporal region, the insula/operculum, and the supplementary motor area. They identified fifteen foci distributed among these regions as key components of the TPN, the network that whose activity was anticorrelated with the “task-negative network” or the DMN. Indeed, in most healthy individuals, activity in the TPN is anticorrelated with activity in the DMN. However, in some individuals with certain psychiatric disorders (such as bipolar disorder), this anticorrelation is absent, but can be restored by mindfulness training (Chou et al., 2022). It should also be noted that one node in the paper by Fox and coworkers describing the TPN, the insula, is a key node in another network, the salience network (SN), and there is evidence that this network serves a distinct purpose in the TPN that is separate from that of the FPCN and the DAN.

            The SN is a brain network that selects stimuli in a context-specific manner and directs access to resources (i.e. other networks) in the brain. In simple terms, as the name suggests, it directs attention towards salient stimuli regardless of the context (i.e. visual, auditory, interoceptive, etc.); as Vinod Menon states, it is responsible for focusing the ‘spotlight of attention’ on behaviorally relevant stimuli (Menon, 2015). Given that the saliency of stimuli is critical for executing many tasks that researchers utilize, it was difficult to isolate (in fMRI scans) the salience network from the other components of the TPN during task execution. Seeley and coworkers (Seeley et al., 2007) were able to do so by analysis of intrinsic connectivity (connectivity of regions measured during rest) of two regions of interest (ROIs). In each individual subject, the first ROI was selected during a working memory task and was centered in a region (the dlPFC) reported to be involved in these working memory tasks. The second ROI was also selected during this task as a region hypothesized to be a component in the SN (it was centered in the right anterior insula). The paper goes on to confirm that this SN is distinct from the network centered in the dlPFC. While activity in the SN was to a degree temporally correlated with activity in the dlPFC-centered network, temporal activity in one network only explained a very small percentage of the variance in temporal activity of the other network, further suggesting that these networks were distinct. Additional evidence that the SN is distinct from other networks that comprise the TPN comes from diffusion tensor imaging, which traces white matter tracts. White matter contains the axons of neurons (axons are the parts of neurons important for the transmission of electrical signals). The other nodes of the TPN are connected by distinct white matter tracts compared to the nodes of the SN. As we will explore in the next article, the SN is a network that constitutes a major target of mindfulness practices. Thus, it is critical for our discussion to outline the key nodes in the SN.

             As was found by Seeley and coworkers (Seeley et al., 2007), one major node of the SN is the insula, particularly the anterior insula (the posterior insula, while not technically part of the SN, is responsible for visceral, interoceptive awareness such as that of the breath, and it is connected to the greater SN). The dorsal (the surface that points upwards in a standing human being) aspect of the anterior insula is most relevant to the SN. This part of the anterior insula is heavily connected to the dorsal anterior cingulate cortex (dACC), the other cortical node of the SN. Both the ACC and the insula are home to a special type of neuron known as the von Economo neuron. These neurons possess axons with very wide diameters, enabling rapid signal transmission to regions inside and outside of the SN (Menon, 2015). In addition to these two cortical nodes within the SN, there are several subcortical nodes, particularly in the limbic regions of the brain (the limbic system is a part of the brain that regulates functions such as memory and emotion). These subcortical nodes include: the amygdala, the ventral striatum (important in reward processing and for mediating motivated behavior), the dorsomedial thalamus, hypothalamus (important for mediating homeostasis within the brain and the whole body), and the substantia nigra/ventral tegmental area (these are the regions in the brain that contain cells which produce the neurotransmitter dopamine).

           The two cortical nodes (the anterior insula and the dACC) have distinct functions in the SN. The anterior insula receives sensory input from other regions of the brain as well as emotional and saliency input from the subcortical nodes of the SN. Its key role is saliency detection; the anterior insula processes information regarding various potentially salient phenomena and determines which of these phenomena are actually salient or deserve the mind’s “spotlight.” In contrast, the dACC receives little sensory input. The dACC is associated with conflict monitoring and action selection. In this context, conflict monitoring refers to conflicts between two competing stimuli, sometimes of different modalities (i.e. the actual color of a color word versus the meaning of that color word, cognition required for a task known as the Stroop task). In this context, the dACC is also responsible for resolution of these conflicts and selection of the appropriate action. Unlike the anterior insula, the dACC is well-connected with motor regions of the brain and the spinal cord. When a salient event that demands our attention (or our action) is detected, the anterior insula sends signals that suppress activity in the DMN and increase activity in the task-related regions of the brain (i.e. those comprising the TPN). Simultaneously, signals from the anterior insula are routed through connections to the dACC, where processing regarding conflicting information and the signals necessary to act on salient information occurs. Thus, the anterior insula node of the SN receives multimodal sensory and motivational input from subcortical nodes of the SN as well as other regions of the brain responsible for sensation, determining the ultimate conscious-level “salience” of stimuli. It routes this information to the dACC, which resolves conflicts in this information and send signals to regions of the brain responsible for action. Meanwhile, both the dACC and the anterior insula divert resources away from elements of the DMN towards regions of the brain necessary for addressing the task at hand (Menon, 2015).

                How do these networks work together in the context of mindfulness? To conceptualize how these networks work together during mindfulness, let us revisit key components of the definition of mindfulness that was put forth by Bishop and coworkers (Bishop et al., 2004): direction of attention towards an “anchor” (usually the breath in focused attention mindfulness, but also bodily sensations or emotions in open monitoring mindfulness) and adoption of an open, accepting orientation towards one’s present experience. The direction of attention piece entails a shift in the salience of the “anchor” component in our present experience – for most of us, normally, the breath is not really a salient stimulus, but when we practice mindfulness, it becomes a salient stimulus. When we notice our minds wandering (one of the functions of the DMN), we bring our attention back to the present. Remember, mindfulness does not entail a total rejection of mind-wandering. Rather, we non-judgmentally observe our minds wandering and bring our attention back to the anchor. Thus, we can reason that decreased activation of the DMN is not the immediate goal of mindfulness practices. Rather, we attempt to increase the degree of control that the SN has over the DMN and elements of the TPN. In directing our attention towards the anchor and the present moment, certain aspects of the TPN are engaged, but the goal, at least for the average practitioner of mindfulness, is not to completely shut off activity in the DMN. When we practice mindfulness, we are really seeking the ability to regulate the balance between activity of the TPN and activity of the DMN through the SN. In the next article, we will explore this three-network concept in more detail by drawing information from the literature and investigate how mindfulness training affects the structure of the brain including key nodes in these networks.

References

Bishop, S. R., Lau, M., Shapiro, S., Carlson, L., Anderson, N. D., Carmody, J., Segal, Z. V., Abbey, S., Speca, M., Velting, D., & Devins, G. (2004). Mindfulness: A proposed operational definition. Clinical Psychology: Science and Practice, 11(3), 230–241. https://doi.org/10.1093/clipsy.bph077

Chou, T., Dougherty, D. D., Nierenberg, A. A., & Deckersbach, T. (2022). Restoration of default mode network and task positive network anti-correlation associated with mindfulness-based cognitive therapy for bipolar disorder. Psychiatry Research. Neuroimaging, 319, 111419. https://doi.org/10.1016/j.pscychresns.2021.111419

Elton, A., & Gao, W. (2015). Task-positive Functional Connectivity of the Default Mode Network Transcends Task Domain. Journal of Cognitive Neuroscience, 27(12), 2369–2381. https://doi.org/10.1162/jocn_a_00859

Fox, M. D., Snyder, A. Z., Vincent, J. L., Corbetta, M., Van Essen, D. C., & Raichle, M. E. (2005). The human brain is intrinsically organized into dynamic, anticorrelated functional networks. Proceedings of the National Academy of Sciences of the United States of America, 102(27), 9673–9678. https://doi.org/10.1073/pnas.0504136102

Gao, W., & Lin, W. (2011). Frontal parietal control network regulates the anti‐correlated default and dorsal attention networks. Human Brain Mapping, 33(1), 192–202. https://doi.org/10.1002/hbm.21204

Kelly, A. M. C., Uddin, L. Q., Biswal, B. B., Castellanos, F. X., & Milham, M. P. (2008). Competition between functional brain networks mediates behavioral variability. NeuroImage, 39(1), 527–537. https://doi.org/10.1016/j.neuroimage.2007.08.008

Menon, V. (2015). Salience Network. In A. W. Toga (Ed.), Brain Mapping (pp. 597–611). Academic Press. https://doi.org/10.1016/B978-0-12-397025-1.00052-X

Menon, V. (2023). 20 years of the default mode network: A review and synthesis. Neuron, 111(16), 2469–2487. https://doi.org/10.1016/j.neuron.2023.04.023

Raichle, M. E. (2015). The brain’s default mode network. Annual Review of Neuroscience, 38, 433–447. https://doi.org/10.1146/annurev-neuro-071013-014030

Raichle, M. E., MacLeod, A. M., Snyder, A. Z., Powers, W. J., Gusnard, D. A., & Shulman, G. L. (2001). A default mode of brain function. Proceedings of the National Academy of Sciences, 98(2), 676–682. https://doi.org/10.1073/pnas.98.2.676

Seeley, W. W., Menon, V., Schatzberg, A. F., Keller, J., Glover, G. H., Kenna, H., Reiss, A. L., & Greicius, M. D. (2007). Dissociable Intrinsic Connectivity Networks for Salience Processing and Executive Control. Journal of Neuroscience, 27(9), 2349–2356. https://doi.org/10.1523/JNEUROSCI.5587-06.2007

Shulman, G. L., Fiez, J. A., Corbetta, M., Buckner, R. L., Miezin, F. M., Raichle, M. E., & Petersen, S. E. (1997). Common Blood Flow Changes across Visual Tasks: II. Decreases in Cerebral Cortex. Journal of Cognitive Neuroscience, 9(5), 648–663. https://doi.org/10.1162/jocn.1997.9.5.648