top of page

A Review of the Networks Trained by Neurofeedback

Updated: Dec 30, 2022

profile of woman with cosmos as brain

Neurofeedback has evolved from training activity at single discrete scalp sites corresponding to discrete Brodmann areas to modifying network communication. Quantitative EEG (qEEG) normative databases can reveal the key parts of a network that need training and the required direction.

With neurofeedback, we want to know where to place active electrodes and what brain activity to train. It's normal to think that the sensor should go over the part of the brain corresponding to a Brodmann area that needs to be up-trained or down-trained. That may work for some conditions that respond to single-channel training. But, one site is connected to another site, and so on, in networks. Networks, which can be functional or structural, mediate connectivity. In functional networks, there is correlated activity between regions over time. Structural networks consist of axonal projections and pathways. Although functional and structural networks can overlap, functional connectivity does not automatically imply an underlying anatomical connection. Fornito et al. (2016) explained the importance of time scale in their convergence: ". . . as functional connectivity is averaged over longer time periods, it may converge onto structural connectivity, although it is important to remember that structural and functional connectivity are different measures and may thus yield connectomes with different values of some topological parameters (Zalesky et al., 2012b)" (p. 25). Interventions as different as immobilizing a limb and neurofeedback can change functional connectivity. Newbold and Dosenbach (2021) put a cast on participants' (non-broken) dominant hand for a few weeks and performed multiple fMRI scans to see how this impacted the functional connectivity of their motor cortices. After just a day or two, functional connectivity was nearly non-existent between the left and right motor cortex, which is a faster timescale than many people had predicted. Recovery was also quite rapid in 2 of the 3 people. The premise of connectivity training is that neurofeedback (EEG and functional MRI) can increase or decrease functional connectivity to improve performance.

Commonsense About Functional Brain Areas

Peterson and Fiez (1993) observed: "A functional area of the brain is not a task area; there is no 'tennis forehand area' to be discovered. Likewise, no area of the brain is devoted to a very complex function; 'attention' or 'language' is not localized in a particular Brodmann area or lobe. Any task or 'function' utilizes a complex and distributed set of brain areas.

The areas involved in performing a particular task are distributed in different locations in the brain, but the processing involved in task performance is not diffusely distributed among them. Each area makes a specific contribution to the performance of the task, and the contribution is determined by where the area resides within its richly connected parallel, distributed hierarchy."

Key Anatomical Terms

Long brain region names can be intimidating. These six descriptions can serve as your decoder ring. Anterior or rostral refers to toward the head end, whereas posterior or caudal means toward the tail. Dorsal means toward the top of the brain, while ventral means toward the bottom of the brain. A gyrus is a ridge of convoluted brain tissue, whereas a sulcus is a furrow (Breedlove & Watson, 2020).

Networks have functional and anatomical names. For example, the dorsal attention network (DAN) corresponds to the dorsal frontoparietal network (D-FPN).

Brodmann Areas

The original Brodmann areas consisted of 47 numbered cytoarchitectural zones of the cerebral cortex based on Nissl staining. They require subdividing certain areas and vary in size and shape across individuals.

Gordon et al. (2016) compared the original Brodmann areas and several other brain atlases using fMRI and concluded: "The Brodmann parcellation (Brodmann 1909) [. . .] does successfully represent structure in the data, but is too underparcellated to represent true cortical areas. This perspective agrees with modern attempts to anatomically parcellate human cortex, which frequently observe more fine-grained architectonic divisions than those reported by Brodmann (e.g., Morris et al. 2000, retrosplenial cortex; Öngür et al. 2003, orbitofrontal cortex; Morosan et al. 2005, superior temporal gyrus; Caspers et al. 2006, inferior parietal cortex; Scheperjans et al. 2008, superior parietal cortex; Kujovic et al. 2013, extrastriate visual cortex)." The more than 100-year-old Brodmann areas lack some of the specificity needed for in vivo and functional studies.

Revised Brodmann areas

The revised Brodmann maps shown above reveal 180 regions, 100 of which were not previously identified (Glasser et al., 2016). The revised maps were contributed by Mark Dow, Research Assistant at the Brain Development Lab at the University of Oregon, to Wikimedia Commons. Cognitive neuroscience tends to use modern brain atlases of defined regions (e.g., Yeo et al., 2011) and exact coordinates like Montreal Neurological Institute (MNI) space and Talairach space (Chau & McIntosh, 2005). Researchers sometimes "convert" these coordinates to corresponding Brodmann areas since they are well-known. Brodmann areas participate in networks. They play an important role in neurofeedback because they help us target functional and structural networks instead of discrete, disconnected scalp sites. Neuroscience reveals the connections between Brodmann areas and how different networks can become active or quiet together during diverse conditions.

For example, attention to an object's location activates sites in four main areas that make up the dorsal attention network (dorsal frontoparietal network) are shown in blue. The five areas comprising the ventral attention network (ventral frontoparietal network) shown in red automatically process unanticipated environmental events. Flexible attention control involves the dynamic interaction of these top-down and bottom-up systems (Vossel et al., 2014).

Dorsal and ventral attention networks

Psychological disorders can be associated with shifts from normal activity in particular brain areas and their connections. For example, the figure below illustrates networks related to unipolar depression.

networks involved in psychological disorders

In the aftermath of traumatic brain injury (TBI), neuronal connections across the entire brain change. For example, surviving long axonal projections no longer target inhibitory neurons (Frankowski et al., 2022).

physician pointing out traumatic brain injury

Current neurofeedback protocols allow us to train networks involved in cognitive functions like attention or psychological disorders like depression using multiple electrodes simultaneously. Connectivity training enables us to increase or decrease communication between brain locations to treat symptoms and improve performance. Activations and deactivations are both important. Quantitative EEG (qEEG) normative databases can reveal the key parts of a network that need training and the required direction. Graphic courtesy of NeuroNavigator’s swLORETA Effective Connectivity Normative Database.

Brain Organization and Dynamics

The brain is organized into interactive functional, distributed networks with spatial, temporal, and content-based relationships. These networks interact through feedback loops and transiently organized aggregates of neurons, all mediated by rhythmic, oscillatory electrical discharges that ultimately produce the EEG. This process is further controlled/informed by selective attention to specific interest categories.

interactive brain networks

Each type of local cognitive, sensory processing, or emotional network produces oscillatory activity and contains internal stabilizing characteristics. These local networks exist within a global dynamic network system that links and provides an interactive capacity to the smaller networks, also operating within an oscillatory framework. Graphic courtesy of Michael W. Cole, WUSTL.

global dynamic network system

Note: This diagram shows 264 human brain regions color-coded by their network affiliation. The central sphere shows flexible hubs and their potential functions. The lines represent task-dependent inter-network communication changes. Boldfaced lines designate the largest-scale changes. A densely-connected lateral prefrontal and posterior parietal cortical network orchestrates responses to novel cognitive tasks using flexible hubs. The frontoparietal network assigns tasks to the most appropriate brain regions and shares information among these regions to master new skills (Cole et al., 2013).

The central nervous system processes incoming content. Separate regions process specialized content (e.g., auditory, kinesthetic, tactile, visual). Content is shared, integrated, compared to previous content, and analyzed. Decisions are made regarding memory and responses. All of this central nervous system activity occurs within interacting networks linked by electrical/chemical signals. Electrical discharges from network activity are recorded from the scalp surface as the EEG.

Network Overview

The networks most relevant to attention include the oculomotor, motor, affective, social, and executive circuits.

Oculomotor Network

The frontal eye field (FEF), in concert with the dorsolateral prefrontal cortex, posterior parietal cortex, basal ganglia, and thalamus, programs and initiates voluntary eye movements, inhibits eye movements toward distracting stimuli, and allows us to return our focus to locations we've experienced in the past (Thompson & Thompson, 2016). Graphic courtesy of Gorges et al. (2018).

oculomotor network

Note: SEF, supplemental eye field; PEF, parietal eye field; CEF, cingulate eye field; IPL, interparietal sulcus; DLPFC, dorsolateral prefrontal cortex; CN, caudate nucleus; SMG, supramarginal gyrus; PCC, posterior cingulate cortex.

Motor Network

The supplementary motor area (SMA), in concert with the premotor cortex, primary motor cortex, sensorimotor cortex, and cerebellum, plans, initiates, and inhibits voluntary movements and muscle contractions (Breedlove & Watson, 2020; Thompson & Thompson, 2016). Graphic courtesy of Sanders and Levitin (2020).

motor network

Note: Solid lines are excitatory, and dashed lines are inhibitory projections.

Affective Network

The pre- and subgenual areas of the anterior cingulate cortex (ACC) participate in affective circuits triggered when we make mistakes (Arnsten, 2009). The dorsal rostral cingulate zone monitors cognitive activity to predict when errors are likely, and greater executive control may be needed (Thompson & Thompson, 2016). The ventromedial prefrontal cortex projects to the amygdala, basal ganglia, hypothalamus, and brainstem arousal and reward pathways.

affective network

Social Network

The orbitofrontal cortex (OFC), along with the basal ganglia and thalamus, orchestrates the highest level of emotional processing in the nervous system. The social network is responsible for socially responsible behavior, empathy, behavioral inhibition, emotional regulation, and sound judgment (Thompson & Thompson, 2016). Graphic courtesy of Han and colleagues (2021).

social network

The blue nodes represent the mentalizing network, including the vmPFC (ventromedial prefrontal cortex), OFC (orbitofrontal cortex), dlPFC (dorsolateral prefrontal cortex), and dmPFC (dorsomedial prefrontal cortex). This network enables us to think about our own and others' mental states (Hoskinson et al., 2019). The orange dots are the mirror network (superior temporal sulcus, STS), which is active during our performance and observations of others' actions. The mirror network supports observational learning and social cognition (Sadeghi et al., 2022). The green dot is the amygdala, which detects salient stimuli. The yellow dot is the entorhinal cortex. Finally, the red dot is the anterior insular cortex, AIC.

Executive Network

The dorsolateral prefrontal cortex plays a critical role in executive functions, which Kropotov (2009) described as "the coordination and control of motor and cognitive actions to attain specific goals." Executive functions include allocation of attention, cognitive inhibition, behavioral inhibition, working memory, and cognitive flexibility. The executive network focuses and maintains continuous attention (Faraone et al., 2015). This network shows reduced activation and connectivity in ADHD.

Executive network

Attentional Processes

Attention is the selection of sensory information or cognition for enhanced processing. We can overtly or covertly attend to stimuli. In overt attention, our attentional focus and sensory orientation coincide. For example, you parse this sentence as you focus your gaze on it. In covert attention, we shift our attentional focus from our sensory orientation. For example, you attend to a reminder on the corner of your screen while you gaze at this sentence. While the midbrain superior colliculus is mainly implicated in overt attention, it may also regulate covert attention (Breedlove & Watson, 2020).