If I were to tell you that you could learn to control your brain’s activity you might be at least somewhat sceptical. However, learning to regulate neural activity is not unlike learning to control other physiological processes. Take your heartbeat for example. If you lay your hand flat on your chest you can easily monitor your current heartbeat. If I were to ask you now to decrease your heart rate, you might spontaneously start trying out different strategies: Taking deep breaths or closing your eyes and thinking about your last holiday at the beach.

This analogy, however, quickly falls short because unlike monitoring your heartbeat, there are no intuitive means of monitoring your brain’s activities. This is where neurofeedback comes into play. Neurofeedback uses neuroimaging techniques, such as electroencephalography (EEG) and functional magnetic resonance imaging (fMRI), to provide people with information about certain aspects of their brain physiology in (near) real-time. Based on this information, individuals can then learn to regulate specific aspects of brain activity in a desired manner.

Due to the vast number of neurofeedback techniques and applications, I will focus on just one exciting field of research that is gaining increasing attention: The application of real-time fMRI (rt-fMRI) neurofeedback for treating anxiety disorders.

The anxious brain

Different types of anxiety disorders are characterised by distinct behavioural syndromes. A prominent feature is an excessive fear-response to particular stimuli. These may be social (social anxiety disorder), trauma-related (posttraumatic stress disorder [PTSD]) or specific objects or animals (specific phobias).

Most anxiety disorders correlate with abnormalities in fear and stress-related brain systems. Most notably, the amygdala has consistently been found to show increased activation levels when confronted with anxiety-provoking stimuli across disorders. Similarly, the insula has also been found to be overreactive in social and specific phobia and less consistently in PTSD (Check out this 3D brain atlas to get an overview of some of the brain regions mentioned throughout this article).

Based on our knowledge about the neural underpinnings of anxiety disorders, one can now start thinking about ways to “normalise” brain activity using rt-fMRI neurofeedback. Most studies to date have focused on improving emotion regulation networks.

Emotion regulation: How we keep our emotions at bay

Emotion regulation describes the active modulation of affective (emotional) responses (positive and negative) by employing various mental or behavioural strategies. For example, if you have a spider phobia you may try to avoid encounters with spiders as best as possible. Alternatively, a more adaptive approach would be to cope with an encounter by humanising the spider and thereby changing the meaning of the stimulus. The latter is an example of reappraisal and is a common element of cognitive behavioural therapies.

On the neural level, reappraisal is generally thought to be accomplished by top-down control of emotion-processing regions, most prominently the amygdala and ventral striatum, by cognitive control regions, including multiple prefrontal and parietal regions.

Boosting emotional control with Neurofeedback

Neurofeedback studies on emotion regulation aim to increase top-down control by providing feedback about this “reappraisal network”. One approach is to display activity levels in relevant brain regions while participants try to regulate their emotional response. Initial studies involving healthy participants yielded some promising results.

Paret and colleagues, for example, asked participants to downregulate their amygdala activation while viewing aversive pictures. Thermometers indicating current activation levels in this region were placed on either side of each picture. The results revealed that while participants in both the experimental group and a sham-control group successfully decreased amygdala activation during training, only the experimental group retained lower activation levels in the right amygdala during a transfer run without neurofeedback.

Others have shown that neurofeedback of left lateral prefrontal cortex activity can also reduce amygdala activity beyond simply implementing reappraisal strategies (though this study only included a no-feedback control group). It should also be mentioned that no study reports any behavioural effects. This clearly calls for more well-controlled studies with sensitive behavioural measurements.

The clinical picture – Neurofeedback in action

While studies on healthy participants are important to establish neurofeedback methodology, we are ultimately interested in applying neurofeedback in clinical practice. Although few in numbers, some studies have shown beneficial effects for patients with anxiety disorders, particularly for specific phobias and PTSD.

For instance, a study by Zilverstand and colleagues investigated the effectiveness of rt-fMRI neurofeedback on spider phobia. They provided spider phobic women with information about their activation levels in the insula and dorsolateral prefrontal cortex via a dual-thermometer display following exposure to pictures of spiders. A control group of spider phobic individuals employed reappraisal strategies without receiving neurofeedback. People in the neurofeedback group reported lower levels of anxiety compared to the control group and this effect increased over the course of training. Although both groups showed engagement of the lateral prefrontal cortex when trying to regulate their emotional response, only the neurofeedback group achieved lower levels of insula activation over time. Follow-up assessments furthermore revealed lower levels of spider fear in both groups, but individual success in insula-regulation predicted lower long-term spider fear in the neurofeedback group.

Does this mean that we should go ahead and offer neurofeedback therapy to patients with anxiety disorders? No – at least not yet. These clinical studies merely point in a certain direction and carefully designed studies on a large scale with follow-up assessments are required to provide certainty about neurofeedback’s effectiveness.

It’s all about team work: From local activation to network responses

All neurofeedback studies highlighted so far focused on altering responsivity of very specific brain regions. We must not forget, however, that anxiety disorders and emotion regulation are characterised by networks comprising various brain regions. Therefore, one should also consider providing feedback about connectivity or (personal) activation patterns within those brain networks.

To the best of my knowledge, studies on downregulation of negative emotions have neither focused on connectivity-based neurofeedback nor on controlling activation patterns in individual anxiety-related brain networks thus far. However, some evidence for the feasibility of connectivity-based feedback comes from a study by Koush and colleagues, who focused on upregulating positive emotions.

In their experiment, participants were asked to engage with a positive social stimulus (upregulation condition) and to simply view neutral pictures (baseline condition). After each trial, participants received feedback about the degree of top-down connectivity from the dorsomedial prefrontal cortex onto the bilateral amygdala (For the interested reader: This feedback was computed by comparing a top-down versus a bottom-up dynamic causal model). The results indicated that the experimental group, but not a sham-control group, increased top-down control over training runs, and this effect was maintained during a transfer run without neurofeedback. Moreover, neurofeedback training also led to more positive ratings of the stimuli after the training sessions. These findings may also translate to downregulation of negative emotions, although up- and downregulation may engage somewhat different brain networks.

What’s next? – New avenues for neurofeedback research

While reappraisal is an important tool in the treatment of anxiety disorders and a prime target for neurofeedback interventions, it is by no means the only way. Other therapeutic approaches make use of associative learning principles that act on a subconscious level. The general idea is to strengthen the association between the fear-inducing stimulus and another non-aversive stimulus by repeatedly presenting the two together. In one iteration of this principle, the non-aversive stimulus takes on the form of a reward. This approach is generally referred to as counterconditioning.

An intriguing new study by Koizumi and colleagues explores the possibility of counterconditioning neural representations of fearful stimuli outside people’s conscious awareness. Participants firstly acquired a fear-response to a coloured grating stimulus (using a classical fear-conditioning technique). During the neurofeedback training sessions, they were then asked to increase the size of a disc, which indicated the size of a monetary reward. Unaware to the participants, the disc size represented the likelihood that the neural activation pattern in their visual cortex corresponded to that of the target grating stimulus. Even though people were unaware of the purpose of the training, they displayed an attenuated skin conductance response – indicating lower fear levels – to the target stimulus afterwards.

These findings, although promising, raise more questions than they answer: Can neural decoders be built for complex stimuli related to anxiety disorders? Do these effects translate to clinical populations? Are the treatment effects stable over time? Future studies will hopefully provide answers to these and other pressing questions.

The big picture

While neurofeedback has been most commonly researched in the context of ADHD and stroke rehabilitation, little is known about other potential clinical applications. We have seen that there is initial evidence that rt-fMRI neurofeedback may help with anxiety disorders, especially when applied together with conventional psychotherapeutic methods. However, whether neurofeedback will be useful in clinical practice depends on future findings of well-controlled neurofeedback studies and the further development of neurocircuit models of anxiety disorders.