fMRI Neurofeedback for Emotional Regulation
fMRI neurofeedback training of amygdala activity for emotional regulation
In past past couple of decades, researchers have studied the applications of real-time functional magnetic resonance imaging neurofeedback training (rtfMRI-nf), which allows user to see their own MRI Blood Oxygenation Level Dependent (BOLD) activity level in order to regulate their brain activity during a given task. One popular area of research is in emotional processing and regulation, which involves regions such as the amygdala, insula, and lateral prefrontal cortex. Previous studies have explored using rtMRI-nf for either upregulating amygdala activity during positive emotions or downregulating amygdala activity during negative emotions (Young et al. 2018). Often, subjects are instructed to upregulate the amygdala hemodynamic response during positive autobiographical memory (AM) recall. For the downregulation of amygdala activity, subjects may be presented with negative stimuli and asked to use emotional reappraisal or reality check (to be discussed in more detail later). The purpose of this review is to analyze the protocols and validity of research studies involving rtfMRI-nf and the amygdala.
Research has suggested that rt-fMRI-nf can be used to modulate emotional experiences in healthy subjects, anxious patients, post-traumatic stress disorder (PTSD) patients, and major depressive disorder (MDD) patients (Young et al. 2018, Zhao et al. 2019, Herwig et al. 2019). As the leading cause of disability worldwide, MDD treatments are highly sought after. Pharmaceutical drugs are common treatments; however, they often cause side effects and can vary in efficacy based on the patient. Cognitive behavioral therapy (CBT) is the most commonly implemented psychological MDD treatment, but only has moderately successful response rates (Young et al. 2018). Since it has been reported that MDD patients “show an exaggerated amygdala response to negative stimuli [or threat] and an attenuated amygdala response to positive stimuli,” it can be hypothesized that targeting the amygdala during rt-fMRI-nf has potential for clinical therapy (Linhartová et al. 2019, Zhao et al. 2019). Additionally, this responsiveness to positive stimuli is inversely correlated with depression severity (Young et al. 2018).
Through various cortical and subcortical networks, the amygdala “assigns salience to stimuli, coordinates adaptive behavioral responses, and modulates cognitive processing, such as perception, attention, memory, and decision-making” (Young et al. 2018). It is believed that the right amygdala plays a role in automatic detection of emotional stimuli, while the left amygdala plays a role in detailed stimulus evaluation. Furthermore, the amygdala plays a role in threat response, which involves top-down control and amygdala downregulation from the prefrontal cortex (PFC) (Zhao et al. 2019). In particular, Sarkheil et al. 2015 showed that upregulation of the lateral prefrontal cortex using fmri-nf caused downregulation of the amygdala (Herwig et al. 2019).
Previous studies vary widely in methodological parameters and study design. This includes emotion regulation instructional and feedback protocols, quantification of regulation effect, and control groups. Firstly, studies differ in providing instruction for participants to help with emotional regulation. Withholding suggestions for regulation allows the subject to find their own strategy; however, this runs a risk that they develop slow, changing, or dysfunctional strategies. For example, both positive and negative emotions can increase ROI activity, so to ensure that the correct valence of regulation is used, it is best to provide some instruction (Young et al. 2018). For upregulation strategies, subjects can be asked to recall positive autobiographical memories (AM). For downregulation strategies, subjects are often instructed to reappraise the emotional situation or stimuli using reality check. Reality check is an “induced cognitive intervention that directs attention onto the ‘real’ features of a situation compared to anticipated or fear-exaggerated features by focusing on the description (but not the interpretation) of the situation” (Herwig et al. 2019).
When it comes to feedback protocol, most studies utilize a block design alternating ‘regulate with nf’ blocks (train), rest blocks, and ‘regulate without nf’ blocks (transfer). (Young et al. 2018) The ROI of interest, often the amygdala, is defined using an anatomic brain atlas. The neurofeedback signal comes from the percent change in the ROI during the training blocks (task-related) compared to its average signal during the rest block. Subjects are shown a scale bar representing the percent change in activity, with a target percent of 0.5-2.0% change (Young et al. 2018). Some studies have used continuous nf presentation, while other studies have used intermittent nf presentation, such as between blocks. While continuous presentation provides more information to the subject, it can be distracting and more cognitively demanding to keep up with the changing signal. Thus, it is generally preferred to use continuous nf presentation (Linhartová et al. 2019).
The efficacy of regulation is measured by the signal change difference between the experimental / regulation group and control group during training blocks (Linhartová et al. 2019). Additionally, signal change during the transfer blocks can demonstrate sustained training effects. A couple of studies included self-reported cognitive or clinical symptoms in their final analyses.
Finally, the control group is highly variable across studies. Poorly designed studies use a control group without any type feedback, making it difficult to determine if experimental differences in brain activity and clinical behavior were specific to amygdala feedback, rather than other cognitive processes, such as attention or motivation (Barreiros et al. 2019). These results would likely lead to false positives. On the other hand, the use of sham neurofeedback has potential to demonstrate amygdala specificity to nf training. Sham feedback can come from another participant’s BOLD activity (termed yoked feedback) or it can come from BOLD activity in a different region of interest (ROI). For example, the intraparietal sulcus has been used as a sham feedback signal, as it is involved in number processing (Young et al. 2018). Despite providing a stronger control group, the use of sham feedback has the potential to induce frustration in participants or even lead to their realization that it is a control signal (Linhartová et al. 2019).
In a literature review including 51 studies, most showed successful upregulation of amygdala activity (particularly the left amygdala) using rtfMRI-nf in depressed and anxious individuals compared to controls (Linhartová et al. 2019). The studies that included transfer runs show that the effects are maintained without nf several days later. The only studies that did not find regulation success mixed both upregulation and downregulation in the protocol, which could have made the task more difficult for the subjects. Some studies also found successful regulation in control groups; however, the experimental groups usually had more pronounced effects than the control (Linhartová et al. 2019). A minority of studies demonstrated mildly successful downregulation of amygdala activity (particularly the right amygdala) in response to negative stimuli as a result of rt-fMRI nf; however, many of these studies lacked a robust control / sham group. Barreiros et al. 2019 demonstrated a lateralization of the amygdala, in which upregulation activates the left amygdala, while downregulation activates the right amygdala. This fits with the theory that the right amygdala is involved in the automatic processing of emotions, and the left amygdala is involved in controlling emotions (Barreiros et al. 2019).
In conjunction with increased amygdala response to positive stimuli following nf, Young et al. demonstrated that MDD subjects showed decreased reaction times and increased accuracy in processing positive stimuli. Furthermore, using a backward masking task where face stimuli were presented below conscious awareness, they reported that the amygdala showed decreased hemodynamic activity in response to negative faces in the nf group compared to the control, “suggesting that amygdala rtfMRI-nf normalizes the amygdala response to emotional stimuli” (Young et al. 2018). So even though the task was to increase response to positive stimuli, emotional processing outcomes generalized to other types of emotional stimuli, proving that the activity was not just nonspecifically increased.
Beyond the amygdala, several studies looked for other brain regions related to task-related modulation. Connectivity between these regions was calculated either by using amygdala-seeded psychophysiological interaction (PPI) analysis or by using a partial correlation between the ROI time series (ex. between amygdala and vlPFC) (Herwig et al. 2019, Zhao et al. 2019). These connectivity values were updated in real time to adjust the scale bar presented to subjects for feedback. Zotev et al. 2011 and 2013 showed that upregulation of the left amygdala was associated with increases in functional connectivity between the amygdala, dmPFC, and ACC, while Herwig et al. 2019 showed that the downregulation of the amygdala was associated with increases in connectivity between the amygdala and vmPFC. This is supported by previous findings that both PFC and rACC connectivity with the amygdala area are reduced in MDD patients, with lower connectivity associated with increased depression severity (Young et al. 2018, Zotev et al. 2020). Further studies are required to determine if this relationship is causal, and if increasing connectivity can cause changes in clinical symptoms.
In an attempt to investigate this, Zhao et al. 2019 recruited high anxiety individuals to complete fMRI-nf training for emotional regulation with the goal of enhancing connectivity between the amygdala and the ventrolateral prefrontal cortex (vlPFC) during threat exposure. Results showed promising effects, as the experimental group showed successful increases in amygdala-vlPFC functional connectivity, along with scaled reduction in self-reported anxiety. Subjects were able to maintain increased connectivity at follow-up testing 3 days later; however, the decreased anxiety levels were not maintained at this time (Zhao et al. 2019). A review article by Young et. al provides supporting evidence that “although changes in the processing of negative stimuli were also evident following neurofeedback training, only changes to positive stimuli were associated with measures of clinical improvement” (Young et al. 2018). Further research is needed to elucidate the effect of various types of nf on clinical outcomes.
Several limitations exist in current fMRI-nf training protocols and analysis. For the experiments using negative or threatening stimuli, researchers run the risk of habituating subjects to these stimuli (Young et al. 2018). In order to prevent this, studies can try using a non-emotional but cognitively engaging task, such as counting, in order to disengage the amygdala between trials (Barreiros et al. 2019). For the connectivity analyses, one limitation to expanding training results to other brain regions is that these regions, such as the PFC, may simply be activated by motivation or learning.
As mentioned previously, future studies are needed in order to determine the dose-response of training sessions and to quantify the duration of outcome measurements in order to evaluate clinical potential. Furthermore, Young et al. 2018 recommends studies to evaluate the potential for fMRI-nf to be used in conjunction with other therapies, such as cognitive behavioral therapy (CBT). It is possible that fMRI-nf has an additive effect on these existing therapies. Finally, there is some evidence to suggest that the length of a subject’s current depressive episode was inversely correlated to the effect of training. In other words, the fMRI-nf worked better on patients earlier in the course of their depressive episode (Young et al. 2018). This needs to be further validated to provide an optimal window for nf training.
Overall, real-time fMRI neurofeedback seems to be a promising method of improving emotional regulation and symptoms in individuals with major depressive disorder. To date, studies have reported varying levels of regulation success with some indication of sustained improvement over time. In order to truly determine the effectiveness of neurofeedback treatment, studies need to include robust control groups and standardize their design protocol. Additionally, as researchers learn more about the brain’s emotional regulation processes and connections, neurofeedback treatment can be altered to most effectively ‘rewire’ the brain.
Barreiros, A. R., Almeida, I., Baía, B. C., & Castelo-Branco, M. (2019). Amygdala Modulation During Emotion Regulation Training With fMRI-Based Neurofeedback. Frontiers in Human Neuroscience, 13. https://www.frontiersin.org/article/10.3389/fnhum.2019.00089
FMRI neurofeedback in emotion regulation: A literature review—ScienceDirect. (n.d.). Retrieved April 24, 2022, from https://www.sciencedirect.com/science/article/pii/S1053811919301788
Herwig, U., Lutz, J., Scherpiet, S., Scheerer, H., Kohlberg, J., Opialla, S., Preuss, A., Steiger, V. R., Sulzer, J., Weidt, S., Stämpfli, P., Rufer, M., Seifritz, E., Jäncke, L., & Brühl, A. B. (2019). Training emotion regulation through real-time fMRI neurofeedback of amygdala activity. NeuroImage, 184, 687–696. https://doi.org/10.1016/j.neuroimage.2018.09.068
Sarkheil, P., Zilverstand, A., Kilian-Hütten, N., Schneider, F., Goebel, R., & Mathiak, K. (2015). FMRI feedback enhances emotion regulation as evidenced by a reduced amygdala response. Behavioural Brain Research, 281, 326–332. https://doi.org/10.1016/j.bbr.2014.11.027
Young, K. D., Zotev, V., Phillips, R., Misaki, M., Drevets, W. C., & Bodurka, J. (2018). Amygdala real-time functional magnetic resonance imaging neurofeedback for major depressive disorder: A review. Psychiatry and Clinical Neurosciences, 72(7), 466–481. https://doi.org/10.1111/pcn.12665
Zhao, Z., Yao, S., Li, K., Sindermann, C., Zhou, F., Zhao, W., Li, J., Lührs, M., Goebel, R., Kendrick, K. M., & Becker, B. (2019). Real-Time Functional Connectivity-Informed Neurofeedback of Amygdala-Frontal Pathways Reduces Anxiety. Psychotherapy and Psychosomatics, 88(1), 5–15. https://doi.org/10.1159/000496057
Zotev, V., Mayeli, A., Misaki, M., & Bodurka, J. (2020). Emotion self-regulation training in major depressive disorder using simultaneous real-time fMRI and EEG neurofeedback. NeuroImage: Clinical, 27, 102331. https://doi.org/10.1016/j.nicl.2020.102331