Success and failure of controlling the real-time functional magnetic resonance imaging neurofeedback signal are reflected in the striatum

Abstract

Introduction: Over the last decades, neurofeedback has been applied in variety of research contexts and therapeutic interventions. Despite this extensive use, its neural mechanisms are still under debate. Several scientific advances have suggested that different networks become jointly active during neurofeedback, including regions generally involved in self-regulation, regions related to the specific mental task driving the neurofeedback and regions generally involved in feedback learning (Sitaram et al., 2017, Nature Reviews Neuroscience, 18, 86).

Methods: To investigate the neural mechanisms specific to neurofeedback but independent from general effects of self-regulation, we compared brain activation as measured with functional magnetic resonance imaging (fMRI) across different mental tasks involving gradual self-regulation with and without providing neurofeedback. Ten participants freely chose one self-regulation task and underwent two training sessions during fMRI scanning, one with and one without receiving neurofeedback. During neurofeedback sessions, feedback signals were provided in real-time based on activity in task-related, individually defined target regions. In both sessions, participants aimed at reaching and holding low, medium, or high brain-activation levels in the target region.

Results: During gradual self-regulation with neurofeedback, a network of cortical control regions as well as regions implicated in reward and feedback processing were activated. Self-regulation with feedback was accompanied by stronger activation within the striatum across different mental tasks. Additional time-resolved single-trial analysis revealed that neurofeedback performance was positively correlated with a delayed brain response in the striatum that reflected the accuracy of self-regulation.

Conclusion: Overall, these findings support that neurofeedback contributes to self-regulation through task-general regions involved in feedback and reward processing.

Keywords: neurofeedback; real-time functional magnetic resonance imaging; self-regulation; striatum.

Conflict of interest statement

None declared.

© 2019 The Authors. Brain and Behavior published by Wiley Periodicals, Inc.

Figures

Figure 1

Absolute distance of achieved activation level to instructed target activation level. Participants evaluated the appropriateness of their mental operation (and therewith their self‐regulation success) based on the visually provided neurofeedback information. They could assess their self‐regulation success by obtaining the absolute distance between the magnitude of the actually achieved activation level (provided neurofeedback information) and the instructed target activation level (indicated by the red rectangular). A smaller and larger distance to the target activation level represented a superior and inferior self‐regulation performance, respectively

Figure 2

Definition of striatum regions of interest. The figure shows the right‐ and left‐hemispheric striatum regions of interest (R = right, L = left) overlaid on the mean of all individual anatomical data sets and slice positions of displayed coronal (orange) and axial (purple) slices. Regions of interests included all voxels in a 3‐mm sphere centered around peak coordinates from a recent meta‐analysis on reward processing representing maximal overlap of BOLD‐signal increase in response to positive reward (Bartra et al., 2013). Provided coordinates are in Talairach space

Figure 3

Time‐resolved analysis of striatum activation in response to self‐regulation success. The figure displays the logic of the performed correlation analysis. Simulated data during gradual self‐regulation is shown: (a) An HRF‐convolved time series of performance indices is created from the absolute distance to the target activation level. Successful self‐regulation (i.e., accurate regulation of the feedback signal to the target activation level) is represented by a low value. (b) When a corresponding activation increase in the striatum ROI is delayed (in this example 6 TR), the activation peak is not paired to the improvement in performance during correlation analysis. (c) Only, when the striatum time‐course is shifted 6 TRs backwards, the increase in striatum activation is aligned to the decrease in absolute distance during correlation analysis

Figure 4

Self‐regulation with neurofeedback compared to passive viewing of neurofeedback. (a) In comparison to the rest condition, self‐regulation with neurofeedback was accompanied by increased activation in prefrontal control regions and regions involved in feedback processing (visual cortices, anterior insula) as well as decreased activation in the default mode network and the posterior insula. (b) An extended increase in subcortical activation was present during self‐regulation with neurofeedback, encompassing the striatum, thalamus, claustrum and the brainstem. The figure shows the whole‐brain RFX contrast map thresholded at FDR corrected q < 0.05 on a sample participant's inflated cortex segmentation (a) and on the average of the individual anatomical data sets (b)

Figure 5

Effect of gradual self‐regulation success on striatum activation (group and single‐subject results). The figure visualizes the BOLD‐signal level within the striatum region of interest ipsilateral to the neurofeedback target region for the two type‐of‐training conditions and across the different target‐level conditions: (a) Mean beta values for each target‐level condition across all participants separately for the no‐feedback (blue) and feedback (red) condition. Error bars represent standard errors of the means. When pooling the data across the target‐level conditions, the difference of mean‐beta values for the two type‐of training conditions (feedback, no‐feedback) was significant (p < 0.05, Bonferroni‐corrected, one‐sided). (b) Single‐subject mean beta values separately for each target‐level and type‐of‐training condition. In 80% of participants (red‐rimmed), the mean striatum activation (i.e., pooled activation across the three target‐level conditions) was higher in the feedback compared to the no‐feedback condition. Remark .Participants with black underline underwent the feedback condition first and no‐feedback condition second. Abbreviations for mental strategies: IS = Inner speech, MO = mental orchestra, VM = visual motion imagery, MD = mental drawing, MS = mental sounds, MR = mental running

Figure 6

Relationship between self‐regulation success and striatum activation level (group and single subject results). Relationship between absolute distance to target activation level and striatum activation separately for the two type‐of‐training conditions. (a) Mean Fisher z‐transformed correlation coefficients between self‐regulation success and striatum activation separately for an early time window (0–3 TR shift, immediate and slightly delayed striatum activation) and a late time window (4–7 TR shift, delayed striatum activation). The difference of the correlation values with respect to the two type‐of‐training conditions (feedback, no feedback) was only significant for the late time window (p < 0.05, Bonferroni‐corrected, one‐sided). (b) Single‐subject results for the late time window. Eighty percent of participants showed a more negative correlation between distance to target‐level and striatum activation during gradual self‐regulation when receiving neurofeedback. Remark. Participants with black underline underwent feedback condition first and no‐feedback condition seconds. Abbreviations for mental tasks: IS = Inner speech, MO = mental orchestra, VM = visual motion imagery, MD = mental drawing, MS = mental sounds, MR = mental running