The technology at the heart of the trial is called real-time functional magnetic resonance imaging amygdala neurofeedback, and it is as conceptually striking as it sounds. While lying inside a 3-Tesla MRI scanner, participants viewed distressing images drawn from standardized affective picture databases, and a custom analysis pipeline computed the difference in activity between their right amygdala and a control region with every two-second scan cycle. That signal was translated into an eight-level thermometer displayed on either side of the image, updating in real time. In the Regulate condition, patients were told to bring the thermometer down using whatever mental strategy they could devise, without any explicit instructions. In the View condition, they simply watched the images passively. Over three neurofeedback sessions, each comprising two training runs of four cycles, participants effectively practiced voluntary control over a brain structure that most people can only influence indirectly, if at all.

The scientific rationale for the experiment rests on a well-documented neural signature. Neuroimaging studies and meta-analyses have consistently shown heightened amygdala activation to aversive stimuli in borderline personality disorder, accompanied by disturbed connectivity between the amygdala and prefrontal and anterior cingulate regions, the networks thought to support emotional modulation. Dialectical behavior therapy, the first-line evidence-based treatment for the condition, has itself been shown to reduce amygdala hyperactivation in parallel with symptom improvement. The researchers reasoned that if both interventions act on the same limbic circuit, neurofeedback might amplify the neural plasticity that psychotherapy induces, in the same way that pharmacological and neuromodulatory augmentations are increasingly being tested to enhance the effects of psychotherapy. Prior uncontrolled pilot studies had shown sustained amygdala reductions and improved emotion regulation after neurofeedback training, but without control groups or long-term follow-up, the findings could not rule out placebo effects or natural symptom fluctuation.

To measure whether the brain training translated into everyday emotional life, the team used ecological momentary assessment, an intensive smartphone-based sampling method that prompted participants roughly once per hour between 9 a.m. and 10 p.m. across four-day assessment windows, delivering thirteen prompts per day through randomized intervals. Each prompt included seventeen items assessing positive and negative affect, inner tension, and dissociative symptoms. The primary endpoint was not the average level of distress but its instability, quantified using the mean squared successive difference, a statistical index of how dramatically a person’s emotional state swings from one moment to the next. This choice matters because affective instability is the defining clinical feature of borderline personality disorder, and laboratory measures often fail to capture the volatile emotional dynamics that unfold in daily life. Compliance was remarkably high, averaging 87.2 percent, which lends credibility to the resulting data.

The results paint a nuanced picture. Participants did successfully downregulate their amygdala activity during neurofeedback training, as indicated by lower hemodynamic response function amplitude and area under the curve during Regulate compared with View trials. The downregulation was strongest during the first four training runs, replicating the temporal pattern reported in earlier single-arm work, before diminishing in later runs, possibly reflecting habituation to repeated exposure to negative stimuli. But when the researchers examined the primary outcome, the daily-life instability of negative affect, they found improvements over time in both groups without consistent superiority of the neurofeedback group. Mean levels of negative affect and inner tension decreased, with significantly lower levels at three-month follow-up than at pre-treatment, but these gains were not larger in the neurofeedback arm. Self-report questionnaires, including the Borderline Symptom List-23, the Difficulties in Emotion Regulation Scale, and the Beck Depression Inventory-II, likewise showed no additional benefit of neurofeedback, and borderline symptom severity actually increased after discharge in both groups.

The neural findings were similarly mixed. In the emotion regulation task, where participants viewed negative and neutral images under Regulate, View Negative, and View Neutral conditions, the critical Regulate greater than View contrast showed no significant group-by-time interaction, suggesting no evidence that neurofeedback selectively enhanced deliberate emotion regulation. Visual inspection of individual conditions hinted at broader changes: amygdala responses during both Regulate and View decreased from pre- to post-assessment in the neurofeedback group but increased in the treatment-as-usual group, pointing to transient habituation rather than selective enhancement of regulation. In the faces task, where participants viewed expressions of disgust, sadness, and fear, amygdala reactivity showed a significant group-by-time interaction. The treatment-as-usual group decreased from pre- to post-assessment and then increased again by follow-up, whereas the neurofeedback group showed no significant change, resulting in higher amygdala reactivity in the treatment-as-usual group at follow-up. This may indicate a more stable neural trajectory after neurofeedback, though the absence of significant within-group change in that arm limits interpretation.

Resting-state functional connectivity analyses revealed a transient pre-to-post increase in amygdala coupling with precuneus and dorsomedial frontal regions in the neurofeedback group, with the largest cluster centered in the precuneus and extending into the supplementary motor area, superior frontal gyrus, and postcentral gyrus. This aligns with prior work implicating amygdala-precuneus coupling in attentional deployment during emotion regulation, and with evidence that altered amygdala-precuneus connectivity relates to childhood trauma in depression. Notably, the neurofeedback group had higher baseline childhood trauma scores, and exploratory analyses adjusting for the Childhood Trauma Questionnaire showed that the resting-state effect persisted, though with a somewhat different spatial distribution. The researchers caution that the small sample and substantial trauma imbalance preclude firm conclusions about the specific contribution of neurofeedback, but the findings suggest a transient shift in amygdala coupling with midline systems, pointing to altered affective processing rather than selective enhancement of voluntary regulation.

Why did a technique that demonstrably engages its neural target fail to produce clear clinical gains? The researchers offer several explanations, all of which carry weight for the broader field of psychotherapy augmentation. The most plausible is that any additive effects of neurofeedback were difficult to detect against the backdrop of intensive residential dialectical behavior therapy. Both groups received a highly active intervention, and if the two treatments partly engage overlapping mechanisms of emotion regulation, the incremental contribution of neurofeedback may be comparatively small and harder to distinguish statistically. A simulation-based sensitivity analysis indicated that the study reached 80 percent power only for group-by-time effects of approximately 1.7 residual standard deviations, meaning smaller additive effects may have gone undetected. The absence of significant between-group differences should therefore not be interpreted as evidence of no additional effect. Recruitment was also challenging, and the originally planned sample size of twenty-two per group was not reached, partly due to strict inclusion criteria and disruptions related to the COVID-19 pandemic.

The study also raises questions about optimal timing. During residential treatment, patients were already learning multiple emotion-regulation skills, which may have limited the added impact of neurofeedback. The post-discharge increase in borderline symptoms in both groups highlights the challenge of maintaining gains during care transitions, a known feature of the condition’s unstable symptom trajectory. The researchers suggest that neurofeedback may be more useful after discharge, when patients can apply it to consolidate and strengthen previously acquired regulation strategies, consistent with evidence from other populations that a single neurofeedback session after cognitive-behavioral therapy increased the perceived efficacy and later use of emotion-regulation strategies practiced during scanning. Future studies should test whether neurofeedback is more effective as a maintenance or booster intervention than as concurrent augmentation during intensive treatment.

The trial was preregistered on ClinicalTrials.gov in 2020 and reported according to the CONSORT 2025 guidelines and the CONSORT extension for randomized pilot and feasibility trials. Analysis code is openly available on the Open Science Framework, while data are accessible under controlled access conditions. The study was funded by the AE Foundation and approved by the Ethics Committee of the Medical Faculty Mannheim at Heidelberg University. For a field increasingly drawn to the promise of combining neurofeedback with psychotherapeutic interventions, the findings represent a methodologically rigorous reality check: voluntary amygdala modulation is achievable, but translating that capacity into robust clinical benefit when added to effective standard treatment remains an open question. The researchers emphasize that the results should be considered preliminary given the limited sample size, and that the absence of demonstrated additive benefit within the constraints of this pilot trial does not rule out a role for amygdala neurofeedback in other configurations, other populations, or other moments in the treatment pathway.

Subject of Research: Amygdala neurofeedback as an adjunct to residential dialectical behavior therapy for borderline personality disorder: a pilot randomized controlled trial

Article Title: Amygdala neurofeedback as an adjunct to residential dialectical behavior therapy for borderline personality disorder: a pilot randomized controlled trial

Article References: Hofmanová, M., Santangelo, P. S., Kleindienst, N., Schmahl, C., & Paret, C. (2026). Amygdala neurofeedback as an adjunct to residential dialectical behavior therapy for borderline personality disorder: a pilot randomized controlled trial. BMC Psychiatry, 26(1), Article 802. https://doi.org/10.1186/s12888-026-08767-7

Image Credits: AI Generated

DOI: 10.1186/s12888-026-08767-7

Keywords: Amygdala, neurofeedback, adjunct, residential, dialectical, behavior, therapy, borderline, personality, disorder, pilot, randomized