The Lack of Neurofeedback Training Regulation Guidance and Process Evaluation May be a Source of Controversy in Post-Traumatic Stress Disorder–Neurofeedback Research: A Systematic Review and Statistical Analysis

Abstract

Objectives:

Neurofeedback (NF) based on brain–computer interface (BCI) is an important direction in adjunctive interventions for post-traumatic stress disorder (PTSD). However, existing research lacks comprehensive methodologies and experimental designs. There are concerns in the field regarding the effectiveness and mechanistic interpretability of NF, prompting this study to conduct a systematic analysis of primary NF techniques and research outcomes in PTSD modulation. The study aims to explore reasons behind these concerns and propose directions for addressing them.

Methods:

A search conducted in the Web of Science database up to December 1, 2023, yielded 111 English articles, of which 80 were excluded based on predetermined criteria irrelevant to this study. The remaining 31 original studies were included in the literature review. A checklist was developed to assess the robustness and credibility of these 31 studies. Subsequently, these original studies were classified into electroencephalogram-based NF (EEG-NF) and functional magnetic resonance imaging-based NF (fMRI-NF) based on BCI type. Data regarding target brain regions, target signals, modulation protocols, control group types, assessment methods, data processing strategies, and reported outcomes were extracted and synthesized. Consensus theories from existing research and directions for future improvements in related studies were distilled.

Results:

Analysis of all included studies revealed that the average sample size of PTSD patients in EEG and fMRI NF studies was 17.4 (SD 7.13) and 14.6 (SD 6.37), respectively. Due to sample and neurofeedback training protocol constraints, 93% of EEG-NF studies and 87.5% of fMRI-NF studies used traditional statistical methods, with minimal utilization of basic machine learning (ML) methods and no studies utilizing deep learning (DL) methods. Apart from approximately 25% of fMRI NF studies supporting exploratory psychoregulatory strategies, the remaining EEG and fMRI studies lacked explicit NF modulation guidance. Only 13% of studies evaluated NF effectiveness methods involving signal classification, decoding during the NF process, and lacking in process monitoring and assessment means.

Conclusion:

In summary, NF holds promise as an adjunctive intervention technique for PTSD, potentially aiding in symptom alleviation for PTSD patients. However, improvements are necessary in the process evaluation mechanisms for PTSD-NF, clarity in NF modulation guidance, and development of ML/DL methods suitable for PTSD-NF with small sample sizes. To address these challenges, it is crucial to adopt more rigorous methodologies for monitoring NF, and future research should focus on the integration of advanced data analysis techniques to enhance the effectiveness and precision of PTSD-NF interventions.

Impact Statement

The implications of this study are to address the limited application of Neurofeedback training (NFT) in post-traumatic stress disorder (PTSD) research, where a significant portion of the approaches, foundational research, and conclusions lack consensus. There is a notable absence of retrospective statistical analyses on NFT interventions for PTSD. This study provides a comprehensive statistical analysis and discussion of existing research, offering valuable insights for future studies. The findings hold significance for researchers, clinicians, and practitioners in the field, providing a foundation for informed, evidence-based interventions for PTSD treatment.

Keywords

neurofeedback process assessment protocol guidance PTSD statistical analysis

Introduction

Post-traumatic stress disorder (PTSD) refers to a mental health disorder that occurs after an individual directly experiences or witnesses one or more traumatic events, such as life-threatening illnesses, sudden disasters, or developmental-inappropriate sexual experiences (Ho et al., 2021). Current treatments for PTSD include psychotherapy, pharmacotherapy, and physical interventions (Grey et al., 2014; Levin, 2015; FOA et al., 2008; Erickson et al., 2014; Cloitre, 2014; Post et al., 2021). However, these methods face significant limitations. Psychotherapy requires a high level of therapeutic expertise, has long treatment durations, lacks convenience, and lacks an objective quantitative assessment system, greatly restricting its widespread clinical application. Pharmacotherapy, on the other hand, suffers from delayed onset of action, high dropout rates, and side effects such as dependency (Friedman, 2013); Regarding physical interventions, further research is needed for precise stimulation parameter adjustment and accurate target localization in PTSD patients (Post et al., 2021). In summary, the existing treatment systems mentioned above all have room for improvement.

Compared with existing treatment methods, neurofeedback (NF) is a form of biofeedback that utilizes various forms of feedback (visual, auditory, or tactile) to provide real-time information about central nervous system activity (such as brain activity collected through functional magnetic resonance imaging [fMRI], electroencephalogram [EEG], functional near-infrared spectroscopy [fNIRS], etc.) to users. It operates on the principles of operant conditioning, allowing users to learn to voluntarily enhance or inhibit neural activity with the goal of improving corresponding brain function or behavioral performance. Numerous studies have demonstrated the neuroplasticity effects of NF on the brain (Collura, 2014).

Currently, NF has become an important noninvasive neuromodulation technique, showing unique advantages in enhancing cognitive and behavioral performance and intervening in neurological or psychiatric disorders, particularly for cognitive and psychological impairments (Collura, 2014). In addition, NF represents one of the earliest applications of brain–computer interfaces (BCIs) (Van der Kolk et al., 2016), playing a crucial role as both a component and key technology of BCI systems (Hull, 2002). It enables users to learn to control their brain activity to generate and strengthen brain signal features that drive BCI systems (Curran and Stokes, 2003).

Currently, there is existing research utilizing neurofeedback training (NFT) interventions for PTSD (Nicholson et al., 2020; Van der Kolk et al., 2016; Leem et al., 2020; Fine et al., 2023; Askovic et al., 2017; Leem et al., 2021; Nooripour, 2017; Shaw et al., 2023; Askovic et al., 2020; Nicholson et al., 2016; Kluetsch et al., 2014; Gapen et al., 2016; Du Bois et al., 2021; Ros et al., 2017; Nicholson et al., 2023; Weaver et al., 2020; Zweerings J et al., 2018; Nicholson et al., 2022; Gerin et al., 2016; Zotev et al., 2018; Lieberman et al., 2023; Zhao et al., 2023; Fruchtman-Steinbok et al., 2021; Misaki et al., 2021; Misaki et al., 2019; Nicholson et al., 2018; Nicholson et al., 2017; Goel et al., 2023; Zweerings et al., 2020; Misaki et al., 2018; Zweerings et al., 2021; Schönenberg et al., 2017; Fruchter et al., 2024; Gurevitch et al., 2024; Choi et al., 2024), yet the efficacy and methodologies of NFT interventions remain contentious (Schönenberg et al., 2017; Cortese et al., 2016). Key points of contention include the neural feedback mechanisms affecting patients and whether the NF process is transparent, along with the certainty of its effectiveness. Therefore, to advance the research on PTSD-NFT further, this study conducted a literature review and statistical analysis of the primary research on NF interventions for PTSD, distilled key techniques, and explored the origins of these controversies to propose avenues for improvement.

Methods

Study selection

As shown in Figure 1, the Sci database was searched comprehensively up to April 1, 2025, using the keywords “Neurofeedback” and “post-traumatic stress disorder (PTSD),” yielding a total of 119 entries. An additional search combining “PTSD” with “BCI,” “EEG,” “fNIRS,” or “fMRI” identified an additional 8 potential studies (n = 8).

FIG. 1.

The inclusion and exclusion process of review.

Inclusion criteria comprised the following: (1) studies using a broad BCI approach, specifically those utilizing fMRI/EEG/fNIRS systems or paradigms for characterizing and modulating brain signals in PTSD patients; studies focusing on noncentral nervous system signals were excluded; and (2) studies involving autonomous regulation of brain signals in PTSD patients during experimental procedures. It is crucial to note that, for the sake of focus, this review defines PTSD-NF as the subjective modification of current brain activity/state by the participants themselves. Therefore, studies investigating natural recovery/change in PTSD patients’ brain states, implicit/exogenous regulation, and follow-up/signal analysis were excluded from this review. In the literature selection process for this study, we included only peer-reviewed journal articles. Although preprints and gray literature may contain valuable information, we opted to include only peer-reviewed publications to ensure the quality and reliability of the research. This selection criterion helps maintain the rigor and reproducibility of the study’s findings.

The study independently screened the titles and abstracts generated from the search based on the inclusion criteria. After removing duplicates, 122 English articles were identified. A total of 87 articles were excluded, including 9 conference articles, 12 dissertations, 37 reviews, 9 case reports, 18 preprints, 1 secondary analysis of data already included in the study, and 1 preliminary study on new regulatory targets for NFT.

The remaining 35 studies underwent full-text screening. Any uncertainties regarding the eligibility of articles were discussed with other reviewers. Through this process, an additional study that did not meet both inclusion criteria was excluded. This study did not primarily focus on NF intervention for PTSD despite mentioning NF in their abstracts or texts (e.g., Dehghani et al., 2023; Cortese et al., 2016), instead emphasizing emotion regulation and stimuli. The final repository consists of 34 original studies.

Statistics and examinations included in the study

The included studies were categorized into EEG-based and fMRI-based NFT interventions. Their key parameters—including target brain regions, signals, paradigms, control conditions, strategies, assessment methods, data processing approaches, and main conclusions—are summarized in Tables 1 and 2. To assess potential methodological issues and design flaws in the included studies, statistical analyses were conducted on extracted parameters. Chi-square goodness-of-fit tests were applied to categorical variables—such as target brain regions, NF paradigms, and control conditions—to evaluate whether their distributions deviated significantly from a uniform or theoretically expected pattern, thereby identifying possible biases or over-represented elements in existing research. In addition, independent-samples t-tests were performed on continuous variables (e.g., number of participants, baseline scores) between experimental and control groups, aiming to examine the appropriateness and balance of sample allocation. These statistical assessments were intended to provide insights into the methodological rigor and potential limitations of current NF intervention studies.

Table 1.

Parameter Statistics of Electroencephalogram Based Neurofeedback Intervention for Post-Traumatic Stress Disorder Research

Study	Target region	Signal	Paradigm	Control size	Sample size	Regulation strategy	Evaluation methods	Data processing methods	Conclusion
Nicholson et al., 2020	DMN and SN	Alpha-rhythm EEG	Alpha desynchronizing feedback	PTSD-CG (n = 18)	PTSD-NF (n = 18)	×	×	fMRI scans for DMN and SN	×
Van der Kolk et al., 2016	Temporal regions	2–6 Hz,22–36 Hz, and 10–13 Hz EEG	Visual feedback video game	PTSD-CG (n = 24)	PTSD-NF (n = 28)	×	Scale assessment and at one month follow-up	Chi-square test	A significantly smaller proportion of NF met criteria for PTSD than the waitlist (WL) condition
Leem et al., 2020	The parietal lobe	Alpha and theta-rhythm EEG	Meditative state with strengthened and audio feedback	PTSD-CG (n = 23)	PTSD -NF (n = 23)	×	Comparing the PCL-5-K scores on week 8. Anxiety	LORETA analysis, Z-scores, t-test, U-test	×
Fine et al., 2023	Amygdala	AmygEFP	3D audio-visual animated scenario	PTSD-CG (n = 27)	PTSD -NF (n = 28)	×	(1) Follow-up; (2) amygdala real-time fMRI-NF	GLM, ANCOVA, ITT, LMM	A marginally significant immediate reduction in PTSD symptoms
Askovic et al., 2017	Sensory motor region and parietal and temporal regions	SMR-rhythm EEG, alpha rhythm	×	×	PTSD -NF (n = 2)	×	Completed VCPT, ERP measures	×	×
Leem et al., 2021	The parietal lobe	Alpha and theta-rhythm EEG	Via audio feedback in an eye-closed state	PTSD-CG (n = 9)	PTSD -NF (n = 10)	×	Evaluating PTSD symptom change	QEEG Analysis, LORETA, t-test, CUA	The outcome was PTSD symptom change
Nooripour, 2017	The parietal lobe	Alpha and theta-rhythm EEG	×	PTSD-CG (n = 15)	PTSD -NF (n = 15)	×	Evaluating PTSD symptom change	SW test, Levene test, Means, SD	Results showed significant improvement for depression and PTSD symptoms
Shaw et al., 2023	The parietal lobe	Alpha rhythm EEG	Visual feedback during the interactive games	PTSD-CG (n = 17)	PTSD-NF (n = 18)	×	Follow-up; fMRI scan with N-back tasks	PSD, ANOVA, Fast ICA, regression analyses	(1) Significant reductions in PTSD symptom severity scores; (2) NFB enhanced engagement of top-down cognitive and emotional control centers and its performance correlated with increased activity in brain regions involved in top-down control
Askovic et al., 2020	Sensory motor region, temporal regions, Frontal	SMR-rhythm EEG, 8–10 Hz or 6–9 Hz EEG, alpha-rhythm EEG	×	PTSD-CG (n = 13); HC was from database	PTSD-NF (n = 13)	×	Self-Report and Cognitive ERP Measures	ANCOVA	NFT demonstrated an increased P3 amplitude and improved behavioral performance
Nicholson et al., 2016	The parietal lobe	Alpha-rhythm EEG	The “Space Race” game	×	PTSD-NF (n = 21)	×	Comparing amygdala complex connectivity using fMRI before and after NF	t-test; short-time Fourier Transformations; ANOVA; Multiple regression analysis	This shift in amygdala complex connectivity was associated with reduced arousal, greater resting alpha synchronization, and was negatively correlated to PTSD symptom severity
Kluetsch et al., 2014	The parietal lobe	Alpha and theta-rhythm EEG	The “Space Race” game	×	PTSD -NF (n = 21)	×	Followed by a resting-state fMRI scan	Spectral analysis, Connectivity analysis, ICA, Multiple regression analysis	A spontaneous EEG “rebound” after neurofeedback, pointing to homeostatic/compensatory mechanisms operating in the brain
Gapen et al., 2016	Temporal regions	Theta-rhythm EEG, 22–36 Hz EEG, SMR-rhythm EEG	×	×	PTSD -NF (n = 17)	×	Comparing Davidson Trauma Scale scores before and after NF	Hierarchical Linear and Non-Linear Modeling, Multilevel mediation analyses	NF significantly reduced PTSD symptoms and preceded gains in affect regulation
du Bois et al., 2021	The parietal lobe	Alpha-rhythm EEG	The “Space Race” game	PTSD (n = 9) control group	PTSD -NF (n = 10); PTSD -MI (n = 10)	×	Comparing band power and PCL-5, HTQ scores	SW-test, FBCSP, LDA, Friedman test	A low-cost, neurotechnological solution for neurofeedback as an effective treatment of PTSD
Ros et al., 2017	The parietal lobe	Alpha-rhythm EEG	Visual feedback with a dynamic bar graph or a “Space Race” game	Healthy controls (n = 30)	PTSD-NF (n = 21)	×	Comparing with HC about LRTCs before and after NFT	ICA, Hilbert transform, MANOVA	An inverse U relationship between strength of LRTC and oscillation amplitude
Nicholson et al., 2023	The parietal lobe	Alpha-rhythm EEG	Continuous visual feedback with a photo	PTSD-CG (n = 18); HC (n = 32)	PTSD-NF (n = 20)	×	Follow-up and comparing with HC and PTSD-NF before and after NFT	Pearson’s chi-squared, Fisher’s exact tests	NFT can rescue pathologically reduced alpha rhythmicity. Repeatedly associated with symptoms of overexcitement and cortical disinhibition in PTSD.
Fruchter et al., 2024	Amygdala	EFP	Prism software	×	PTSD -NF (n = 66)	×	Follow-up	ANOVA	Statistically significant improvement at 8 weeks and 3-month follow-up
Gurevitch et al., 2024	Amygdala	AmygEFP	Auditory/audiovisual	HC (n = 30); FM (n = 19)	PTSD (n = 48)	×	Self-report assessments; Post-training fMRI	LME	Increased neurofeedback training (NFT) ability enhances AmygEFP regulation and emotional processing
Choi et al., 2024	×	Alpha waves	Auditory	PTSD (n = 30)	PTSD (n = 28)	×	Self-report assessments; EEG assessment before and after testing	ANOVA	NFT significantly improved mental health, emotional stability, and stress resilience in the experimental group, but no significant difference was found in trauma symptoms or left-side stress resilience

EEG, electroencephalogram; AmygEFP, amygdala-electrical fingerprint; PTSD-CG, post-traumatic stress disorder control group; NF, neurofeedback; fMRI, functional magnetic resonance imaging; HC, healthy control.

Table 2.

Parameter Statistics of Functional Magnetic Resonance Imaging Based Neurofeedback Intervention for Post-Traumatic Stress Disorder Research

Study	Target region	Signal	Paradigm	Control size	Sample size	Regulation strategy	Evaluation methods	Data processing methods	Conclusion
Weaver et al., 2020	Whole-brain	BOLD	A-NF	HC (n = 9)	PTSD -NF (n = 10)	×	Assess the effectiveness, cost-utility, and safety	MVPA and SVM	×
Zweerings et al., 2018	ACC	BOLD	The face from neutral to a wide smile	HC (n = 9)	PTSD -NF (n = 9)	Recalling positive emotional autobiographic memories, imagining to perform a hobby, concentrating on a specific physiological perception	Calculating the signal changes before and after NFT	GLM: Pearson product moment correlation coefficient	Both groups achieved regulation, and learning rate was lower in patients, correlating with symptom severity.
Nicholson et al., 2022	PCC	BOLD	Respond to personalized trauma-related words and neutral words	HC (n = 15)	PTSD -NF (n = 14)	×	(1) Comparing the differential activation and follow-up comparisons from two groups. (2) Classify brain activation during NFT.	ANOVA, LRA, MKL	(1) Both groups reduced relieving symptoms. (2) The PTSD group reduced symptoms of distress. (3) Both groups were able to downregulate the PCC.
Gerin et al., 2016	Amygdala	rsFC	Respond to trauma Scripts		PTSD -NF (n = 3)	×	Measure symptoms and rsFC changes before and after NFT.	Functional Connectivity	Showed clinically meaningful improvements.
Zotev et al., 2018	Amygdala- prefrontal	BOLD	Explore various other happy autobiographical memories	PTSD-CG (n = 11)	PTSD -NF (n = 20)	Thinking of and write down five happy autobiographical memories	Comparing CAPS ratings, functional connectivity, and alpha EEG coherence before and after NFT.	GLM, Functional Connectivity, R², coherence analysis	80% of the EG participants demonstrated clinically meaningful reductions in CAPS ratings, compared with 38% in the CG. No significant difference in the CAPS rating changes was observed between the groups.
Lieberman et al., 2023	PCC	×	Trauma/emotion provocation paradigm	HC (n = 15)	PTSD -NF (n = 14)	×	(1) Comparing functional connectivity and alpha EEG coherence before and after the NFT. (2) Comparing how PCC connectivity changed during NFT.	PPI analysis	The PTSD group showed significantly greater PCC connectivity.
Zhao et al., 2023	Amygdala	Averaged voxels	Experiencing the anxiety induced by the scripts	PTSD -CG (n = 11)	PTSD -NF (n = 14)	×	(1) Comparing the ability to downregulate amygdala activation level. (2) Comparing CAPS-5 score.	GLM, t-tests, Levene’s test, rsFC	(1) Both groups showed improvements in symptom scores; however, it has no statistical significance. (2) The finding of greater improvement in amygdala control.
Fruchtman-Steinbok et al., 2021	Amygdala	AmygEFP	A neutral musical excerpt and a multimodal animated scenario	PTSD -CG (n = 13)	PTSD: Trauma-NF (n = 14), PTSD: Neutral-NF (n = 13)	×	Follow-up clinical assessment	The AmygEFP model	Both NF groups learned to volitionally downregulate AmygEFP signal and showed greater reduction in PTSD symptoms and improved downregulation of the amygdala.
Misaki et al., 2021	LA	BOLD	Recall a positive autobiographical memory	HC (n = 43), PTSD -CG (n = 9)	PTSD -NF (n = 20)	Recall a positive autobiographical memory	Comparing and evaluating the hippocampal subfield volumes.	R², LMA	Rt-fMRI-NF positive emotional training increased the hippocampus volume among people with PTSD.
Misaki et al., 2019	LA	BOLD	Recall a positive autobiographical memory.	PTSD -CG (n = 10)	PTSD -NF (n = 21)	Recalling positive autobiographical memories	SEMM analyses	GLM, SEMM	(1) There was a trend toward less activation in these regions for the EG compared with the CG. (2) Symptom reduction due to the NFT was not specifically mediated by the NF target activity.
Nicholson et al., 2018	Amygdala	BOLD	Viewing a personalized trauma word	×	PTSD -NF (n = 14)	×	Comparing and evaluating the ICN changes of three conditions (regulate, view, and neutral).	ANOVA, ICA, PCA, SSA	Downregulation of the amygdala using rt-fMRI-NFB in PTSD is associated with dynamic changes in ICN.
Nicholson et al., 2016	Amygdala	BOLD	Viewing a personalized trauma word	×	PTSD -NF (n = 10)	×	The gPPI and DCM analyses	ANOVA, t-tests, Regression analysis, DCM, gPPI	Successful downregulation of the amygdala using rt-fMRI-NF in PTSD, which was critically sustained in a subsequent transfer run without NF.
Goel et al., 2023	×	×	Subjective stress responses to the VR-like stimuli	×	PTSD -NF (n = 8)	×	×	SVM, SVR, MLP	We accurately classified the 8-class self-reported stress responses.
Zweerings et al., 2020	Amygdala	BOLD	Eliciting negative affect	HC (n = 21)	PTSD -NF (n = 20)	Imagining that the situation improves in the future, generating alternative explanations, imagining that the situation is not real	Comparing the cognitive reappraisal	GLM, ROI analysis, ANOVA	Beneficial effects of the NF training indicated by reduced amygdala.
Misaki et al., 2018	Amygdala	BOLD	×	HC (n = 17); PTSD-CG (n = 9)	PTSD -NF (n = 21)	×	Comparing the longitudinal MDMR before and after the NFT	multivariate distance matrix regression(MDMR), LME	The rt-fMRI-NF training effect was not limited to a feedback target region, and symptom relief could be mediated by brain modulation in several regions
Zweerings et al., 2021	Localized brain regions	BOLD	×	HC (n = 21)	PTSD (n = 9)	×	×	ISC, GLM	ISC enables a novel set of research questions that can guide future NF and neuroimaging investigations

ACC, anterior cingulate cortex; HC, healthy control; PTSD-CG, post-traumatic stress disorder control group; NF, neurofeedback; NFT, neurofeedback training; rsFC, resting-state functional connectivity.

In addition, this study provides a checklist (Table 3) for evaluating the credibility and robustness of the included research and assesses each study against these criteria. Evaluation indicators include whether the study presents a clear theoretical foundation, whether the sample size meets the estimated value required for statistical power, and if not, how the sample size was determined. Specifically, the sample size was estimated based on the expected effect size, desired statistical power (e.g., 80%), and significance level (e.g., 0.05), using specific software/tool [e.g., G*Power (Faul et al., 2007)]. The formula for sample size estimation is shown below:

n = \frac{2 * {(Z_{\frac{α}{2}} + Z_{β})}^{2} * σ^{2}}{Δ^{2}}

(1)

Where $n$ is the sample size required per group, $Z_{\frac{α}{2}}$ is the critical value corresponding to the significance level $α$ , $Z_{β}$ is the critical value corresponding to the desired statistical power, $σ^{2}$ is the population variance (or group variance), which can be estimated from prior studies or pilot data, and $Δ$ is the effect size, representing the expected mean difference between the groups.

Table 3.

Comparison of Classification Results

Study				Above the average level of studies included	Average level of studies included	Below the average level of studies included
Assessment		Y	N	Above the average level of studies included	Average level of studies included	Below the average level of studies included
Theory	Has the research foundation been clearly defined?
Sample	Are sample size estimates met?
CS	Are there PTSD controls?
CS	Are there healthy controls?
Regulation Strategy	Are there regulatory strategies?
Regulation Strategy	Is there an explanation provided for the presence or absence of strategies?
Evaluation methods	Follow-up
Evaluation methods	Are nonclinical indicators incorporated?
Data processing methods	Methods with explanatory power
Conclusion	Supported by statistical results
General comments

Each subitem is scored on a scale of 1–5 points. If the reviewed study adequately meets the criterion (Y), it scores 5 points; if it does not (N), it scores 1 point. If a reviewed study does not fully meet the criterion, but is comparable to the average level of included studies, scores are assigned ranging from 4 to 2 points based on its comparative level. Comparable items include Sample and CS. Finally, the total score is divided by the number of criteria to obtain the average score.

In cases where the estimated sample size was not met, this study compares the included studies horizontally by comparing them to the average sample size of the studies in the review. Furthermore, evaluation criteria include the selection of control groups, guidance on regulation strategies, rigor of assessment methods, interpretability of data processing methods, credibility of conclusions, and finally, an overall assessment based on the completion of the checklist.

It should be noted that certain studies (Leem et al., 2020; Gerin et al., 2016) were included in the study despite some prominent issues such as severely inadequate sample sizes or unclear conclusions. This inclusion was based on noteworthy aspects of their methods or research processes that warrant discussion.

Results

Summary of parameters for included studies

Following the criteria and checklist established by this study, the parameters of included studies in EEG-based and fMRI-based NFT groups are summarized in Tables 1 and 2. There are 18 studies included in EEG-NFT research and 16 in fMRI-NFT research. In EEG-NFT studies, four studies did not clearly report conclusive findings and one study reported negative results (no statistically significant difference between the NFT group and the control group). In fMRI-NFT studies, two studies reported negative results. The remaining studies (79.4% of the total) concluded on the effectiveness of NFT intervention for PTSD across various aspects. In addition, recent reviews have further underscored the effectiveness of NF in addressing both psychological and physical conditions, such as chronic pain, highlighting its broader therapeutic potential (Diotaiuti et al., 2024a).

Target brain areas, target signals, and experimental paradigm statistics

As shown in Figure 2, we conducted a statistical analysis of the target brain areas and target signals in the included studies, categorized into EEG-based and fMRI-based NFT groups. The statistical results indicate that for the EEG-NFT group, 5.6% of studies targeted the default mode network (DMN) and salience network (SN), whereas 55.6% targeted the parietal lobe. In addition, 16.7% targeted the Amygdala, and 11.1% targeted the sensory motor region. In terms of temporal regions, 22.2% of studies focused on them, and 5.6% targeted the frontal region. A chi-square goodness-of-fit test revealed a significant difference in the distribution of target brain regions (χ² = 16.429, p = 0.006), suggesting that current studies exhibit a nonuniform selection of target areas. Standardized residuals further indicated a notable preference for the parietal lobe (residual = 3.47), highlighting a potential bias toward targeting this region in EEG-based NFT studies.

FIG. 2.

Target brain areas, target signals, and experimental paradigm statistics. (a) NF regulation targets based on EEG; (b) NF regulation signal based on EEG; (c) NF regulation paradigm based on EEG; (d) NF regulation targets based on fMRI; (e) cNF regulation signal based on fMRI; (f) NF regulation paradigm based on fMRI. NF, neurofeedback; EEG, electroencephalogram; fMRI, functional magnetic resonance imaging.

Regarding the main regulatory signals in the included studies, the majority were segmented by EEG signal rhythms, with 72.2% involving Alpha (8–12Hz) rhythms. Other studies included low-frequency (2–6Hz) EEG signals (5.6%), high-frequency (22–36 Hz) EEG signals (11.1%), and theta-rhythm (4–8 Hz), sensory motor rhythm (12–15Hz), and 10–13 Hz signals, accounting for 27.8%, 16.7%, and 5.6%, respectively. Furthermore, 16.7% of studies focused on the amygdala-electrical fingerprint (AmygEFP) signal model (Fine et al., 2023). A chi-square goodness-of-fit test revealed a significant deviation in the distribution of regulatory signal types (χ² = 26.500, p < 0.001). Notably, the standardized residual for the Alpha rhythm was 4.50, suggesting a substantial overrepresentation of this signal type, indicating a marked bias toward Alpha rhythm regulation in current EEG-based NF research.

For the fMRI-NFT group, the distribution of targeted brain regions in the included studies is as follows: Whole brain, anterior cingulate cortex (ACC), and Amygdala-prefrontal, each accounting for 6.3%. Studies involving regulation of the posterior cingulate cortex (PCC) and lateral amygdala (LA) regions each constitute 12.5% of the included studies. Amygdala regulation is reported in 43.8% of the included studies. A chi-square goodness-of-fit test indicated a significant difference in the selection frequency of targeted brain regions (χ² = 11.714, p = 0.039), with standardized residuals revealing that the Amygdala was notably overrepresented (residual = 3.06), suggesting a marked tendency to target this region in fMRI-based NFT studies. Regarding the main regulatory signals in the included studies, BOLD signals account for 68.8%, whereas studies involving the AmygEFP, Averaged voxels, and resting-state functional connectivity each constitute 6.3%. A separate chi-square test on the distribution of signal types also showed a significant deviation from uniformity (χ² = 21.429, p < 0.001). The standardized residual for the BOLD signal was 4.01, indicating a strong tendency toward using BOLD signals in current fMRI-NFT research.

For the types of NFT paradigms in the EEG-NFT group, paradigms were categorized into Audio/Visual scene and Game groups. They constituted 33.3% and 33% of the included studies, respectively. In anticipation of the potential of VR technology, a 3D/VR scene group was included, accounting for 5.6% of the studies. The remaining paradigms were categorized as Other, including meditation paradigms, and accounted for 22.2% of the included studies. A chi-square goodness-of-fit test showed no significant deviation in the distribution of NFT paradigms (χ² = 3.941, p = 0.268), with standardized residuals revealing a slight underrepresentation of the 3D/VR scene category (residual = −1.58) and a slight overrepresentation of the Audio/Visual scene and Game categories (residual = 0.85). This suggests that the selection of paradigms across studies is relatively balanced, with no marked preference for any specific type.

In the fMRI-NFT group, paradigms primarily belonged to the Eliciting effect category, which constituted 68.8% of the included studies. Audio/Visual scene, Game, and Other groups each accounted for 6.3% of the included studies. A chi-square goodness-of-fit test revealed a significant deviation in the distribution of NFT paradigms (χ² = 21.429, p < 0.001). Standardized residuals showed a marked overrepresentation of the Eliciting effect paradigm (residual = 4.01), suggesting a strong preference for this type of paradigm in fMRI-based NFT studies. In contrast, the Audio/Visual scene, 3D/VR scene, and Other categories were underrepresented (residuals = −1.34).

Statistical analysis of study parameters

As shown in Figure 3., we conducted a statistical analysis of the included studies categorized into EEG-based and fMRI-based NFT research. The statistical results indicate that for the EEG-NFT group, approximately 77.8% of the studies included a control group. Among these, 22.2% used healthy controls, 61.1% used PTSD patient controls, 5.6% had dual experimental groups, and 16.7% had dual control groups. The average number of participants in the PTSD control groups was 18.5 (SEM = 6.93) and in the NFT experimental groups, it was 22.7 (SEM = 14.99). There was no significant statistical difference between the PTSD control and NFT experimental groups (p = 0.3882), suggesting that the control group setups in the included studies are reasonable.

Due to the limited use of healthy control samples and the inclusion of studies (such as Mirjana Askovic [Askovic et al., 2020]) using matched data from databases, this study provides only a reference value of approximately 31 individuals for healthy controls. Among the studies included, none provided detailed regulatory strategies for PTSD-NFT processes. About one-third of the studies provided reasons (Fine et al., 2023; Nicholson et al., 2016; Kluetsch et al., 2014; Ros et al., 2017; Shaw et al., 2023), which were similar, aiming to prevent demand characteristics from influencing training. These studies did not provide specific regulatory strategy instructions to participants, encouraging a psychological strategy of free exploration.

Using the chi-square test statistic to examine the parameters related to the evaluation standards and methods of the EEG-NF group (χ² = 39, p < 0.001), the following results were obtained: Specific NFT regulation paradigms were provided in 61.1% of the studies. Data processing in 94.4% of the studies used traditional statistical methods, whereas only 16.7% involved simple machine learning (ML) methods such as independent component analysis (ICA) and linear discriminant analysis (LDA) (Weaver et al., 2020; Ros et al., 2017). No studies utilized deep learning (DL) methods. All included studies used pre–post-test comparative assessments to verify the effectiveness of NFT. Among EEG-NFT studies, process evaluations were not used in any study, whereas 38.9% used follow-ups and algorithms or physiological indicators beyond clinical scales, such as long-range temporal correlations (LRTCs), event-related potentials (ERP), and fMRI measures (Ros et al., 2017; Askovic et al., 2017; Fine et al., 2023). Standardized residuals indicated that traditional statistical methods (residual = 2.67) and pre–post assessments (residual = 3.00) were significantly overrepresented, whereas guidance strategies and process evaluations were notably underrepresented (residuals = −3.00). The use of ML/DL methods also appeared underutilized (residual = −2.00), suggesting a general tendency in current EEG-NFT studies to rely on conventional analytical approaches and outcome assessments, with limited integration of more dynamic or process-based evaluations.

In the fMRI-NFT group, 75% of the included studies included a control group, with 12.5% having dual control groups and 6.3% having dual experimental groups. Half of the studies used healthy controls, whereas 37.5% used PTSD patient controls. The average number of participants in the NFT experimental groups was 14.6 (SD 6.37) and 10.5 (SD 1.52) for the PTSD patient control groups and 18.8 (SD 10.82) for the healthy control groups. Independent samples t-tests showed no significant statistical differences among the three groups as follows: between NFT experimental groups and PTSD patient control groups (p = 0.0913), between NFT experimental groups and healthy control groups (p = 0.1372), and between PTSD patient control groups and healthy control groups (p = 0.2495), indicating that the control group setups in the included studies are reasonable.

Among the studies included, 25% provided detailed regulatory strategies for NFT processes, and 87.5% specified NFT regulation paradigms. In terms of data processing, 87.5% of the studies used traditional statistical methods, whereas 25% utilized simple ML methods. No studies involved DL methods. All studies used pre–post-test comparative assessment methods to validate the effectiveness of NFT. Statistically, for the included fMRI-NFT studies, 13% included process evaluations (Nicholson et al., 2022), and 31% used algorithms or physiological indicators beyond clinical scales in data processing, such as gPPI and alpha EEG coherence (Nicholson et al., 2017; Zotev et al., 2018). A chi-square goodness-of-fit test revealed a significant deviation in the distribution of evaluation-related parameters among fMRI-NFT studies (χ² = 26.625, p < 0.001). Standardized residuals suggested that the use of clear experimental paradigms (residual = 2.12), traditional statistical methods (residual = 2.12), and pre–post assessments (residual = 2.83) was notably more frequent than expected, whereas process evaluations were underrepresented (residual = −2.12). The use of guidance strategies and ML/DL methods also appeared relatively limited (residuals = −1.41), indicating a preference for conventional and outcome-based evaluation strategies over dynamic or process-focused approaches.

To improve the clarity and organization of the statistical techniques used in the included studies, we have summarized the methods in Table 4 –6. The studies are categorized according to the statistical approaches used, including hypothesis testing, ML, DL, neuroimaging methods, signal processing, and additional methods. This categorization allows for a clearer understanding of the variety of statistical methods applied in the included studies.

FIG. 3.

Statistical Analysis of Study Parameters. (a) Control group statistics; (b) Double control group statistics; (c) Control group characteristic statistics; (d) Statistics on guiding strategies; (e) Statistics of data processing method types; (f) Statistical analysis of evaluation method types; (g) Statistical analysis of the number of subjects based on EEG neurofeedback research; (h) Statistical analysis of the number of participants based on fMRI neurofeedback research. EEG, electroencephalogram; fMRI, functional magnetic resonance imaging. *Note: WCG, Wait control group; NCG, No control group; DEG, Double experimental group; DCG, Double control group; HCG, Healthy control group; NFT-CG, NFT control group; GS, Guidance Strategy; CEP, Clear Experimental Paradigm; CSM, Conventional Statistical Methods; ML/DL, Machine Learning/Deep Learning; PPCE, Pretest–Post-test Comparative Evaluation; PE, Process Evaluation.

Table 4.

Overview of Data Analysis Methods for Electroencephalogram-based Neurofeedback

Study	Data processing methods	Hypothesis testing	Machine learning	Deep learning	Neuroimaging methods	Signal processing	Additional methods
Nicholson et al., 2020	fMRI scans for DMN and SN				√
Van der Kolk et al., 2016	Chi-square test	√
Leem et al., 2020	LORETA analysis, Z-scores, t-test, U-test	√			√
Fine et al., 2023	GLM, ANCOVA, ITT, LMM	√					√
Askovic et al., 2017	×	×	×	×	×	×	×
Leem et al., 2021	QEEG Analysis, LORETA, t-test, CUA	√			√		√
Nooripour, 2017	SW-test, Levene test, Means, SD	√					√
Shaw et al., 2023	PSD, ANOVA, Fast ICA, regression analyses	√	√			√
Askovic et al., 2020	ANCOVA	√
Nicholson et al., 2016	t-test; short-time Fouriertransformations; ANOVA; Multiple regression analysis	√				√
Kluetsch et al., 2014	Spectral analysis, Connectivity analysis, ICA, Multiple regression analysis		√		√	√
Gapen et al., 2016	Hierarchical Linear and Non-Linear Modeling, Multilevel mediation analyses	√
du Bois et al., 2021	SW-test, FBCSP, LDA, Friedman test	√	√
Ros et al., 2017	ICA, Hilbert transform, MANOVA	√	√			√
Nicholson et al., 2023	Pearson’s chi-squared,Fisher’s exact tests	√

fMRI, functional magnetic resonance imaging.

Table 5.

Overview of Data Analysis Methods for Functional Magnetic Resonance Imaging based Neurofeedback

Study	Data processing methods	Hypothesis testing	Machine learning	Deep learning	Neuroimaging methods	Signal processing	Additional methods
Weaver et al., 2020	MVPA and SVM	√
Zweerings et al., 2018	GLM: Pearson product moment correlation coefficient	√
Nicholson et al., 2021	ANOVA, LRA, MKL	√	√
Gerin et al., 2016	Functional Connectivity				√
Zotev et al., 2018	GLM, Functional Connectivity, R², coherence analysis	√			√	√	√
Lieberman et al., 2023	PPI analysis				√
Zhao et al., 2023	GLM, t-tests, Levene’s test, rsFC	√				√
Fruchtman-Steinbok et al., 2021	The AmygEFP model				√
Misaki et al., 2021	R², LMA					√	√
Misaki et al., 2019	GLM, SEMM	√					√
Nicholson et al., 2018	ANOVA, ICA, PCA, SSA	√				√	√
Nicholson et al., 2016	ANOVA, t-tests, Regression analysis, DCM, gPPI	√			√
Goel et al., 2023	SVM, SVR, MLP		√	√
Zweerings et al., 2020	GLM, ROI analysis, ANOVA	√			√
Misaki et al., 2018	MDMR, LME		√				√
Zweerings et al., 2021	ISC, GLM	√			√

Table 6.

Categorization of Statistical and Analytical Methods

Category	Methods
Hypothesis Testing	GLM, Pearson product moment correlation coefficient, ANOVA, t-tests, Levene’s test, Chi-square test, LORETA analysis, Z-scores, U-test, ANCOVA, Fisher’s exact tests, SW-test, Means, SD, MANOVA
Machine Learning	MVPA, SVM, SVR, LRA, MKL, SVM, SVR, MLP, LDA, FBCSP
Deep Learning	MLP
Neuroimaging Methods	Functional Connectivity, PPI analysis, rsFC, The AmygEFP model, DCM, gPPI, ISC, fMRI scans for DMN and SN, QEEG Analysis, LORETA, ROI analysis
Signal Processing	Coherence analysis, PSD, Fast ICA, short-time Fourier transformations, Spectral analysis, Connectivity analysis, Hilbert transform, Fast ICA
Additional Methods	R², LMA, SEMM, MDMR, LME, Hierarchical Linear and Non-Linear Modeling, Multilevel mediation analyses, SW-test, Friedman test

Discussion

This study screened, examined, and statistically analyzed existing research in the PTSD-NFT field, integrating study designs and various indicators related to PTSD-NFT studies, and presented statistical findings. Based on the current statistical results, this study summarizes key technical issues and suggests the need for further research in the following areas. To visually illustrate the key technical issues and research gaps identified in this study, Figure 4 presents an overview of the interrelationships between ML/DL methods, process evaluation, regulatory strategies, and paradigm innovation in the context of PTSD-NFT research. The figure highlights the current shortcomings in existing studies across these areas and serves as a foundation for the following discussion, which delves deeper into the specific challenges and research needs that need to be addressed for future advancements in the field.

FIG. 4.

Key Technical issues and Research Needs in PTSD-NFT. PTSD, post-traumatic stress disorder; NFT, neurofeedback training.

The value of fNIRS technology in target brain regions, target signals, and experimental paradigms for NFT

Existing research has demonstrated that the pathogenesis of PTSD involves disruptions in the central nervous system’s memory processes related to stress information, leading to difficulties in inhibiting conditioned fear responses or excessive suppression thereof. PET studies (Bremner et al., 2003) have shown severe reductions in regional cerebral blood flow in PTSD patients, including the orbitofrontal cortex, anterior cingulate gyrus, anterior medial frontal cortex (Brodmann’s areas 2, 9), fusiform gyrus/temporal subcortex, and increased activation in the posterior cingulate gyrus, left subcortical areas, and motor cortex, which are associated with memory circuits. PET and functional MRI confirm heightened reactivity of the amygdala and anterior insular cortex to traumatic stimuli, whereas reduced reactivity is observed in the ACC and prefrontal regions.

In addition, EEG studies (Jokic-Begic and Begic, 2003) indicate that PTSD patients exhibit decreased alpha waves and increased beta waves. Increased beta I excitations extend beyond the medial frontal cortex plane and left parietal regions, whereas beta II excitations manifest in the frontal lobe. Theta wave abnormalities extend beyond the medial frontal cortex, suggesting cortical overexcitation and prolonged arousal, indicative of dysregulation in frontal lobe activation.

Statistical analysis of existing PTSD-NFT studies reveals that these NFT paradigms, target regions, and target signals indeed strictly adhere to the neurophysiological findings mentioned above. Due to the amygdala’s deep location in the brain, fMRI-NFT studies focus on regulating the amygdala by downregulating it to attenuate its and the anterior insular cortex’s heightened response to traumatic stimuli. In contrast, EEG-NFT studies concentrate on modulating alpha waves and theta activity.

However, it is crucial to note that the frontal lobe plays a significant role in the pathogenesis of PTSD, making it an important target for regulation. Yet statistical analysis of included studies suggests that there is relatively less focus on regulating the frontal lobe in current research. One possible reason is that in fMRI-NFT studies, the amygdala is often prioritized, whereas in EEG-NFT studies, alpha waves are typically targeted from the parietal lobe and theta waves from the temporal areas, thus making the frontal lobe not a primary area for regulation.

Introducing another noninvasive BCI technology—fNIRS BCI, statistical analysis of included studies shows very few PTSD-NFT studies based on fNIRS. However, for regulation of the frontal lobe, hemodynamic responses are excellent target signals for fNIRS, and given the depth of measurement of fNIRS, the frontal region is highly suitable. In fact, in other areas of PTSD research, studies using fNIRS BCI technology have been conducted. For example, Balters et al. (Balters et al., 2021) utilized the NimStim dataset to construct neutral/fearful facial expression recognition tasks, activating the frontal-amygdala circuit in PTS. Thus, based on fNIRS, regulating the hemodynamic response in the frontal lobe appears to be an expandable option for PTSD-NFT.

Nevertheless, it is important to acknowledge why this promising technology has not yet seen widespread adoption. One key limitation is the current lack of well-established NF markers that can guide individuals in voluntarily modulating frontal lobe activity through fNIRS. Unlike EEG, which offers relatively direct and intuitive feedback targets such as alpha or theta rhythms, the hemodynamic signals recorded by fNIRS are less straightforward for real-time self-regulation. Moreover, the mechanisms involved in regulating cerebral blood oxygenation differ fundamentally from those of neural electrical activity and remain less understood. These limitations present challenges in protocol design and user feedback effectiveness, necessitating further research to optimize fNIRS-based NF strategies.

The assessment of NF processes as a crucial link in the evidence chain of PTSD-NFT studies

The inability to clearly define assessment methods for NFT processes and participant regulation strategies has created a black box in the broader control system of NFT. This also makes it challenging for PTSD-NFT research to adopt controlled scientific methods to explain and model the PTSD-NFT process. Even though some included studies attempt to use fMRI technology to compare brain states before and after NFT sessions, while this is highly valuable for demonstrating the efficacy of NFT in treating PTSD, it still does not explain how NFT achieves these effects.

In simpler terms, researchers cannot explain what exactly patients do during NFT to cause changes in brain states that lead to final experimental results. Moreover, the underlying neurophysiological mechanisms of this process or the mathematical and physical processes from a control engineering perspective remain unclear. However, there is a lack of improvement in the monitoring and evaluation system of neural feedback processes in existing research, which may be the reason for the insufficient evidence chain.

The necessity of clear regulation guidance: Worth considering

Statistical analysis of included studies indicates that current PTSD-NFT research provides limited guidance on regulation strategies. While free exploration aligns with core NF principles, studies suggest that clear strategy guidance may enhance intervention outcomes (Chen et al., 2022). Furthermore, this lack of guidance not only limits the participant’s ability to engage meaningfully in the task but also hinders the development of reliable evaluation methods. This lack of guidance ultimately reinforces the challenge of interpreting the NF process. Similar concerns have been raised in broader NF and biofeedback applications, where integrated methodological frameworks have been proposed to improve protocol clarity (Tosti et al., 2024).

The critical role of clear regulation strategies and improved process evaluation methods in data processing approaches

Existing PTSD-NFT studies rely heavily on pre–post statistical comparisons to evaluate outcomes; yet, these methods often fail to capture the underlying regulation dynamics. Despite the significant utility and potential demonstrated by ML and DL in the interdisciplinary field of medical engineering, statistical analysis of included studies indicates that their adoption in current PTSD-NFT research is minimal. In other populations, such as university students, studies have demonstrated that well-structured NF protocols can significantly enhance cognitive functions like working memory while reducing anxiety, supporting the value of precise evaluation approaches (Diotaiuti et al., 2024b). It appears that these methods fail to fully leverage the capabilities of ML and DL.

One possible reason, as discussed in the previous section, could be the unclear nature of exploratory tasks in the NFT process. This implies that most experimental signals collected during research lack precise semantic labels. Whether using data-driven DL methods or model-driven ML theories, it becomes challenging to design a stable model capable of processing such data effectively. Therefore, the primary opinion of this study is to advance process evaluation approaches for PTSD-related NFT by introducing algorithmic tools that complement traditional pre–post assessments. Although this process inherently involves a certain level of mechanistic interpretation, the focus remains on improving the methodological framework for evaluating NF procedures in a more continuous and dynamic manner.

To achieve this, future studies could incorporate time window-based feature extraction strategies, coupled with adaptive filtering algorithms, to dynamically model neural feedback signals and uncover regulation patterns. In addition, adopting a “semi-guided regulation” approach could help provide participants with structured strategy references (e.g., focusing attention or emotional association) while preserving the essence of self-regulation. Moreover, integrating interpretable ML models, such as decision trees or Shapley-based XGBoost, would enable identification of meaningful patterns in neural changes, thus supporting individualized interventions and enhancing clinical outcomes.

Limitations and challenges

The studies included in this review were primarily based on literature retrieved from specific databases within a defined time frame, which may introduce selection bias and limit the comprehensiveness of the review’s findings, potentially overlooking certain perspectives and discoveries in the field. Furthermore, the criteria and process for selecting studies could affect the representativeness of the included literature. While efforts were made to control for these differences during the review process, it is important to acknowledge that these factors may have a potential impact on the results of the study.

Conclusion

This study conducted a literature review and statistical analysis of research in the PTSD-NFT field. Statistical synthesis of included studies suggests that controversies in PTSD-NFT research likely stem from the following two main issues within existing NFT assessment frameworks: the lack of designed process evaluations and the absence of clear evidence regarding the effectiveness of regulation strategies. Through statistical analysis of experimental outcomes and regulated brain regions, this study highlights that fNIRS-NFT may be a valuable avenue for further research.

The perspectives presented in this study are grounded in interpretations of statistical findings from included studies, aiming to contribute to the development of the PTSD-NFT field.

Authors’ Contributions

Conceptualization, P.D.; methodology, P.D., A.G., W.N., and Y.F.; software, P.D.; validation, P.D., L.T., and H.P.; formal analysis, P.D.; investigation, L.T.; data curation, P.D., L.T., and H.P.; writing—original draft preparation, P.D.; writing—review and editing, P.D.; visualization, P.D.; supervision, A.G., W.N., and Y.F.; project administration, Y.F.; and funding acquisition, Y.F. All authors have read and agreed to the published version of the article.

Footnotes

Author Disclosure Statement

The authors declare no conflicts of interest.

Funding Information

This research was funded by the National Natural Science Foundation of China (82172058).

References

Askovic

, Watters

, Aroche

, et al. Neurofeedback as an adjunct therapy for treatment of chronic posttraumatic stress disorder related to refugee trauma and torture experiences: Two case studies. Australas Psychiatry, 2017; 25(4):358–363; doi: 10.1177/1039856217715988

Askovic

, Watters

, Coello

, et al. Evaluation of neurofeedback for posttraumatic stress disorder related to refugee experiences using self-report and cognitive ERP measures. Clin EEG Neurosci, 2020; 51(2):79–86; doi: 10.1177/1550059419849170

Balters

, Li

, Espil

, et al. Functional near-infrared spectroscopy brain imaging predicts symptom severity in youth exposed to traumatic stress. J Psychiatr Res, 2021; 144:494–502; doi: 10.1016/j.jpsychires.2021.10.020

Bremner

, Vythilingam

, Vermetten

, et al. Neural correlates of declarative for emotional valenced words in women with posttraumatic stress disorder related to early childhood sexual abuse. Biol Psychiatry, 2003; 53(10):879–889; doi: 10.1016/s0006-3223(02)01891-7

Chen

, Wang

, Chang

, et al. Effects of the function-specific instruction approach to neurofeedback training on frontal midline theta waves and golf putting performance. Psychology of Sport and Exercise, 2022; 61:102211; doi: 10.1016/j.psychsport.2022.102211

Choi

, Cho

, Lee

. Feasibility of a mobile app for traumatic stress management using neurofeedback-based meditation and binaural beat music: A pilot randomized controlled trial. Digit Health, 2024; 10:20552076241308986; doi: 10.1177/20552076241308986

Cloitre

. Alternative intensive therapy for PTSD. Am J Psychiatry, 2014; 171(3):249–251; doi: 10.1176/appi.ajp.2013.13121695

Collura

. Technical Foundations Of Neurofeedback. Routledge; 2014; doi: 10.4324/9780203795132

Cortese

, Ferrin

, Brandeis

, et al.; European ADHD Guidelines Group (EAGG). Neurofeedback for attention-deficit/hyperactivity disorder: Meta-analysis of clinical and neuropsychological outcomes from randomized controlled trials. J Am Acad Child Adolesc Psychiatry, 2016; 55(6):444–455; doi: 10.1016/j.jaac.2016.03.007

10.

Curran

, Stokes

. Learning to control brain activity: A review of the production and control of EEG components for driving brain-computer interface (BCI) systems. Brain Cogn, 2003; 51(3):326–336; doi: 10.1016/S0278-2626(03)00036-8

11.

Dehghani

, Soltanian-Zadeh

, Hossein-Zadeh

. Neural modulation enhancement using connectivity-based EEG neurofeedback with simultaneous fMRI for emotion regulation. Neuroimage, 2023; 279:120320; doi: 10.1016/j.neuroimage.2023.120320

12.

Diotaiuti

, Corrado

, Tosti

, et al. Evaluating the effectiveness of neurofeedback in chronic pain management: A narrative review. Front Psychol, 2024a;15:1369487; doi: 10.3389/fpsyg.2024.1369487

13.

Diotaiuti

, Valente

, Corrado

, et al. Enhancing working memory and reducing anxiety in university students: A neurofeedback approach. Brain Sci, 2024b;14(6):578; doi: 10.3390/brainsci14060578

14.

Du Bois

, Bigirimana

, Korik

, et al. Neurofeedback with low-cost, wearable electroencephalography (EEG) reduces symptoms in chronic Post-Traumatic Stress Disorder. J Affect Disord, 2021; 295:1319–1334; doi: 10.1016/j.jad.2021.08.071

15.

Erickson

, Hurley

, Taber

. Psychotherapy for PTSD: Neuroimaging of recovery processes. J Neuropsychiatry Clin Neurosci, 2014; 26(3):188–195; doi: 10.1176/appi.neuropsych.260301

16.

Faul

, Erdfelder

, Lang

, et al. G*Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behav Res Methods, 2007; 39(2):175–191; doi: 10.3758/BF03193146

17.

Fine

, Helpman

, Armon

, et al. Amygdala-related EEG Neuro-feedback as an add-on therapy for treatment-resistant childhood sexual abuse PTSD: Feasibility study. Psychiatry and Clinical Neurosciences, 2023; doi: 10.1111/pcn.13591

18.

Foa

, Keane

, Friedman

, et al. Treatment Guidelines, Effective Treatment for PTSD: Practice Guidelines from the International Society for Traumatic Stress Studies, 2nd Edition. Jason Aronson: New York; 2008; doi: 10.1080/03069885.2010.550798

19.

Friedman

. PTSD: Pharmacotherapeutic Approaches. FOC, 2013; 11(3):315–320; doi: 10.1176/appi.focus.11.3.315

20.

Fruchter

, Goldenthal

, Adler

, et al. Amygdala-derived-EEG-fMRI-pattern neurofeedback for the treatment of chronic post-traumatic stress disorder. A prospective, multicenter, multinational study evaluating clinical efficacy. Psychiatry Res, 2024; 333:115711; doi: 10.1016/j.psychres.2023.115711

21.

Fruchtman-Steinbok

, Keynan

, Cohen

, et al. Amygdala electrical-finger-print (AmygEFP) NeuroFeedback guided by individually-tailored Trauma script for post-traumatic stress disorder: Proof-of-concept. Neuroimage Clin, 2021; 32:102859; doi: 10.1016/j.nicl.2021.102859

22.

Gapen

, van der Kolk

, Hamlin

, et al. A pilot study of neurofeedback for chronic PTSD. Appl Psychophysiol Biofeedback, 2016; 41(3):251–261; doi: 10.1007/s10484-015-9326-5

23.

Gerin

, Fichtenholtz

, Roy

, et al. Real-time fMRI neurofeedback with war veterans with chronic PTSD: A feasibility study. Front Psychiatry, 2016; 7:111; doi: 10.3389/fpsyt.2016.00111

24.

Goel

, Tse

, Smith

, et al. Framework for accurate classification of self-reported stress from multisession functional MRI data of veterans with posttraumatic stress. Chronic Stress (Thousand Oaks), 2023; 7:24705470231203655; doi: 10.1177/24705470231203655

25.

Grey

, Ehlers

, Hackmann

, et al. A randomized controlled trial of 7-day intensive and standard weekly cognitive therapy for PTSD and emotion-focused supportive therapy. AJP, 2014; 171(3):294–304; doi: 10.1176/appi.ajp.2013.13040552

26.

Gurevitch

, Lubianiker

, Markovits

, et al. Amygdala self-neuromodulation capacity as a window for process-related network recruitment. Philos Trans R Soc Lond B Biol Sci, 2024; 379(1915):20240186; doi: 10.1098/rstb.2024.0186

27.

GWK

, Karatzias

, Vallières

, et al. Complex PTSD symptoms mediate the association between childhood trauma and physical health problems. J Psychosom Res, 2021; 142(3):110358; doi: 10.1016/j.jpsychores.2021.110358’

28.

Hull

. Neuroimaging findings in post-traumatic stress disorder: Systematic review. Br J Psychiatry, 2002; 181(2):102–110; doi: 10.1034/j.1399-5618.2002.01220.x

29.

Jokic-Begic

, Begic

. Quantitative electroencephalogram (qEEG) in combat Veterans with PTSD. Nord J Psychiatry, 2003; 57(5):351–355; doi: 10.1080/08039480310002688

30.

Kluetsch

, Ros

, Théberge

, et al. Plastic modulation of PTSD resting‐state networks and subjective wellbeing by EEG neurofeedback. Acta Psychiatr Scand, 2014; 130(2):123–136; doi: 10.1111/acps.12229

31.

Leem

, Cheong

, Lee

, et al. Effectiveness, cost-utility, and safety of neurofeedback self-regulating training in patients with post-traumatic stress disorder: A randomized controlled trial. Healthcare (Basel), 2021; 9(10):1351; doi: 10.3390/healthcare9101351

32.

Leem

, Cheong

, Yoon

, et al. Neurofeedback self-regulating training in patients with post traumatic stress disorder: A randomized controlled trial study protocol. Integr Med Res, 2020; 9(4):100464; doi: 10.1016/j.imr.2020.100464

33.

Levin

. Trial of interpersonal therapy may open new door to treat PTSD. PN, 2015; 50(8):1–1; doi: 10.1176/appi.pn.2015.4b1

34.

Lieberman

, Rabellino

, Densmore

, et al. Posterior cingulate cortex targeted real‐time fMRI neurofeedback recalibrates functional connectivity with the amygdala, posterior insula, and default‐mode network in PTSD. Brain Behav, 2023; 13(3):e2883; doi: 10.1002/brb3.2883

35.

Misaki

, Mulyana

, Zotev

, et al. Hippocampal volume recovery with real-time functional MRI amygdala neurofeedback emotional training for posttraumatic stress disorder. J Affect Disord, 2021; 283:229–235; doi: 10.1016/j.jad.2021.01.058

36.

Misaki

, Phillips

, Zotev

, et al. Brain activity mediators of PTSD symptom reduction during real-time fMRI amygdala neurofeedback emotional training. Neuroimage Clin, 2019; 24:102047; doi: 10.1016/j.nicl.2019.102047

37.

Misaki

, Phillips

, Zotev

, et al. Real-time fMRI amygdala neurofeedback positive emotional training normalized resting-state functional connectivity in combat veterans with and without PTSD: A connectome-wide investigation. Neuroimage Clin, 2018; 20:543–555; doi: 10.1016/j.nicl.2018.08.025

38.

Nicholson

, Densmore

, Frewen

, et al. Homeostatic normalization of alpha brain rhythms within the default-mode network and reduced symptoms in post-traumatic stress disorder following a randomized controlled trial of electroencephalogram neurofeedback. Brain Commun, 2023; 5(2):fcad068; doi: 10.1093/braincomms/fcad068

39.

Nicholson

, Rabellino

, Densmore

, et al. Differential mechanisms of posterior cingulate cortex downregulation and symptom decreases in posttraumatic stress disorder and healthy individuals using real‐time fMRI neurofeedback. Brain Behav, 2022; 12(1):e2441; doi: 10.1002/brb3.2441

40.

Nicholson

, Rabellino

, Densmore

, et al. Intrinsic connectivity network dynamics in PTSD during amygdala downregulation using real‐time fMRI neurofeedback: A preliminary analysis. Hum Brain Mapp, 2018; 39(11):4258–4275; doi: 10.1002/hbm.24244

41.

Nicholson

, Rabellino

, Densmore

, et al. The neurobiology of emotion regulation in posttraumatic stress disorder: Amygdala downregulation via real‐time fMRI neurofeedback. Hum Brain Mapp, 2017; 38(1):541–560; doi: 10.1002/hbm.23402

42.

Nicholson

, Ros

, Densmore

, et al. A randomized, controlled trial of alpha-rhythm EEG neurofeedback in posttraumatic stress disorder: A preliminary investigation showing evidence of decreased PTSD symptoms and restored default mode and salience network connectivity using fMRI. Neuroimage Clin, 2020; 28:102490; doi: 10.1016/j.nicl.2020.102490

43.

Nicholson

, Ros

, Frewen

, et al. Alpha oscillation neurofeedback modulates amygdala complex connectivity and arousal in posttraumatic stress disorder. Neuroimage Clin, 2016; 12:506–516; doi: 10.1016/j.nicl.2016.07.006

44.

Nooripour R , Noohi S , Miraghaie A M ,et al.Effectiveness of neuro-feedback treatment with alpha/theta method on PTSD symptoms and their executing function[J].Allied Academies, 2017(5).

45.

Post

, Youngstrom

, Connell

, et al. Transdiagnostic emotion regulation processes explain how emotion-related factors affect co-occurring PTSD and MDD in relation to trauma. J Anxiety Disord, 2021; 78:102367; doi: 10.1016/j.janxdis.2021.102367

46.

Ros

, Frewen

, Theberge

, et al. Neurofeedback tunes scale-free dynamics in spontaneous brain activity. Cereb Cortex, 2017; 27(10):4911–4922; doi: 10.1093/cercor/bhw285

47.

Schönenberg

, Wiedemann

, Schneidt

, et al. Neurofeedback, sham neurofeedback, and cognitive-behavioural group therapy in adults with attention-deficit hyperactivity disorder: A triple-blind, randomised, controlled trial. Lancet Psychiatry, 2017; 4(9):673–684; doi: 10.1016/S2215-0366(17)30291-2

48.

Shaw

, Nicholson

, Ros

, et al. Increased top-down control of emotions during symptom provocation working memory tasks following a RCT of alpha-down neurofeedback in PTSD. Neuroimage Clin, 2023; 37:103313; doi: 10.1016/j.nicl.2023.103313

49.

Tosti

, Corrado

, Mancone

, et al. Integrated use of biofeedback and neurofeedback techniques in treating pathological conditions and improving performance: A narrative review. Front Neurosci, 2024; 18:1358481; doi: 10.3389/fnins.2024.1358481

50.

Van der Kolk

, Hodgdon

, Gapen

, et al. A randomized controlled study of neurofeedback for chronic PTSD. PLoS One, 2016; 11(12):e0166752; doi: 10.1371/journal.pone.0166752

51.

Weaver

, Birn

, Cisler

. A pilot adaptive neurofeedback investigation of the neural mechanisms of implicit emotion regulation among women with PTSD. Front Syst Neurosci, 2020; 14:40; doi: 10.3389/fnsys.2020.00040

52.

Zhao

, Duek

, Seidemann

, et al. Amygdala downregulation training using fMRI neurofeedback in post-traumatic stress disorder: A randomized, double-blind trial. Transl Psychiatry, 2023; 13(1):177; doi: 10.1038/s41398-023-02467-6

53.

Zotev

, Phillips

, Misaki

, et al. Real-time fMRI neurofeedback training of the amygdala activity with simultaneous EEG in veterans with combat-related PTSD. Neuroimage Clin, 2018; 19:106–121; doi: 10.1016/j.nicl.2018.04.010

54.

Zweerings

, Pflieger

, Mathiak

, et al. Impaired voluntary control in PTSD: Probing self-regulation of the ACC with real-time fMRI. Front Psychiatry, 2018; 9:219; doi: 10.3389/fpsyt.2018.00219

55.

Zweerings

, Sarasjärvi

, Mathiak

, et al. Data-driven approach to the analysis of real-time fMRI neurofeedback data: Disorder-specific brain synchrony in PTSD. Int J Neural Syst, 2021; 31(11):2150043; doi: 10.1142/S012906572150043X

56.

Zweerings

, Sarkheil

, Keller

, et al. Rt-fMRI neurofeedback-guided cognitive reappraisal training modulates amygdala responsivity in posttraumatic stress disorder. Neuroimage Clin, 2020; 28:102483; doi: 10.1016/j.nicl.2020.102483