Posterior cingulate cortex targeted real‐time fMRI neurofeedback recalibrates functional connectivity with the amygdala, posterior insula, and default‐mode network in PTSD

Abstract Background Alterations within large‐scale brain networks—namely, the default mode (DMN) and salience networks (SN)—are present among individuals with posttraumatic stress disorder (PTSD). Previous real‐time functional magnetic resonance imaging (fMRI) and electroencephalography neurofeedback studies suggest that regulating posterior cingulate cortex (PCC; the primary hub of the posterior DMN) activity may reduce PTSD symptoms and recalibrate altered network dynamics. However, PCC connectivity to the DMN and SN during PCC‐targeted fMRI neurofeedback remains unexamined and may help to elucidate neurophysiological mechanisms through which these symptom improvements may occur. Methods Using a trauma/emotion provocation paradigm, we investigated psychophysiological interactions over a single session of neurofeedback among PTSD (n = 14) and healthy control (n = 15) participants. We compared PCC functional connectivity between regulate (in which participants downregulated PCC activity) and view (in which participants did not exert regulatory control) conditions across the whole‐brain as well as in a priori specified regions‐of‐interest. Results During regulate as compared to view conditions, only the PTSD group showed significant PCC connectivity with anterior DMN (dmPFC, vmPFC) and SN (posterior insula) regions, whereas both groups displayed PCC connectivity with other posterior DMN areas (precuneus/cuneus). Additionally, as compared with controls, the PTSD group showed significantly greater PCC connectivity with the SN (amygdala) during regulate as compared to view conditions. Moreover, linear regression analyses revealed that during regulate as compared to view conditions, PCC connectivity to DMN and SN regions was positively correlated to psychiatric symptoms across all participants. Conclusion In summary, observations of PCC connectivity to the DMN and SN provide emerging evidence of neural mechanisms underlying PCC‐targeted fMRI neurofeedback among individuals with PTSD. This supports the use of PCC‐targeted neurofeedback as a means by which to recalibrate PTSD‐associated alterations in neural connectivity within the DMN and SN, which together, may help to facilitate improved emotion regulation abilities in PTSD.


INTRODUCTION
Posttraumatic stress disorder (PTSD) is a severely debilitating psychiatric condition that may develop among individuals in the aftermath of trauma and includes symptoms of trauma-related intrusive recollections and flashbacks, hyperarousal/hyperreactivity, avoidance behaviors, and negative alterations in cognitions and mood (American Psychiatric Association, 2022). Concerningly, a significant proportion of individuals with PTSD have difficulty tolerating and/or do not experience clinically significant symptom reductions from gold-standard psychotherapeutic and pharmacotherapeutic treatments (Bradley et al., 2005;Haagen et al., 2015;Kline et al., 2018;Krystal et al., 2017;Ravindran & Stein, 2009;Stein et al., 2006). As a result, many individuals remain burdened with persistent symptoms. In PTSD, it is well documented that there are strong associations between symptom presentations and multiple functional disruptions in the brain (Fenster et al., 2018;Lanius et al., 2015;Tursich et al., 2015). As such, emerging research supports the notion that recalibrating PTSD-associated functional disruptions within neural circuits constitutes a promising treatment approach that may complement existing psycho-and pharmacotherapies (Chiba et al., 2019;Koek et al., 2019;Nicholson et al., 2021;Panisch & Hai, 2020;Rogel et al., 2020).
More specifically, the DMN is comprised of several midline brain regions, including the posterior cingulate cortex (PCC), the precuneus, and the medial prefrontal cortex (mPFC), which together subserve functions of self-referential processing, social cognition, and autobiographical memory Qin & Northoff, 2011). Individuals with PTSD have repeatedly been found to display decreased coupling between major DMN nodes at rest, where the magnitude of decrease has been associated with greater PTSD symptom severity (Bluhm et al., 2009;di et al., 2012;Koch et al., 2016;Sripada et al., 2012). Among individuals with PTSD, functional disruptions in the DMN have been associated with traumatic/negative autobiographical memories, as well as distorted and fragmented self-referential processing (Akiki et al., 2017;Bluhm et al., 2009;Daniels et al., 2010;Fenster et al., 2018;Lanius et al., 2015;Tursich et al., 2015). Graph theoretical analyses have found spared connectivity within posterior DMN nodes (i.e., PCC, precuneus), relative to decreased connectivity within anterior DMN nodes, including the mPFC, among patients with PTSD during rest (Akiki et al., 2018;Holmes et al., 2018;Kennis et al., 2016;Shang et al., 2014). On the other hand, during working memory tasks, individuals with PTSD have been found to display maladaptive increased connectivity of the PCC with other DMN areas, as compared to increased connectivity with CEN and SN regions observed in healthy individuals (Daniels et al., 2010). This may reflect the difficulty faced by patients with PTSD in disengaging attention from self-referential processing to focus on more task-relevant stimuli (Aupperle et al., 2012). Furthermore, there is a substantial body of literature establishing enhanced activation within the PCC and other DMN regions among individuals with PTSD during the reliving and reexperiencing of trauma-related autobiographical memories (Awasthi et al., 2020;Fenster et al., 2018;Frewen et al., 2011;Hopper et al., 2007;Liberzon & Abelson, 2016;Mickleborough et al., 2011;Ramage et al., 2013;Thome et al., 2020). Indeed, findings of enhanced DMN recruitment in response to trauma reminders is not surprising given that individuals living with PTSD can often report feeling as though their sense-of-self is inextricably fused with their experiences of trauma (Fenster et al., 2018;Frewen et al., 2011;Lanius et al., 2020;Liberzon & Abelson, 2016;Terpou et al., 2019;Terpou et al., 2020;Thome et al., 2020).
Regions of the SN are also critically involved in the innate alarm system, a network of brain regions that together enable the rapid detection and response to threatening environmental stimuli via the integration of emotion and sensory information, which has critical implications for chronic stress exposure and PTSD (Lanius et al., 2017;Szeszko & Yehuda, 2019). Among individuals with PTSD, alterations in restingstate SN activity have been associated with symptoms of hyperarousal, hypervigilance, avoidance, and altered interoception (Akiki et al., 2017;Harricharan et al., 2020;Koch et al., 2016;McCurry et al., 2020;Rabinak et al., 2011;Sripada et al., 2012;Tursich et al., 2015;Yehuda et al., 2015). Specifically, studies have found that individuals with PTSD exhibit increased coupling of the anterior insula within the SN, as well as reduced connectivity between the SN and the dorsolateral prefrontal cortex (dlPFC) which is involved in emotion regulation Jeong et al., 2019;Koch et al., 2016;Lanius et al., 2015;Sripada et al., 2012). With regard to the amygdala, a limbic region centrally involved in emotion generation/processing and the innate alarm system, hyperactivity has been repeatedly demonstrated in association with PTSD psychopathology (Aghajani et al., 2016;Birn et al., 2014;Etkin et al., 2011;Fenster et al., 2018;Fitzgerald et al., 2018;Koch et al., 2016;Lanius et al., 2015;Mickleborough et al., 2011;Patel et al., 2012;Schulze et al., 2019;Yehuda et al., 2015). Indeed, the role of altered SN activity in PTSD is unsurprising given that many PTSD symptoms (e.g., hyperarousal, hypervigilance, irritability, aggression) often involve maladaptive coding of salience and thus misassignment of the appropriate degree of imminence and threat to external stimuli (Szeszko & Yehuda, 2019). Interestingly, in a narrative review of MRI studies that attempted to predict PTSD treatment response to psychotherapies (i.e., cognitive behavioral therapy and exposure-based therapies), it was found that better responses following psychotherapeutic treatment were associated with reduced amygdala and anterior insula activity along with increased ACC activity (Szeszko & Yehuda, 2019), thus indicating that recalibrating altered SN connectivity may be a critical path towards improvement of PTSD symptoms. Taken together, it is apparent that altered connectivity patterns within both the DMN and SN are highly associated with PTSD symptoms. For this reason, interventions that restore DMN and SN functioning may represent a promising treatment avenue for reducing PTSD symptomatology (Koek et al., 2019;Szeszko & Yehuda, 2019). Of importance, emerging evidence from neurofeedback studies in PTSD suggest that it may be effective in recalibrating aberrant DMN and SN dynamics and reducing associated symptoms (Bell et al., 2019;Gerin et al., 2016;Kluetsch et al., 2014;Misaki et al., 2019;Misaki et al., 2018;Nicholson et al., 2021;Nicholson et al., 2018;Zotev et al., 2018).

Restoring DMN and SN connectivity in PTSD with neurofeedback
Neurofeedback is an emerging adjunctive treatment approach that has been investigated across a wide range of psychiatric disorders (Li et al., 2013;Linden et al., 2012;Mehler et al., 2018;Schoenberg & David, 2014;Young et al., 2017), including PTSD (Bois et al., 2021;Chiba et al., 2019;Gerin et al., 2016;Kluetsch et al., 2014;Misaki et al., 2019;Misaki et al., 2018;Nicholson et al., 2021;Nicholson et al., 2018;van der Kolk et al., 2016;Weaver et al., 2020;Zotev et al., 2018;Zweerings et al., 2018). During neurofeedback, neural activity is measured and presented to participants in real-time, thus allowing them to noninvasively selfregulate brain activity (Sitaram et al., 2017). As such, individuals with PTSD can learn to regulate disrupted brain activity that is associated with the manifestation and maintenance of their symptoms. Importantly, recalibrating activity within main ICN hubs may help restore optimal interactions within-and between-ICNs. As altered connectivity within the DMN and SN are highly associated with PTSD symptoms, utilizing neurofeedback to restore DMN and SN connectivity may be a promising approach for reducing PTSD symptoms. Importantly, preliminary evidence suggests that neurofeedback may be effective in recalibrating aberrant DMN and SN dynamics and reducing PTSD symptoms Nicholson et al., 2021;Nicholson et al., 2018; via homeostatic neuroplasticity (Bois et al., 2021;Sitaram et al., 2017;Nicholson et al., 2023) and, as such, merits further investigation.
Previous electroencephalography (EEG) neurofeedback studies have been conducted in PTSD, including single-session mechanistic studies  and a 20-week randomized controlled trial Nicholson et al., 2023;Shaw and Nicholson et al., 2023), targeting alpha oscillations over the PCC. Alpha oscillations are correlated with activation of the DMN, including the PCC (Clancy et al., 2020;Jann et al., 2009;Mantini et al., 2007), where reduced resting-state alpha-rhythms in PTSD are thought to reflect chronic hyperarousal associated with SN hyperactivity and dysregulated DMN dynamics Clancy et al., 2020;Clancy et al., 2017;Liberzon & Abelson, 2016;Sitaram et al., 2017). We found previously that a single session of alpha-based EEG neurofeedback recalibrated DMN and SN resting-state functional connectivity, which was associated with reduced hyperarousal symptoms . Similarly, in our randomized controlled trial of alpha-based EEG neurofeedback, significantly decreased PTSD severity scores in the experimental group were associated with restored patterns of resting-state connectivity within the DMN (i.e., increased dmPFC connectivity and decreased PCC/precuneus connectivity with the DMN) and SN (i.e., decreased anterior insula connectivity with the SN) . Importantly, at a 3-month follow-up, over 60% of experimental group participants no longer met diagnostic criteria for PTSD , a remission rate comparable to currently available psychotherapies and pharmacotherapies for PTSD (Berger et al., 2009;Bradley et al., 2005;Haagen et al., 2015;Krystal et al., 2017;Ravindran & Stein, 2009;Stein et al., 2006). Taken together, these results indicate that regulating PCC/posterior DMN associated brain signals (i.e., alpha rhythms) may be effective in reducing PTSD symptoms and recalibrating altered DMN and SN connectivity among individuals with PTSD Nicholson et al., 2023;Shaw and Nicholson et al., 2023). In comparison with EEG neurofeedback, real-time functional magnetic resonance imaging (fMRI) neurofeedback (rt-fMRI-NFB) affords superior spatial resolution and is therefore a critical modality for generating evidence to further elucidate neurophysiological mechanisms underlying treatment success.
Real-time fMRI-NFB has been successfully implemented in PTSD populations, where a number of previous studies have targeted regulation of the amygdala (Chiba et al., 2019;Gerin et al., 2016;Misaki et al., 2019;Misaki et al., 2018;Nicholson et al., 2018;Zotev et al., 2018). Indeed, in a previous study by our group, it was shown that among PTSD participants exposed to personalized trauma words, downregulating the amygdala was associated with increased activity and connectivity of the dlPFC and vlPFC, brain regions associated with emotion regulation , a finding that was replicated in other amygdala-targeted rt-fMRI-NFB studies in PTSD Zotev et al., 2018). Moreover, in a network-based analysis, amygdala downregulation was also found to induce neuroplastic changes within ICNs, including the DMN and SN, over neurofeedback training (Nicholson et al., 2018), an effect similar to those found in EEG-based neurofeedback investigations .
Most recently, our group utilized rt-fMRI-NFB to train individuals with PTSD and healthy controls to downregulate PCC activity during the same trauma/emotion provocation paradigm in which participants viewed personalized trauma/distressing words (Nicholson et al., 2021).
The PCC was targeted for neurofeedback as it shows hyperactivity during the reliving and reexperiencing of trauma memories among individuals with PTSD (Awasthi et al., 2020;Fenster et al., 2018;Frewen et al., 2011;Hopper et al., 2007;Liberzon & Abelson, 2016;Mickleborough et al., 2011;Ramage et al., 2013;Thome et al., 2020). In a previously published analysis, we reported that individuals in the PTSD group showed reduced symptoms of reliving and distress in response to trauma-related stimuli over neurofeedback training (Nicholson et al., 2021), an effect not observed during the single-session rt-fMRI-NFB study targeting the amygdala (Nicholson et al., 2018;. While both groups were able to downregulate their PCC activity with similar success over neurofeedback training, downregulation was associated with unique within-PTSD group decreases in neural activity in several brain regions including those of the DMN (i.e., bilateral dmPFC, hippocampus) and SN (i.e., amygdala, mid-cingulate cortex) (Nicholson et al., 2021). Interestingly, these findings add to a growing body of research that highlight the critical involvement of the PCC in emotion regulation, including associations between normalized PCC activity and positive clinical outcomes following psychotherapy (Fresco et al., 2017;Garrett et al., 2019;Scult et al., 2019), as well as positive correlations between PCC-amygdala connectivity and PTSD symptom severity following trauma exposure Zhou et al., 2012). The PCC is also critical in neural circuits supporting cognitive reappraisal, attentional shifting, and acceptance (Ferri et al., 2016;Kanske et al., 2011;King et al., 2016;Ma et al., 2017;Messina et al., 2021), all of which are key volitional emotion regulation strategies. Notably, however, functional connectivity of the PCC during rt-fMRI-NFB has not previously been investigated among individuals with PTSD. In the present analysis, our goal was to further examine how PCC-targeted fMRI neurofeedback may recalibrate connectivity within large-scale brain networks. In doing so, we were able to characterize neural mechanisms that might underlie improvements in emotion regulation among PTSD participants that were previously  reported by our group (Nicholson et al., 2021) and also evaluate whether findings are convergent with that of previous EEG-based neurofeedback studies Nicholson et al., 2018). Given the accumulation of functional neuroimaging findings of altered DMN and SN activity in PTSD, as well as studies which have demonstrated a recalibration of DMN and SN connectivity as a function of successful neurofeedback treatment for PTSD, examination of functional connectivity between the PCC and key hubs of both networks was of particular interest in this study.

Current study
In the current rt-fMRI-NFB study, individuals with PTSD and healthy controls trained to downregulate their PCC during a trauma/emotion provocation paradigm, wherein participants viewed traumarelated/distressing words. In the present analysis, we utilized the psychophysiological interaction (PPI) method to elucidate within-and between-group differences in PCC functional connectivity during neurofeedback training. During regulation, we hypothesized that the PCC would display unique functional connectivity to key DMN and SN hubs within the PTSD group. Furthermore, we also expected to observe positive associations between psychiatric symptoms and functional connectivity between the PCC and both DMN and SN hubs.

Participants
Our total sample (n = 30) consisted of n = 15 participants who met the criteria for a current primary diagnosis of PTSD, as determined by the Clinician-Administered PTSD Scale (CAPS-5) (Weathers et al., 2018) and the Structured Clinical Interview for DSM-5 (SCID) (First, 2015), and n = 15 healthy, nontrauma-exposed controls (see Table 1).
One participant in the PTSD group was not included in the analysis as they self-reported falling asleep in the scanner during the transfer run. Hence, the final sample size was n = 14 in the PTSD group and n = 15 in the healthy control group. The PTSD and healthy control groups had nonsignificant differences with respect to biological sex.
The mean age of PTSD participants was significantly higher than that of healthy control participants. As such, we included age as a covariate in our analyses and, importantly, did not find any significant differences in PCC functional connectivity results. Similarly, given its potential impact on brain connectivity, participant handedness was also included as a covariate in our analyses and was found not to yield any significant differences in PCC functional connectivity. Notably, no study participants reported a significant history of head injuries without loss of consciousness. While all analyses presented in this paper are novel, the imaging data attained from this participant group was previously analyzed in a recently published work (Nicholson et al., 2021). Participants were included in the PTSD group if they had a current primary diagnosis of PTSD as measured by the CAPS-5 and SCID.
PTSD participants currently receiving psychotropic medication were all on a stable dose for at least 1 month prior to their participation in the current study. PTSD participants were excluded from the study if they had ongoing and/or recent (within previous 3 months) alcohol or substance use disorders, suicidal ideations, or self-injurious behaviors requiring medical attention. Lifetime diagnoses of bipolar or psychotic disorders were additional exclusion criteria for PTSD participants. Control participants were excluded from the study if they had a F I G U R E 1 Depiction of the rt-fMRI-NFB set-up. While participants were inside the scanner, they were presented with a neurofeedback signal in the form of a virtual thermometer that increased/decreased in response to fluctuating activity within the neurofeedback target region (PCC). Participants completed three neurofeedback training runs, followed by a transfer run, in which they were not presented with the neurofeedback signal. Figure (Table S1) for information pertaining to the history of trauma exposure in each group.

Neurofeedback paradigm
This study used an experimental protocol and neurofeedback paradigm as described in our previously published work (Nicholson et al., 2021;Nicholson et al., 2018;) (see Figure 1). During neurofeedback training runs, participants were shown a neurofeedback signal corresponding to activation in their PCC. The neurofeedback signal was made to appear as two identical thermometers on either side of a screen projected to participants while they were inside the scanner. The bars on the thermometer increased/decreased in correspondence to changing BOLD signal from the PCC target region.
Each segment within the thermometer corresponded to a 0.2% activation change in the PCC, with a maximal increase range of 2.8% and a maximal decrease range of 1.2% from baseline activation (Nicholson et al., 2021;Paret et al., 2014;Paret et al., 2016). At the onset of each trial, the mean of the preceding four data points was taken as the baseline and displayed to participants as an orange line on the thermometer (Nicholson et al., 2021;Paret et al., 2014;Paret et al., 2016). Participants were not provided with specific instructions on regulation strategies. Rather, participants were told that they would be "regulating an area of the brain related to emotional experience." Participants were instructed to focus their vision on the word for the entire duration of each condition, while using their peripheral vision to monitor the thermometers.
Before the first run, participants were informed about the temporal delay in neurofeedback signal due to the BOLD signal delay.
Our neurofeedback protocol consisted of three conditions: regulate, view, and neutral. In the regulate condition, participants were instructed to decrease the neurofeedback signal, corresponding to activity in the PCC seed region, while viewing a personalized traumarelated word (PTSD group) or a matched distressing word (healthy control group). In the view condition, participants viewed the chosen words but were instructed to respond naturally and not attempt to change the neurofeedback signal. In the neutral condition, participants in both groups viewed personalized neutral words and were instructed to respond naturally. Participants selected personalized trauma/distressing words (n = 10) and neutral words (n = 10) under the guidance of a trauma-informed clinician.

Real-time signal processing for neurofeedback
To present real-time PCC neural activation to participants via thermometer display, anatomical scans were imported into BrainVoyager (version QX2.4; Brain Innovations, Maastrict, Netherlands), then skullstripped and transformed into Talairach space. Subsequently, normalization parameters were input into TurboBrainVoyager (TBV) (version 3.0, Brain Innovations, Maastricht, Netherlands). A 4-mm full-widthhalf-maximum (FWHM) Gaussian kernel was implemented in TBV to allow for motion correction and spatial smoothing. The first two volumes of the functional scans were removed prior to real-time processing. The neurofeedback target, the PCC, was defined using a 6 mm sphere over the coordinate (MNI: 0 −50 20) (Bluhm et al., 2009), and the "best voxel selection" tool in TBV was used to calculate the BOLD signal amplitude in this target area. This method identifies the 33% most active voxels (i.e., highest beta-values) for the view > neutral contrast. The first two trials comprising each neurofeedback run were the view and neutral conditions to enable an initial voxel selection based on the view > neutral contrast, which was dynamically updated throughout the duration of training.

Neurofeedback PCC downregulation analysis and state changes in emotional experience
In order to evaluate PCC downregulation (neurofeedback success), we examined the event-related BOLD response from the PCC target sphere during the regulate and view training conditions (Nicholson et al., 2021). We also examined state changes in subjective response to traumatic/distressing stimuli over the neurofeedback training experiment, as measured by RSDI subscales (Nicholson et al., 2021). As these analyses have been previously published by our group (Nicholson et al., 2021), methodological details are included in the supplementary material (see supplementary methods s2).

PPI analysis
Novel analyses in the current manuscript included examining taskdependent changes in PCC functional connectivity during neurofeedback training via PPI analyses as implemented in SPM12. A PPI analysis measures the context-sensitive change in connectivity between one or more brain regions by comparing connectivity of the seed region during a particular experimental condition versus connectivity of the seed region during a different condition (Friston et al., 1997). In this case, we investigated the change in PCC connectivity during the view con- The PPI interaction terms were created by deconvolving the BOLD signal of the physiological regressor (i.e., timeseries of the selected PCC sphere) by the hemodynamic response function and multiplying by the psychological regressor (view > regulate), thus creating a series of interaction terms. These interaction terms were then reconvolved with the hemodynamic response function prior to being passed on to the second-level for within-and between-group analyses. For the left and right PCC seed, across all training runs combined, we conducted a priori planned one-and two-sample t-tests to examine within-and between-group PCC connectivity across the whole-brain. We were also interested in assessing how PCC connectivity changed over the course of neurofeedback training and therefore conducted additional one-sample t-tests for both groups to examine PCC connectivity during training run 3 as compared to training run 1, and vice versa. Additionally, for the left and right PCC seed, we also conducted a priori planned one-and two-sample t-tests for the transfer run separately. Previous research by our group, including a randomized controlled trial, showed that neurofeedback is associated with normalized DMN and SN activity among individuals with PTSD . As such, we chose to also perform a region-of-interest (ROI) analysis using a single mask consisting of 2 main hubs from each of the DMN (dmPFC, vmPFC) and SN (bilateral amygdala, bilateral insula).
Coordinates for the midline vmPFC (x, y, z = −2, 36, −2) and dmPFC (x, All coordinates are reported in MNI space. This single ROI mask was applied to all aforementioned one-and two-sample t-tests. Both wholebrain and ROI analyses were corrected for multiple comparisons using a peak-level family-wise error (FWE) threshold at p < .05, k = 10, with an initial cluster defining threshold at p < .0001, k = 10 (Eklund et al., 2016;Roiser et al., 2016).

Clinical correlations
Linear regression analyses were conducted across all subjects, to evaluate correlations between clinical scores and PCC connectivity over neurofeedback training runs. Interaction term parameters were correlated with participant scores for CAPS-5 total, DERS, CTQ, BDI, and MDI. Results were corrected for multiple comparisons using the same peak-level FWE threshold at p < .05, k = 10, with an initial cluster defining threshold at p < .0001, k = 10 (Eklund et al., 2016;Roiser et al., 2016).

Regulation strategies
As previously mentioned, participants were not provided with any spe-

PCC downregulation with neurofeedback and state changes in emotional experience
Individuals in both the PTSD and healthy control groups were similarly able to significantly decrease activity within their PCC (neurofeedback target area) during regulate as compared to view conditions during all three neurofeedback training runs, as well as for the transfer run (see Figure 2, as also reported elsewhere) (Nicholson et al., 2021).
Indeed, there were no significant differences between the two groups when comparing the PCC BOLD responses during regulate as compared to view conditions for each of the neurofeedback training runs and transfer run (Nicholson et al., 2021). In other words, the PTSD and healthy control participants were able to regulate PCC activity with similar success (Nicholson et al., 2021). Additionally, over neurofeedback training and in response to a trauma/emotion provocation paradigm, both the PTSD and healthy control groups showed significantly reduced reliving symptoms, whereas only the PTSD group showed significant reductions in distress symptoms, as measured by the RSDI scale (see Figure 3, as also reported elsewhere) (Nicholson et al., 2021).

Neurofeedback training runs
Both the PTSD and healthy control groups demonstrated significant whole-brain PCC connectivity with the bilateral precuneus/cuneus during regulate as compared to view neurofeedback training conditions (see Table 2 and Figures 4a and b). However, only the PTSD group showed significant whole-brain PCC connectivity with the left parietal/central operculum during regulate as compared to view conditions during neurofeedback training (see Table 2 and Figure 4b). Note: Within-and between-group differences in functional connectivity of the PCC (left and right seeds) during the neurofeedback training runs. Reported results for whole-brain and ROI analyses are at a significance threshold of p-FWE peak-level < .05, k = 10. The comparison column lists each group comparison, and the contrast column indicates the direction in which the regulate and view conditions are being contrasted (i.e., view > regulate or regulate > view). The brain region, hemisphere of the region (H), cluster size (k), MNI coordinates (x, y, z), t-statistic (t-stat.), z-Score, and significance ( Interestingly, the ROI analyses revealed that only the PTSD group displayed significant PCC functional connectivity to the DMN and SN during regulate as compared to view neurofeedback training conditions. Both the left and right PCC seeds were found to yield significant functional connectivity with the left posterior insula during regulate as compared to view conditions (see Table 2 and Figure 4b). Additionally, both PCC seeds were also found to yield significant PCC connectivity with PFC regions-the left PCC to the dorsomedial prefrontal cortex (dmPFC) and the right PCC to the ventromedial prefrontal cortex (vmPFC)-during regulate as compared to view conditions (see Table 2 and Figure 4b). Furthermore, in the between-group ROI analysis, the right PCC was found to display significantly greater connectivity with the right amygdala for the PTSD group as compared to the control group, during regulate as compared to view neurofeedback training conditions (see Table 2 and Figure 4c). Lastly, we did not detect any significant changes in PCC connectivity over the course of neurofeedback training (i.e., run 3 as compared to run 1, and vice versa) during regulate as compared to view, and vice versa.

Transfer run
Within-group whole-brain analyses of PCC functional connectivity during the transfer run revealed that the healthy control group showed left PCC connectivity with the bilateral calcarine cortex/cuneus during regulate as compared to view conditions (see Table 3 and Figure 4d). For both whole-brain and ROI analyses, there were nonsignificant differences in PCC connectivity within the PTSD group and when comparing between the PTSD and healthy control groups.

Clinical correlations
We found several significant results when conducting linear regression analyses between PCC functional connectivity and psychiatric clinical symptoms across all participants during neurofeedback training runs (see Table 4 and Figure 5). In the whole-brain analysis, a positive cor-

Regulation strategies
In the present study, participants were not provided with any instructions regarding strategies for neurofeedback regulation. Interestingly, we observed the recurrence of several consistent themes across the self-reported regulation strategies utilized by participants. Indeed, the most commonly reported strategy was relating to the monitoring and/or alteration of the breath (e.g., deep breathing, breath as an anchor, etc.). Other common strategies reported by participants include self-talk, and the use of visual imagery. Notably, the majority of participants reported utilizing two or more regulatory strategies throughout the session.

DISCUSSION
In our previously published analysis, we reported that a single session of successful PCC-downregulation with rt-fMRI-NFB facilitated emotion regulation among individuals with PTSD, where psychological symptoms of reliving and distress were reduced during a trauma/emotion provocation paradigm (Nicholson et al., 2021). Indeed, this corresponded to unique patterns of brain activation among individuals with PTSD and healthy controls during neurofeedback training (Nicholson et al., 2021). In the present analysis, comparing functional connectivity of the PCC during regulate as compared to view conditions permits further insight into neurobiological mechanisms that may Note: Clinical correlations from the linear regression analyses between clinical scores and PCC functional connectivity during the neurofeedback training runs. The linear regression analyses were conducted across all study participants. Reported results for whole-brain and ROI analyses are at a significance threshold of p-FWE peak-level < .05, k = 10. The measure column lists the specific clinical inventory or questionnaire administered. The direction of correlation (±), PCC seed, brain region, hemisphere of the region (H), cluster size (k), MNI coordinates (x, y, z), t-statistic (t-stat.), z-Score, and significance ( These linear regression results may indicate that individuals with more severe symptoms maintain a greater need for network recalibration in comparison with those with milder symptoms. Importantly, observed patterns of DMN and SN connectivity during neurofeedback, as well as correlations with psychiatric symptoms, are concordant with connectivity results from both our pilot study  and randomized controlled trial  of alpha-rhythm EEG neurofeedback.
During neurofeedback training, we observed both commonalities and differences between the PTSD and healthy control groups with respect to PCC-DMN connectivity. Both groups were found to exhibit PCC-precuneus/cuneus connectivity during regulate as compared to view conditions. Conversely, there was unique within-PTSD group connectivity between the PCC and both the dmPFC and vmPFC dur- This may represent a neurofeedback-mediated recalibration of PTSDassociated alterations in DMN connectivity. These findings are also in alignment with results from our group's randomized controlled trial of EEG neurofeedback, wherein participants trained to downregulate alpha amplitude using real-time EEG feedback signals from the PCC/posterior DMN . In this study, decreased PTSD severity scores coincided with increased resting-state connectivity between the DMN and the dmPFC after NFB training, which may also reflect a recalibration of aberrant connectivity patterns associated with PTSD.
Although dorsomedial and ventromedial DMN communities have differences with regard to cytoarchitecture and anatomical connectivity, as well as some degree of functional segregation (Raichle, 2015), both are consistently implicated in PTSD psychopathology (Bluhm et al., 2009;Daniels et al., 2010;. Indeed, the dmPFC and vmPFC have both been repeatedly demonstrated to be hypoactive during the reliving and reexperiencing of traumatic experiences, as well as the processing of fearful facial expressions, among individuals with PTSD (Killgore et al., 2014;Lanius et al., 2007;Peres et al., 2011;Shin et al., 2005;Williams et al., 2006). Interestingly, the dmPFC was also identified in our linear regression analysis, where there was a positive correlation between PCC-dmPFC connectivity and CTQ scores during regulate as compared to view conditions. Severe and/or prolonged childhood trauma has been shown to have a destabilizing impact on the developmental differentiation of the DMN, particularly with regard to integration between the anterior and posterior DMN subsystems (Bluhm et al., 2009;Daniels et al., 2011) (Akiki et al., 2017;Bluhm et al., 2009;Daniels et al., 2010;Fenster et al., 2018;Lanius et al., 2015;Tursich et al., 2015).
During neurofeedback training, we observed unique patterns of PCC-SN connectivity among PTSD participants. The within-PTSD group comparison revealed significant PCC-left posterior insula connectivity during regulate as compared to view conditions. Additionally, the linear regression analysis revealed a positive correlation between right PCC-left anterior insula connectivity and CAPS scores during regulate compared to view. The insula, a core cortical region of the SN, is integral to the processing of emotional states and can be parcellated into functionally differentiated anterior, mid-, and posterior subregions (Craig, 2009, Chang et al., 2013Couto et al., 2013;Craig, 2002). Interestingly, a posterior-to-mid-to-anterior integration of interoceptive information (i.e., signals related to the internal state of the body) within the insular cortex has been proposed and substantiated by functional neuroimaging research (Craig, 2009, Cauda et al., 2012Chang et al., 2013;Couto et al., 2013;Craig, 2002;Simmons et al., 2013). According to this model, primary interoceptive signals are first represented in the posterior insula and integrated with external (environmental) and sensory information along the processing pathway (Craig, 2009, Cauda et al., 2012Chang et al., 2013;Couto et al., 2013;Craig, 2002;Simmons et al., 2013). Subsequently, the anterior insula is critically involved in attributing salience and facilitating higher-order emotional and perceptive processing via interactions with frontal brain regions, including particularly strong functional interactions with the anterior cingulate cortex (Craig, 2009, Beissner et al., 2013Cauda et al., 2011;Uddin, 2015). In PTSD, dysregulation within the insular cortex-namely, posterior insula hypoactivity and anterior insula hyperactivity Koch et al., 2016;Patel et al., 2012;Wang et al., 2016)-may reflect a disrupted balance between interoceptive and exteroceptive processing, as well as maladaptive emotional reactivity to traumarelated stimuli and altered bodily self-consciousness. Indeed, extensive research has linked altered connectivity within subregions of the insular cortex to particular PTSD symptoms (Akiki et al., 2017;Koch et al., 2016;McCurry et al., 2020;Rabinak et al., 2011;Sripada et al., 2012;Tursich et al., 2015;Yehuda et al., 2015). In one study, Tursich et al. (2015) analyzed network connectivity via an independent component analysis and found decreased left posterior insula integration within the SN as a function of hyperarousal symptoms, along with opposing trend-level SN associations of overlapping clusters, including the anterior insula, as a function of reexperiencing (increased connectivity) and avoidance/numbing (decreased connectivity). Additionally, our observation of PCC connectivity with only the left (as opposed to bilateral or only the right) posterior insula may be consistent with research suggesting that affective processing may entail critical involvement of the left posterior insula in particular (Duerden et al., 2013), although future research in this area is warranted. Taken together, regulating PCC activity via neurofeedback may be an effective approach in recalibrating PTSD-associated alterations within the insular cortex.
With regard to the linear regression analysis, another critical positive correlation was found between left PCC-right ventral striatum/nucleus accumbens (VS/NAcc) connectivity and CTQ scores during regulate compared with view neurofeedback training conditions.
Although not specified as a ROI ahead of the present analysis, the VS/NAcc is a key hub of the SN and is critically involved in rewardbased brain networks (Menon, 2011;Peters et al., 2016;Seeley et al., 2007). Significantly, childhood adversity and/or trauma has been shown to alter neural connectivity, including in the VS/NAcc, pertaining to an individual's seeking system (Birnie et al., 2020;McLaughlin et al., 2019), a system whose activation determines one's attitude and/or disposition towards their environment (Alcaro & Panksepp, 2011). Indeed, such alterations in neural connectivity may explain the occurrence of negatively valenced seeking behaviors that are commonly observed among individuals with PTSD, including elevated risk-taking (e.g., thrill seeking, aggression) (Contractor et al., 2017;Seidemann et al., 2021;Strom et al., 2012;Weiss et al., 2013), impulsivity (Kim & Choi, 2020;Mahoney et al., 2020;Weiss et al., 2012), and substance use (Mahoney et al., 2020;Michaels et al., 2021;Norman et al., 2018). Interestingly, convergent evidence has also indicated a modulatory effect of the VS/NAcc on the insular cortex by providing incentive signals as part of the integration of salience from sensory stimuli (Leong et al., 2016;Leong et al., 2021;Menon & Levitin, 2005;Perry et al., 2019).
Most importantly, however, results from the between-group comparison further supported the notion that regulating the PCC via neurofeedback may recalibrate SN connectivity. Specifically, we observed greater PCC-right amygdala connectivity for the PTSD as compared to the healthy control group during regulate as compared to view neurofeedback training conditions. Hyperactivity within the amygdala-a brain region associated with emotion generation and processing-has been repeatedly implicated in PTSD psychopathology (Aghajani et al., 2016;Birn et al., 2014;Etkin et al., 2011;Fenster et al., 2018;Fitzgerald et al., 2018;Koch et al., 2016;Lanius et al., 2015;Mickleborough et al., 2011;Patel et al., 2012;Yehuda et al., 2015). Moreover, alterations in amygdala connectivity are involved in multiple PTSD symptom domains, including avoidance, reexperiencing, and altered perception of valence (Fenster et al., 2018), in addition to its well-established association with hyperarousal and hyperreactivity symptoms. As such, downregulating the amygdala has been the most common target of previous rt-fMRI-NFB studies in PTSD (Chiba et al., 2019;Gerin et al., 2016;Misaki et al., 2019;Misaki et al., 2018;Nicholson et al., 2018;Zotev et al., 2018). Notably, in a previous EEG-based neurofeedback study by our group, we demonstrated a shift in amygdala connectivity from brain areas involved in defensive, emotional, and fear processing/memory retrieval (i.e., PAG and hippocampus) toward prefrontal areas that facilitate emotion regulation and executive functioning (i.e., vmPFC) . Indeed, given the central involvement of the amygdala in PTSD psychopathology, and in emotion generation/processing generally, observed PCC-amygdala functional connec-

Expanding neurobiological models of emotion regulation in PTSD
The manifestation and maintenance of PTSD symptoms have been linked to impaired emotion regulation capacities among individuals with this psychiatric disorder. As such, understanding the neural mechanisms underlying emotion regulation has been a priority for PTSD researchers and significant progress has been made. In PTSD, hypoactivity within the prefrontal cortex has been extensively reported and has been linked to impairments in emotion regulation (Andrewes & Jenkins, 2019;Fenster et al., 2018;Koenigs & Grafman, 2009;Lobo et al., 2011;Sripada et al., 2012). Indeed, such impairments among individuals with PTSD are commonly understood to involve reduced inhibition from prefrontal cognitive control systems on the subcortical limbic regions (i.e., the amygdala) that generate emotion (Comte et al., 2016;Gross, 1998;Ochsner & Gross, 2005;Ochsner et al., 2012). More recently, neurobiological evidence indicates that the DMN and the PCC are intimately linked with the limbic system and emotion generation areas and, as such, are critical targets for enabling healthy emotion regulation in the context of PTSD.
For example, in a recent meta-analysis, functional deactivations in the PCC were found to constitute the specific neural substrate underlying emotional acceptance -an emotion regulation strategy promoted by a wide variety of psychotherapeutic practices (Messina et al., 2021).
Interestingly, in a rt-fMRI-NFB study targeting PCC downregulation, mindfulness-based meditation was found to be associated with deactivation of the PCC (Garrison et al., 2013a;Garrison et al., 2013b). In a different study, increased PCC-ACC and PCC-anterior insula functional connectivity were associated with greater decentering, a specific metacognitive strategy for emotion regulation, among individuals with co-occurring anxiety and depressive disorders (Fresco et al., 2017).
With regard to PTSD specifically, a recent study reported decreases in both PCC and amygdala activation in conjunction with reduced PTSD symptoms among adolescents following trauma-focused cognitive behavioral therapy (Garrett et al., 2019). Additionally, the strength of PCC-amygdala connectivity has been demonstrated to have predictive value both in terms of positively correlating to PTSD symptom severity Zhou et al., 2012) and negatively correlating to treatment response among patients with PTSD post-intervention (Sheynin et al., 2020). Furthermore, neurofeedback-mediated regulation of the PCC has previously been shown to facilitate improvements in emotion regulation during a single-session Nicholson et al., 2021) and reduce PTSD symptoms in a 20-week randomized controlled trial . Taking these findings into consideration, downregulating the PCC may represent a fruitful approach for recalibrating aberrant limbic connectivity and facilitating emotion regulation in the context of PTSD.
Indeed, in the current study we observed PCC functional connectivity to subcortical limbic regions (i.e., amygdala) and limbic-related cortices (i.e., posterior insula, prefrontal cortex) that was unique to PTSD participants during rt-fMRI-NFB training. The observed functional connectivity results suggest that by regaining control over PCC activity, individuals with PTSD may be able to recalibrate aberrant connectivity within their subcortical limbic system and related cortical brain structures/regions. In doing so, they may be better able to regulate negative emotions that are central to PTSD symptomatology as evidenced by the observed decreases in reliving and distress symptoms. Taken together, existing neural mechanistic models of emotion regulation may benefit from expanding to include the involvement of the PCC/DMN in facilitating emotion regulation among individuals with PTSD.

Future directions and limitations
First, despite concordance with previous neurofeedback investigations in PTSD, we are unable to definitively conclude that these study findings are specifically attributable to PCC-targeted neurofeedback (i.e., neurophysiological specificity), as we did not include a shamcontrol condition. Second, we did not collect data on, and were thus unable to control for socio-economic status (including educational attainment) and the effect of previous or current psychotropic medication and/or psychotherapy on neurofeedback training success and brain network connectivity. A follow-up study to this PPI analysis should investigate the directed connectivity of the PCC during rt-fMRI-NFB using a generative modeling approach, such as dynamic causal modeling. Additionally, as most previous rt-fMRI-NFB studies have included only a single session of training, future studies should investigate PCC downregulation across multiple sessions which would help to elucidate the optimal 'dose' of neurofeedback for the treatment of PTSD. Furthermore, there are subgroups of clinical presentations of PTSD, such as the dissociative subtype, which should be considered separately by ideally powered studies to determine whether they are associated with unique responses to neurofeedback training. Last, further research is needed to investigate the involvement of the posterior DMN (i.e., PCC) in emotion regulation which may represent a fruitful research avenue in clinical neuroscience and psychiatry. Indeed, in the present study, there are several plausible emotion regulation strategies (e.g., effortless awareness, cognitive reappraisal), which may be associated with increased control over the PCC. Future research studies should seek to further clarify the efficacy of specific strategies.

CONCLUSION
In summary, we found that rt-fMRI-NFB targeting PCC downregulation led to within-and between-group differences in functional connectivity between the PCC and DMN and SN hubs during regulate as compared to view neurofeedback training conditions. While

CONFLICT OF INTEREST
All authors declare no financial interests or potential conflicts of interest with regard to the current study.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.

PEER REVIEW
The peer review history for this article is available at https://publons. com/publon/10.1002/brb3.2883