Emotional intensity and visual imagery in trauma nightmares map to posterior hippocampus connectivity

Abstract

Distressing dreams or nightmares about a traumatic experience are common among trauma-exposed individuals and contribute to the maintenance of posttraumatic stress disorder (PTSD) symptoms. Like other intrusive reexperiencing symptoms, trauma-related nightmares are characterized by vivid sensory details and heightened emotional intensity. However, little is known about their underlying neurobiology, and whether it overlaps with that of diurnal reexperiencing symptoms. Therefore, this study aimed to elucidate neurobiological correlates of the phenomenology of trauma-related nightmares. Sixty-six trauma-exposed adults (54 female) completed daily ecological momentary assessments (EMAs) of the phenomenological properties of their trauma-related nightmares for two weeks, as well as resting-state functional magnetic resonance imaging. Pearson correlations examined associations between nightmare properties and functional connectivity within two a priori circuits previously implicated in trauma-related intrusions and episodic memory: 1) an affective circuit between the anterior hippocampus (aHPC) and amygdala (Amyg), and 2) a sensory circuit between the posterior hippocampus (pHPC) and visual cortex (VC). There were no significant associations between nightmare properties and aHPC-Amyg connectivity. In contrast, stronger pHPC-VC connectivity was associated with greater visual detail and emotional intensity of nightmares. These findings align with the established roles of the pHPC and VC in the detailed experience of episodic memory and suggest a novel role for these regions in the affective features of trauma nightmares. Overall, these findings provide support for overlapping neural correlates of the properties of trauma nightmares and diurnal reexperiencing symptoms. They offer additional evidence for a pHPC-VC circuit as a promising neurotherapeutic target for trauma reexperiencing symptoms.

Introduction

Trauma-related distressing dreams or nightmares are highly prevalent among individuals exposed to trauma and reported by 70–90% of those diagnosed with posttraumatic stress disorder (PTSD) [1–3]. Network models of PTSD symptomatology have found that trauma-related nightmares, along with other intrusive reexperiencing symptoms, exhibit high centrality, or influence, on the expression of other PTSD symptoms [4–6]. This has prompted theories that position nightmares as a central symptom of PTSD rather than a secondary consequence, warranting direct targeting in treatment [4, 5, 7]. To facilitate the precise therapeutic targeting of trauma-related nightmares, a deeper understanding of their mechanisms is needed.

Insights into mechanisms of trauma-related nightmares may be found in their markedly intense phenomenological properties [8, 9]. As in other intrusive reexperiencing symptoms, nightmares often involve images or scenes [10, 11] that either replicate or thematically represent the traumatic event [12, 13]. Like trauma-related intrusive memories that occur during waking hours, these nightmares are characterized by a strong negative emotional component and vivid sensory details recalled upon awakening [14]. The intensity of these affective and sensory properties is linked to overall PTSD symptom severity [11], especially intrusion symptoms [15].

The shared phenomenological properties between trauma-related nightmares and diurnal intrusions suggest overlapping neural mechanisms [9]. Established neurobiological models of trauma reexperiencing symptoms have focused heavily on a circuit involving the amygdala (Amyg), hippocampus (HPC), and prefrontal cortex circuit, which supports conditioned fear memories [16]. Independent lines of research on the neural mechanisms of idiopathic nightmares also implicate this circuit [9, 17, 18]. These models highlight impaired top-down control from prefrontal regions over the Amyg, alongside dysfunctional HPC engagement, hindering adaptive memory integration and extinction of fear memories. During nightmares, a hyperactive Amyg is thought to amplify the emotional intensity of threat-related memory content [17]. Rather than facilitating coherent contextualization, the HPC may fail to integrate memories adaptively, instead reactivating fragmented sensory imagery that is relayed to the Amyg and fuels a fear response [17, 19]. Taken together, nightmares are believed to be maintained by a hyperactive Amyg in concert with the resurgence of fearful, arousing memory fragments provided by the HPC [12].

Recent findings from our group suggest that an expansion of this Amyg-HPC circuit may better account for the exaggerated sensory-perceptual vividness and affective intensity of trauma reexperiencing symptoms [20]. Our work builds on literature identifying distinct functional roles and connectivity patterns of the anterior and posterior segments of the HPC during episodic memory recall [21–23]. Studies in healthy adults show that the anterior hippocampus (aHPC) is more involved in retrieving higher-order affective and conceptual information, while the posterior hippocampus (pHPC) reinstates lower-order information such as detailed sensory and spatial information [24]. Functional connectivity and co-activation between the aHPC and the Amyg have been associated with recalling the affective components and schematic “gist” of a memory [25]. Connectivity between the pHPC and visual cortex (VC) has been found to support visuospatial imagery of a memory [26]. In PTSD, greater symptom severity has been associated with preferential connectivity between affective brain regions and the aHPC, compared to the pHPC [27]. In our prior work, we found that more robust resting-state functional coactivation between the pHPC and sensory cortices, particularly the VC, was linked to more intense sensory-perceptual features of trauma-related intrusive memories [20]. Together, these findings suggest that connectivity in two functional circuits – aHPC-Amyg and pHPC VC – may differentially contribute to the affective and sensory-perceptual qualities of trauma nightmares, respectively.

Building on this literature and drawing from the same participant sample as our prior study on daytime intrusive memories [20], we sought to examine whether trauma-related nightmare phenomenology reflects functional connectivity within these two circuits. Moreover, we examined the clinical relevance of nightmare phenomenology by linking self-reported nightmare qualities to clinician ratings of symptom severity, thus offering insights into intermediate processes (i.e., nightmare qualities) that may affect clinical symptoms. Integrating evidence from canonical models of PTSD, neurocognitive models of idiopathic nightmares, and emerging evidence on the neural correlates of trauma reexperiencing phenomenology, we hypothesized that: (1) stronger aHPC-Amyg resting-state connectivity would be associated with greater emotional intensity of nightmares; (2) stronger pHPC-VC connectivity would be associated with more vivid sensory (visual) detail of nightmares.

Methods

Participants and procedures

One hundred thirty-three trauma-exposed adults (101 female; 18 – 65 years old) completed a study of trauma reexperiencing symptoms [20, 28, 29]. Participants completed a study visit involving clinical interviews and magnetic resonance imaging (MRI), followed by two weeks of smartphone-delivered ecological momentary assessments (EMAs) of trauma-related memories and nightmares. Participants were recruited via advertisements in the Boston metropolitan area. Inclusion criteria were 1) exposure to a DSM-5 criterion A traumatic event and 2) experiencing at least two trauma-related intrusive memories per week over the past month. Exclusion criteria were left-handedness, medical conditions that could confound results (e.g., uncontrolled neurological disorder), a history of moderate to severe traumatic brain injury, current (past month) moderate-to-severe alcohol or substance use disorder, and current psychotic disorder or manic or mixed mood episode. While a history of psychotic or bipolar disorders was not exclusionary, no study participants met DSM-5 criteria for a lifetime diagnosis for these disorders. Study procedures were approved by the Mass General Brigham Human Research Committee, and all participants provided written informed consent.

Data from 17 participants were excluded due to MRI confounds or poor quality: excessive motion (n = 12), falling asleep during the scan (n = 2), and artifacts or anatomical defects (n = 3). An additional 50 participants were excluded for reporting less than 2 trauma-related nightmares over the EMA period. The final analyzed sample comprised 66 participants (54 female; mean age: 33.5 ± 10.6 years), and their demographic and clinical characteristics are summarized in Table 1.

Table 1.

Demographic and clinical characteristics of the study sample.

Total Sample (N = 66)
Age (years)	33.52 ± 10.62
Sex assigned at birth
Female	54 (82%)
Male	12 (18%)
Gender
Woman	45 (68%)
Man	15 (23%)
Gender fluid/non-binary	6 (9%)
Race
White	48 (73%)
Multiracial	11 (17%)
Asian	4 (6%)
Black	2 (3%)
Missing	1 (2%)
Ethnicity
Non-Hispanic	62 (94%)
Hispanic	4 (6%)
Current Psychotropic Medication	35 (53%)
Antipsychotic	1 (2%)
Antidepressant	28 (42%)
Mood Stabilizer	2 (3%)
Sedative/Hypnotic	1 (2%)
Stimulant	3 (5%)
CAPS-5 Nightmare Symptom Severity	1.98 ± 1.13
CAPS-5 Total Score	36.03 ± 11.25
CAPS-5 PTSD Diagnosis	56 (85%)

Values are presented as Mean ± Standard Deviation or N (%). CAPS-5 Clinician-Administered PTSD Scale for DSM-5. Sex assigned at birth, gender identity, race, and ethnicity were self-reported.

Ecological momentary assessments (EMAs)

Participants completed EMAs to capture the occurrence and phenomenological properties of trauma-related nightmares. Each morning, within one hour of the participant’s individualized wake time, a survey was delivered via the MetricWire smartphone app (Kitchener, Ontario, Canada). Participants first indicated whether they had experienced a trauma-related dream or nightmare during the previous night. If endorsed, they responded to 17 prompts about the specific phenomenological properties of the nightmare. These prompts were adapted from the Autobiographical Memory Questionnaire (AMQ) [30, 31] and rated on a 5-point (0–4) Likert scale. Based on prior work, nightmare properties were grouped into five dimensions: vividness, visual detail, emotional intensity, reliving, and fragmentation (i.e., extent to which the dream was experienced as “bits and pieces”) [32]. Ratings were averaged across the two-week EMA period. Zero-order correlations among nightmare properties are reported in Table 2. While our hypotheses focused on the visual and emotional qualities of nightmares, we included the other a priori defined features to assess specificity and determine which nightmare properties were most strongly related to clinical symptom severity.

Table 2.

Means, standard deviations, and zero-order correlations among nightmare properties assessed with ecological momentary assessment.

Variable	M	SD	1	2	3	4
1. Visual Detail	2.06	1.01	–
2. Emotional Intensity	2.39	0.94	0.58**	–
3. Vividness	1.83	0.97	0.74**	0.46**	–
4. Nowness	1.57	1.07	0.56**	0.55**	0.35**	–
5. Fragmentation	2.07	1.12	−0.32**	0.01	−0.39**	−0.19

M and SD represent mean and standard deviation, respectively. **p < 0.01.

Clinician-Administered PTSD scale for DSM-5 (CAPS-5)

The CAPS-5 [33], the gold-standard diagnostic interview for PTSD, was administered by doctoral-level clinicians after completion of the EMA period. This interview consists of 30 items assessing the onset, duration, and impact of PTSD symptoms, yielding a determination of PTSD diagnosis and symptom severity. Based on the CAPS-5, 85% of analyzed participants (n = 56) met diagnostic criteria for PTSD.

We used the nightmare symptom (Criterion B, Item 2) severity rating from the CAPS-5 to examine associations with nightmare properties, allowing us to determine which were most relevant to clinical severity and thus highlighting the clinical relevance of nightmare phenomenology and their underlying neural correlates.

Functional magnetic resonance imaging

Eyes-open resting-state fMRI data (13 min, 976 volumes) were acquired on a 3 T Siemens Prisma scanner with a 64-channel head coil using the Human Connectome Project (HCP) Lifespan protocol (T2-weighted echoplanar images; TR/TE: 800/37 ms, in-plane resolution: 2 mm; voxels: 2 mm isotropic; multiband factor = 8; anterior-posterior phase encoding). In addition, T1-weighted 3D MPRAGE anatomical images were obtained using the HCP 0.8 mm resolution sequence (TR/TEs: 2500/1.81/3.6/5.39/7.18; flip angle: 8 deg; FOV: 256 × 240; voxel size: 0.8 mm isotropic). MRI data were preprocessed using fMRIPrep version 20.2.7 [34]. Additional denoising steps included regressing out white matter and cerebrospinal fluid signals, scrubbing of motion outliers (framewise displacement > 0.5 mm) [35], and high pass (0.01 Hz) filtering using CONN toolbox scripts [36]. Participants were excluded from analyses if their mean framewise displacement (FD) exceeded 0.5 mm or if more than 20% of volumes exceeded the 0.5 mm threshold (n = 12).

Cleaned timeseries were extracted from the right aHPC and pHPC regions of interest (ROIs), defined using a study-specific probabilistic template of these regions [29]. Additional ROIs consisted of a right Amyg ROI and bilateral cuneus (referred to as VC hereon) from the Harvard-Oxford Atlas [37]. Right-lateralized HPC and Amyg ROIs were used given the parent study’s a priori focus on the right HPC, based on prior work demonstrating reexperiencing symptom associations specific to the right, but not left, HPC [38]. Results for the left HPC and Amyg are also reported for thoroughness but were not examined for hypothesis testing. Pearson correlations were performed between aHPC and Amyg ROI timeseries and pHPC and VC ROI timeseries to estimate functional connectivity (FC) within an affective circuit (aHPC-Amyg) and a sensory circuit (pHPC-VC). Additional whole-brain seed-to-voxel correlations were performed with the a/pHPC ROIs to compute whole-brain functional connectivity maps. ROI- and seed-based FC values were Fisher’s z-transformed prior to further statistical analyses.

Statistical analyses

To evaluate the clinical significance of nightmare properties, particularly visual detail and emotional intensity, we first examined their association with clinician-rated nightmare symptom severity (CAPS-5 Item B2) using Pearson correlations. Bonferroni correction was applied to adjust for multiple comparisons across 5 properties (p < 0.01). To ascertain the unique contribution of individual nightmare properties when controlling for their shared variance, a linear regression was performed with all nightmare properties simultaneously entered as predictors of clinician-rated nightmare symptom severity.

Next, we tested associations between nightmare properties and a/pHPC connectivity. Pearson correlations were performed between each property and connectivity within the right aHPC-Amyg and right pHPC-VC circuits. Bonferroni correction for multiple comparisons was applied to adjust for 10 comparisons (5 properties x 2 circuits = 10 tests; p < 0.05/10 = p < 0.005). For all significant correlations, we conducted follow-up linear regressions controlling for Age and Sex to evaluate robustness to these commonly used covariates. We then performed whole-brain analyses to assess the robustness and spatial specificity of significant effects, regressing the relevant property onto the whole-brain connectivity map seeded in the a/pHPC ROI. Whole-brain results were examined using contrasts consistent with the direction of the ROI-ROI results (i.e., positive contrast if the ROI-ROI FC correlation was positive), which were thresholded at a voxel-wise level of p < 0.001 (uncorrected), and significance was determined using a cluster-level false discovery rate (FDR) correction at q < 0.05.

To test the association between the HPC circuits and conventional clinical assessments, Pearson correlation analyses were performed between clinician-rated nightmare severity and connectivity within the aHPC-Amyg and pHPC-VC circuits. Additionally, Poisson regressions examined the association between nightmare frequency during the EMA period (i.e., count of number of reported nightmares) and HPC connectivity.

Finally, to examine the potential moderating effect of PTSD diagnosis or total PTSD symptom severity, additional linear regressions were performed for significant property-connectivity associations. Models included either PTSD diagnosis (yes/no = 1/0) or PTSD total symptom severity (CAPS-5 Total Score) entered as an interaction term with the predictor.

Results

Linking nightmare phenomenology to symptom severity

Clinician-rated nightmare symptom severity (CAPS-5 Item B2) was significantly associated with both the visual detail (r = 0.45, p < 0.001) and emotional intensity (r = 0.36, p = 0.003) of nightmares reported during the EMA period. Additional associations were seen with nightmare vividness (r = 0.28, p = 0.025) and reliving (r = 0.27, p = 0.028), though these did not survive Bonferroni correction for multiple comparisons. There was no significant association between nightmare severity and fragmentation (r = −0.12, p = 0.357).

In the multiple linear regression entering all five nightmare properties as simultaneous predictors, only visual detail was a significant predictor of clinician-rated nightmare symptom severity (β = 0.48, p = 0.020), indicating its specific relevance beyond shared variance across properties.

Linking nightmare phenomenology to hippocampal connectivity

Functional connectivity within the right aHPC-Amyg circuit showed marginal associations with visual detail (r = −0.24, p = 0.065) and fragmentation (r = 0.26, p = 0.041), which did not survive correction for multiple comparisons (Fig. 1A).

Fig. 1. Associations between nightmare properties and hippocampal connectivity.

A Pearson correlation coefficients between nightmare properties and resting-state functional connectivity within two a priori circuits: posterior hippocampus–visual cortex (pHPC–VC) and anterior hippocampus–amygdala (aHPC–Amyg). Significant correlations surviving corrections for multiple comparisons were observed between right pHPC-VC connectivity and the visual detail and emotional intensity of trauma nightmares. B Scatter plots and linear fits illustrating significant associations that survived correction for multiple comparisons. *p < 0.05, **p < 0.01, ***p < 0.0005.

In contrast, functional connectivity in the right pHPC-VC circuit was significantly associated with both visual detail (r = 0.46, p = 0.0001) and emotional intensity (r = 0.43, p = 0.0003; Fig. 1A, B). These associations remained significant after controlling for Age and Sex (visual detail: β = 0.50, p = 0.00001; emotional intensity: β = 0.42, p < 0.0003). Additionally, there were associations between right pHPC-VC and vividness (r = 0.33, p = 0.006) and reliving (r = 0.26, p = 0.032) which did not survive correction for multiple comparisons.

There were no effects with the left a/pHPC (p’s > 0.057; Fig. 1A), further highlighting the potentially unique role of the right pHPC in intrusive re-experiencing symptoms.

Regressing visual detail ratings onto whole-brain pHPC-seeded functional connectivity maps revealed a large, contiguous cluster spanning the bilateral parietooccipital sulcus and dorsal cuneus (peak = −14, −74, 28; t = 4.90; k = 360 voxels, cluster qFDR < 0.001; Fig. 2A). Regressing emotional intensity of nightmares onto the same pHPC functional connectivity map revealed four clusters within the visual cortex (Fig. 2B): the ventral visual stream (peak = −12, −54, −2; t = 5.05, k = 677, qFDR < 0.001), left dorsal cuneus (peak = −8, −84, 30; t = 4.46, k = 268, qFDR < 0.001), right dorsal cuneus (peak = 14, −82, 32; t = 4.30, k = 66, qFDR = 0.003), and lingual gyrus (peak = −4, −88, −10; t = 4.05, k = 66, qFDR = 0.020).

Fig. 2. Spatial specificity of posterior hippocampus (pHPC) functional connectivity associations with nightmare properties.

Regressing nightmare properties onto whole-brain functional connectivity maps seeded in the pHPC revealed associations with regions of the visual cortex. (A) Greater visual detail of nightmares was associated with stronger pHPC connectivity to bilateral parieto-occipital sulcus, cuneus, and lingual gyrus (MNI peak coordinate = −14, −74, 28); (B) Greater emotional intensity of nightmares was associated with stronger pHPC connectivity to overlapping visual areas and an additional cluster in retrosplenial cortex (MNI peak = −12, −54, −2). Results are thresholded at voxel-wise p < 0.001, with a minimum cluster extent of k = 60 voxels for visualization.

No association between hippocampal connectivity and nightmare severity or frequency

There were no significant associations between clinician-rated nightmare severity and right pHPC-VC (r = 0.21, p = 0.083), right aHPC-Amyg (r = 0.04, p = 0.762), left pHPC-VC (r = 0.14, p = 0.272), or left aHPC-Amyg (r = −0.16, p = 0.202) connectivity. Additionally, HPC connectivities were not significantly correlated with the total number of nightmares experienced during the EMA period (p’s > 0.140).

No moderating effect of PTSD diagnosis or symptom severity

PTSD diagnosis did not significantly moderate the association between right pHPC-VC connectivity and visual detail (p = 0.243) or emotional intensity (p = 0.117) of nightmares. Likewise, total PTSD symptom severity did not moderate these associations (visual detail: p = 0.536; emotional intensity: p = 0.899).

Discussion

Like other PTSD intrusion symptoms, trauma-related nightmares are marked by intense sensory-perceptual and affective properties that often contribute to a vivid sense of reliving the traumatic event and to clinically-significant distress [39]. These properties may reflect neurobiological correlates akin to other reexperiencing symptoms (e.g., intrusive traumatic memories) and idiopathic nightmares [22, 26, 40–44], such as functioning within aHPC-Amyg and pHPC-VC circuits. Building on this literature, we hypothesized that functional connectivity of the aHPC-Amyg circuit would be correlated with the emotional intensity of trauma-related nightmares, while pHPC-VC connectivity would relate to their visual detail. Partially supporting these hypotheses, we found that greater emotional intensity and visual detail were both associated with stronger pHPC-VC functional connectivity. In contrast, aHPC-Amyg connectivity was not correlated with any nightmare properties.

Our findings align with prior studies linking functional connectivity between the pHPC and VC to episodic memory functioning [26, 45–47], including trauma-related intrusive memories [20]. Across these studies, a pHPC-VC circuit has been linked to the visual detail and vivid reliving of memories. The present finding that stronger pHPC-VC functional connectivity was related to greater visual detail of nightmares extends prior evidence to memory-related imagery generated during sleep. Whole-brain analyses further demonstrated the spatial specificity of this finding, such that associations between greater visual detail and pHPC functional connectivity were constrained to the visual cortices. Interestingly, clusters emerged both within the primary VC as well as regions along the ventral visual stream involved in object recognition. This builds upon prior literature implicating interactions between the pHPC and cortical networks supporting visuospatial processing in memory-related processes [48].

The neurobiology of memory-related dreaming more broadly has potential ties to the noradrenergic system. Dreaming is thought to occur predominantly, although not exclusively, during REM sleep [49, 50]. Achieving REM sleep requires the effective cycling through non-REM (NREM) sleep stages. This cycling is regulated in part by the dynamic switching of activity within the locus coeruleus (LC), the primary hub of the noradrenergic system, such that LC activity decreases during the transition from NREM to REM and is subsequently silenced during REM sleep [51–53]. However, studies in rodent models and humans with PTSD suggest a maladaptive persistence of LC activity during REM sleep, resulting in disrupted regulation of NREM-REM cycles and impairments in associated neurocognitive processes [54, 55]. In humans, PTSD is associated with noradrenergic dysregulation, particularly in relation to hypervigilance and reexperiencing symptoms such as nightmares [56, 57]. Prazosin, an alpha-1 adrenergic receptor antagonist, is often used in the treatment of trauma-related nightmares [58, 59], albeit with mixed efficacy [60, 61]. Relatedly, the noradrenergic system is tied to memory consolidation and reactivation by facilitating sleep-dependent memory consolidation [62, 63] and enhancing the expression of prioritized memory representations in cortical regions [64]. In addition to regulating sleep cycles, activation of LC-limbic circuitry is believed to drive a reactivation of sensory-perceptual cortices, such as the visual cortex, involved in the encoding of salient memories [65–67], resulting in an arousal-mediated reactivation of sensory memory traces [68]. Interestingly, recent evidence has demonstrated that dreaming is associated with altered activity within the occipitoparietal cortex [69], overlapping with our visual cortex clusters and regions previously tied to pHPC co-activation and trauma memory features [20]. Therefore, it is possible that noradrenergic dysregulation may play a mechanistic role in the identified pHPC-VC circuit through driving reactivation of cortical networks tied to the trauma memory representation, even during sleep. Future work should examine the influence of LC activity on the pHPC-VC circuit and trauma-related nightmare properties.

Unexpectedly, pHPC-VC connectivity was also associated with the emotional intensity of trauma nightmares. Given the strong correlation between visual detail and emotional intensity (r = 0.58), this finding may reflect the interconnected nature of these nightmare components. Specifically, the emotional intensity of nightmares may arise partly from the intensity of its visual imagery, rather than solely from the recall of emotions from the original trauma. This interpretation is consistent with theoretical models suggesting an active role of visual detail in the emotional intensity of nightmares [44, 70, 71]. Consistent with this notion, visual detail emerged as an independent and specific predictor of clinician-rated nightmare symptom severity, above and beyond emotional intensity. At the neural level, however, whole-brain analyses revealed very similar connectivity patterns for visual detail and emotional intensity, reinforcing the possibility of shared underlying neural substrates. This interpretation is consistent with accumulating evidence that sensory cortices, including the VC, play an active role in threat processing [72–75], contributing to the early evaluation and long-term storage of threat information, rather than serving merely as passive relays of sensory information. Further research into this circuit, as well as broader cortical networks, may illuminate the differences and similarities between visual and emotional dimensions, and the degree to which they reflect distinct versus overlapping reexperiencing processes.

While prior research has implicated the aHPC-Amyg circuit in the emotional quality of episodic memory [25], we did not observe associations between aHPC-Amyg connectivity and any nightmare properties. This could indicate that connectivity within this circuit is more relevant to the emotional processing of diurnal memories, and that the emotional quality of traumatic nightmares is supported by alternative circuits, such as systems that also contribute to visual detail and imagery. Alternatively, the aHPC and Amyg may contribute to trauma nightmares through connectivity with broader cortical networks rather than through direct aHPC-Amyg coupling. For example, prior work on episodic memory suggests the HPC interacts with multiple regions, such as the orbitofrontal cortex, temporal pole, and components of the language, somatomotor, and default mode networks, forming a complex neural architecture of episodic memory that extends beyond a simple HPC-Amyg circuit [76]. In addition, the Amyg is structurally connected to the ventral visual stream, a pathway that is highly implicated in trauma-related symptoms [77] and that may provide a route for emotional amplification of visual imagery during nightmares.

Although emotional intensity and visual detail correlated with clinician-rated nightmare symptom severity, aHPC-Amyg and pHPC-VC connectivity did not. The HPC-Amyg circuit has been extensively implicated in PTSD [27, 40, 78, 79] and idiopathic nightmares [17, 80, 81]. Prior investigations have found patterns of Amyg and HPC activity that are associated with traumatic nightmare severity [17, 82, 83], but these studies did not examine nightmare phenomenology. Conversely, connectivity patterns of the a/pHPC with cortical networks have been associated with the properties of intrusive memories, but not with clinician-rated reexperiencing symptom severity [20]. This dissociation suggests that pHPC-VC connectivity indexes specific phenomenological properties of nightmares rather than overall clinician-rated symptom severity. Notably, symptom severity scores from clinician-administered interviews often aggregate multiple factors such as frequency and duration. Therefore, the pHPC-VC circuit may represent a low-level, mechanistically-specific target to more directly address distinct features of trauma-related nightmares that may be more malleable to intervention. Whether modifying these nightmares features, such as their visual details, subsequently leads to reductions in overall PTSD symptom severity remains to be tested.

The notion that trauma-related nightmares and intrusive memories share sensory-perceptual mechanisms is supported by their converging treatment approaches. Exposure-based interventions, such as prolonged exposure and written exposure therapies, involve repeated and detailed retrieval of the sensory-perceptual features of trauma memories to promote extinction in a safe context [84, 85]. Similarly, Imagery Rehearsal Therapy is an empirically-supported intervention for trauma-related and idiopathic nightmares that alleviates their distressing qualities by targeting and rehearsing mental imagery [86]. Patients learn to rescript distressing dream content into alternative, non-distressing narrative and imagery, and to then repeatedly rehearse the modified dream narrative and imagery during waking hours [87]. These treatments, involving immersion in mental imagery, may potentially recruit the identified pHPC-VC circuit.

By contrast, cognitive-behavioral interventions that target cognitive restructuring or the reduction of behavioral avoidance may engage sensory-perceptual processes underlying nightmares less directly [88]. While effective for other PTSD symptoms through their engagement of cognitive-affective processes, such treatments may not adequately recruit the sensory-perceptual processes that drive nightmares. This could help explain why such treatments are often less effective for nightmares and result in residual sleep disturbances [89, 90]. Alternatively, non-cognitive approaches to treat trauma-related sleep disturbances such as the use of morning blue light therapy have demonstrated efficacy in addressing sleep-related symptoms in PTSD, including nightmare severity [91]. By stabilizing circadian rhythmicity and improving sleep duration, these therapies may indirectly engage neurocognitive processes that are linked to intrusive reexperiencing symptoms, such as fear extinction retention [92], and have been associated with changes in limbic-cortical connectivity [93]. Whether they affect intrusive reexperiencing symptoms and modulate the identified pHPC-VC circuit specifically remains unknown. More broadly, emerging treatments for nightmares in PTSD are expanding, with diverse therapeutic mechanisms of action [94]. If the pHPC-VC circuit is a critical driver of the sensory-emotional vividness of nightmares, then treatments modulating this system may yield better outcomes for nightmare reduction. Evaluating this circuit as a potential therapeutic target will require studies examining the extent to which trauma- or nightmare-focused psychotherapies affect nightmare features and the associated pHPC-VC connectivity. We did not assess for current or past therapeutic interventions in the present study and therefore cannot determine whether individuals engaged in targeted psychotherapies differ in their nightmare properties or pHPC-VC connectivity. This may be especially relevant for nightmare-focused psychotherapy, given its emphasis on imagery. Future studies examining this circuit should therefore attend to the potential influence of such therapeutic interventions, which may provide further insight into the utility of this circuit as a therapeutic target.

While our study offers novel insights into the neurobiological correlates of trauma-related nightmares, several limitations should be acknowledged. First, the analyzed sample was a subset from a larger study that focused on intrusive trauma memories and did not specifically recruit individuals with frequent trauma-related nightmares. To sufficiently capture nightmare phenomenology, we excluded participants who reported fewer than two nightmares over the two-week EMA period, yielding a sample of 66 participants that may have been under-powered to detect more subtle effects. Notably, we were powered to detect correlations of r = 0.33 or greater [95], an effect size comparable to the median reported for brain-symptom associations [96]. Second, participants were predominantly white, non-Hispanic, and female, limiting generalizability to other demographic groups. Future research in more diverse, representative samples is needed to determine whether findings replicate across biological sex, gender identity, race, and ethnicity, and if these findings are influenced by cumulative life stressors. Third, the present study examined trauma-related nightmares specifically and did not assess idiopathic nightmares; therefore, we cannot determine if the identified pHPC-VC connectivity pattern is specific to trauma-related nightmares or reflects a more general neural mechanism underlying nightmares broadly. Similarly, we did not acquire reports of nightmares preceding participants' index traumas, precluding any statements about whether the nightmare features and associated pHPC-VC connectivity are a consequence of trauma exposure or a pre-existing risk factor. Future studies should therefore incorporate assessments of non-trauma related nightmares, either within- or between-subjects, to determine the specificity of these findings to post-traumatic nightmares. Finally, the use of resting-state fMRI data precludes causal inference about neural activity occurring during nightmares. Task-based fMRI studies incorporating memory recall paradigms, or non-invasive neuroimaging methods conducted during sleep, like polysomnography, would be necessary to test this research question. Relatedly, our study utilized an eyes-open resting-state paradigm which, while considered the standard for examining intrinsic functional connectivity of neural networks, may reflect a brain state less analogous to sleep than eyes-closed rest. In addition to task-based studies, future work examining the state-dependence of the identified circuits and their link to nightmare phenomenology and severity is needed.

Overall, this study offers novel insights into the neurobiological correlates underlying the phenomenological properties of trauma-related nightmares. Intrinsic functional connectivity within the pHPC-VC circuit emerged as a unique correlate of both visual detail and emotional intensity, implicating it as a potential neuropathophysiological mechanism of the intense sensory-emotional experience of trauma nightmares. Together with prior evidence implicating pHPC-VC functional connectivity in the vivid details of episodic memory and the reliving of trauma-related intrusive memories, this adds to accruing evidence for a potential role of exaggerated connectivity between the pHPC and VC in trauma reexperiencing symptoms. Given that the visual features of trauma-related nightmares were independently and specifically associated with clinical severity of the symptom, future mechanistic and therapeutic studies should test whether interventions that target this pHPC-VC pathway can mitigate intrusive reexperiencing symptoms by reducing their sensory detail.

Competing interests

The authors declare no competing interest.