PTSD is a chronic psychiatric condition that can develop after traumatic experience, marked by intrusive re-experiencing, hyperarousal and enduring avoidance. Despite decades of clinical work, the underlying neurobiology remains incompletely understood, and existing interventions often deliver limited relief. Growing evidence points to the extended amygdala, and particularly the BNST, as central to the transition from acute, short-lived fear to the prolonged, sustained threat responses that define the disorder. Unlike the amygdala proper, which specializes in phasic fear, the BNST is preferentially recruited by uncertain, sustained and anticipatory threat, making it an especially relevant target for studying trauma that lingers.

The ovBNST is a small, discrete compartment within the anterolateral BNST complex, and it receives a strong and anatomically defined input from the posterior basolateral amygdala (pBLA), a region that integrates socially and emotionally salient contextual information. In the new work, the researchers combined viral tracing with RNAscope single-molecule in situ hybridization to characterize this pathway. They found that roughly 77 percent of pBLA neurons projecting to the ovBNST express the gene encoding the cannabinoid type-1 receptor (CB1R), the presynaptic receptor through which endocannabinoids exert their inhibitory effects. Moreover, about 92 percent of these projecting neurons were glutamatergic, meaning the pathway delivers excitatory signals that endocannabinoids are positioned to dial down.

To model PTSD, the team developed a modified five-day chronic social defeat stress (CSDS) paradigm in which mice are repeatedly exposed to aggressive CD-1 residents. Stressed mice subsequently avoided unfamiliar mice, spent most of their time huddled in corners away from social targets, showed reduced exploration of open and central spaces in anxiety tests, and displayed anhedonia and despair-like behaviors in sucrose preference and forced swim assays. Notably, the shortened five-day protocol induced social avoidance more efficiently than the conventional ten-day version, and it produced comparable behavioral changes in female mice, strengthening the model’s relevance.

In vivo fiber photometry revealed that ovBNST neurons are robustly activated by both acute defeat and chronic stress, and that stressed animals show exaggerated ovBNST responses upon re-encountering an aggressor. Whole-cell electrophysiology in brain slices uncovered the synaptic correlate: miniature excitatory postsynaptic current frequency was elevated after CSDS while amplitude was unchanged, indicating increased presynaptic glutamate release. Critically, blocking CB1Rs with the antagonist AM251 increased excitatory drive in both stressed and control neurons, but the effect was markedly smaller in stressed tissue, showing that the endogenous cannabinoid-mediated restraint on excitatory transmission had been substantially weakened by chronic stress.

At the molecular level, chronic stress reduced Cnr1 mRNA in the pBLA, cutting the proportion of positive neurons from 9.1 to 4.2 percent, and lowered CB1R protein in the ovBNST by about 28 percent. Using a genetically encoded fluorescent sensor called eCB2.0, the researchers directly visualized endocannabinoid release from pBLA terminals in the ovBNST in real time. Acute social defeat triggered a sustained endocannabinoid surge lasting more than fifty seconds, and after chronic stress, subsequent social re-exposure evoked an even larger release. Together, these data suggest the system tries to compensate: postsynaptic neurons pump out more endocannabinoid, but the presynaptic receptors that should receive the signal are disappearing.

The causal tests were striking. Infusing JZL195, which blocks the enzymes that degrade both major endocannabinoids, directly into the ovBNST before behavioral testing normalized social approach, restored anxiety-like measures, and rescued the shortened duration of individual social interactions in stressed mice. A more selective inhibitor, JZL184, which targets MAGL and thereby preserves the abundant endocannabinoid 2-arachidonoylglycerol (2-AG), produced the same benefits. Strikingly, systemic administration of JZL184, which acts throughout the brain, gave minimal improvement, underscoring that the therapeutic effect depends on acting within the right circuit rather than globally amplifying cannabinoid signaling.

The team then went further with projection-specific tools. Using optoCB1R, a light-activated version of the cannabinoid receptor expressed selectively in pBLA neurons, they showed that shining light onto pBLA terminals in the ovBNST during trauma re-exposure pushed stressed mice back to control-level performance, with social interaction ratios exceeding 1.2 and doubled social engagement. Conversely, using a Cre-dependent CRISPR-Cas9 strategy to knock down Dagla, the gene encoding the 2-AG synthesis enzyme DAGLα, only in ovBNST cells innervated by pBLA inputs made mice more vulnerable: under a truncated three-day stress regimen that normally causes no symptoms, DAGLα-depleted mice developed social avoidance and anxiety-like behavior. The knockdown reduced Dagla mRNA to 22 percent of control levels and had no behavioral effect in unstressed animals, confirming the interaction with stress exposure.

The mechanistic picture that emerges is one of a negative feedback loop turned pathological. During acute social threat, postsynaptic ovBNST neurons release 2-AG, which travels backward across the synapse to activate presynaptic CB1Rs on glutamatergic pBLA terminals, dampening further excitation and preventing overgeneralization of threat. With chronic stress, sustained endocannabinoid demand and receptor downregulation dismantle this brake, allowing recurrent amygdala excitation to drive pathological ovBNST hyperactivity, which manifests behaviorally as persistent avoidance and anxiety. This explains why human PET studies of CB1R availability in PTSD have yielded seemingly contradictory results, since receptor changes are region- and symptom-dependent rather than uniform across the brain.

Therapeutically, the work argues for circuit-targeted rather than systemic cannabinoid modulation. Broad CB1R activation carries known risks of cognitive and motor impairment and receptor desensitization, whereas restoring signaling only where it has failed could minimize off-target effects. The authors also caution that important questions remain: the mouse model captures behavioral dimensions rather than the full diagnostic syndrome, the eCB2.0 sensor cannot distinguish 2-AG from anandamide, and PTSD prevalence, BNST organization and endocannabinoid function all differ by sex, while the full circuit-level analysis was performed mainly in males. Still, by identifying a discrete synapse where the brain’s native cannabis-like chemistry gates the persistence of trauma-like states, the study offers a concrete, testable target for the next generation of PTSD treatments.

Subject of Research: Endocannabinoid signaling in the oval bed nucleus of the stria terminalis and PTSD-like behaviors in mice

Article Title: The oval bed nucleus of the stria terminalis endocannabinoid signaling constrains post-traumatic stress disorder-like behaviors

Article References: Zuo, H., Li, J., Li, J., Zhang, X., & Zhang, B. (2026). The oval bed nucleus of the stria terminalis endocannabinoid signaling constrains post-traumatic stress disorder-like behaviors. Journal of Advanced Research. https://doi.org/10.1016/j.jare.2026.10.015

Image Credits: AI Generated

DOI: 10.1016/j.jare.2026.10.015

Keywords: PTSD, endocannabinoid, ovBNST, bed nucleus of the stria terminalis, CB1 receptor, 2-AG, social defeat stress, pBLA-ovBNST circuit, anxiety, social avoidance, fiber photometry, optogenetics

The research team, led by Yihang Du of Capital University of Physical Education and Sports together with colleagues at Wuhan Sports University and WLSA Shanghai School, measured each construct with validated instruments. Physical exercise was assessed with the Physical Activity Rating Scale-3, regulatory emotional self-efficacy with the Regulatory Emotional Self-Efficacy Scale, social avoidance with the corresponding dimension of the Social Avoidance and Distress Scale, and social anxiety with the Social Anxiety Scale for Adolescents. All participants were junior high students, and the study received ethics approval from the Academic Committee of Capital University of Physical Education and Sports, with informed consent obtained from both the students and their guardians.

The correlational results were striking in their consistency. Social anxiety was negatively correlated with physical exercise, with a correlation coefficient of r = -0.320, and with regulatory emotional self-efficacy, at r = -0.330, both statistically significant at p < 0.001. In other words, students who exercised more and who felt more capable of regulating their emotions tended to report lower social anxiety. The strongest single correlate, however, was social avoidance: the correlation between social avoidance and social anxiety was r = 0.434, also highly significant. Students who actively steered away from social contact were markedly more likely to report the fear of scrutiny, embarrassment and negative judgment that defines social anxiety.

Correlations alone can be misleading, so the researchers went further. Using a two-step hierarchical multiple linear regression with HC3 robust standard errors, a technique that guards against distortion from outliers and heteroscedasticity in survey data, they adjusted for sex, grade and household registration. Even after these adjustments, physical exercise retained a negative association with social anxiety (β = -0.174, p < 0.001), as did regulatory emotional self-efficacy (β = -0.182, p < 0.001), while social avoidance remained positively associated (β = 0.322, p < 0.001). Together, the final model explained 26.6 percent of the variance in social anxiety scores, a substantial share for a single model in psychological survey research and a sign that these three factors capture a meaningful portion of what differentiates anxious from less anxious students.

One of the study’s most technically interesting contributions is its decomposition of regulatory emotional self-efficacy into subcomponents. The construct, rooted in Albert Bandura’s social cognitive theory, refers not to emotional regulation itself but to a person’s belief in their own capacity to regulate emotions. The researchers found that perceived efficacy in managing despondency or distress and perceived efficacy in managing anger or irritation were both negatively associated with social anxiety. Perceived efficacy in expressing positive affect, by contrast, was not independently significant once the other factors were accounted for. This asymmetry suggests that what matters for socially anxious adolescents is not general emotional confidence but specifically the belief that one can contain and manage negative emotional states in the moment they arise.

The team also subjected its findings to robustness checks designed to test whether the results were artifacts of particular statistical choices. When the researchers log-transformed the physical exercise variable to reduce skew, and when they excluded potentially influential observations from the analysis, the directions of the main associations were unchanged. In a supplementary analysis using an alternative outcome measure, social avoidance remained associated with fear of negative evaluation, one of the core cognitive components of social anxiety. These checks do not prove causation, which no cross-sectional design can, but they strengthen confidence that the reported associations are not fragile statistical accidents.

The causal question deserves emphasis. Because the data were collected at a single point in time, the study cannot determine whether low exercise and low emotional self-efficacy contribute to social anxiety, whether socially anxious students withdraw from exercise and lose confidence in their emotional coping, or whether some third factor drives all three. Each direction is plausible. Exercise is known to reduce physiological arousal and improve mood, which could buffer against social fear. Conversely, a student terrified of being watched may quit team sports, gym class and playground activity, eroding both fitness and opportunities for positive social feedback. Similarly, repeated failures to manage distress in social situations could erode regulatory self-efficacy, while strong self-efficacy could make social challenges feel manageable rather than threatening.

What the design does establish is that these factors are independently associated with social anxiety even after controlling for demographic characteristics, and that they are not simply proxies for one another. The fact that physical exercise, emotional self-efficacy and social avoidance each contributed uniquely to the regression model indicates that social anxiety in this population sits at the intersection of behavior, self-belief and social strategy. A student could be physically active yet still avoidant, or emotionally confident yet sedentary, and the model treats each configuration as informative rather than collapsing them into a single risk score.

For schools, the authors argue, the practical message is about differentiated assessment and support. Screening that focuses only on anxiety symptoms may miss students whose primary signal is withdrawal from social settings, or whose anxiety is intertwined with low confidence in handling anger and distress. Conversely, interventions that boost exercise participation or build emotional regulation skills, such as structured physical education programs and self-efficacy-focused counseling, may offer complementary routes to reducing social anxiety, even though the present study cannot confirm that such interventions would work. The findings also suggest that social avoidance itself, not merely the anxiety behind it, is a legitimate target for attention, given that it was the strongest correlate in the dataset.

The study’s setting matters for interpretation. All participants came from junior high schools in a single prefecture in Henan Province, and the measures were self-reported, which introduces the usual possibility of response bias. Cultural context may also shape how social anxiety, exercise and emotional expression manifest in Chinese junior high students compared with adolescents elsewhere. Still, the scale of the sample, the use of validated scales, the robust regression approach and the consistency of the results across sensitivity analyses make this a useful data point in a growing literature linking adolescent mental health to everyday behavior. As the authors conclude, social anxiety among junior high students appears to be associated with multiple behavioral and psychological domains rather than a single factor, and effective support will need to consider exercise participation, confidence in managing difficult emotions and avoidance-related behavior alongside anxiety symptoms themselves.

Subject of Research: Cross-sectional associations of physical exercise, regulatory emotional self-efficacy and social avoidance with social anxiety among junior high school students

Article Title: Social anxiety, physical exercise, regulatory emotional self-efficacy, and social avoidance among junior high school students: a cross-sectional study

Article References: Du, Y., Zhang, K., Wu, W., & Liu, M. (2026). Social anxiety, physical exercise, regulatory emotional self-efficacy, and social avoidance among junior high school students: a cross-sectional study. BMC Psychology. https://doi.org/10.1186/s40359-026-05742-5

Image Credits: AI Generated

DOI: 10.1186/s40359-026-05742-5

Keywords: social anxiety, adolescents, physical exercise, regulatory emotional self-efficacy, social avoidance, junior high school students, cross-sectional study, BMC Psychology, mental health, China, fear of negative evaluation, school psychology