Amygdala and Anxiety: Threat Probability, Not Uncertainty, Drives Neural Response

New Findings: Threat Probability, Not Uncertainty, Governs Amygdala Function

A 2026 peer-reviewed study (OpenAlex W7220368713) has demonstrated that the extended amygdala (EA)—including the central nucleus of the amygdala (Ce) and bed nucleus of the stria terminalis (BST)—is more responsive to the probability of a threat than to uncertainty itself. This directly challenges the canonical model of anxiety, which has long posited that these brain regions are primarily activated by uncertainty about threats. Using a computational framework and a racially diverse sample of 104 individuals, the study found that self-reported anxiety, physiological responses, and neural activity all tracked closely with threat probability, not with uncertainty as previously assumed.

Mechanistic Insights: Revisiting the Extended Amygdala

The extended amygdala is a set of interconnected brain regions implicated in emotional processing and defensive behaviors. The canonical model, based on earlier experiments, suggested that the EA generates anxiety in response to uncertain threats—situations where the likelihood of harm is ambiguous. However, the new study reveals a confound in those foundational experiments: threat probability and uncertainty were not adequately separated. By precisely distinguishing these variables, the researchers showed that EA activity, especially in the BST, is most reliably associated with the likelihood of a threat occurring, rather than with how uncertain that threat is.

Functional magnetic resonance imaging (fMRI) data indicated that the BST is particularly sensitive to low-probability threats, while the Ce responds more generally to threat probability. Lesion studies in rhesus monkeys (Macaca mulatta) further demonstrated that the BST causally contributes to defensive behaviors in low-probability threat contexts. This cross-species evidence supports a shift in the theoretical understanding of anxiety's neurobiological underpinnings.

Implications for Clinical Trials and Translational Research

This reframing of anxiety neurobiology has immediate implications for the design and interpretation of clinical trials, especially those evaluating new treatments—including psychedelic-assisted therapies. Most current trials use endpoints and biomarkers based on the assumption that uncertainty drives anxiety-related neural activity. The new evidence suggests that future trials should instead focus on how interventions modulate threat probability processing in the EA.

• Trial Design: Outcome measures may need to shift from uncertainty-based paradigms to those that manipulate threat probability, potentially requiring new behavioral tasks and imaging protocols.
• Biomarker Development: Biomarkers derived from EA activity should be validated specifically for sensitivity to threat probability, not just general threat or uncertainty.
• Translational Consistency: The study's cross-species approach and computational rigor provide a blueprint for harmonizing animal and human research, which is critical for regulatory acceptance and translational success.

For psychedelic research, this insight could guide the selection of patient populations, endpoints, and imaging strategies, ensuring that interventions are evaluated against the most relevant neurobiological targets.

Risks, Limitations, and Remaining Unknowns

While the study presents robust evidence, several limitations remain. The sample size, though diverse, is moderate, and findings may not generalize to all anxiety disorders or demographic groups. The computational framework, while precise, may not capture the full complexity of real-world threat processing, where probability and uncertainty often co-occur. The lesion studies in rhesus monkeys, though compelling, may not fully translate to human anxiety pathology.

Another risk is the premature overhaul of existing clinical trial paradigms before these findings are replicated and extended. Regulatory agencies such as the U.S. Food and Drug Administration (FDA) and the European Medicines Agency (EMA) may require additional validation before accepting new biomarkers or endpoints based on threat probability. Moreover, the focus on probability may overlook the role of uncertainty in other affective states or psychiatric conditions, necessitating a nuanced approach in future research.

Looking Forward: Toward Targeted Anxiety Interventions

The discovery that the extended amygdala is governed by threat probability rather than uncertainty marks a significant paradigm shift in anxiety neuroscience. This insight opens new avenues for the development of targeted interventions, including pharmacological and psychedelic therapies, that modulate specific neural circuits implicated in threat evaluation. For researchers and clinicians, the challenge will be to integrate these findings into experimental design, biomarker validation, and ultimately, patient care.

One non-obvious implication is that interventions previously deemed ineffective in "uncertainty-based" models might show efficacy when re-evaluated with probability-focused paradigms. This could lead to the revival of compounds or approaches previously set aside, provided they are tested with the right mechanistic endpoints. As the field moves forward, coordinated efforts between basic scientists, clinical researchers, and regulatory bodies will be essential to translate these insights into tangible benefits for patients with anxiety disorders.

How we research: This article was written and reviewed by Dr. Alex J. Morgan, PhD (Neuroscience), Psychedelic Research Journal editor, on 2026-10-07. Primary source: OpenAlex W7220368713.