Fear, anxiety, and the extended amygdala- Absence of evidence for strict functional segregation

A harmonized mega-analysis of fMRI data challenges the prevailing view of strict functional segregation within the extended amygdala, revealing that both the central nucleus and bed nucleus of the stria terminalis respond similarly to threat regardless of temporal certainty, while frontocortical regions specifically process uncertain threats.

The Gist

Imagine your brain has a specialized "Threat Detection Team" responsible for keeping you safe. For decades, scientists believed this team was divided into two distinct squads with very specific jobs:

1. The "Sprint Squad" (The Central Amygdala or Ce): This group was thought to handle immediate, certain dangers. Think of a car swerving right in front of you. You know exactly what's happening and when it's happening. This squad triggers a quick, sharp "FIGHT OR FLIGHT" panic response.
2. The "Stalk Squad" (The Bed Nucleus of the Stria Terminalis or BST): This group was thought to handle vague, uncertain dangers. Think of walking through a dark alley where you hear a noise but don't know if it's a cat or a mugger. You don't know when it will happen, only that it might. This squad triggers a slow-burning, nagging "ANXIETY" feeling.

This idea was so popular it became the standard rulebook for how we understand fear and anxiety disorders. It was like a strict traffic law: "If the threat is certain, use Squad A. If the threat is uncertain, use Squad B."

The Big Experiment
A team of researchers, led by Dr. Alexander Shackman, decided to test this rulebook. They didn't just look at one or two people; they gathered data from 295 different adults (a huge group!) and combined all their brain scans into one massive study.

They used a game called the "Maryland Threat Countdown."

• Certain Threat: Participants saw a countdown timer. They knew for sure a mild, unpleasant shock was coming at the exact moment the timer hit zero.
• Uncertain Threat: Participants saw a timer, but it was randomized. They knew a shock was coming, but they had no idea when it would hit. It could be in 5 seconds or 30 seconds.

The Surprise Discovery
When the researchers looked at the brain scans, they found that the "Traffic Law" was wrong.

Instead of two separate squads doing different jobs, they found that both the "Sprint Squad" (Ce) and the "Stalk Squad" (BST) were working overtime on both types of threats.

• When the threat was certain? Both squads were active.
• When the threat was uncertain? Both squads were active.

In fact, the activity levels in these two brain regions were so similar that they were practically indistinguishable. It's as if you checked the security logs of two different security guards, and they both showed up for every single shift, regardless of whether the alarm was a scheduled drill or a random noise.

The Real Difference Was Elsewhere
So, if these two brain regions aren't the ones distinguishing between "fear" and "anxiety," who is?

The study found that the front part of the brain (the "Frontocortical" region) was the one making the distinction. This area lit up much more during the uncertain threats. You can think of this part of the brain as the Worrying Manager. When the threat is vague and unpredictable, the Manager has to work harder, constantly calculating probabilities and trying to figure out "What if?" This extra mental load is what creates that feeling of anxiety.

Why Does This Matter?
This discovery is a big deal for a few reasons:

1. It breaks the old model: We can no longer say, "Fear is in this part of the brain, and anxiety is in that part." They are more like a team working together on the same problem, just with different levels of intensity.
2. It helps treat anxiety: Many current treatments are based on the old idea that fear and anxiety are totally different biological processes. If they are actually processed by the same brain circuits, we might need to rethink how we treat anxiety disorders. Maybe we need to target the whole "Threat Team" rather than trying to fix just one specific "squad."
3. It validates the struggle: The study confirms that uncertainty is a massive stressor. The fact that the "Worrying Manager" in the front of the brain works so hard during uncertain times explains why not knowing when something bad might happen feels so much worse than knowing exactly when it will happen.

The Bottom Line
For a long time, we thought our brains had a "Fear Switch" and an "Anxiety Switch." This study shows that it's actually more like a dimmer switch. The same brain circuits handle both fear and anxiety; the difference lies in how hard the rest of the brain has to work to figure out what's going on. The "Threat Detection Team" doesn't care if the danger is certain or uncertain; they are ready to spring into action either way.

Technical Summary

1. Problem Statement

The distinction between fear (response to immediate, certain threat) and anxiety (response to potential, uncertain threat) is a foundational concept in clinical psychology (DSM-5-TR) and neuroscience (Research Domain Criteria, RDoC). A dominant theoretical framework, the double-dissociation model, posits that these states are mediated by two strictly segregated neural systems within the extended amygdala (EA):

• Central Nucleus of the Amygdala (Ce): Mediates responses to certain (imminent) threat (Fear).
• Bed Nucleus of the Stria Terminalis (BST): Mediates responses to uncertain (distal/probabilistic) threat (Anxiety).

However, recent rodent mechanistic data and conflicting human neuroimaging findings have challenged this strict segregation. The authors aim to adjudicate this debate by determining whether the Ce and BST show distinct functional profiles regarding threat certainty in humans, or if they function as a unified system.

2. Methodology

The study employed a harmonized mega-analysis of functional MRI (fMRI) data to maximize statistical power and anatomical resolution.

• Data Sources: Combined data from two previously published studies ($N=295$ adults) using identical paradigms and scanner parameters:
  1. 220 psychiatrically healthy university students.
  2. 75 tobacco smokers from the community.
  • Demographics: Racially diverse sample (52.2% White, 19.0% African American, 16.6% Asian, etc.), ages 18–40.
• Experimental Paradigm: The Maryland Threat Countdown (MTC) task.
  • Design: 2 (Valence: Threat vs. Safety) $\times$ 2 (Temporal Certainty: Certain vs. Uncertain).
  • Certain-Threat: A 18.75s countdown followed by a guaranteed noxious stimulus (shock, image, audio).
  • Uncertain-Threat: A randomized countdown (8.75–30s) where a noxious stimulus was guaranteed but the timing was unknown.
  • Control Conditions: Certain-Safety and Uncertain-Safety (neutral stimuli).
• Imaging & Processing:
  • Acquired on a Siemens 3T scanner using a multiband EPI sequence.
  • Harmonization: All data were reprocessed using a single, best-practices pipeline (SPM12) to eliminate inter-study variability.
  • Analysis Strategy:
    • Voxelwise: Whole-brain GLMs to identify distributed networks.
    • Region-of-Interest (ROI): Anatomical ROIs for the Ce and BST were used with spatially unsmoothed data to prevent signal blurring and ensure unbiased effect size estimates.
    • Statistical Framework: Used both frequentist statistics (Cohen's $d$) and Bayesian inference (Bayes Factors, $BF_{10}$) to quantify evidence for the null hypothesis (equivalence) versus the alternative (difference). This allowed the authors to test for statistical equivalence rather than just failing to reject the null.

3. Key Contributions

• Mega-Analysis Scale: The study leverages a large sample ($N=295$), providing 80% power to detect small effect sizes ($d \geq 0.16$), which is critical for detecting subtle regional differences in subcortical structures.
• Methodological Rigor: The use of unsmoothed data within anatomical ROIs avoids the spatial smoothing artifacts that often obscure fine-grained subcortical differences in standard fMRI analyses.
• Bayesian Approach: The application of Bayes Factors provides strong evidence for the null hypothesis (i.e., that the regions are functionally equivalent regarding threat certainty), moving beyond the limitations of traditional null-hypothesis significance testing (NHST).
• Direct Adjudication: The study directly tests the "double-dissociation" hypothesis head-to-head in a human sample, rather than relying on indirect inferences or coordinate-based meta-analyses.

4. Results

A. Behavioral and Physiological Validation

• The MTC paradigm successfully elicited subjective distress and objective arousal (skin conductance).
• Uncertain threat elicited significantly higher fear/anxiety ratings and arousal than certain threat (large effect sizes, $BF_{10} > 10^{20}$), confirming the paradigm's validity as a probe for anxiety.

B. Whole-Brain Voxelwise Findings

• Both certain and uncertain threat anticipation recruited a common distributed cortico-subcortical network, including the midcingulate cortex (MCC), anterior insula (AI), dorsolateral prefrontal cortex (dlPFC), periaqueductal gray (PAG), BST, and dorsal amygdala (Ce).
• Direct Comparison: When contrasting Uncertain vs. Certain threat, frontocortical regions (MCC, AI, dlPFC) showed significantly greater activation for uncertain threat.
• Subcortical Regions: The BST, Ce, and PAG showed negligible differences in activation between certain and uncertain threat conditions.

C. ROI Analysis (Ce vs. BST)

• Activation Magnitude: Both the Ce and BST were significantly activated during threat anticipation compared to safety.
• Functional Segregation Test:
  • The interaction between Region (Ce vs. BST) and Threat Certainty (Certain vs. Uncertain) was not significant ($F(1,294)=0.12, p=0.73$).
  • Effect Size: The difference in selectivity was negligible ($d = -0.02$).
  • Bayesian Evidence: The Bayes Factor ($BF_{10} = 0.07$) indicated that the null hypothesis (no difference) was 14.3 times more likely than the alternative hypothesis.
  • Double Dissociation: Only ~50% of participants showed the hypothesized double-dissociation pattern (Ce > BST for certain; BST > Ce for uncertain), which is no better than chance ($Z_{Sign} = 0.12, p=0.91$).
• Main Effect: The BST showed generally higher reactivity to both types of threat compared to the Ce, but neither region discriminated between threat certainty.

5. Significance and Implications

• Refutation of Strict Segregation: The findings provide robust evidence against the popular double-dissociation model that strictly separates fear (Ce) and anxiety (BST) based on threat certainty. Instead, the Ce and BST appear to function as a unified system (Extended Amygdala) that responds to both certain and uncertain threats.
• Reformulation of Models: The authors argue for reformulating conceptual models of fear and anxiety. The distinction between fear and anxiety may not lie in the anatomical segregation of the Ce and BST, but rather in:
  1. Frontocortical modulation: Greater recruitment of prefrontal regions during uncertain threat (potentially reflecting cognitive load or regulatory efforts).
  2. Temporal dynamics: Differences in the phasic vs. sustained nature of activation (though the current "boxcar" model did not detect this, other work suggests similar dynamics).
  3. Individual differences: The BST may be more sensitive to "nature" (heritable traits) while the Ce is more sensitive to "nurture" (early experience), but both respond to the threat itself.
• Clinical Relevance: Understanding that the EA functions as an integrated unit challenges the development of treatments targeting specific subnuclei for specific disorders. It suggests that therapeutic interventions (pharmacological or neuromodulatory) may need to target the broader EA network or the frontocortical regulation of this network to effectively treat anxiety disorders.
• Translational Bridge: The results align with recent rodent mechanistic studies showing that Ce-BST projections are necessary for responses to uncertain threat, bridging the gap between animal models and human neuroimaging.

In conclusion, this mega-analysis demonstrates that the human extended amygdala does not strictly segregate the processing of certain and uncertain threats, urging a shift toward more integrated models of fear and anxiety neurobiology.