fMRI, Threat, and the Amygdala: Fear Learning, Extinction, and Zone 3

With the PDF provided, we now have enough material for a more complete BrainLatam2026 reading. One important technical adjustment: this article is not primarily a functional fMRI study. It uses structural MRI for neuronavigation, TUS — transcranial ultrasound stimulation, SCR — skin conductance response, and computational modeling. fMRI enters better as a future bridge to map brain networks involved in threat, extinction, and Zone 3.

We usually think of fear as just an emotion. But the study by Meijer and colleagues shows something deeper: fear is also a learning state. When the body learns threat, it learns fast. And often, it forgets slowly.

The scientific question of the article is direct and very important: is the human amygdala causally necessary for rapidly learning a conditioned threat and forming memories that are difficult to extinguish? This is an old question in neuroscience, but strong causal evidence in healthy humans was still missing.

The study deserves praise because it did not remain at the level of correlation. Instead of only observing the amygdala with neuroimaging and inferring function, the researchers used transcranial ultrasound stimulation — TUS, a noninvasive technique capable of modulating deep brain structures with spatial and temporal precision. This allowed them to causally test the role of the amygdala during threat learning.

The experimental design was very elegant. In one experiment, the authors applied bilateral TUS to the amygdala during Pavlovian threat conditioning. In another experiment, they applied TUS to the posterior hippocampus, used as a control target. Each experiment included 25 healthy participants in the final analysis. In this way, the study compared amygdala-TUS, hippocampus-TUS, and sham conditions.

The task was a classic threat-conditioning paradigm. Participants viewed images of snakes. One snake was the threat stimulus, CS+, paired with a mild shock in 50% of the trials. Another snake was the safety stimulus, CS−, never paired with shock. The main physiological response was the skin conductance response — SCR, used as a trial-by-trial index of conditioned threat responding.

Figure 1 of the article clearly shows the logic of the experiment: threat, safety, shock, TUS/sham, and autonomic response. First comes threat acquisition. Then retention, extinction, recovery after reinstatement with unexpected shocks, and re-extinction. In other words, the study does not measure only “fear”; it measures how the body learns, maintains, extinguishes, and recovers threat.

Stimulation was applied during acquisition. Each trial began with TUS 200 ms before the snake image and continued for 1000 ms, covering the full association window between the visual cue and possible shock. This is essential: the amygdala was modulated exactly when the body needed to learn whether that snake meant danger.

The study used individual structural MRI to guide stimulation. Imaging was performed on a 3T Magnetom Skyra scanner, Siemens, with a 32-channel head coil. T1, T2, and PETRA/UTE scans were acquired for anatomical segmentation, skull reconstruction, acoustic simulations, and neuronavigation. Figure 2 shows the basolateral amygdala and hippocampal targets, as well as acoustic simulations and the bilateral TUS protocol.

TUS was delivered with the NeuroFUS Pro system from Sonic Concepts Inc., with supplier/support from Brainbox Ltd. The article reports two four-element piezoelectric ultrasound transducers, models CTX-250-001 and CTX-250-026, driven by TPO-203-035 and TPO-105-010 modules. The fundamental frequency was 250 kHz, with a 64 mm aperture, 90 ms pulses, 30 ms ramp, 5 Hz repetition, 1000 ms total duration, and 61.5 mm focal depth. This level of technical detail is very important for BrainLatam/Brain Support because it shows an emerging frontier of noninvasive deep-brain neuromodulation.

The central result was strong: amygdala-TUS slowed early threat learning. In other words, when amygdala function was modulated during acquisition, participants took longer to differentiate threat from safety in the first trials. This effect did not appear in the same way under hippocampus-TUS.

This confirms the idea that the human amygdala participates in a critical temporal window: the beginning of threat learning. This is when the body tries to rapidly discover: “Is this danger or not?” When the amygdala is modulated, this fast learning loses strength.

But the most beautiful finding comes afterward. Memories formed under amygdala-TUS extinguished faster. In other words, when threat was learned with less amygdala participation, it became less resistant to extinction. The body returned to safety-cue levels a few trials earlier compared with sham.

The computational model summarized this brilliantly: the amygdala creates an emotional learning state of learn fast, forget slow. When the amygdala was modulated, this emotional bias was reduced: threat was learned more slowly and forgotten more quickly.

In BrainLatam2026 language, this speaks directly to Zone 3. Zone 3 can be understood as a state in which the body becomes captured by threat, rigidity, defense, hypervigilance, or aversive memory. The person is not merely “thinking wrong.” The body has learned danger quickly and has started to forget safety slowly.

We can say that Zone 3 works like a Stone-like functional connectome. Not in the sense of a fixed anatomical connectome, but as a rigid bodily and neural state in which the organism reduces openness to the new and prioritizes fast survival responses: attack, flee, freeze, or replicate already learned patterns. It is a form of defensive “thinking fast,” close to the idea popularized by Daniel Kahneman about fast, automatic, and less reflective responses.

In this state, criticality and creativity are reduced because the body is occupied with surviving, predicting threat, and avoiding error. The new stops being possibility and becomes risk. For BrainLatam2026, the difference is that this “thinking fast” is not only cognitive; it is interoceptive, proprioceptive, autonomic, and territorial.

The study by Meijer and colleagues helps give materiality to this reading: the amygdala participates in an emotional learning state that favors learning threat fast and forgetting threat slowly. This is very close to what we call Zone 3: a body organized for defense, with less flexibility to update safety.

The way out of Zone 3 is not simply to “think better.” It is to recover bodily conditions so the nervous system can update safety again. In Zone 2, the body gains more space for Fruition, Metacognition, criticality, and creativity. Threat stops hijacking the whole field of action, and the person can learn the new without reducing the world to danger, enemy, or repetition.

Here, the Damasian Mind becomes central: interoception and proprioception form the bodily field where threat becomes real. The shock, the snake image, expectation, sweating, autonomic acceleration, and defensive readiness are not separate events. They form a bodily state of threat.

SCR is very important in this sense. It does not measure an opinion about fear; it measures an autonomic skin response. This brings the study closer to the materiality of the body. For BrainLatam2026, threat is not only narrative. It appears in skin conductance, associative memory, value updating, and the difficulty of returning to safety.

The avatar-lens for this blog can be Tekoha with DANA. Tekoha perceives the internal territory of threat: the body that learns danger. DANA reminds us that a healthy system needs to create rituals, policies, and technologies that allow the body to return to regulation without transforming fear into an existential prison.

We can also use APUS: threat is never only internal. The snake, the shock, the experimental environment, the sound, the body, and expectation form a threatening body-territory. When this APUS becomes captured, the person enters Zone 3: they begin to react to the world as if danger were still present.

The BrainLatam2026 question would be: when does a threat memory become Zone 3, and what prevents the body from updating safety? The article suggests one answer: the amygdala helps form threat memories that resist extinction. But the BrainLatam2026 question expands this: which social, educational, family, religious, digital, and economic states keep the body learning threat fast and forgetting safety slowly?

A future experimental design could combine TUS + fMRI + EEG + fNIRS + HRV/RMSSD + respiration + GSR/SCR + EMG. TUS modulating the amygdala would allow causal testing. fMRI would show deep and cortical networks: amygdala, vmPFC, hippocampus, insula, striatum, and cingulate cortex. EEG would show the rapid temporality of prediction and error. fNIRS could follow prefrontal cortex in more ecological tasks. HRV, respiration, GSR, and EMG would show the whole body entering or leaving threat.

This type of study would be excellent for investigating Zone 2 versus Zone 3. In Zone 2, the body can update: “this was threat, but now it is safety.” In Zone 3, the body continues responding as if threat persists. Extinction is not simply erasing fear; it is recovering bodily flexibility.

The generous decolonial critique is that neuroscience often studies threat in highly controlled environments, with snakes, shocks, and screens. This is necessary for experimental rigor. But in real life, threat also appears as poverty, racism, urban violence, abuse of power, school humiliation, food insecurity, fear-based algorithms, and politics organized around enemies. The body learns threat in the territory.

The bridge with DREX Cidadão appears strongly here. If the body learns threat fast and forgets slowly, an unjust society produces collective Zone 3. People living without economic security, belonging, and predictability tend to have less bodily space for Fruition and Metacognition. DREX Cidadão, as a policy of social metabolism, can be thought of as an attempt to reduce basal threat and create conditions for the body to return to Zone 2.

This study also shows something ethical: it is not enough to ask someone to “overcome fear.” Threat memory can be sustained by deep circuits and persistent autonomic states. Extinction requires time, safety, repetition, context, and perhaps, in the future, neuromodulation technologies used with great care.

Although the article is not primarily a functional fMRI study, it opens a clear avenue for fMRI. The next question would be: how does TUS modulation of the amygdala alter connectivity between the amygdala, vmPFC, hippocampus, and insula during acquisition and extinction? This could show how the brain leaves Zone 3 and recovers the ability to update safety.

Closing
This study shows that the human amygdala is not merely a “fear center.” It participates in an emotional learning state: learning threat fast and forgetting threat slowly. For BrainLatam2026, this helps us think of Zone 3 as a bodily memory of threat that resists updating. The Stone-like functional connectome is this body that attacks, flees, freezes, or replicates, with less criticality and less creativity to learn the new. The way out is not to deny fear, but to create body, territory, public policy, and technology so that safety can be relearned. When the body learns safety again, life can leave defense and return to Fruition and Metacognition.

References

Meijer, S., Carpino, E., Kop, B. R., Lam, J., de Voogd, L. D., Roelofs, K., & Verhagen, L. (2026). The human amygdala in threat learning and extinction. Science Advances, 12, eaea8233. doi:10.1126/sciadv.aea8233.

Kahneman, D. (2011). Thinking, Fast and Slow. Farrar, Straus and Giroux.

Damasio, A. R. (1999). The Feeling of What Happens: Body and Emotion in the Making of Consciousness. Harcourt.

LeDoux, J. E. (1996). The Emotional Brain: The Mysterious Underpinnings of Emotional Life. Simon & Schuster.