Summary: For the first time, scientists have identified the electrophysiological brain signals associated with extinguishing fear memories in humans. Using implanted electrodes and representational similarity analysis (RSA), researchers found that theta-frequency activity in the amygdala increases when cues previously linked to unpleasant events are relearned as safe. These extinction memories proved to be strongly context-dependent, which helps explain why fear often returns after therapy ends. The findings point toward new directions for treating fear-related conditions such as post-traumatic stress disorder (PTSD) and anxiety.
Extinction learning—the process by which a learned fear response is suppressed through new learning—is central to adaptive behavior and to treatments for anxiety disorders. Animal studies have long suggested that extinction depends on forming new, context-specific memories that inhibit the original fear trace, and that theta (4–12 Hz) oscillations in the amygdala and hippocampus play a key role. Until now, direct evidence for these mechanisms in the human brain was lacking.
Key Facts:
- Amygdala theta activity: During extinction learning, theta oscillations in the amygdala increased when previously aversive cues were presented in a safety context, consistent with a safety signal.
- Context dependence: Extinction memories were highly tied to the learning context, which helps explain the common clinical phenomenon of fear returning outside the therapeutic setting.
- Human validation: This study provides direct intracranial evidence in humans that confirms mechanisms observed in animal models of fear learning and extinction.
Source: UAB
Background: Suppressing fear-related memories after unpleasant experiences is crucial for healthy functioning. When suppression fails or is incomplete, maladaptive responses can contribute to anxiety, depression, and other psychiatric problems. Contemporary theories propose that extinction does not erase the original fear memory but instead creates a new, highly context-specific trace that inhibits the fear response when retrieved in the same context.
Electrophysiological studies in rodents support this view and link theta oscillations in the amygdala and hippocampus to both fear acquisition and extinction. To test whether similar neural dynamics occur in humans, researchers at the Universitat Autònoma de Barcelona (UAB) and Ruhr-Universität Bochum (RUB) combined intracranial EEG recordings with representational similarity analysis, a method that characterizes how brain regions encode specific cues and contexts.
The study recruited 49 patients with epilepsy who already had depth electrodes implanted for clinical reasons. Those electrodes provided direct access to deep structures involved in emotional memory. Participants viewed a series of neutral everyday images—a hair dryer, a fan and a toaster—while some of those images were paired with an unpleasant sound during the acquisition phase. In a subsequent extinction phase, the images were presented again without the aversive sound so that extinction learning could take place. Brain activity was recorded throughout these phases, enabling precise measurement of oscillatory dynamics and representational patterns associated with learning and extinction.
Using representational similarity analysis, the researchers examined how neural patterns for specific cues and contexts changed across acquisition, extinction and later testing. They found that items previously paired with aversive sounds showed greater representational similarity, consistent with a generalized signature for unpleasant memories. Importantly, during extinction learning the amygdala exhibited increased theta oscillations when previously punished stimuli were re-presented in the extinction context, indicating that amygdala theta activity signals safety rather than threat.
The study also demonstrated that extinction memory traces are stable yet highly context-specific. When the extinction context was strongly represented during learning, retrieval of the safety memory was more likely; conversely, when the extinction context was less pronounced, fear memory traces were more likely to reemerge during testing. This competition between context-bound extinction traces and the original fear traces provides a mechanistic explanation for phenomena such as fear renewal and the frequent relapse of symptoms when patients leave the therapeutic environment.
Daniel Pacheco-Estefan, first author and researcher at the UAB Department of Basic, Developmental and Educational Psychology, notes that RSA offers a finer-grained, mechanistic view of episodic memories than approaches that look only at regional activation levels. Nikolai Axmacher, coordinating researcher at RUB, highlights that patients may experience the safe therapeutic setting as an exceptional episode, which helps explain why extinction memories do not automatically generalize to other situations.
Overall, these pioneering human intracranial and representational findings clarify how extinction learning is encoded in the brain and how context shapes the balance between fear and safety memories. The results not only validate animal-based hypotheses in humans but also suggest new avenues for improving therapeutic strategies for PTSD, anxiety disorders and related conditions by promoting broader generalization of extinction memories across contexts.
About this memory and neuroscience research news
Author: Octavi Lopez
Source: UAB
Contact: Octavi Lopez – UAB
Image: Image credited to Neuroscience News
Original Research: Open access. Title: “Representational dynamics during extinction of fear memories in the human brain” by Daniel Pacheco-Estefan et al., published in Nature Human Behaviour.
Abstract (condensed)
Extinction learning—the suppression of a previously acquired fear response—is critical for adaptive behaviour and for understanding anxiety disorders. Rodent electrophysiology implicates theta oscillations in amygdala and hippocampus and suggests extinction depends on novel, context-dependent memory traces. Combining intracranial EEG in human patients with representational similarity analysis, this study shows that amygdala theta during extinction signals safety and that extinction memory traces are stable yet highly context-specific and coordinated across the extinction network. Context specificity during extinction predicts the later reappearance of fear traces during a test period, while reoccurrence of extinction traces predicts safety responses. These results reveal neural mechanisms underlying context-dependent extinction learning in humans and provide a mechanistic account for clinically relevant phenomena such as fear renewal and extinction retrieval.