How Should We Think About the Brain’s Response to Threat?

Post by Kasey Hemington

What's the science?

A threat is something with a high probability of causing either mental or physical damage. When we encounter a threat, many related processes occur in the brain, such as detecting the threat, learning to associate a cue with the impending threat, remembering what cues (and in what contexts) predict the threat, updating how or whether certain cues predict the threat over time, and deciding on the best behavioural response. These processes are typically studied independently and are each considered to be disrupted as distinct entities in threat-related disorders like anxiety and post-traumatic stress disorder (PTSD). This week in Trends in Cognitive Sciences, Levy and Schiller reviewed the neural basis of threat and proposed that we aim to understand threat in a more holistic manner; by studying the processes that make up the threat experience as interconnected phases of threat with common underlying neural computations.

What do we already know?

Though scientists often attempt to study them separately, it can be difficult to isolate the neural correlates of each aspect of the threat experience because each brain region known to be involved in these processes is involved in multiple processes. For example, the hippocampus, amygdala, and ventral striatum are involved not only in associative learning but also in decision-making, while areas of the prefrontal and parietal cortices are involved in decision-making but also learning. The insula is known to be involved in decision-making and learning, in addition to physiological reactivity, while the periaqueductal grey also plays a role in physiological reactivity, alongside providing threat-related signals to the amygdala. When it comes to understanding the brain’s response to threat, it may be more accurate to refer to the aforementioned regions as being part of one unified, global brain network.

What’s new?

Instead of studying each threat-related-process separately, the authors consider how different brain regions play a role in different ‘phases’ of the threat experience while asking the same neural computation-related question at each phase: in an uncertain and volatile environment, how does the brain use cues to predict outcomes?

For example, consider a person who witnesses a threatening event; the explosion of a blue car at close range. The authors divide this experience into five phases: 1) initial encounter (witnessing the explosion), 2) learning (that a blue car could signal danger), 3) post-association learning (e.g. learning whether the association should be generalized to cars of all colours), 4) memory retrieval and potential updating (remembering the danger in response to seeing another blue car, potentially stabilizing or destabilizing the threatening memory depending on the events and perception at the time of retrieval) and 5) decision making (e.g. choosing between whether to drive a car or ride a bicycle).  

The authors consider the neural correlates and clinical implications (for threat-related disorders) at each phase. 

Phase 1: As a threat becomes more imminent (for example, a predator moving into the field of view of its prey) there is a shift in brain activity from the prefrontal cortex to midbrain areas, and a corresponding shift in behavioural response from anxiety and fear to panic, freezing or fleeing. This pattern is mirrored in individuals with anxiety or PTSD; high anxiety is often experienced during anticipation of a threat before it is imminent. 

Phase 2: Learning the association between a cue and a threat occurs via prediction error in the brain; there is a difference between the expected and observed outcomes predicted by a cue, so the brain learns to update predictions. Synaptic plasticity in the amygdala results in the storage of threat memories. In individuals with anxiety disorders, learning may be overgeneralized in a maladaptive way to include cues that do not predict threat.

Phase 3: Extinction learning can counteract threat conditioning. It’s when repeated exposures elicit smaller and smaller responses to a stimulus over time. In the brain, the ventral tegmental area helps to compute a prediction error between expected and observed outcomes and sends a signal to other brain areas including the amygdala in order to update the memory with new extinction memories. In PTSD, defensive responses can linger following a threat for longer than they typically would.

Phase 4: When a memory is reactivated, this provides an opportunity to destabilize the memory (a cascade of cellular and molecular processes that put it in an unstable state) and potentially modify it in this unstable state. A clinical goal for PTSD and anxiety is to modify these threatening memories long-term, by reactivating the memory, alternating the memory to include a more adaptive emotional response, and ultimately altering the way an individual engages with the world. 

Phase 5: Decisions such as avoidance of a cue indicative of a threat are made based on subjective valuation of a potential outcome, for which the ventromedial prefrontal cortex and ventral striatum, in particular, are responsible. The uncertainty of the potential outcome is also encoded in the brain, including the ventral striatum, posterior parietal cortex, and anterior insula among other regions. Finally, the expected risk or reward and an individual’s overall tolerance for ambiguity are also weighed in the decision-making process. In PTSD and anxiety disorders, a decreased tolerance for ambiguity is observed.

What's the bottom line?

This review highlights learning, memory, and decision-making together as they relate to threat experiences and threat-related disorders. At the neural level, a response to threat can be thought of as computations that predict outcomes from cues. New associations become memories, which can be updated as behaviour and environments change. Finally, decisions can be made that incorporate these associations. This way of thinking about threats and related processes can help us to study the neural correlates of threat-related disorders like anxiety disorders and PTSD more holistically.

Levy and Schiller. Neural Computations of Threat. Trends in Cognitive Sciences (2020). Access the original scientific publication here.