Threat-Dependent Neural Control

Threat-Dependent Neural Control: How Threat Characteristics Shape the Dynamic Organization of Human Defensive Behavior

J.Michelle

Abstract

Human survival behavior is often described through categorical responses such as fight, flight, or freeze. Contemporary neuroscience, however, increasingly suggests that defensive behavior cannot be adequately explained by a single threat center, fixed neural hierarchy, or universal sequence of responses. Instead, threat processing appears to emerge from dynamic interactions among distributed cortical, subcortical, autonomic, motor, and interoceptive systems whose relative recruitment changes according to characteristics of the threat and the organism encountering it.

Recent human neuroimaging demonstrates that threat imminence and temporal certainty alter the dynamics of a shared threat-anticipation circuit encompassing the central extended amygdala, periaqueductal gray (PAG), midcingulate cortex, and anterior insula. Other human studies indicate that qualitatively different threats—including anticipated pain, predator attack, and aggressive interpersonal attack—produce distinguishable activation patterns within the amygdala and hypothalamus. Experimental work examining approach–avoidance conflict further demonstrates interaction among amygdala, striatal, prefrontal, cingulate, motor, and autonomic processes during behavioral selection under threat. Together, these findings support a model in which defensive responding reflects context-dependent network reconfiguration rather than transfer of control to a single brain structure. (DOI) (Abend, 2023; Tseng et al., 2023).

This paper examines evidence published primarily between 2021 and 2026 concerning threat type, imminence, certainty, controllability, action availability, internal physiological state, and defensive behavior. It proposes that a more productive scientific question than Which part of the brain takes control during threat? is: How does the relative contribution and interaction of neural systems change as the characteristics of a perceived threat and the behavioral demands imposed by it change?

—

1. Introduction

A human being facing uncertain danger, an approaching threat, physical attack, unavoidable harm, and an opportunity for escape is not confronting the same computational problem.

Uncertainty requires monitoring and prediction. An approaching threat introduces temporal urgency. Physical attack may require immediate defensive action. Escape requires selection and execution of coordinated movement. Situations containing both potential harm and potential reward require arbitration between competing behavioral possibilities. The nervous system must continuously integrate information about the environment with information about the body’s current physiological condition.

It would therefore be surprising if every threatening event produced an identical neural state.

Yet popular descriptions of survival biology frequently imply precisely that. The amygdala is often characterized as an alarm that “takes over” the brain, while fight, flight, and freeze are presented as relatively discrete outputs of a singular defensive system. Contemporary neuroscience supports a considerably more complex architecture.

Defensive behavior depends upon interaction among systems involved in threat detection, salience, interoception, contextual memory, autonomic regulation, action selection, motor preparation, valuation, and executive control. Relevant structures include the amygdala and extended amygdala, hypothalamus, PAG, anterior insula, anterior and midcingulate cortices, hippocampus, thalamus, striatum, prefrontal regions, sensorimotor systems, brainstem structures, and peripheral physiological systems. These structures are neither functionally isolated nor exclusively dedicated to threat. (Nature)

The emerging question is therefore not which structure universally controls survival.

It is how control is dynamically organized across interacting systems when the survival problem changes.

—

2. From a Survival Center to a Survival Network

No contemporary evidence establishes a single human brain region as the command center for survival behavior.

The amygdala occupies an important position in threat processing, but describing it as the brain’s universal fear or survival center obscures both its broader functions and the contribution of other systems. The PAG participates in defensive behavior but also cannot independently explain the diversity of human responses to threat. Prefrontal activity does not simply disappear whenever subcortical defensive systems become engaged. Nor can conscious cognition and automatic defensive processing be reduced to mutually exclusive neural states.

Instead, defensive behavior emerges from communication among systems.

A major 2023 review in Nature Reviews Neuroscience describes defensive behavior as the product of integrated brain–body interactions. Threat imminence changes defensive organization, while internal physiological conditions can modify behavioral responses to the same external danger. Endocrine, immune, gastrointestinal, reproductive, and other peripheral signals participate in shaping defensive behavior alongside central neural circuitry. (Nature)

This produces an important conceptual shift:

The nervous system does not merely identify danger. It must determine what the organism can do about it.

Threat processing and action selection are therefore inseparable components of adaptive survival.

—

3. Threat Imminence Changes Neural Recruitment

One of the strongest organizing variables in contemporary defensive neuroscience is threat imminence: how close a potential threat is in space or time.

Human neuroimaging demonstrates that neural activity evolves as threat approaches.

In research examining threat and reward imminence, temporally increasing activity was observed across multiple regions, including the bed nucleus of the stria terminalis (BST/BNST), PAG, ventral striatum, dorsal anterior insula, and anterior midcingulate cortex. Rather than revealing a single region progressively assuming control, the results demonstrated distributed and temporally evolving recruitment across cortical and subcortical systems. (PubMed Central (PMC))

This distinction matters.

A distant possibility of danger permits prediction, environmental sampling, contextual interpretation, and prospective action planning. As danger becomes increasingly proximal, the time available for deliberation contracts while the need for rapid behavioral preparation increases.

But even this continuum should not be interpreted as a rigid handoff from “thinking brain” to “survival brain.”

The evidence instead suggests changing relative participation within an interconnected system.

—

4. Certainty Changes Timing More Than Anatomy

One of the most consequential recent findings concerns the distinction between uncertain/distal and certain/imminent threats.

Older accounts frequently proposed a relatively clean anatomical division: the BNST was associated with sustained responses to uncertain threat, whereas the central amygdala was associated with short-lived responses to immediate threat.

Recent human evidence substantially complicates that interpretation.

A 2025 Journal of Neuroscience study used fMRI and theory-driven temporal modeling in 220 adults. Certain/imminent and uncertain/distal threat anticipation recruited a shared circuit encompassing the central extended amygdala, PAG, midcingulate cortex, and anterior insula.

What differed prominently was the temporal pattern of recruitment.

Uncertain and distal threat produced sustained elevations across the circuit. Certain and imminent threat produced transient surges as the anticipated encounter approached. Importantly, the BST and central-amygdala regions showed statistically indistinguishable threat dynamics, contradicting a strict anatomical division between “anxiety circuitry” and “fear circuitry.” (DOI)

A subsequent harmonized human fMRI mega-analysis of 295 adults similarly found evidence against strict functional segregation of the BST and central amygdala during certain versus uncertain threat anticipation. (PubMed Central (PMC))

This leads to a critical principle:

Different defensive states may sometimes be distinguished less by which structures are present than by when, how strongly, and in what configuration shared structures are recruited.

The architecture may overlap while its dynamics change.

—

5. The Nature of the Threat Also Matters

Imminence and certainty are not the only relevant variables.

The kind of threat appears capable of altering neural recruitment.

A human fMRI experiment compared conditioned anticipation of three biologically meaningful threats: painful electrical stimulation, an attacking predator, and an attacking human. Forty healthy women participated.

The researchers identified differential activation within the bilateral amygdala and left hypothalamus depending upon threat category. Anticipated pain, predator attack, and aggressive human attack did not produce identical subcortical activation patterns. (Frontiers)

These findings are important but must be interpreted conservatively.

Participants were anticipating laboratory stimuli while lying in an MRI scanner. They were not experiencing actual physical assault, predation, or uncontrolled injury. The experiment therefore provides evidence that threat category can influence human neural activity, not a complete map of how the human brain behaves during real-world life-threatening events.

Much of the fine-grained causal knowledge concerning threat-specific hypothalamic, amygdala, and PAG circuits continues to come from animal models. Translation to humans is biologically plausible and scientifically informative, but anatomical conservation does not justify assuming identical human psychological or behavioral organization.

The appropriate conclusion is consequently narrower:

Human neural processing is sensitive not only to whether something is threatening, but also to what kind of threat is anticipated.

—

6. Controllability and Available Action

A threat presents two related questions:

What is happening?

and

What can I do about it?

The second question introduces controllability and action availability.

A situation in which escape is possible presents a different behavioral problem from one in which escape is unavailable. Likewise, a threat requiring movement toward safety differs from one requiring inhibition of movement, and both differ from situations in which approaching something potentially rewarding simultaneously increases exposure to danger.

Neural systems involved in threat processing therefore interact with systems responsible for valuation, action selection, motor preparation, and execution.

This is particularly visible in experimental approach–avoidance research.

In a 2024 human fMRI study, 58 participants made decisions involving varying monetary rewards and threats of electrical shock while cardiac activity was simultaneously measured. Approach–avoidance decisions recruited an amygdala–striatal–prefrontal circuit. Freezing-related bradycardia was associated with increased threat-induced avoidance and altered reward–threat comparison, with specific involvement of the amygdala and dorsal anterior cingulate cortex in integrating physiological state with decision variables. (Nature)

This finding challenges another oversimplification:

Freezing does not necessarily represent absence of processing or preparation for action.

Defensive cardiac states can coexist with valuation, decision-making, and preparation for subsequent behavior.

The nervous system can inhibit immediate movement while remaining actively engaged in determining what happens next.

—

7. The Body Participates in the Decision

Threat is not evaluated by the brain in isolation from the body.

Heart rate, respiration, hormonal state, metabolic demands, interoceptive signals, and other physiological conditions provide information that can influence behavioral selection.

The 2024 approach–avoidance findings are particularly instructive because trial-to-trial changes in bradycardia were associated with changes in decision behavior and neural computation. Stronger bradycardia was associated with greater shock-induced avoidance, while interactions between bodily state and reward–threat comparison involved the dACC and supplementary motor area. (Nature)

The broader literature similarly emphasizes continuous integration between internal physiological conditions and external threat information. (Nature)

Consequently, two people exposed to superficially similar threats—or the same person encountering a similar threat on two occasions—need not produce identical neural or behavioral responses.

The state of the organism is itself part of the information being processed.

—

8. Defensive Behavior Does Not Require a Singular Conscious Experience of Fear

Another important distinction concerns subjective experience.

Threat processing, defensive physiology, behavioral preparation, conscious fear, and the subjective sense of controlling an action are related phenomena, but they are not synonymous.

This distinction is essential because ordinary language often assumes that an extreme defensive response must be preceded by an equally extreme conscious feeling of fear.

Neuroscience does not justify that assumption.

Threat-responsive neural and physiological processes can be studied independently from self-reported emotional experience. Likewise, the subjective sense of agency—the feeling that one is controlling one’s own actions—is itself a constructed component of experience rather than a direct readout of whether motor behavior occurred.

A 2025 meta-analysis in Psychological Bulletin synthesized 75 studies examining intentional binding, a commonly used experimental measure related to sense of agency. The findings support contributions from both prospective motor processes and retrospective information in constructing the experience of agency. (UCL Discovery)

This does not establish that survival behavior during real-world trauma routinely occurs without conscious authorship.

It does establish something more fundamental:

Action, subjective authorship of action, conscious emotional experience, and defensive neural processing are scientifically distinguishable variables.

That distinction creates an important avenue for future investigation.

—

9. A Dynamic-Control Account of Human Defensive Behavior

Taken together, contemporary findings support a dynamic rather than hierarchical account of neural organization during threat.

The available evidence suggests that defensive behavior depends upon several interacting dimensions:

Threat characteristics — what kind of danger is present.

Imminence — how soon or how close the threat is.

Certainty and predictability — whether, when, and how the threat is expected to occur.

Controllability — whether behavior can meaningfully alter the outcome.

Action availability — whether escape, defense, inhibition, approach, or other responses are possible.

Context and prior learning — what previous experience predicts about the present situation.

Internal physiological state — what the body is communicating about its own current condition.

These variables do not appear to select a single corresponding brain structure.

Instead, they modify the relative recruitment, timing, connectivity, and functional contribution of overlapping systems.

A useful conceptual representation is therefore:

Threat characteristics + organism state + available actions

→ dynamic neural and physiological network recruitment

→ behavioral selection and preparation

→ action and environmental consequences

→ continuous sensory and interoceptive feedback

This is not proposed as a newly established biological circuit. It is a synthesis of the organizational principle emerging from contemporary findings and should be tested rather than treated as demonstrated fact.

—

10. What Current Research Does Not Yet Establish

The limitations of the evidence are as important as its findings.

Current neuroscience cannot reliably examine the human brain with fMRI during genuine uncontrolled life-threatening attacks. Ethical and methodological constraints mean that much human research uses conditioned threat, anticipated electrical shock, simulated predators or aggressors, virtual environments, approach–avoidance tasks, and other experimental approximations.

Animal models provide considerably greater causal and circuit-level precision but cannot establish the phenomenology of human consciousness.

BOLD fMRI measures changes associated with blood oxygenation rather than directly measuring neuronal firing or establishing causal control.

Activation of a structure does not demonstrate that the structure is “in charge.”

Greater activation does not necessarily mean greater behavioral control.

The presence of a region in multiple experimental conditions does not mean that it performs the same computation in each.

And the absence of measurable activation in a particular analysis does not establish that a region was functionally irrelevant.

These limitations are especially important when attempting to explain complex experiences such as physical assault, accidents, combat, entrapment, interpersonal violence, or sudden medical emergencies.

Current research therefore supports probabilistic network-level inference, not retrospective neural certainty about what occurred inside a particular person’s brain during a past event.

—

11. Implications and Future Research

The evidence reviewed here produces a more interesting question than the traditional fight–flight–freeze formulation.

If human defensive behavior emerges from dynamically interacting neural and physiological systems, then future research should investigate how the configuration of those systems changes as the survival problem changes.

Particularly important questions include:

1. How does neural organization differ when danger develops gradually versus appears without warning?

2. How does interpersonal threat differ from environmental or injury-related threat when imminence and intensity are controlled?

3. What changes when escape is perceived as possible, impossible, or uncertain?

4. How does prior experience alter the interpretation of otherwise comparable present threats?

5. Can conscious appraisal remain available while voluntary motor output is inhibited?

6. Can highly organized defensive motor behavior occur while subjective agency or conscious deliberation is diminished?

7. How tightly coupled are subjective fear, autonomic defense, neural threat processing, and defensive action?

8. What determines transitions between monitoring, freezing, avoidance, escape, defensive aggression, and other behaviors?

9. How does the nervous system transition back from defensive organization once the threat has ended?

10. Which findings derived from animal defensive circuitry genuinely generalize to complex human threat?

These questions move the field away from attempting to locate a singular survival switch and toward understanding the dynamic allocation of neural resources during danger.

—

12. Conclusion

The human brain does not appear to possess a single neural structure that universally assumes control during threat.

Instead, current evidence supports a distributed and dynamic architecture in which overlapping neural and physiological systems alter their relative recruitment according to the characteristics of the threat, its temporal proximity and certainty, the actions available to the organism, contextual learning, and internal physiological state.

Recent human research provides particularly strong evidence that certain/imminent and uncertain/distal threats can recruit the same broad threat-anticipation circuit while producing markedly different temporal patterns within it. Other findings demonstrate that threat category can alter amygdala and hypothalamic activity and that defensive physiological states interact with neural systems responsible for valuation and behavioral selection. (DOI)

Thus, the most scientifically defensible question is not:

Which part of the brain takes control when a person is threatened?

It is:

How does the relative contribution and interaction of neural systems change according to the nature of the threat, the state of the organism, and the actions available for survival?

Current neuroscience has begun to answer that question.

It has not finished.

And that uncertainty is not a weakness in the science. It identifies the next question the science needs to solve.

—

Evidence and Citation Boundary

Current literature supports dynamic, context-sensitive defensive organization and threat-imminence effects. The manuscript avoids a single survival-center account and treats network-level organization as dynamic rather than deterministic.

Core evidence base: (Abend, 2023; Tseng, 2023; Survival in a world of complex dangers. (2024). Neuroscience & Biobehavioral Reviews, 2024; Xavier, 2024).

References

Abend, R. (2023). Understanding anxiety symptoms as aberrant defensive responding along the threat imminence continuum. Neuroscience & Biobehavioral Reviews, 152, 105305. https://doi.org/10.1016/j.neubiorev.2023.105305

Tseng, Y.-T., Schaefke, B., Wei, P., & Wang, L. (2023). Defensive responses: Behaviour, the brain and the body. Nature Reviews Neuroscience, 24(11), 655–671. https://doi.org/10.1038/s41583-023-00736-3

Survival in a world of complex dangers. (2024). Neuroscience & Biobehavioral Reviews, 166, 105924. https://doi.org/10.1016/j.neubiorev.2024.105924

Xavier, M., et al. (2024). Threat directionality modulates defensive reactions in humans. Cognition and Emotion, 38(6), 954–962. https://doi.org/10.1080/02699931.2024.2335535