Understanding anxiety symptoms as aberrant defensive responding along the threat imminence continuum

Abstract

Threat-anticipatory defensive responses have evolved to promote survival in a dynamic world. While inherently adaptive, aberrant expression of defensive responses to potential threat could manifest as pathological anxiety, which is prevalent, impairing, and associated with adverse outcomes. Extensive translational neuroscience research indicates that normative defensive responses are organized by threat imminence, such that distinct response patterns are observed in each phase of threat encounter and orchestrated by partially conserved neural circuitry. Anxiety symptoms, such as excessive and pervasive worry, physiological arousal, and avoidance behavior, may reflect aberrant expression of otherwise normative defensive responses, and therefore follow the same imminence-based organization. Here, empirical evidence linking aberrant expression of specific, imminence-dependent defensive responding to distinct anxiety symptoms is reviewed, and plausible contributing neural circuitry is highlighted. Drawing from translational and clinical research, the proposed framework informs our understanding of pathological anxiety by grounding anxiety symptoms in conserved psychobiological mechanisms. Potential implications for research and treatment are discussed.

Through the course of evolution, dedicated systems have emerged to support motivational behaviors, including defensive responding for coping with potential threats^1-5. As species evolved, these systems became increasingly complex, allowing for more nuanced threat detection and assessment and mobilization of behaviors to promote survival when faced with a larger variety of threats^2,3,6. While threat-anticipatory defensive responses have evolved to favor a low threshold for initiation⁷, a high rate of false alarms confers the risk for excessive and persistent execution of responses that can lead to pathological anxiety^7-9. Anxiety symptoms may therefore reflect aberrant expression of otherwise normative defensive responses.

The goal of the present review is to link specific anxiety symptoms to the aberrant expression of distinct defensive responses. Through this work, an organizing framework for anxiety symptoms emerges, describing when different symptoms, such as excessive vigilance, worry, physiological arousal, or avoidance behavior, should arise. While all are prominently featured in pathological anxiety¹⁰, contextualizing these symptoms within the repertoire of normative defensive responses¹¹ could guide our expectations as to which symptoms to expect in which situations. Specifically, the proposed framework extends influential conceptualizations rooted in translational neuroscience research that highlight the role of threat imminence in determining the expression of distinct defensive responses and identify key elements in the neural circuitry driving them. Systematically linking anxiety symptoms to perturbations in psychobiologically-grounded patterns of conserved defensive responding could begin to address limitations in phenomenology-based clinical conceptualizations of anxiety^12,13, and thus inform our understanding of pathological anxiety and its treatment^14,15.

1. Defining features of pathological anxiety

Anxiety disorders are prevalent, chronic, impairing, and associated with adverse outcomes^16-19. In psychiatric nosology, they encompass several diagnoses, with diagnostic distinctions centering primarily on the types of objects or settings that are perceived as threatening, and which evoke a range of symptoms such as tension and physiological arousal, hypervigilance and worry cognitions, and avoidance behaviors^10,20. For example, symptoms may be evoked by social situations that involve potential scrutiny (social anxiety disorder), by circumscribed objects or situations (e.g., specific phobias), or in relation to separation from attachment figures (separation anxiety disorder)¹⁰.

Yet, despite these diagnostic distinctions, clinical observations and empirical work consistently indicate limited disorder specificity in symptom presentation (i.e., significant homotypic and heterotypic comorbidity), diagnostic reliability and validity, familial aggregation and genetic loci, pathophysiology, and treatment targets or response^19,21-28. These lines of evidence suggest that anxiety disorders likely share common mechanistic perturbations, etiology, and phenomenology, potentially arising from a shared biological or psychosocial diathesis^26,29-32. Identifying common mechanisms and patterns of symptom manifestation in pathological anxiety, rather than focusing on relatively weak diagnostic distinctions, could significantly promote our understanding of ‘core’ pathological processes; this, in turn, could guide mechanistic research towards treatment development^30,33.

What might constitute a core process in pathological anxiety? Psychiatric nosology generally highlights reported symptoms pertaining to “excessive fear and anxiety” as defining features and diagnostic criteria of anxiety disorders¹⁰. While the terms ‘fear’ and ‘anxiety’ are broadly used to distinguish emotional responses to imminent threat from anticipation and preparation for future threat, respectively^29,34-38, the precise nature of this distinction in terms of theoretical considerations, underlying neural circuitry, clinical expression, and other facets, is still very much a matter of debate^36-39. Elusive distinctions limit the utility of these terms as the basis for describing core processes in pathological anxiety, especially when it comes to linking quantitative, psychobiological research in lab models to clinical classification³⁶.

Instead, ‘fear’ and ‘anxiety’ symptoms may be understood as commonly reflecting maladaptive defensive responses to potential threat (or stressors) observed across anxiety disorders. Considerable translational research indicates that objects or settings signaling potential threat, such as danger to one’s life and health, or loss of caregivers, social standing, or resources^40,41, elicit a cascade of threat-anticipatory defensive responses observed across multiple levels (e.g., molecular, synaptic, physiological, and cognitive) that is subserved by partially conserved neural circuitry, and promotes survival by driving defensive behaviors^3,5,19,41-44. Thus, we may begin to link specific clinical symptoms of anxiety to specific patterns of exaggerated and pervasive expression of otherwise normative anticipatory, defensive responses to potential threat^{8,9,19,29,45-47}. The focus on responses to potential threat is further underscored by recent work associating pathological anxiety with aberrant anticipation, rather than ultimate experience, of aversive events^8,48,49. Linking anxiety symptoms to patterns of conserved defensive responses may help bridge clinical terminology and psychobiological processes increasingly delineated in translational neuroscience research.

Among threat-anticipatory responses, emerging evidence indicates the centrality of excessive cognitive (e.g., vigilance, worry cognitions) and physiological (autonomic arousal) defensive responses in anxiety symptom presentation, regardless of specific source of potential threat^{13,29,34,35,38,43,50,51}; their excessive expression may then promote exaggerated execution of defensive behaviors (e.g., avoidance) observed across anxiety disorders^29,30,52-54. In other words, while different objects and situations may be perceived as posing potential threat, the exaggerated expression of otherwise normative cognitive, physiological, and behavioral defensive responses is shared across disorders^{4,10,31,40,45,55,56}. Moreover, individuals who tend to show enhanced cognitive responses to threats also show enhanced physiological responses⁵², suggesting these are parts of a shared defensive mechanism⁴⁵. Together, it may be posited that pathological anxiety reflects a tendency for enhanced defensive responsivity, manifesting particularly in the physiological, cognitive, and behavioral domains^45,57; these responses are expressed too strongly or pervasively that they become maladaptive, distressing, and impairing, constituting what is referred to clinically significant anxiety symptoms^9,45,58.

2. Threat imminence continuum

Considerable cross-species research demonstrates that defensive responses to threat are organized along a threat imminence continuum, typically comprising a pre-encounter, post-encounter, and circa-strike phases^36,47,59-63. Thus, distinct responses have evolved to anticipate threat and minimize harm, such as vigilance, acute physiological responding, and avoidance behavior; these are differentially and dynamically expressed as a function of physical or temporal threat proximity, and are orchestrated and carried out by conserved neural circuitry^2,5,47,64. Importantly, recent work begins to extend such insights to humans, providing a strong translational, empirical and conceptual framework for describing human defensive responding^58,65-69.

Under the assumption that pathological anxiety reflects the propensity for exaggerated expression of normative responses to potential threat^{9,45,58,65,70}, and that such responses may be organized by threat imminence^58-60,66, distinct anxiety symptoms could be systematically linked to aberrant expression of imminence-dependent defensive responses, providing a model for dynamically contextualizing symptoms within a wider translational framework of response to threat^29,62,67,71. An organization of anxiety symptoms that is grounded in conserved psychobiological mechanisms may be useful in two key respects. First, drawing from considerable translational research, it may begin to systematically describe the manifestation of pathological anxiety with greater temporal precision^12,13,72. Anxiety symptom presentation, much like physiological, behavioral, and subjective responses to aversive events, shows temporal variability^73-75. The fact that symptom presentation is not static complicates phenomenological descriptions of pathological anxiety, as reflected in its relatively loose nosological definition^10,20; it is also acknowledged by influential theories on the dynamic nature of emotion and in the emergence of denser symptom assessment strategies (e.g., ecological momentary assessment)^74,76-79. Thus, an anxious individual will not chronically exhibit a specific set of symptoms, such as worry and arousal; however, current nosology insufficiently specifies which and when specific symptoms arise. The proposed framework anchors symptoms to specific threats and the imminence-dependent defensive responses they evoke, delineating which symptoms are expected at different times. For instance, an important public speaking event (conferring the threat of potential social rejection) is expected to differentially elicit exaggerated physiological tension, worry, passive and active avoidance behaviors, acute physiological arousal, and even panic, as the threat becomes increasingly imminent. As such, extant nosology may be complemented by predictions of when different symptoms are expected to emerge.

Second, considerable translational research and theory delineate the conserved mechanisms orchestrating and triggering imminence-dependent defensive responses^36,47,59,80. Harnessing such explanatory power towards providing a sound psychobiological basis for anxiety symptoms could substantially improve our mechanistic understanding of pathological anxiety guided to date primarily by categorical definitions provided by psychiatric nosology^13,28,81. Stronger mapping of symptoms to underlying circuitry could provide more tractable and robust biomarkers much needed for neuroscience-guided research on pathophysiological mechanisms and treatment^15,71,82-84.

In the next sections, a framework linking anxiety symptoms to aberrant expression of imminence-dependent defensive responses is described. The focus in defensive responding is on the cognitive, physiological, and behavioral domains, due to the centrality of such responses in anxiety symptomatology and the relative ease with which these can be assessed both in the lab and by patients. First, changes in cognitive and physiological responses along the threat imminence continuum, and the distinct defensive behaviors they drive, are characterized. Next, findings linking exaggerated expression of imminence-dependent defensive responses to pathological anxiety are reviewed. Finally, key elements in the neural circuitry underlying defensive responding are described.

3. Threat imminence and defensive responses

Extensive translational research delineates several phases of encounter with potential threat^47,59-61. These phases are associated with specific defensive responses, as described next.

3.1. Pre-encounter.

Pre-encounter constitutes a state in which no concrete threat is currently present, but one is possible and thus anticipated^85,86. As such, it is a context-activating state reflecting greater likelihood of threat encounter in the current spatial or temporal context, initiated through species-specific innate responses (e.g., by open field, darkness, or predator odor) or learned through past experiences (i.e., context conditioning)^87,88. It is typically modeled experimentally in humans using paradigms involving context-based induction of uncertain-threat delivery, such as context-conditioning and threat-of-shock paradigms in which subjects are verbally informed about context-dependent, temporally-unpredictable electric shocks^32,87,89,90.

Cognitively, the pre-encounter phase is characterized by vigilance^67,91, facilitated by increased sensory gain⁹², to facilitate detection of potential threats. Such responding is coupled by increased tonic physiological arousal, e.g., elevated skin conductance levels, increased muscle tone, and potentiation of startle reflexes, likely to facilitate rapid encounter phase responding (see below)^88,93-95. A sustained pre-encounter state may manifest behaviorally in restlessness and tension reflecting prolonged preparation for action, and attempted avoidance of the threat context, and may be accompanied by reduction in appetitive behaviors such as food intake or sexual activities which may be deemed as lower priority^47,67,96,97. Thus, vigilance promotes detection of potential danger in the environment (e.g., a moving object), while sustained, elevated tonic physiological activation facilitates responses to detected objects^{67,88,89,97,98}.

3.2. Encounter.

When one is in a pre-encounter context, and an object conferring potential threat is spotted, the encounter phase is initiated, interrupting and “resetting” ongoing behavior^99,100; phasic arousal then promotes rapid assessment of potential threat value towards execution of defensive behaviors, if necessary^4,31,99,101. For example, when an animal in an open field spots a potential predator, or when a person walking down a dark alley notices movement. Thus, the encounter phase promotes state transition from pre- to post-encounter with a specific threat^4,35,102-105. Of note, translational literature typically combines the encounter and post-encounter phases described here; here, these phases are separated since specific anxiety-relevant processes in humans may be identified in each.

Arousal and action preparation are accompanied by phasic physiological responding, e.g., increased perspiration, heart rate, breathing, and muscle tone, driven by the sympathetic branch of the autonomic nervous system^99,106,107; such insight is derived primarily from cross-species research probing threat learning processes that involve conditioned, experience-driven changes in threat value ascribed to cues^{46,80,101,108}. Time-course analysis further suggests that encounter-driven phasic physiological arousal may precede threat assessment^67,109,110, potentially indicating that this initial response plays a role in promoting threat assessment and preparation for action, if such is needed^104,107. Aside from reflexive responses (e.g., startle)⁸⁰, encounter in animals and humans involves initial behavior inhibition. Such ‘attentive defensive freezing’ (inhibition of overt behaviors) may serve to evade detection by potential predators but also provide the opportunity for detailed assessment and selection of appropriate defensive behaviors^{31,58,64,67,96,99}. Elaborate threat assessment depends on attention allocation which facilitates processing object features^4,111,112; this enables estimation of threat value computed from factors such as its type, proximity, magnitude (or cost), probability of striking, and broader context^{4,58,64,67,113,114}.

The evolved ability to generate and maintain powerful and realistic future representations of potential threat through mental simulation of future action-outcomes⁸⁰ is adaptive as it enables preparation for more distal danger in complex settings. Such imagery and simulation have been shown to elicit robust cognitive and emotional responses, impact neural plasticity, and activate relevant sensory-motor brain regions^115-117. Thus, invoked imagery of potential negative outcomes (e.g., social embarrassment during public speaking) can serve as robust threats even without physically perceiving the relevant situations^35,118-120. As such, we may consider such imagined and stimulated threat as encountered threat, requiring assessment of features such as threat magnitude (cost) and probability of occurring, which accordingly elicit defensive responses^118-121 (see post-encounter). While threat assessment of an object that is physically perceived may be relatively simpler as relevant physical features are clearly observed, assessment of distal, simulated threats may be complicated by difficulty in accurately estimating such features, including occurrence probability. The inherent uncertainty in such estimation⁸, especially given limited knowledge, could prolong this process and call for its repeated initiation.

More broadly, the encounter phase may mediate the resolution of approach-avoid conflicts, arguably more common than threat-only settings, by weighing estimated potential threat and reward values to determine the adaptive motivated behavior to pursue given a certain situation^{31,64,104,122}. Thus, desired things invoke approach behavior which increases the likelihood of attaining them^1,3, but in many cases this also entails potential threat and avoidance motivation that conflict with the approach motivation (e.g., eating sugar-rich foods). The arbitration of such conflicts may be particularly nuanced and prolonged in humans, given the expanded capacity for generating multi-dimensional, internal representations of multifaceted, abstract, and often ambiguous, stimuli and settings^31,64. Internal weighing of approach and avoid action outcomes determines eventual behavioral outcomes, such that greater threat value may result in passive avoidance, a defensive response that prioritizes inaction at the cost of forgone rewards.

3.3. Post-encounter.

Encounter phase computations of threat value guide the post-encounter execution of appropriate defensive behaviors given available options (e.g., the presence of escape routes) and current goals^4,47,60. For example, a concrete threat of high proximity or magnitude calls for acute active avoidance, a prototypical, highly-conserved defensive response that reduces the likelihood of harm by preventing contact with the threat through actively removing oneself from the source of imminent danger^3,31,123-125. Given adequate space and escape route, animals will flee an approaching predator¹²⁵; likewise, humans will create distance from a potential attacker. Cross-species research indicates that the likelihood of eventual execution of behavior scales with proximity and magnitude of the identified threat^{47,66,67,110,125,126}. This scaling is accompanied by changes in autonomic responding reflecting sympathetic activation (e.g., increasing skin conductance, and muscle tone) alongside inhibitory parasympathetic activation (e.g., bradycardia, hypoventilation, and motor inhibition), and these are moderated by available behavioral options^{67,109,110,127-129}. Parasympathetic dominance is understood to act as a brake, allowing for increasing action preparation, but not execution; a threshold-based shift to sympathetic dominance (and parasympathetic withdrawal) occurs as behavior execution is necessary, enabling its initiation^{99,114,130,131}.

The ability to maintain representations of more distal threats that do not yet require immediate, acute active avoidance provides space and time for more complex forms of avoidance to diminish the danger^80,99,132. For example, animals may use defensive burying to create distance from threat¹³³; in other settings, animals and humans may exhibit defensive aggression or submissive behaviors to diminish the chances of being attacked^134-137. The evolved ability to mentally simulate potential threats that are fully imagined, but assessed to pose sufficient danger to invoke defensive behaviors (e.g., social embarrassment during public speaking), provides opportunities for more instrumental, deliberative behaviors in the service of diminishing the future danger^{47,65,80,120,138}. Such behaviors are often more elaborate, slow, and require action simulation to determine expected outcomes^80,132, such as preparing and rehearsing a planned presentation (to avoid social embarrassment and rejection) or washing one’s hands before a meal (to avoid infection). The time afforded by mental simulation of future threats allows for Iterative (encounter) threat assessment to diminish the chosen behavior if it was successful, or switch to simple active avoidance if mitigation failed and the threat has become highly proximal^99,104. As noted above, within the context of motivational conflict, reward-driven approach behaviors may be intertwined with instrumental active avoidance, generating cautious approach (risky foraging)^31,68,125.

3.4. Contact.

Situations in which the threat is already in direct contact (i.e., attack; also referred to as circa-strike¹) and preventative avoidance behaviors were not pursued, invoke the immediate execution of acute, conserved, and stereotypic defensive behaviors, such as escape (uncoordinated flight burst, differing from active avoidance), defensive attack, and tonic immobility (“playing dead”, as opposed to risk assessment-associated freezing during encounter³¹); i.e., flight, fight, or freezing, depending on available options^{31,47,61,62,67,96,139}. The all-or-none nature of activation of these simple, conserved behaviors suggests a threshold that is crossed following rapid, coarse computation^47,131, and which favors enduring many false-positives for a rare, actual danger⁷. While post-encounter involves monotonically increasing physiological preparation for active avoidance execution, cross-species research indicates that threat contact is accompanied by phasic, systemic sympathetic discharge of physiological responses to support these acute behaviors, such as perspiration, hyperventilation, and tachycardia^29,40,86. In contrast, cognitive responses may be dampened since there is not enough time to carry out more sophisticated threat computations and nuanced goal-directed behaviors^31,47,140.

In summary, evidence gleaned from cross-species research supports a dynamic organization of defensive responses by threat imminence, whereby as a potential threat becomes increasingly imminent, different patterns of defensive responses are expressed at different phases. Of note, some responses may be observed in multiple phases, reflecting common required functions; for example, increased physiological arousal, to facilitate behavior execution, may be observed in all phases, but shows distinct temporal dynamics across phases (e.g., tonic, phasic, or monotonically increasing).

4. Anxiety symptoms and excessive expression of defensive responses

According to the proposed framework, pathological anxiety reflects a tendency for exaggerated expression (i.e., greater magnitude and persistence) of these otherwise normative defensive responses^9,45,58. As such, as a threat becomes increasingly imminent, an anxious individual will show excessive expression of expected imminence-dependent defensive responses which follow the same organizing scheme (Fig. 1; Box 1). In other words, pathological anxiety as a disposition will dynamically vary in expression as a function of imminence of relevant threats. In the next section, empirical work linking specific anxiety symptoms to imminence-dependent defensive responses is reviewed.

Figure 1.

Phases of threat imminence (defined along horizontal axis) are described in terms of magnitude of expression (vertical axis; from low to high magnitude) of cognitive (blue) and physiological (red) defensive responses, the behaviors they drive (green), and how these may be linked to anxiety symptoms and related responses (purple). Note that some symptoms may potentially occur in more than one phase. Solid and dashed lines describe expected intensity for individuals characterized by high and low levels of anxiety, respectively (note that anxiety severity, and magnitude of defensive responses, likely reflects a continuum).

Box 1: Empirical data on anxiety and threat imminence.

As an example of research linking imminence-dependent defensive responding and anxiety, in recent work¹⁰⁹, we examined physiological responding in 50 youths who were either free of any psychiatric diagnoses (healthy group) or diagnosed with social, separation, and/or generalized anxiety disorder (unmedicated, treatment-seeking; anxiety group). All participants completed a task whereby instructed cues signaled the upcoming delivery of non-painful (safety) or highly painful (threat) thermal stimulation. This design enabled us to compare groups in terms of responses at different threat-imminence phases, as shown in the accompanying Figure. Physiological response was quantified in bins as average skin conductance signal relative to its tonic level.

During pre-encounter, when cue onset was expected, the anxiety group showed greater responding than the healthy group. This group effect became stronger during the encounter (that is, following the onset of cues signaling threat or safety), but no cue effect was observed, suggesting that anxious individuals' enhanced arousal is relatively non-specific and occurs prior to threat differentiation. The presence of a group effect but no cue effect highlights exaggerated responding at the cost of many false-positives (safety trials). During the post-encounter epoch, the groups showed different patterns of arousal. Among anxious individuals, arousal increased as threat became more proximal, and decreased as the safe outcome became more proximal, with maximal threat differentiation becoming maximal at the moment of greatest imminence. This pattern was substantially attenuated in the healthy group. Importantly, group effects on physiological responses also manifested as dimensional associations with continuous symptom severity scores. Finally, response to the painful stimulation ("contact") were significantly greater in the anxiety group. Together, these findings underscore the value of considering dynamic responses across different stages of the threat imminence continuum for understanding pathological anxiety.

Changes in physiological response by threat imminence in individuals with and without anxiety disorders.

Threat-imminence phases were defined as epochs (demarcated in gray boxes) relative to cue and thermal stimulation onset in which skin conductance was indexed. Data points and lines reflect response to threat (highly painful heat; solid lines) and safety (non-painful heat; dashed line), for the anxiety (red) and no-anxiety comparison (blue) groups. Data presented reflect physiological activation already accounting for individual differences in local tonic (baseline) responding. Distance between X-axis ticks is 2 seconds.

Note: *, p<0.05, **, p<0.01, ***, p<0.001. Colored asterisks reflect within-group changes; black asterisks reflect between-group effects. Data adapted from prior work⁹³.

Reviews and meta-analyses examine links between anxiety severity and magnitude of specific cognitive and physiological responses during different phases of threat encounter. Pre-encounter is typically modeled using sustained unpredictable-threat (relative to safety) paradigms that robustly elicit elevated subjective anxiety^72,90. Meta-analysis of functional imaging studies indicates an overlap between such induced states and findings in anxiety patients that manifests primarily in anterior and mid-cingulate and inferior frontal gyrus (IFG)/insula cortices³². Given the purported roles of cingulate and IFG in selection and control of behavior²⁷² and of the insula in representing salience and mediating physiological homeostasis²⁷³, such findings may potentially reflect anxiety-related enhanced motor and somatic preparation for action should salient stimuli be detected in a context in which potential threat might occur³².

Several meta-analyses examine cognitive responding in anxiety that may correspond to the encounter phase (although it may be difficult to sufficiently distinguish those from post-encounter processes). Anxiety was found to moderately relate to bias towards threat in early allocation of attention measured using reaction-time (ß=0.08 in 1,291 youths¹⁷⁸; d=0.45 assessed across 2,263 individuals and various anxiety disorders⁵¹) and (eye-tracking-derived) initial fixations (g=0.47, N=1,085¹⁴¹) measures; with a bias away from threat in sustained attention processes (free-viewing dwell time in youths with anxiety disorders, g=0.26, N=798¹⁸²); with interpretation bias (self-generated text to ambiguous scenarios in youths with anxiety disorders; d=0.62, N=11,507²⁷⁴); and with behavioral measures of conditioned fear over-generalization (g=0.24, N=867 across anxiety disorders²⁷⁵).

Physiologically, comparison of anxiety patients to controls (N>2,000) indicated excessive responding (SCR or EMG startle response) to conditioned safety, d=0.30, and threat, d=0.35, stimuli during the conditioning and extinction phases of fear learning, respectively¹⁷⁴, which may primarily reflect learned encounter responses. That is, across fear learning paradigms, the assessment of threat value ascribed to reinforced cues changes (increasing threat during conditioning; decreasing threat during extinction) in an anxiety-dependent manner. Neurally, meta-analyses of fMRI studies examining neural responses to anxiogenic stimuli (Ns=1,993, 873, and 327 including patients and controls) reveal differences in circuitry components implicated in encounter responses, including amygdala, insula, vmPFC, and ACC, as well as parietal cortex (which may reflect attentional components)^238,276,277. fMRI meta-analyses of responses to conditioned and extinguished threat cues in healthy individuals likewise reveal consistent ACC, insula, and vmPFC, as well as PAG, hippocampus, and additional structures implicated in threat assessment^237,278, but a meta-analysis comparing anxious and healthy individuals is needed to identify specific pathophysiology.

Meta-analysis of acute social stress, comparable to imminent-threat post-encounter, indicated sex-dependent effects on cortisol release in 372 anxiety patients and controls (SMD=0.50²⁷⁹). Here, as well, imaging studies on specific post-encounter and contact processes in anxious individuals are needed to enable meta-analytic inference on neural circuitry. Together, extant meta-analytic findings using behavior, eye-tracking, physiology, and neuroimaging (where available) indicate consistent anxiety-related differences in patterns of imminence-specific responses to potential threat.

4.1. Pre-encounter.

Studies indicate that anxiety is associated with enhanced expression of defensive responses associated with pre-encounter states. Specifically, findings demonstrate exaggerated cognitive (e.g., hypervigilance, worry) and physiological (elevated autonomic levels) responding in such contexts, potentially reflecting excessive efforts to predict, detect, and prepare for aversive events^{8,35,88,89,109,141-144}. Thus, key clinical features of anxiety disorders such as heightened vigilance and muscle tension “in preparation for future danger”¹⁰ may be framed as aberrant pre-encounter responses. Likewise, excessive avoidance of contexts perceived as conferring threat is prominent in anxiety, such as avoidance of separation from caregivers or social settings^10,96,97,145. Related maladaptive behaviors that arise during anticipation of potential threat, such as restlessness, difficulty relaxing, and tonic irritability, are observed in anxiety and may potentially be attributed to this phase^10,146,147. Relatedly, key aspects of “stress” may correspond to preparation for threat encounter¹⁴⁸; stress-induced reduction in appetitive motivated behaviors^149-151, reflecting prioritization of survival functions, is observed in anxiety and has been proposed as a possible mechanism linking anxiety and depression^152-154. Enhanced pre-encounter responding may potentially promote stronger subsequent encounter and post-encounter responses (i.e., biased threat assessment leading to exaggerated avoidance behaviors), as reflected in context-potentiated startle responses^{89,92,143,155,156}. Such a prepotent downstream effect is inherently adaptive as the cost of false-positive response cascade initiation is relatively low given the threat associated with the context^7,40.

States of sustained anticipation of potential threat can also impair some cognitive functions such as working memory and attention control in healthy individuals, and could potentially account for such cognitive deficits observed in anxiety patients^143,157-160. As noted above, these deleterious effects may reflect the prioritization of threat detection over other goal-directed processes^98,161, although it remains a matter of debate whether such effects are specific to processing of threat (vs. non-threat) stimuli and which specific functions are affected^143,162-165. For example, some work finds unpredictable-threat states to impair attention control and other cognitive processing while other research finds improvement^161,166-168. Finally, intolerance of the uncertainty associated with occurrence of future threat (e.g., if and when it will materialize) is a trait found to strongly relate to anxiety severity, suggesting a role for biased certainty and confidence in threat estimation in pathological anxiety^8,169-172.

4.2. Encounter.

Different forms of excessive physiological, cognitive, and behavioral responses to threat observed in pathological anxiety may be attributed to aberrant defensive responses associated with the encounter phase. Stronger phasic physiological responses and attentive freezing in response to encounter with cues signaling potential threat have been linked to anxiety severity, particularly in fear-learning and unpredictable-threat paradigms (via responses to conditioned cues and startle probes, respectively)^{99,106,173-175}. Recent work further indicates that anxious individuals show stronger initial physiological responses to detection of potential-threat cues, occurring prior to threat assessment^49,109; such findings support the notion of enhanced encounter responsivity in pathological anxiety, at the cost of excessively high rates of unnecessary, false-positive initiation of the defensive response cascade⁷.

Observations of threat-related biases in otherwise normative cognitive processes underlying threat detection and assessment have inspired influential cognitive models of pathological anxiety, e.g.^112,176,177. For example, evidence from emotional face paradigms suggests associations between anxiety severity and excessive allocation of attention to, and visual processing of, encountered threat cues, although inference is complicated due to inconsistent findings and effect sizes (Box 1)^{51,112,141,178-181}. Considering threat imminence may begin to resolve such discrepancies, as early attention allocation during threat encounters may reflect enhanced threat vigilance¹⁴¹, followed by threat avoidance during post-encounter (sustained attention)^182,183. Biases in several processes involved in threat valuation and probability assessment are commonly observed across anxiety disorders. These include, for example, inflated estimates of negative outcome likelihood and cost, and interpretation of ambiguous cues as conferring greater threat value^8,184-186. Such biases have conceivably been “rewarded” through evolution as they ultimately promote fitness; however, in most situations, they result in the frequent and unnecessary defensive mobilization observed in anxiety⁷.

As noted, we may begin to consider mental imagery and simulation of aversive outcomes as eliciting encounter responses within the threat-imminence framework. Much like physically perceived objects, imagery of future, simulated outcomes calls for assessment of threat levels to determine appropriate defensive responses. Cognitive biases⁸ could lead to inflated estimations of threat magnitude and probability of occurrence which then manifest as pessimistic expectations and catastrophizing cognitions that underlie the clinical symptom of excessive worry observed across anxiety disorders^{8,34,116,120,187,188}. Difficulty in accurately estimating future threat magnitude and probability, especially in situations of significant uncertainty, could lead to repetitive invocation of threat assessment processes, contributing to the persistent and intrusive nature of worry in anxiety^{34,65,187,188}. Similarly, worry has been suggested to reflect maladaptive attempts to simulate actions that will sufficiently mitigate the threat (problem solving)¹⁸⁹. Prioritized, repetitive threat assessment and preparation for defensive behavior may further be associated with observed deficits in cognitive flexibility and decision making in anxiety^190,191, and could potentially underlie obsessive tendencies common in obsessive-compulsive disorder (OCD)¹⁰; indeed, OCD is highly comorbid with anxiety disorders¹⁹².

4.3. Post-encounter.

Anxiety-related biases in encounter-phase threat appraisal, such as inflated threat cost and probability⁸, could lead to stronger downstream mobilization of defensive behaviors^193,194. Indeed, excessive expression of post-encounter defensive responses and their behavioral outcomes is a central feature across anxiety disorders^{10,29,34,52,187}.

As noted, increasing proximity of a concrete, identified threat is associated with changes in sympathetic physiological responding to facilitate execution of active avoidance⁶⁷. Indeed, excessive phycological responses to proximal threats are defining features of pathological anxiety¹⁰. For example, experimental manipulations of post-encounter threat proximity show that approaching internal/interoceptive threats (i.e., induced dyspnea) and external/exteroceptive threats (i.e., noxious thermal stimulation) lead to exaggerated increases in physiological responses and execution of active avoidance in anxiety patients^67,109,127. Aside from acute forms of active avoidance, pathological anxiety is also associated with excessive expression of other, more complex and nuanced behaviors aimed at diminishing threat, such as defensive aggression, and submissive behavior^{10,87,119,137,138,195}.

A more distal negative outcome that does not require immediate coping offers time for more complex forms of instrumental avoidance that will mitigate the threat¹⁹⁶. Excessive execution of such behaviors is observed in pathological anxiety and may disrupt normative appetitive behaviors^10,197; for example, in an effort to minimize potential public embarrassment, anxious individuals may spend excessive amounts of time in preparation and rehearsal before public speaking¹⁰. Similarly, perfectionism in school settings may be viewed as excessive means to diminish academic threat (e.g., not achieving academic goals) and is highly correlated with anxiety severity¹⁹⁸. As with repetitive threat assessment, and perhaps due to it, excessive execution of certain instrumental defensive responses (e.g., excessive hand washing) may underlie OCD-related compulsions^10,192.

Alternatively, rather than actively utilizing instrumental actions to mitigate future threat, individuals often engage strategies for effortful emotion regulation, i.e., sets of actions aimed to influence one’s own emotional experience¹⁹⁹. Anxiety is specifically associated with increased use of maladaptive regulation strategies, such as distraction, emotional avoidance, and suppression^200-202. These may reflect forms of post-encounter, ‘internal’ active avoidance behavior to diminish defensive responsivity to the threat which may be experienced as distressing, rather than actions to cope with the threat itself.

Within the context of approach-avoid conflicts, exaggerated threat valuation during the encounter phase⁸ should increase the tendency for passive avoidance behavior (i.e., diminished approach behavior)^4,8,189,203. Indeed, excessive passive avoidance is a salient and impairing feature of pathological anxiety, whereby individuals prefer to forgo potential rewards (e.g., social interactions, making friends) due to the risk of aversive outcomes (e.g., social rejection)^{10,104,189,204,205}. Such maladaptive behavioral tendencies, especially to novel objects or situations, are a key feature of childhood behavioral inhibition temperament^96,206,207 which confers greater risk for later pathological anxiety^208,209.

From a reinforcement learning perspective, major theories further propose that excessive and pervasive use of avoidance behavior, both in passive and active forms, precludes determination of (likely) eventual safety and the integration of such information into future threat assessments^124,210-212. This impedes normative extinction of learned threat contingencies, or correction of inaccurate beliefs, contributing to the persistence and maintenance of heightened threat responding in pathological anxiety^{4,132,213,214}. For example, a non-anxious child bitten once by a dog will extinguish their acquired fear of dogs through subsequent encounters with friendly dogs; an anxious child is more likely to remain fearful since subsequent encounters elicit avoidance, precluding linking dogs with the experience of safety.

4.4. Contact:

Acute physiological discharge and stereotypic defensive behaviors associated with threat contact are referred to clinically as panic attacks^{10,34,58,67,86}. In line with the notion of pathological anxiety reflecting excessive defensive responding, panic attacks indeed feature prominently across anxiety disorders in situations where danger is perceived as immediate and all but certain^10,34,86,214. As such, panic attacks may reflect excessive instantiations of perceived threat contact. The selection of defensive behavior to pursue has been shown to depend on available options. Thus, laboratory studies demonstrate greater rates of escape behavior in anxiety patients in face of immediate, but escapable, threat^67,86. In contrast, excessive tonic immobility (or defensive immobilization) in anxiety is associated with threat contact situations perceived as inescapable²¹⁵; relatedly, inescapable peritraumatic tonic immobility predicts later post-traumatic stress disorder (PTSD) symptoms^216-218. Defensive attack may correspond to reactive aggression and rage episodes observed in individuals with anxiety, as well as related conditions such as (phasic) irritability and PTSD^10,219.

5. Neural circuitry supporting defensive responding

The conserved nature of defensive responding offers opportunities for leveraging insight from translational, cross-species neuroscience research to identify potential pathophysiological mechanisms in anxiety^{38,59,66,72,80}; but see^38,220. As the scope of potentially relevant brain regions is wide³⁹, this review is necessarily selective, and will focus on the most intensively scrutinized circuits (Fig. 2). It should be noted that inference on phase-specific functions is still limited due to experimental designs that do not allow for clear dissociations.

Figure 2.

Brain structures and regions hypothesized to play key roles in defensive responses associated with each phase of threat imminence. Additional structures and regions are likely involved; the subset depicted here has been specifically associated with anxiety. Note: dmPFC = dorsomedial prefrontal cortex; PFC = prefrontal cortex; BNST = bed nucleus of the stria terminalis; Ant. = anterior; Post. = posterior; PAG = periaqueductal gray.

5.1. Pre-encounter.

Translational research in rodents and humans focusing on contexts conferring potential threats implicates a network that includes hippocampus and amygdala, and contiguous cortices, in encoding and retrieving associations between environmental, interoceptive, and temporal contexts and aversive outcomes^4,32,221-223. Studies in humans invoking induced, sustained potential-threat anticipation compared to safe settings point to key involvement of amygdala and bed nucleus of stria terminalis (BNST) in such states, and their modulation of sensory gain in other regions to facilitate responsiveness to incoming information^{67,88,224-226}. Electrophysiology work shows, for example, anxiety-related enhanced early processing of presented auditory stimuli, evidenced through mismatch negativity-evoked responses linked to auditory cortices and brainstem processing, and visual stimuli, assessed via effects on early-visual components to presented visual stimuli measured over occipital visual areas^67,143. Importantly, anxiety-related effects on early processing in threat contexts are observed to both threat and neutral stimuli^67,143, suggesting a potential-threat-induced hypervigilant state that impacts general sensory processing of incoming stimuli.

Complementing this work, recruitment of anterior insula and dorsomedial PFC and anterior cingulate cortex (ACC), as well as dorsal attention network nodes (e.g., dorsolateral PFC and parietal cortex), has been observed in humans and non-human primates when anticipating unpredictable threat, with variation in their coactivation linked to anxiety and modulation of physiological responses^{72,98,227,228}. As such, it has been implicated in adaptively promoting and maintaining attentional vigilance and preparation for action, while its excessive activation has been suggested to contribute to pathological anxiety^9,32,87,143. Despite these findings, additional research is required to directly link neural circuitry function to physiological and behavioral pre-encounter responses.

5.2. Encounter.

A large body of research, primarily in rodents and humans, examines the circuitry subserving encounter responses. Initial object detection initiates activation of the sympathetic branch, primarily via widespread ascending and descending locus coeruleus (LC) noradrenergic projections^99,100. This facilitates phasic physiological responding, such as perspiration and increased heart rate, through interactions among LC and amygdala with hypothalamus and other downstream structures, that accompany preparation for immediate mobilization of acute defensive behaviors subserved by amygdala-dorsal periaqueductal gray (PAG) and LC-spinal cord paths^{42,99,100,229}. Concurrently, LC and amygdala nuclei modulate cognitive functions by promoting cortical excitability that facilitates arousal, perception, and attention allocation^99,100,230. These activating effects are counteracted by parasympathetic branch function involving basal forebrain and ventral PAG pathways, providing rapid inhibition of responding and attentive freezing behavior^99,231,232. This sympathetic-parasympathetic balance enables more elaborate assessment of threat levels and nuanced selection of behavior when threat is ambiguous or during approach-avoid conflicts (see below)⁹⁹.

A key role in detailed processing of threat-relevant cues (e.g., conditioned cues) is attributed to basolateral and central amygdala activity, particularly in conjunction with ventromedial prefrontal cortex (vmPFC), as revealed from research in rodents, non-human primates, and humans^{80,226,233-238}. This circuit is part of a broader network mediating threat valuation and selection of motivated behaviors that considers multiple object features given internal and external contexts, with key subcortical nodes including amygdala nuclei, ventral striatum, hypothalamus, and ventral hippocampus^{101,122,230,235}. Research in rodents, non-human primates, and humans further suggests the integration of subcortical input in vmPFC/orbitofrontal cortex and septum, anterior cingulate, and anterior insula during valuation-driven cognitive functions such as encoding of future aversive and appetitive outcomes and probability, conflict monitoring, and mediation of approach-avoid conflict resolution and broader decision making^{8,64,85,102,113,114,239-242}. Activation of posterior cingulate cortex during encounter^{129,237,243-245} may reflect the relevance of environmental context and spatial information during threat assessment^246,247. Aberrant function in this circuitry has been linked to threat-related biases in these cognitive functions^{8,203,205,248} as well as to anxiety, and particularly in aberrant amygdala connectivity with vmPFC and dorsal PFC, potentially indicating dysregulated threat assessment^{9,96,205,249-252}. Cross-species work further suggests that perturbations that aberrant reinforcement-based updating of threat value, subserved in part by amygdala-vmPFC function, is linked to anxiety, and hypothesized to play a key role in its etiology and treatment^{9,70,175,234,253}.

Limited literature directly examines mental imagery of potential negative outcomes associated with abstract or distal potential threats. Extant work on negative imagery and worry cognitions suggests the involvement of nodes of the ‘default mode network’ (DMN), primarily including anterior and posterior cingulate cortices, and their interaction with emotion regulatory cortical regions such as lateral PFC^54,254-256. Additional work further implicates amygdala and PAG activation and functional connectivity with DMN hubs and lateral PFC during induced threat imagery and worry^257,258.

5.3. Post-encounter.

Expression of physiological responses to a specific, proximal threat have been robustly and causally linked to vmPFC, and particularly its caudal part, via its regulation of downstream structures including amygdala, hypothalamus, and brainstem nuclei, as shown in non-human primates and humans^{9,96,252,259,260}. The maintenance of parasympathetic dominance until acute active avoidance execution has been associated with ACC and adjacent mPFC regions^99,126,261. Complementing such work, rodent and human research suggests that the amygdala and perigenual cingulate cortex, adjacent to vmPFC and ACC, mediate the shift from parasympathetic to sympathetic dominance which facilitates the execution of active avoidance^99,104,126. vmPFC has likewise been linked to gating of acute active avoidance through projections to ventral striatum, a key structure implicated in executing motivated behaviors, as part of a circuit that also includes amygdala nuclei, anterior insula, multiple cingulate sectors, and premotor cortices^{37,44,47,69,104,262}. Indeed, function in different nodes in this network has been shown to be dynamic and sensitive with respect to threat proximity^{44,47,59,67,129}. For example, research across species shows that as a threat is perceived to get closer, amygdala, BNST, and anterior/mid-cingulate and motor cortices are increasingly engaged, potentially to support the expected execution of the prepared active avoidance behavior^{44,67,110,129,140,263-265}; conversely, some work shows parametrically decreasing activation in vmPFC and posterior cingulate cortex with threat proximity^129,243. Of note, studies implicate PAG in preparation and execution of active avoidance in humans^{47,110,126,129,160}, potentially capturing ventral inhibition of lateral divisions mediating active behaviors¹³⁹; however, these need to be reconciled with animal work linking PAG specifically to contact-driven escape²⁶².

Anxiety severity has been found to moderate the association between amygdala-vmPFC intrinsic functional connectivity and magnitude of imminence-dependent increases in physiological response¹⁰⁹. Similarly, anxiety moderates connectivity among amygdala, BNST, and vmPFC during cued anticipation of unpredictable outcomes^266,267. Converging work in animals shows that induced vmPFC over-activation (mimicking pathophysiological states) results in increased amygdala and hypothalamus activity accompanied by altered physiological responses to post-encounter threat²⁵⁹, highlighting the importance of this region in orchestrating the expression of active avoidance.

5.4. Contact.

The nature of acute, contact-driven defensive behaviors is similar across a variety of species, suggesting the involvement of highly-conserved circuitry⁴⁷. PAG, through amygdala and hypothalamus paths, has been robustly implicated in generating escape, aggression, and tonic immobility across species^{47,62,131,139,140}. Electrical stimulation of PAG in humans evokes acute physiological responding (e.g., increased sweating, heart rate, and respiratory rate), as well as fear of death and desire to flee, which closely resemble panic attacks²⁶⁸. Further, PAG structural and neurochemical abnormalities are associated with a proneness for panic attacks in patients²⁶⁹. Medial sections of PFC, including ACC, and insula additionally project to PAG²⁷⁰; given their roles in threat assessment (see above), these regions may mediate the initiation of contact-driven defensive behaviors⁹⁹.

The findings reviewed above suggest the involvement of several conserved neural circuits in imminence-dependent defensive responding, and indications for anxiety moderation of aspects of these circuits. These findings also show that some brain structures, as well as activated biological systems (e.g., autonomic physiology), are involved in multiple imminence phases, potentially pointing to shared original functionality that, over time, became more intricately controlled to support distinct functions³. Additional research is needed to link anxiety to specific aberrant function in defensive response circuitry, and test different theoretical models. For example, a central, restricted mechanistic perturbation (e.g., amygdala dysfunction¹⁰¹) may give rise to aberrant function of multiple defensive responses; alternatively, it may be the case that functional perturbation independently spans multiple, distributed brain systems⁸. Considering threat imminence in study designs may allow for stronger inference on the involvement of specific circuitry in different forms of aberrant defensive responding (Box 2).

Box 2: Implications for research.

Practical considerations for research can be drawn from the organization of anxiety symptoms by threat imminence. Given that specific symptoms may reflect excessive expression of distinct imminence-dependent defensive responses, it is critical to qualify which phase or process is evoked and modeled in experimental paradigms. For example, paradigms examining cued (learned or instructed) physiological responses to threat typically index responses as an aggregate across a window of several seconds, which may encompass multiple effects related to pre-encounter, encounter, and post-encounter. Collapsing responses across several phases may diminish sensitivity to detect expected pathological anxiety effects on conditioned responses to threat (vs. safety) cues, potentially contributing to inconsistent findings observed across studies^{49,173,174,280}; analysis of conditioned response dynamics could potentially improve detection of anxiety effects¹⁰⁹ (Box 1).

Relatedly, studies commonly aim to engage “threat processing” neural circuitry, broadly defined, by presenting pictures of different emotional faces. Yet, such manipulations may primarily elicit encounter responses characterized by a general, non-specific initial response to cue detection, with threat-based differentiation in responses emerging only later^109,281,282; this may account for issues relating to small effect sizes²⁸³ in neuroimaging studies contrasting responses to emotional faces. Modeling neural activation in a phase-specific manner (e.g., using a phasic impulse model for encounter responses, and a sustained-response model for post-encounter activation) could potentially better capture the underlying psychobiological processes⁴⁴.

Physiological responses provide primary readouts in research on defensive responding¹⁰⁶. Given phase-specific response patterns (e.g., tonic vs phasic responding), it is important to consider which readout to use to capture the process of interest. For example, eye-blink startle response has been consistently used for indexing anxiety effects during evoked threat-anticipatory states or contexts, such as variants of threat-of-shock tasks^72,88. In contrast, skin conductance response and pupil dilation may capture phasic responses to onset of cues signaling potential threat, and thus may be particularly useful in paradigms targeting encounter responses^106,284. Temporal changes (e.g., increase or decrease) in skin conductance levels, pupil dilation, muscle tone, and heart rate may be useful in capturing response dynamics during post-encounter epochs in which the threat is becoming increasingly proximal^67,109,282. Different models of neural (BOLD) responses may likewise be differentially suited to capture processes of interest⁴⁴.

In addition to basic and clinical research on normative and pathological defensive response mechanisms, the proposed framework can potentially guide research on treatment for anxiety⁷². For example, cognitive-behavioral therapy (CBT) is a first-line treatment for anxiety but shows variable response rates²⁸⁵, highlighting the need for improving treatment efficacy. Given that CBT emphasizes maladaptive cognitions and behaviors, it may be complemented by a symptom framework that enables patients and therapists to better anticipate, and target, cognitive and behavioral symptoms when these are expected to manifest as a function of imminence of a relevant, identified threat. This approach is in line with therapeutic protocols that integrate different treatment modules depending on symptom presentation²⁸⁶, as different modules may be applied to phase-specific symptoms. Further, it could pave the way for standardized, multi-modal measurements of symptom severity, previously proposed as means to generate more accurate assessment of functioning and treatment response²⁸⁵. Finally, given the conserved nature of defensive responding, considerable translational research aims to improve anxiety treatment through cross-species behavioral and pharmacological research^88,212. Specifying mechanisms that are central to the aberrant expression of defensive responses in each threat-imminence phase could inform research on developing targeted treatments (e.g., drugs that diminish pre-encounter potentiation of encounter responses that may drive further “downstream” excessive responding).

6. Summary

Here, it is posited that anxiety symptoms could be understood as reflecting the excessive, maladaptive expression of otherwise normative threat-anticipatory defensive responses, and as such could be organized by threat imminence. Such an organization could enhance our theoretical understanding of pathological anxiety by providing a testable framework within which different aspects of symptom presentation and psychobiological correlates are hypothesized to occur and interact.

From a clinical perspective, this framework may begin to resolve challenges in anxiety phenomenology (e.g., individuals are not expected to continually express a certain subset of symptoms; rather, different symptoms are expressed over time as a specific threat is becoming, or is perceived as becoming, more imminent), imprecise terminology (e.g., ‘fear’ vs ‘anxiety’), and high comorbidity (since anxiety may reflect a tendency for defensive responsivity, similar symptom expression patterns are expected for different sources of harm). By generating more specific expectations of symptom expression, treatments may be developed to target specific symptoms with greater precision (Box 2).

A symptom framework rooted in partially conserved biological mechanisms also lends itself to translational neuroscience research on pathophysiology (Box 2). Anxiety research in humans may benefit from focusing on symptoms that correspond to conserved patterns of defensive responses as these may track more closely with function in specific neural circuits^9,72. For example, a circuit containing the amygdala, hippocampus, and vmPFC has been robustly linked to defensive responses in animals and humans and to pathological anxiety^{9,234,236,252}, but further refinement is needed in order to link function within it to distinct, symptom-relevant responses. Likewise, induction of context-based pre-encounter states elicits robust insula, ACC, and BNST activation in humans and animals³²; such findings could guide research linking function in this circuitry to pre-encounter symptoms such as hypervigilance and muscle tension¹⁴³. Reliable cross-species biomarkers may then pave the way for development of novel targeted interventions^15,82.

Finally, it should be noted that the proposed framework does not attempt to replace other models of pathological anxiety; rather, it is intended to complement and integrate other conceptualizations of psychopathology that emphasize different theoretical and methodological considerations. Thus, this framework attempts to link the detailed phenomenology offered by psychiatric nosology¹⁰, with biologically-inspired approaches, such as Research Domain Criteria (RDoC)^13,271 which aims to classify psychopathology based on dimensions of observable behavior and neurobiological measures. These perspectives are considered in light of comparative biology and evolutionary perspectives on anxiety^7,59,66,101, as well as cognitive and motivational-behavioral models of anxiety^{31,34,188,203}. Finally, this framework aligns with the rationale behind reductionist classification approaches that attempt to identify core, shared elements among different symptom manifestations, such as the Hierarchical Taxonomy of Psychopathology (HiTOP)²⁷. Together, this framework draws from the many perspectives on anxiety and the unique advantages they each offer.

Highlights.

• Pathological anxiety reflects aberrant expression of defensive responses.

• Translational research shows defensive responses are organized by threat imminence.

• Anxiety symptoms could therefore be linked to specific imminence-dependent responses.

• This novel framework proposes a temporal organization for anxiety symptom expression.

• Grounding anxiety in conserved neural mechanisms, it links clinical and neuroscience research.