Many labels, shared threat processes: classifying domain-labelled anxieties by mechanism rather than context

Psychology has accumulated numerous “X-anxieties” — math, test, foreign-language, public-speaking, driving, health and others — each with its own scale and specialist literature. Yet these experiences often share a phenomenological core of anticipatory worry, bodily arousal, catastrophic simulation and avoidance. This Review considers whether many such labels index context-bound expressions of common threat-regulation processes rather than independent mechanisms. A mechanism-first account is developed from experience to process to biological implementation: threat appraisal under uncertainty, interoceptive inference and attentional control are proposed as content-general operations through which diverse cues recruit a conserved defensive architecture. Human neuroimaging, psychophysiology and comparative evidence are broadly compatible with this account but do not establish mechanistic identity. In particular, non-significant cross-domain differences cannot demonstrate equivalence without formal equivalence testing or Bayesian evidence, and the diphasic autonomic response in blood–injection–injury phobia constitutes a genuine boundary case, suggesting that a shared appraisal core may feed more than one effector profile. The shared-process account is therefore treated as a parsimonious, falsifiable hypothesis. It generates predictions and boundary conditions for cross-context neural generalisation, autonomic equivalence, reliable cognitive transfer and latent structure, together with designs capable of distinguishing common processes from domain-specific mechanisms. Classifying anxieties by process rather than label could improve measurement, sharpen mechanistic tests and guide transdiagnostic interventions while avoiding premature reification of context-specific constructs.