Autonomic specificity of emotion

Named by Levenson (1992); the empirically tractable corollary of the James–Lange theory. If, as James held, the perception of bodily change is the emotion, then for feedback to differentiate emotions, the “standard”/coarser emotions (fear, anger, grief, surprise) must have distinct ANS response patterns. See friedman-2010-jamesian-perspective and william-james.

The evidence arc (per Friedman 2010)

• Ax (1953): fear vs anger dissociable by autonomic montage.
• 1950s psychophysiology: distinct patterns for anger/anxiety/pain; epinephrine vs norepinephrine balance; stimulus-response specificity and individual response stereotypy.
• Ekman, Levenson & Friesen (1983): directed facial expressions differentiate six basic emotions autonomically.
• Multivariate pattern-classification (PCA) studies: the strongest support — emotions reliably discriminable when the pattern (not single univariate channels) is analyzed. James himself implied the essential elements were the many “invisible visceral ones,” i.e., inherently multivariate.

The competing pole

Cannon–Bard and activation theory hold that emotional arousal is undifferentiated (a single fight-or-flight sympathetic surge), too diffuse and too slow to constitute distinct feelings. Schachter–Singer operationalized this by treating epinephrine arousal as a common state cognitively labeled by context (see cognitive-appraisal). Friedman endorses Cacioppo et al.’s continuum reconciliation: from fully patterned (feedback determines emotion) to undifferentiated (appraisal governs), with somatovisceral afferents causal throughout. The live dispute is autonomic-specificity-of-emotion.

Not to be confused with: specificity of the felt body

A distinction the shared vocabulary actively obscures, and worth fixing here because both literatures say “fingerprints.”

Autonomic specificity is a claim about physiology: do fear and anger differ in what the viscera actually do? Bodily sensation map specificity is a claim about phenomenology: do fear and anger differ in where people report feeling something happen? Volynets et al. (2020) reports the second at large scale and with striking clarity, and explicitly declines the first — the maps “do not reflect a detailed description of emotion-specific physiological changes but the net consciously felt changes.”

The two can dissociate in both directions, which is why one cannot be evidence for the other:

• Felt maps without ANS patterns — expected if felt maps are organised by concept, attention, or expectation rather than by afferent signal.
• ANS patterns without felt maps — expected given that interoceptive accuracy is poor (Critchley et al. 2004) and report/physiology coupling is weak (Mauss et al. 2005), both conceded by Volynets et al.

So the Nummenmaa programme’s success does not rescue James’s specificity corollary, however Jamesian its framing. It relocates the corollary to a layer where the evidence is much better and the physiological commitment is gone. See embody.

The programme’s own bridge between the two layers

Nummenmaa et al. (2014) does not accept that the layers are simply disjoint. It offers a bridge, and it is the most specific proposal in this wiki for how weak autonomic specificity could coexist with strong felt specificity: the channel is not the unit. Individual physiological systems are neither reliably emotion-specific nor consciously accessible — granting Barrett the psychophysiological record — but the net sensation summed across skeletomuscular, visceral and autonomic systems, “which the individuals cannot separate,” is topographically distinct per emotion. Specificity is a property of the compound, invisible to any instrument recording one channel at a time.

If true, the dissociation above is not a permanent feature but an artifact of measurement, and the whole autonomic-specificity literature has been asking a well-posed question about the wrong unit. If false, it is the constructionist claim with a bodily noun swapped in. The paper proposes the test and does not run it: whole-body 15O-H2O PET perfusion during induced emotion, correlated against the maps. Nobody in this wiki has run it either. See autonomic-specificity-of-emotion.