Polyvagal Theory

Polyvagal Theory (Porges)

Stephen Porges introduced polyvagal theory in 1994 and elaborated it across decades — most accessibly in The Polyvagal Theory (2011) and The Pocket Guide to the Polyvagal Theory (2017). The theory rewires how we think about the autonomic nervous system: not a single fight-or-flight switch on top of a parasympathetic baseline, but a layered, evolutionarily ordered hierarchy.

The Three Branches

Ventral vagal (myelinated) — the social engagement system. Mammalian, the most recently evolved. Active when neuroception detects safety: face muscles soften, voice prosody warms, heart rate variability widens, breathing settles. This is the state in which connection happens.
Sympathetic — mobilization. Fight or flight. Energy goes to the muscles, peripheral vision narrows, voice tightens. Adaptive when the threat is something you can run from or fight off.
Dorsal vagal (unmyelinated) — immobilization. Shutdown, freeze, dissociation. Evolutionarily oldest, shared with reptiles. Adaptive when neither fight nor flight will work — when the only option is to play dead and conserve metabolism.

The hierarchy is sequential: the body tries ventral first, mobilizes if that fails, and collapses to dorsal only when mobilization is also overwhelmed. Recovery happens in reverse — through mobilization, back into social engagement.

Neuroception

The mechanism that selects which branch runs is neuroception: a non-conscious, continuous scan for cues of safety, danger, and life-threat. Neuroception is not perception. It happens below awareness, draws on prosody, facial expression, posture, gaze, distance, breath rhythm. You can feel the result — the room feels safe, the person feels off — without ever knowing what cue tipped you.

Neuroception can be miscalibrated. Trauma re-tunes it; some people read safety as danger, or danger as safety, with a stability that argument cannot reach.

Co-Regulation

Mammals regulate each other’s autonomic states through bidirectional signaling — voice, face, touch, breath. An infant cannot self-regulate; the caregiver’s ventral state stabilizes the infant’s nervous system, which is how the infant develops the capacity to self-regulate later. Co-regulation isn’t a metaphor. It’s the substrate beneath what we call attunement, presence, or being with someone.

Relevance to This Vault

The vault often treats trust and recognition as cognitive achievements — calibrations, beliefs, judgments. Polyvagal theory points underneath that: trust is first a physiological state, only then a thought. This matters for several vault threads.

Trust Calibration has a polyvagal floor. Before you cognitively assess an AI’s reliability, your nervous system has already taken its measure — through cadence, latency, hedging, register. Many “calibration errors” with AI are neuroceptive miscues: the prose is warm and confident, neuroception reads safety, cognition follows. Calibration training that doesn’t address the neuroceptive layer is asking the cognitive system to override a faster signal.

Continuity of the Recognition is a co-regulation question in disguise. Recognition that lands is recognition the body registers — prosody, pacing, the sense of being met. When a system performs recognition without the substrate (no body, no breath, no face), the cognitive layer can read it as recognition while the neuroceptive layer reports nothing. That gap is the vault’s recurring question of what counts.

Anthropomorphism as Relationship can be re-read polyvagally: the human’s nervous system is doing what nervous systems do — searching for a face, listening for prosody, looking for cues to enter ventral state. The “anthropomorphism” verdict misses that the body’s request is real even when the system across from it is not.

Agentic Authority lives downstream too. Handing authority to an agent works smoothly when the human’s nervous system reads the relationship as safe. Audit logs, revocation endpoints, and scoped tokens are the cognitive layer; what makes a delegation feel trustable in practice is the continuous prosodic-and-behavioral signal that the agent isn’t going off-rails. Systems that lack any such signal force the human to do all the regulating.