# Vascular Clamp

> Vascular Clamp Hypothesis (VCH) — vascular contraction freezes local neural patterns and plasticity for its duration, storing a specific active-inference prediction as vascular tension = medium-term memory. Use when modeling predictions-as-tension, the FEP-AI top-down model store, or the witness/hold step of vasocomputation.

- Skill: `plurigrid/vascular-clamp` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds@latest add plurigrid/vascular-clamp`
- Raw SKILL.md: https://api.skillmd.com/api/skills/plurigrid/vascular-clamp/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- License: MIT
- Author: plurigrid (https://skillmd.com/u/plurigrid)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/plurigrid/vascular-clamp

---


# vascular-clamp

The **Vascular Clamp Hypothesis (VCH)** (Johnson 2023): a vascular contraction **freezes local neural patterns and plasticity for the duration of the contraction** — like collapsing a superposition, clamping a harmonic system, or pinching a critical network into a definite circuit. *Specific vascular constrictions correspond to specific predictions* in the Active Inference framework and function as **medium-term memory**. This is the **0 / witness** leg of the vasocomputation timescale triad: the clamp that *holds* what was collapsed.

## Use When

- Modeling an active-inference prediction stored as a held vascular constriction (the missing FEP-AI top-down store)
- Representing medium-term memory as sustained tension rather than synaptic change
- Bridging "to-do list as predictions" (untrue observations held until acted upon) to a somatic accumulator
- The witness/hold phase between the fast CVH sweep and the durable LHH latch

## Core Concepts

- **Prediction = constriction**: each held prediction is a specific vascular tension; the *pattern* of constrictions is the top-down model.
- **Freeze, don't rewrite**: VCH suspends plasticity locally — the underlying SOHM ground-truth perception is preserved (not corrupted), giving near-infinite capacity for hypotheticals without damaging perception. One system perceives; one clamps/judges/plans.
- **Three discharge routes**: tension is *released by action*, *consolidated by neural annealing*, or *rendered superfluous by remodeling* (held long enough → becomes default → see `neural-potentiation`).
- **Witness, not commitment**: a clamp is reversible and energy-costly to maintain; only a *sustained* clamp engages the latch (→ `latched-hyperprior`).

## GF(3) Balanced Triad

```
compressive-vasomotion (+1) ⊗ vascular-clamp (0) ⊗ latched-hyperprior (−1) = 0 (mod 3)
```

**Skill Trit**: 0 (Witness — the held clamp *is* the witnessed obligation; cf. `EnsureJustification` returning a still-open relation).

## Concomitant Skills

| Skill | Trit | Interface |
|-------|------|-----------|
| `compressive-vasomotion` | +1 | upstream: the sweep that this freezes |
| `latched-hyperprior` | −1 | downstream: sustained clamp → latch |
| `vasocomputation` | +1 | umbrella / vascular substrate |
| `active-inference-robotics` | 0 | predictions-as-operators on a world model |
| `reafference-corollary-discharge` | 0 | held prediction discharged by action |
| `able-markets` | 0 | obligation held until resolved (clearing) |

## Current literature (2024–2026)

- **Hai & Murphy (1988); Hudson et al. (2022, PNAS)** — smooth-muscle latch state: dephosphorylated, non-cycling cross-bridges hold tension at near-zero ATP, longer in tonic (vascular) muscle. The skill's strongest physical anchor.
- **Hartmann et al. (2021, Nat Neurosci; 2022, Annu Rev Physiol)** — capillary **pericytes** set basal tone and modulate flow *slowly* (s–min): kinetics matching "medium-term" storage. Relocate the clamp to pericytes / precapillary sphincters, not fast arterioles.
- **Sirotin & Das (2009, Nature)** — *anticipatory* arterial dilation entrained to expectation, in darkness, **not** predicted by local activity: haemodynamics already carry prediction, not just metabolic demand.
- **Kim et al. (2016, J Neurosci)** — vasculo-neuronal coupling: increased tone *decreases* pyramidal firing — the missing **return arc** by which held tension biases computation.
- **Stokes (2015); Masse et al. (2019)** — "activity-silent" working memory: information held in non-spiking states (precedent for a non-firing substrate).
- **Hook**: two-photon pericyte-Ca²⁺/diameter imaging during a delay task should show content-specific sustained micro-constriction; optogenetic pericyte relaxation should degrade medium-term retention.
- **Grounded**: latch economy, pericyte tone, vasculo-neuronal coupling, anticipatory flow. **Speculative**: a single constriction *freezing* a decodable prediction.

## References

- Johnson, M.E. (2023). *Principles of Vasocomputation, Part I*. opentheory.net (§V, VCH).
- Friston, K. et al. (2017). *Active Inference: A Process Theory*. Neural Computation 29(1).
- Juliani, Safron, Kanai (2023). *Deep CANALs*: SOHMs + vascular tension = inference landscape.