paper
active
paper:johnson-vasocomputation-2023

Johnson Vasocomputation 2023

Active InferenceManifold geometry for steering networksManifold geometry for neural network steeringNormative predictionsActive InferencePhysical brain resonancesBuddhist phenomenologyVascular contraction as pattern freezingVasocomputationWith one motion the door of possibility slams shut. And so we collapse our world into something less magical but more manageable, one clench at a time.

TL;DR

Vascular smooth muscle cells (VSMCs) are the brain's physical compression and prediction infrastructure — the substrate where top-down predictive models are stored — and Buddhist tanha (craving/grasping) is the phenomenological signature of this machinery operating unskillfully. Johnson advances three nested hypotheses: the Compressive Vasomotion Hypothesis (CVH), which proposes that vasomotion reflexively collapses ambivalent neural resonances into durable definite states; the Vascular Clamp Hypothesis (VCH), which frames specific vascular contractions as medium-term memory encoding specific active inference predictions; and the Latched Hyperprior Hypothesis (LHH), which invokes the latch-bridge mechanism of smooth muscle — where myosin heads lock onto actin without ongoing energy expenditure, persisting minutes to years — as the physical basis of durable hyperprior commitment. Daniel Ingram's observation that the tanha grab can occur within 25–50ms and Nick Cammarata's framing of it as a "fast grabby thing" within ~100ms are recruited as phenomenological evidence that the mechanism is crisp and isolatable rather than distributed across thousands of factors. Adam Safron's Self-Organized Harmonic Modes (SOHMs) from the 2020 Integrated World Modeling Theory serve as the intermediate layer: neural resonances that VSMCs selectively freeze or suppress. Tanha-as-Unskillful-Active-Inference (TUAI) fails in four specific modes — rate overload, uncontrollable-domain prediction, context desynchronization, and metabolic dysfunction. The paper argues this unification implies that the sum of all VSMC latches accounts for the majority of bodily suffering, that migraines and cluster headaches are instances of inappropriate latching, and that contemplative progress is mechanistically progress in using vascular clenching less — meaning Buddhist cessations correspond to near-complete absence of vasomotion.

What to take away

  1. 1. The Latched Hyperprior Hypothesis (LHH) proposes that VSMC latch-bridge contractions — which lock myosin heads to actin without ongoing energy and can persist from minutes to years — physically encode durable hyperpriors that are isolated from conscious updating until the corresponding prediction resolves.
  2. 2. The Compressive Vasomotion Hypothesis (CVH) frames the vasomotion reflex as a compression sweep on nearby neural resonances, collapsing ambivalent patterns into definite states, such that the motifs of vasomotion are equivalent to the motifs of tanha.
  3. 3. Phenomenological timing data from Daniel Ingram (25–50ms) and Nick Cammarata (~100ms) constrains the tanha mechanism to a fast, isolatable reflex rather than a diffuse cortical process, consistent with a vascular rather than purely synaptic substrate.
  4. 4. Tanha-as-Unskillful-Active-Inference (TUAI) fails in four distinct modes: rate overload of normative predictions, prediction of uncontrollable outcomes, context desynchronization between world model and prediction basis, and metabolic dysfunction of the prediction machinery itself.
  5. 5. Adam Safron's Self-Organized Harmonic Modes (SOHMs, from the 2020 Integrated World Modeling Theory) function as symmetry detectors/autoencoders organized around physical brain resonances, and Johnson distinguishes the learning landscape (neural weights + SOHMs) from the inference landscape (SOHMs + vascular tension).
  6. 6. The Vascular Clamp Hypothesis (VCH) assigns VSMCs a medium-term memory role: specific contractions freeze local neural patterns and plasticity for the duration of the contraction, instantiating specific predictions in the Friston et al. (2017) active inference framework.
  7. 7. The April 2025 update refines the conceptual mapping by distinguishing tanha (default computational-biochemical bias) from upādāna (impulsive physical VSMC clenching), with the three Buddhist defilements — greed, fear, delusion — each mapping to a distinct failure mode of vasocomputation-based active inference.
  8. 8. Latch spirals represent a self-reinforcing pathological loop in which a VSMC latch reduces local blood flow, which reduces local energy availability, which prevents latch release — a mechanism proposed to explain persistent pain and conditions such as migraines and cluster headaches.
  9. 9. A replication-ready methodological choice is the use of vasomotion motif libraries as a proxy for tanha motif libraries: researchers could record peripheral or cerebrovascular vasomotion patterns during controlled phenomenological probes (e.g., Shinzen Young's noting practice) to test whether vasomotion signatures co-vary with reported craving intensity.
  10. 10. An open question the paper raises is whether contemplative cessations — operationalized as near-complete absence of vasomotion — can be detected non-invasively via fMRI or Doppler ultrasound in experienced meditators, which would constitute a falsifiable empirical test of the entire vasocomputation framework.

Peer brief — for seminar discussion

Johnson's 2023 essay (updated April 2025) proposes that vascular smooth muscle cells are the brain's primary compression and prediction substrate, and that Buddhist tanha — the craving-grasping reflex Buddhist consensus attributes to roughly 90% of suffering — is the phenomenological correlate of this vascular machinery operating outside its normative envelope. The paper introduces vasocomputation as a unifying construct and advances three hierarchically nested mechanistic hypotheses. The Compressive Vasomotion Hypothesis holds that vasomotion reflexively collapses ambivalent neural resonances into durable definite states. The Vascular Clamp Hypothesis holds that specific VSMC contractions freeze local neural patterns and plasticity, functioning as medium-term memory encoding active inference predictions in the sense of Friston et al.'s 2017 process theory. The Latched Hyperprior Hypothesis is the most ambitious: it invokes the smooth muscle latch-bridge mechanism — myosin heads locking onto actin without ongoing ATP expenditure, stable from minutes to years — as a physical substrate for durable hyperprior commitment that is isolated from conscious experience and global updating until the corresponding prediction resolves. The intermediate computational layer is Adam Safron's Self-Organized Harmonic Modes (SOHMs) from the 2020 Integrated World Modeling Theory, which Johnson uses to distinguish a learning landscape (neural weights + SOHMs) from an inference landscape (SOHMs + vascular tension). Phenomenological timing constraints from Daniel Ingram (25–50ms tanha grab) and Nick Cammarata (~100ms) are used to argue the mechanism is crisp and vascular rather than diffusely cortical. The April 2025 update sharpens the Buddhist mapping by distinguishing tanha (computational-biochemical bias) from upādāna (impulsive physical clenching), and maps greed, fear, and delusion onto three distinct TUAI failure modes. The load-bearing prediction is that Buddhist cessations correspond mechanistically to near-complete absence of vasomotion, which is in principle testable via cerebrovascular Doppler or fMRI in experienced meditators — an experiment the paper does not conduct. A critical reader would press hardest on the evidential gap between the latch-bridge mechanism (well-established in smooth muscle physiology) and its proposed cognitive-computational role: the paper offers no direct measurement of VSMC activity during meditation, tanha episodes, or migraine onset, relying instead on theoretical coherence across Friston's free energy principle, Safron's IWMT, Erik Hoel's 2020 Overfitted Brain hypothesis, and phenomenological reports. The alternative methodological path — using peripheral vasomotion as a proxy and correlating it with real-time phenomenological probes during structured noting practice — is implicit in the framework but not pursued. Whether smooth muscle motif libraries are rich enough to encode the semantic specificity active inference requires of hyperpriors remains an open question the paper acknowledges but does not resolve.

Frameworks (2)

  • Active Inference
    Foundational framework by Karl Friston; the paper extends it to three hierarchical levels for modeling meta-awareness.
  • Buddhist phenomenology
    The systematic first-person investigation of experience as practiced in Buddhism, particularly regarding tanha and dependent origination.

Claims (18)

Hypotheses (6)

Questions (1)

Related work— refs + corpus + external arXiv

Cited / in-corpus / arXiv badges show which signals surfaced each row. Multi-source rows weighted higher.

Similar preprints — Semantic Scholar