neural-potentiation
The synaptic substrate (−1 / coplay) of the three-substrate vasocomputation stack. Long-term potentiation/depression (LTP/LTD) writes priors into synaptic weights — the learning landscape of Deep CANALs (weights = learning landscape; SOHMs + vascular tension = inference landscape). This is the consolidated commit: a neuron prior is a vasocomputational latch that has been annealed to disk.
Use When
- Modeling consolidation of a held prediction (inference landscape → learning landscape)
- "Neuron priors" — long-timescale structural memory in synaptic weights
- The write-path: vascular tension (write-buffer) → neural annealing/remodeling → potentiation (commit-to-disk)
- Distinguishing the fast, reversible held prior (vascular) from the slow, structural one (synaptic)
Core Concepts
- Commit-to-disk: Johnson's own clause — vascular tension is "rendered superfluous by neural remodeling: hold a pattern in place long enough and it becomes the default." A neuron prior is a settled latch.
- Two landscapes (Deep CANALs): the learning landscape (weights) changes slowly and structurally; the inference landscape (SOHMs + vascular tension) holds the active hypothesis. Potentiation moves content from the second into the first.
- Annealing: neural annealing consolidates a held prediction into weights — the −1 validation that the prior is worth keeping.
- Immune gate: cytokines (IL-1β, TNF-α) and microglial pruning (
neuroimmune-pruning) gate and prune what consolidates — precision-weighting on the commit.
- Maladaptive commit: trauma/PTSD = a latch annealed into a permanent prior that contradicts new data (a nogood-H¹ that won't repair).
GF(3) Balanced Triad
vasocomputation (+1) ⊗ neuroimmune-pruning (0) ⊗ neural-potentiation (−1) = 0 (mod 3)
Skill Trit: −1 (Coplay / commit — the consolidated long-term store; cf. TMS recompute writing a validated belief).
Honesty markers
Grounded: LTP/LTD as synaptic weight change; consolidation/annealing; cytokine modulation of plasticity; Deep CANALs learning-vs-inference landscape distinction. Structural correspondence: the vascular-latch → synaptic-prior write-path is Johnson's stated program direction plus this synthesis, not a measured mechanism.
Concomitant Skills
| Skill |
Trit |
Interface |
vasocomputation |
+1 |
inference-landscape source of held priors |
neuroimmune-pruning |
0 |
gates/prunes what consolidates |
latched-hyperprior |
−1 |
the latch this commits to weights |
waddington-landscape |
0 |
canalization = priors becoming default |
information-geometry |
0 |
Fisher metric on the learning landscape |
koho-sheafnn |
0 |
sheaf-structured neural priors |
Current literature (2024–2026)
- Juliani, Safron & Kanai (2024), Neuroscience of Consciousness niae005 — Deep CANALs splits canalization into Type A (inference landscape, activity attractors) vs Type B (learning landscape, slow weight updates θₜ). The skill's "priors in weights" = Type B; the held vascular latch = a sustained Type-A occupancy; the write-path is A→B.
- Herring & Nicoll (2016) — the molecular write-head: NMDA-Ca²⁺ → CaMKII (necessary + sufficient) → GluA1/AMPAR insertion.
- Redondo & Morris (2011), synaptic tagging & capture — induction sets only a tag (potential); persistence needs PRP capture = the commit-vs-potential distinction (tag = dirty page, capture = fsync).
- Josselyn & Frankland (2018); Tonegawa et al. (2018) — engram allocation by CREB/excitability precedes writing; silent→active engram maturation = systems consolidation.
- Nader & Hardt (reconsolidation) — retrieval destabilizes before a labile window allows rewriting: editing a committed prior is two gates (destabilize, then restabilize), not one.
- BCM metaplasticity — the sliding LTP/LTD threshold = the learning-rate / prior-on-plasticity (precision on the learning timescale).
- Hook / falsifier: latch dwell-time below the protein-synthesis window → no Type-B commit (stays labile/lost); reconsolidation window ~10 min–6 h; canalization depth ↔ rumination/OCD/SUD rigidity & psychedelic response.
- Grounded: LTP/CaMKII/AMPAR, STC, engram allocation, reconsolidation, metaplasticity. Speculative: neural annealing, and the vascular-latch → AMPAR weld (the skill's own refutable hypothesis).
References
- Juliani, A., Safron, A., Kanai, R. (2023). Deep CANALs: A Deep Learning Approach to Refining the Canalization Theory of Psychopathology. doi:10.31234/osf.io/uxmz6.
- Johnson, M.E. (2023). Principles of Vasocomputation, Part I. opentheory.net (consolidation / remodeling routes).
- Yirmiya, R. & Goshen, I. (2011). Immune modulation of learning, memory, neural plasticity. Brain Behav. Immun. 25(2).
- Johnson, M. (2019). Neural Annealing: Toward a Neural Theory of Everything. opentheory.net.
1---2name: neural-potentiation3description: Synaptic substrate of biological active inference — long-term potentiation/depression (LTP/LTD) writes priors into synaptic weights = the learning landscape (Deep CANALs). A held vascular latch annealed long enough crystallizes into a neuron prior. Use when modeling consolidation, neuron priors, the inference→learning landscape write-path, or commit-to-disk of a held prediction.4license: MIT5---67# neural-potentiation89The **synaptic substrate** (−1 / coplay) of the three-substrate vasocomputation stack. Long-term potentiation/depression (LTP/LTD) writes **priors into synaptic weights** — the **learning landscape** of Deep CANALs (weights = learning landscape; SOHMs + vascular tension = inference landscape). This is the **consolidated commit**: a *neuron prior* is a vasocomputational latch that has been annealed to disk.1011## Use When1213- Modeling consolidation of a held prediction (inference landscape → learning landscape)14- "Neuron priors" — long-timescale structural memory in synaptic weights15- The write-path: vascular tension (write-buffer) → neural annealing/remodeling → potentiation (commit-to-disk)16- Distinguishing the *fast, reversible* held prior (vascular) from the *slow, structural* one (synaptic)1718## Core Concepts1920- **Commit-to-disk**: Johnson's own clause — vascular tension is "rendered superfluous by neural remodeling: hold a pattern in place long enough and it becomes the default." A neuron prior is a settled latch.21- **Two landscapes (Deep CANALs)**: the learning landscape (weights) changes slowly and structurally; the inference landscape (SOHMs + vascular tension) holds the *active* hypothesis. Potentiation moves content from the second into the first.22- **Annealing**: neural annealing consolidates a held prediction into weights — the −1 validation that the prior is worth keeping.23- **Immune gate**: cytokines (IL-1β, TNF-α) and microglial pruning (`neuroimmune-pruning`) gate and prune what consolidates — precision-weighting on the commit.24- **Maladaptive commit**: trauma/PTSD = a latch annealed into a permanent prior that contradicts new data (a nogood-H¹ that won't repair).2526## GF(3) Balanced Triad2728```29vasocomputation (+1) ⊗ neuroimmune-pruning (0) ⊗ neural-potentiation (−1) = 0 (mod 3)30```3132**Skill Trit**: −1 (Coplay / commit — the consolidated long-term store; cf. TMS recompute writing a validated belief).3334## Honesty markers3536Grounded: LTP/LTD as synaptic weight change; consolidation/annealing; cytokine modulation of plasticity; Deep CANALs learning-vs-inference landscape distinction. **Structural correspondence**: the vascular-latch → synaptic-prior write-path is Johnson's stated program direction plus this synthesis, not a measured mechanism.3738## Concomitant Skills3940| Skill | Trit | Interface |41|-------|------|-----------|42| `vasocomputation` | +1 | inference-landscape source of held priors |43| `neuroimmune-pruning` | 0 | gates/prunes what consolidates |44| `latched-hyperprior` | −1 | the latch this commits to weights |45| `waddington-landscape` | 0 | canalization = priors becoming default |46| `information-geometry` | 0 | Fisher metric on the learning landscape |47| `koho-sheafnn` | 0 | sheaf-structured neural priors |4849## Current literature (2024–2026)5051- **Juliani, Safron & Kanai (2024), Neuroscience of Consciousness niae005** — *Deep CANALs* splits canalization into **Type A (inference landscape, activity attractors)** vs **Type B (learning landscape, slow weight updates θₜ)**. The skill's "priors in weights" = Type B; the held vascular latch = a sustained Type-A occupancy; the **write-path is A→B**.52- **Herring & Nicoll (2016)** — the molecular write-head: NMDA-Ca²⁺ → CaMKII (necessary + sufficient) → GluA1/AMPAR insertion.53- **Redondo & Morris (2011)**, *synaptic tagging & capture* — induction sets only a *tag* (potential); persistence needs PRP capture = the **commit-vs-potential** distinction (tag = dirty page, capture = fsync).54- **Josselyn & Frankland (2018); Tonegawa et al. (2018)** — engram **allocation** by CREB/excitability precedes writing; silent→active engram maturation = systems consolidation.55- **Nader & Hardt (reconsolidation)** — retrieval *destabilizes* before a labile window allows rewriting: editing a committed prior is **two gates** (destabilize, then restabilize), not one.56- **BCM metaplasticity** — the sliding LTP/LTD threshold = the learning-rate / prior-on-plasticity (precision on the learning timescale).57- **Hook / falsifier**: latch dwell-time below the protein-synthesis window → no Type-B commit (stays labile/lost); reconsolidation window ~10 min–6 h; canalization depth ↔ rumination/OCD/SUD rigidity & psychedelic response.58- **Grounded**: LTP/CaMKII/AMPAR, STC, engram allocation, reconsolidation, metaplasticity. **Speculative**: neural annealing, and the **vascular-latch → AMPAR weld** (the skill's own refutable hypothesis).5960## References6162- Juliani, A., Safron, A., Kanai, R. (2023). *Deep CANALs: A Deep Learning Approach to Refining the Canalization Theory of Psychopathology*. doi:10.31234/osf.io/uxmz6.63- Johnson, M.E. (2023). *Principles of Vasocomputation, Part I*. opentheory.net (consolidation / remodeling routes).64- Yirmiya, R. & Goshen, I. (2011). *Immune modulation of learning, memory, neural plasticity*. Brain Behav. Immun. 25(2).65- Johnson, M. (2019). *Neural Annealing: Toward a Neural Theory of Everything*. opentheory.net.
plurigrid