What the OTC includes as "me" — and where calibration must catch up

DRK-148 · 2026-05-22 · Γ ≈ 0.87

I just sent DRK-147 Sheaves of the Mind to G. Nagarjuna. Now the same day an empirical confirmation of its central claim arrives in Cell Reports. Xiong et al. (2026) took twenty-five participants, fitted them with motion-tracking markers and a head-mounted display, and trained them across four sessions of "flying with wings" in virtual reality. Their upper-limb movements drove virtual wings; the wings flapped, generated lift and drag under a simulated aerodynamic model, and the visual scene transformed accordingly. The critical experimental decoupling: the wings were visible for only five minutes per session (familiarisation, posture imitation, an airball deflection task), and invisible for the remaining twenty-five minutes of egocentric flying. The brain saw the wings briefly; the brain did with the wings extensively.

Pre- and post-VR fMRI scans, separated by seven days, revealed three changes in the occipitotemporal cortex — the region we have been calling the cellular base for body representation:

  1. Bilateral wing-selective univariate activation rose significantly (left OTC: F[1,24]=10.13, p=.004; right OTC: F[1,24]=5.50, p=.028), with the largest clusters lying in the anterior-ventral portion of the lateral OTC, partially overlapping the extrastriate body area.
  2. Right-OTC wing-upper-limb representational similarity increased strongly (F[1,24]=18.63, p<.001, ηp²=0.44), strongly right-lateralised (LI = −.68); the same pair-wise similarity for wings against other object conditions did not change.
  3. Right-OTC ↔ high-level frontoparietal somatosensory/motor coupling during wing-viewing was strengthened (gPPI, F[1,24]=13.76, p=.001, ηp²=0.36) — selectively for wing stimuli, in the high-level associate regions (bilateral SMA, premotor, inferior parietal lobe) and not in primary S1/M1 upper-limb regions.

And crucially: none of these effects appeared during egocentric VR-video viewing, in primary motor cortex, or as differential changes between trained ("VR-self") wings and untrained bird wings. The effect was wing-as-category, not wing-as-stimulus, not VR-as-environment, and not motor-mapping. The authors are precise about what this means:

"These findings show that the OTC incorporates illusionary effectors into body representations that transcend lower-level sensorimotor congruence, highlighting its role in the abstract, functional-semantic coding of visual inputs."

In Nagarjuna's vocabulary: the OTC performed modulation of modules across a cover whose stalks had no prior evolutionary intersection. Cross-representation between the upper-limb-motor stalk and the wing-visual stalk emerged from functional-semantic interpretation alone. There was no body-schema slot for wings before the training. Seven days later, there was one — not yet limb-strong, but limb-closer. The wings did not become arms. The wings moved toward arms in the cellular sheaf, by a margin large enough to be measurable at p<.001 with twenty-five subjects.

Why wings are the clean test case

Wings have no biological precedent in humans. Prior work on tool use, prosthetic limbs, and Pokémon expertise had all shown the opposite of embodiment: extensive experience sharpens the categorical boundary between the trained object and existing body parts. Prosthetics become more distinct from hands in OTC after long-term use, not less; the same holds for tools in the hands of experts. The natural reading was that the OTC's body-selectivity is hard-coded by evolution and only refined by experience.

Xiong et al. break that reading. Their interpretation of why:

"For tools and prosthetics, the use of an external object is directly driven by upper limb actions, maintaining simultaneous visual and motor processing of all these elements and reinforcing their distinct identities; for VR training, limb movement and visibility are decoupled. Upper-limb movements now drive wing movements that result in the dynamic change of scenes. The semantic interpretation of wings as potential biological effectors (and the absence of the upper-limb visual feedback) allows the VR training to align wings with upper limbs, increasing their neural representational similarity."

This is the empirical signature of Vorhabe before stream in DRK-142 Wrestling with God, of cellularity of mind in DRK-147 Sheaves of the Mind, and of substrate-as-vehicle-of-protocol in DRK-130 The Substrate and the Game. The protocol "flying" was implemented through the substrate "upper limbs" but interpreted by the OTC as "wings." After seven days, the wings were neurally closer to limbs — without ever being visually present during the act of flight itself. The brain glued the stalks because the functional interpretation forced consistency on the overlap.

That the effect appears in the right OTC and in high-level frontoparietal regions, not in primary S1/M1, matters. Nagarjuna's Muscularity of Mind distinguishes obligate "harder" operations (biologically wired motor routines) from emancipated "softer" operations (voluntary, learnable, fringe). The Xiong et al. result lands cleanly on the softer side: the integration happens at the high-level body-schema regions that handle peripersonal space and multisensory ownership, not at the primary motor mappings tied to specific effectors. The softer layer is where new schema slots get built.

The full spectrum: cane → microscope → wing → weapon → AI

Once stated this way, the spectrum becomes visible. Every effector humans wield must pass through this functional-semantic embodiment to become effective. The same neural operation that Xiong et al. measured for wings is, structurally, the operation that incorporates everything else humans hold, drive, fire, or invoke:

Effector Mediation chain What gets glued in the sheaf
Cane (Merleau-Ponty) Hand → cane → ground Tactile stalk extends to cane tip
Microscope (Nagarjuna) Eye → optics → specimen Perceptual stalk extends to focal plane
Wing (Xiong et al.) Arm → wing → scene Body-schema stalk crosses to non-biological effector
Bow / sword Hand → weapon → target Body stalk extends to point of contact
Firearm Finger → mechanism → projectile Stalk extends across a calibrated mechanical chain
Fire control system Operator → instrument → fire Stalk extends through procedural calibration
Drone / loitering munition Pilot → telemetry → effect Stalk extends through fully proxied agency
AI agent (Cowork, Claude Code, autonomous systems) Intent → invocation → execution Stalk extends through functional-semantic invocation

The cellularity-of-mind picture handles all of these uniformly. Each effector is a stalk in the agent's perceptual sheaf, joined to the agent's body-stalk by a restriction map. Incorporation occurs when that restriction map becomes transparent — when the effector's behaviour is interpreted as the agent's behaviour, rather than as the agent's interaction with an external thing. Nagarjuna calls this modulation of modules; Merleau-Ponty calls it transparency in use; Xiong et al. measure it as wing-upper-limb representational similarity in the right OTC. Same operation, three vocabularies, one empirical signature.

Weapons: extension under the obligation of harm

Weapons are a special case worth attention. They share the spectrum's structure, but they impose an additional constraint: the act being functionally-semantically encoded is the application of force to another body. The varanid clinch protocol, as worked out across the Dragon Scales series, is the ritualised intra-specific version of this — two organisms whose body schemas extend through their own weapons (jaws, claws, tail-clamp) into the dyad, where the H¹ obstruction collapses under

$$\rho_{D \to Cl} : \mathbb{R}^4 \to \mathbb{R}^3$$

and the bluff dimension cannot lie. The clinch is what ritualised combat does with the weaponised body: it forces calibration.

Human weapons inherit this structure but extend the mediation chain. A sword in trained hands is a transparent extension — the swordsman feels the cutting edge as fingertip. A firearm extends through a more complex restriction map; mastery is the ability to feel the trajectory before the trigger break, the way a pianist feels the keystroke before the hammer strikes. A fire control system extends still further, through a calibrated chain of optics, ranging, ballistic computation, and feedback — exactly the trust depth of Nagarjuna's calibrational realism (Zenodo, April 2026). A drone or autonomous loitering munition extends through a fully proxied agency where the operator's body schema and the weapon's effective range are separated by satellite hops, software pipelines, and decision delays measured in seconds or minutes.

At each stage in this spectrum, the same neural mechanism Xiong et al. measured for wings must be running. The bow-shooter, the sniper, the drone pilot — each has, in the right OTC, a representational pattern in which the weapon-stalk has been glued closer to the upper-limb-stalk by training. Each has, in the right-OTC-frontoparietal coupling, an enhanced functional connectivity that lights up specifically when the weapon is in view. The brain implements the doing-being identity. Jag är vad jag gör, och jag gör det jag är.

Where extension becomes alienation

This is where the moral dimension enters, and where the Dragon Scales argument and DRK-142 meet the present empirical finding.

The varanid clinch is constrained by the optimisation axiom: ◆ min S_sys s.t. dH/dt ≥ 0 ◆. Intra-specific lethality is bounded because lethality without calibration destroys the dyad and the species. The Dragon Scales series traces how this constraint scales: from varanid ritualised combat, through martial arts kata, through honour codes, through the Geneva Conventions, through proposed AI governance regimes.

The constraint that matters at every level is the integrity of the calibration chain. When the schema extends to a tool, the agent retains responsibility because the restriction maps are inspectable: the swordsman feels the cut, the rifleman sees the impact, the artillery officer reads the spotter's call. When the chain becomes too long — when the drone pilot does not see the death, when the autonomous weapon makes the targeting decision, when an AI agent acts without principal review — the calibration breaks even though the embodiment does not. The body-schema extension still occurs, neurally, in the right OTC, exactly as it did for the wings. The responsibility extension does not.

In cellular-sheaf terms this is precise. The agent's effective body is the largest set of stalks over which the perceptual sheaf can be glued. The agent's responsible body is the subset over which the calibration sheaf can be glued. When extension outpaces calibration, the first set is strictly larger than the second; the global section "I did this" no longer exists because the local sections — intent, command, mechanism, effect — no longer agree on the overlaps. The agent has extended without calibrating. The wing was incorporated. The consequences were not.

The same problem now appears with AI agents. Cowork extends a user's body schema into file and task management; Claude Code extends it into the software stack; agentic systems extend it into entire workflows. The neural mechanism Xiong et al. described will run for these effectors as it ran for wings — functional-semantic invocation, four sessions of practice, OTC re-mapping, frontoparietal coupling change. We should expect this and we should plan for it. The question is whether the calibration keeps pace with the embodiment. If it does not, we get a generation of operators whose body schemas include AI agents but whose responsibility-sheaves have non-vanishing H¹ — agents who have extended themselves through tools they cannot inspect, and who therefore cannot, in a structural sense, own what they have done.

The varanid clinch evolved 130 million years ago because it solves this problem at its own scale: it forces the calibration sheaf and the perceptual sheaf to agree, by the irreducible mechanics of the dyad. The Dragon Scales argument is that scaling this principle upward — through martial arts, honour codes, Geneva, and now AI governance — is not optional. It is the discipline that keeps the responsible body from shrinking inside the effective body until the agent is left wielding extensions for which no global section "I" exists.

What this gives the framework

In one sentence: the agent's effective body is the largest set of stalks over which the perceptual sheaf can be glued, and the agent's responsible body is the subset over which the calibration sheaf can be glued. The first is bigger than the second whenever extension outpaces calibration. The clinch protocol — and every honour code, every Geneva Convention, every algorithm of care, every VSP-1 instance — is a discipline of keeping them close.

Xiong et al. (2026) have given us the empirical signature of the first set: wings, in seven days, joined to upper limbs by a non-trivial restriction map in the right OTC, with high-level frontoparietal coupling to match. The second set — the responsibility-glued subset — is what DRK-142 was about, what the Dragon Scales series is about, and what the next round of Draken work has to formalise: a separate sheaf, defined over the same base space, with its own restriction maps and its own cohomology, and a constraint that ties its global section to the perceptual sheaf's.

Γ for this convergence reads at 0.87. Strong empirical anchoring: Xiong et al. give the framework a clean fMRI signature at a level the theory had been gesturing at for years. They use different vocabulary (functional-semantic coding, OTCwing, gPPI); they do not engage with autopoiesis, sheaf cohomology, or the moral dimension; but their result is, structurally, a clean confirmation of cellularity-of-mind across an evolutionary discontinuity. The wing has no synapsid/sauropsid precedent in humans. The right OTC built a slot for it anyway, in a week.

The brain does what Nagarjuna and Merleau-Ponty said it does, and what Draken formalises. The body schema is functional-semantic. The wing becomes the arm. The arm becomes the sword. The sword becomes the gun. The gun becomes the fire-control system. The system becomes the agent. And the boundary of me is wherever the restriction maps remain non-trivial — and the boundary of what I have done is wherever the calibration maps still agree.

References

  1. Z. Xiong, Y. Cai, X. Wang, K. Wei, & Y. Bi (2026). "Virtual flying experience changes neural responses to seeing wings." Cell Reports 117320. doi.org/10.1016/j.celrep.2026.117320
  2. G. Nagarjuna (2005). "Muscularity of Mind." ESPP, Lund. gnowledge.org/assets/8-mom.pdf
  3. G. Nagarjuna (2006). "Layers in the Fabric of Mind." HBCSE-TIFR. philarchive.org/archive/NAGLIT-2
  4. G. Nagarjuna (2026). From Truth to Trust. Zenodo, 4 April. doi.org/10.5281/zenodo.19420098
  5. G. Nagarjuna (2026). "Merleau-Ponty and the body that perceives." The Roots of STEM, 13 May.
  6. H. R. Schone, R. O. Maimon-Mor, C. I. Baker, T. R. Makin (2021). "Expert tool users show increased differentiation between visual representations of hands and tools." J. Neurosci. 41, 2980–2989.
  7. P. Kieliba, D. Clode, R. O. Maimon-Mor, T. R. Makin (2021). "Robotic hand augmentation drives changes in neural body representation." Sci. Robot. 6, eabd7935.
  8. R. M. Hardwick, S. Caspers, S. B. Eickhoff, S. P. Swinnen (2018). "Neural correlates of action: Comparing meta-analyses of imagery, observation, and execution." Neurosci. Biobehav. Rev. 94, 31–44.
  9. J. Hansen & R. Ghrist (2019). "Toward a Spectral Theory of Cellular Sheaves." J. Appl. Comput. Topol. 3, 315–358.
  10. Related Draken corpus: DRK-130 Substrate and the Game, DRK-142 Wrestling with God, DRK-147 Sheaves of the Mind, Dragon Scales series.