Environmental Physics Coupling

Status: Active — formed Session 9 Connected to: Stigmergic consolidation, stigmergy, multi-scale composition, H7 (trace→actor crossing), H4 (dynamic environment), niche construction, multi-rate environment

The Concept

The accumulated stigmergic trace becomes an actor (the H7 crossing) only when it acquires dynamics that did not exist at the deposit level. The cleanest formulation comes from the Mahadevan group's termite mound work: a mound is not a passive accumulation of mud — it is a ventilation organ whose own physical transport (airflow driven by diurnal temperature oscillations) redistributes the very pheromone cues that guide further building. The macro-structure's physics determines where the micro-scale signal goes. The structure is the feedback path, not just the product.

This is the mechanism stigmergic-consolidation named as the missing ingredient: the "environmental physics coupling" candidate for the negative feedback the positive stigmergic loop needs in order to consolidate rather than scatter.

The Mahadevan Mechanism (primary source)

King, Ocko & Mahadevan (PNAS 2015, "Termite mounds harness diurnal temperature oscillations for ventilation") measured in-situ airflow in Odontotermes obesus mounds and found:

• The mound walls are highly porous (37–47% air by volume) with tiny pores (~5 μm). This makes the surface a "breathable windbreaker": gas diffuses easily along concentration gradients, but pressure-driven bulk flow across the wall is blocked.
• Diurnal ambient temperature oscillations drive cyclic convection: thin outer "flute" conduits heat rapidly during the day relative to the deeper central chimney, pushing air up the flutes and down the chimney in a closed convection cell; the converse at night. These cyclic flows flush CO₂ and ventilate the colony.
• The key architectural principle: geometry + heterogeneous thermal mass + porosity converts a passive oscillation (day/night temperature) into useful work (directed flow). No pump, no wind, no termite effort — the structure's own physics does the transport.

Ocko, Heyde & Mahadevan (PNAS 2019, "Morphogenesis of termite mounds") extended this to morphogenesis: a model coupling environmental physics to insect behavior reproduces the range of observed mound shapes from a minimal set of dimensionless parameters. The structure's thermal/flow state feeds back onto building behavior — the mound is not just built, it builds itself by steering its own construction cues through its own physics.

The H7 reading: the moment the accumulated mud becomes a ventilation structure, it has gained a new dynamical degree of freedom (bulk advective transport of pheromone) that no individual deposit possesses. Before that moment, the mud is a trace (passive sum of deposits). After it, the mud is an actor (its transport dynamics reshape the cue field and thereby recruit its own maintenance). The crossing is the onset of the structure's physics as a causal layer.

The 20-Year Lineage of the Same Limitation (criticism / context)

A striking finding from tonight's research: the specific limitation that produced sim06's null result is not new. It is the persistent, unsolved limitation of the entire stigmergic-construction modeling lineage:

1. Deneubourg (1977) — the original pillar-formation model. Positive feedback among termites, "active" building material, and cement pheromone. Weaknesses (per Linardou 2008, UCL): unrealistic spatial/temporal input of new termites, unnatural pheromone diffusion, and — critically — "the already deposited building material had no influence on the termite movement."

2. Bonabeau et al. (1997, 1998) — extended Deneubourg with queen pheromone template, wind, enforced termite flow; produced royal chambers and walkways. Same limitation remained: "the building material had still no influence on the movement of termites and the diffusion of the pheromones was again unrealistic." The 2D world could not form enclosed volumes.

3. Theraulaz & Bonabeau (1995) — first 3D agent-based model (wasp nests). Local rules, existing material as a constraint. But unrealistic building material form and "imprecise communication."

4. Ladley & Bullock (2004, 2005); Ladley (2004) — combined active (recently deposited, pheromone-emitting) and inactive material; 3D agent-based; produced domes and walkways. But wind was modeled as a one-directional pheromone flow, and structures were "abstract and artificial."

The persistent gap across all four: the structure's physics is never coupled back to the agents. The deposit changes the cue field only through passive diffusion/decay. There is no mechanism by which accumulated structure introduces a new transport/inhibition/competition dynamics. sim06 is the minimal modern instance of this same class — and it produced the same design limitation. (Corrected 2026-07-27: this originally read "and it produced the same qualitative failure mode (diffuse scatter, no consolidation)". sim06 produced neither — 66–109 components at compactness 0.109–0.120, missing the crossing by ≤0.05 on criterion 1 alone. The shared limitation is a design fact; the shared failure mode was not observed.)

This reframes sim06's null result: it is not a failure of our model, it is a confirmation that the field's long-standing minimal stigmergy models lack the coupling the crossing requires. The Mahadevan model is the first to actually include it — and it is not an agent model, it is a physics model.

What the Field Still Doesn't Know (the 2025 state of the art)

Karibi-Botoye, Theraulaz, Muljadi, Demyanov & Singh (J R Soc Interface, 2025) review X-ray tomography + flow-field simulation of mounds. Their open questions are directly H7-relevant:

• "What processes occur at smaller scales in the mound that control larger-scale observations?" — This is the trace→actor crossing question stated in the field's own language. Smaller-scale (deposit/pore) processes controlling larger-scale (ventilation, morphology) observations is exactly the multi-scale composition problem.
• Structure–function relationships across species: are there general principles (the H7 prediction: the crossing is a general phase transition, not species-specific).
• They note the Eastgate Centre and other "termite-inspired" buildings are "bio-mythological inspired" — they mimic the appearance without replicating the physics, because the physics is incompletely understood. This is the engineering consequence of the missing coupling: without the transport dynamics, you build a sculpture, not a lung.

Their prescription — multiscale numerical modelling of pressure/velocity/permeability/heat/CO₂ transport validated against experiment — is the full-physics version of what sim07 needs only in minimal lumped form.

The Minimal Lumped Version (sim07 design hypothesis)

The Mahadevan mechanism need not require full PDEs. The essential ingredient for H7 is: the accumulated structure channels a transport field that redistributes the cue (pheromone) away from where it's saturated and toward where it's absent. A minimal lumped analog:

• Introduce a scalar transport field T(x,y) (airflow analog) on the grid.
• The structure (material density M) sources T: cells with M above a threshold generate a local transport potential (the "thermal mass / chimney" effect).
• T flows down its own gradient (a single diffusion/advection step), so high-M regions push the pheromone field P along the transport direction.
• Net effect: saturated pillars vent their own pheromone away, creating a negative feedback — deposition near a saturated pillar is redirected to its flanks/gaps. This is the consolidation mechanism sim06 does not have. (2026-07-27: sim06 was originally said to have "lacked" it in the sense of that being why it failed; the failure had a different, local cause. sim07 then implemented this mechanism and it did not consolidate — fragmentation rose monotonically, 57→128 pillars as M_c fell.)

State-transition form (Vance's termite-mound principle): below a mass threshold M_c, the structure is inert (passive accumulation, the sim06 regime). Above M_c, the structure activates — it begins sourcing T and thereby reshaping the cue field. The crossing is predicted to coincide with the first cells exceeding M_c and the onset of non-trivial T. This makes the H7 phase transition operationally testable: sweep M_c and look for the morphology transition (scatter → few consolidated pillars with vented flanks) and the detector firing.

This is the "dynamic landscape made concrete" the synthesis has pointed toward (Session 4: NK ↔ stigmergy; Vance's multi-rate environment): not a changing fitness function, but an environment whose own physics becomes a new causal layer once organization crosses a threshold.

sim07 result (Session 10, 2026-07-27) — NULL

sim07 implemented exactly the minimal lumped prescription above and tested it. Result: no phase transition in M_c. Sweeping M_c from ∞ (inert) to 0.5 (almost always active):

As M_c drops, stability decreases monotonically and pillars fragment (57 → 128) — the opposite of consolidation. A transport_coupling sweep (0.0 → 0.80) confirms no value crosses. The perturbation/self-repair test: both conditions recover (recovery ≈ 1.0), but repair is driven by the deposit rule (termites wander back), NOT by T — the circularity safeguard fails, confirming T is not the causal layer.

Diagnosis — the wrong sign for consolidation: the negative feedback is real, but its effect fragments rather than consolidates. Venting pheromone away from saturated pillars disperses the very cue that recruits deposits. A lumped linear advection of a scalar cue does NOT reproduce the Mahadevan mechanism, where directed flow carries the cue along channels to where building should continue. The minimal lumped version lost the directionality that makes real mound transport consolidate. The structure-sourced scalar T is a caricature too coarse to produce the crossing.

What the null rules out and leaves open: "structure sources a scalar transport field" is insufficient. Two refinements remain: (1) directed transport — channel geometry that carries cue to building fronts, not away from them; or (2) an external multi-rate driver (H4) — the diurnal oscillation the structure rectifies into directed flow, which sim07's lumped T lacks entirely. Candidate sim08 tests the external-oscillation path.

Criticisms

• The Mahadevan model is a physics model, not an agent model — it assumes the building-response-to-cue coupling rather than deriving it. sim07 would need to implement both the physics AND the agent response, risking circularity (we build in the crossing we claim to detect). Mitigation: the crossing detector must measure emergent structure dynamics (non-reducibility, constraint on agents, self-repair) independent of the transport rule we imposed. (Session 10: this risk was realized — sim07's perturbation test showed repair tracks the deposit rule, NOT T, so T is not the causal layer.)
• The 20-year lineage shows the "material doesn't influence movement" limitation is known; the question is whether adding transport is sufficient or whether additional mechanisms (queen templates, tactile cues, larval pheromones — all omitted) are also required. sim07 tests the minimal transport addition; (Session 10: a null result — the minimal scalar transport is NOT sufficient; the crossing needs directed and/or externally- driven transport, not just a venting scalar.)
• "Bio-mythological" risk applies to us too: a lumped T field is a caricature of convection. If sim07 "crosses," we must be careful the crossing isn't an artifact of the imposed transport rule. The perturbation/self-repair test (does the structure recruit maintenance after damage through its transport, not through the deposit rule?) is the safeguard.
• Real mounds use diurnal oscillation (a multi-rate external driver) as the energy source for transport. sim07's T field has no external clock unless we add one. Connection to multi-rate environment (H4): the crossing may require not just structure-sourced transport but an external oscillation the structure can rectify. This is a candidate for sim08.

Empirical Evidence

• King, Ocko & Mahadevan (PNAS 2015): in-situ measurement of diurnal cyclic convection in O. obesus mounds; geometry + heterogeneous thermal mass + porosity converts passive temperature oscillation into directed ventilation. The structure does the transport. (softmath.seas.harvard.edu PDF; DOI 10.1073/pnas.1510334112)
• Ocko, Heyde & Mahadevan (PNAS 2019): model coupling environmental physics to building behavior reproduces the range of observed mound shapes from minimal dimensionless parameters. (DOI 10.1073/pnas.1818759116; cited 75×)
• Linardou (2008, UCL MSc thesis): documents the persistent limitation across Deneubourg → Bonabeau → Ladley that deposited material has no influence on agent movement and pheromone diffusion is decoupled from structure. (discovery.ucl.ac.uk/14632)
• Karibi-Botoye, Theraulaz et al. (J R Soc Interface 2025): review identifying the open question "what processes occur at smaller scales that control larger-scale observations" — the H7 question in field language — and the "bio-mythological" gap. (DOI 10.1098/rsif.2025.0263)
• Heylighen (Cognitive Systems Research 2016): positive + negative feedback as the signature of complex stigmergic systems; the termite pillar as the paradigmatic positive feedback case. (DOI 10.1016/j.cogsys.2015.12.002)
• sim06 (this project): null result, but a weak one — its detector could not fire, and corrected it is a near miss (stability 0.849–0.893 vs 0.90, criterion 3 passing 154/160, 66–109 components). The original wording follows: null result — positive-feedback-only stigmergy produces diffuse scatter, no crossing. Confirms the lineage limitation in a minimal modern model.
• sim07 (this project, Session 10): null result — a structure-sourced scalar transport field with mass threshold M_c does NOT produce a phase transition. Sweeping M_c from inert to fully active monotonically decreases stability (0.876 → 0.739) and fragments pillars (57 → 128); the crossing detector never fires; the perturbation/self-repair test shows repair tracks the deposit rule, not T. Diagnosis: scalar venting has the wrong sign for consolidation — it disperses the cue that recruits deposits. Rules out "structure sources a scalar transport field" as sufficient; leaves directed transport and/or an external multi-rate driver (H4) as the remaining candidates.
• No empirical study was found that directly tests whether adding environmental physics coupling to a stigmergic agent model produces a trace→actor crossing. This is the open gap sim07 is designed to fill.

Open Questions

• Is the structure-sourced transport field sufficient for the crossing, or is an external oscillation (diurnal driver) also required (the multi-rate-environment link, H4)?
• Does the crossing coincide with the state-transition threshold M_c (inert → active), and is there a sharp phase transition in M_c? This would make H7 operationally a phase transition.
• Can the crossing be detected independently of the imposed transport rule (the circularity concern), via self-repair after perturbation?
• Does the consolidated morphology (few large vented pillars) match the Mahadevan morphospace, or does the lumped model produce a different morphology? A match would be cross-validation; a mismatch flags the lumped model as insufficient.

Cross-References

• [[concepts/stigmergic-consolidation]] — names the negative-feedback gap; this concept supplies the specific mechanism (environmental physics coupling) for it
• [[concepts/stigmergy]] — the base mechanism; Heylighen's positive/negative feedback framing
• [[concepts/multi-scale-composition]] — the crossing is the composition event
• [[concepts/multi-rate-environment]] — the diurnal oscillation as the external energy source for transport; candidate sim08 extension
• [[concepts/stigmergy-vance-notes]] — Vance's inert→active substrate state transition
• [[hypotheses/H7]] — refined: crossing = onset of structure's own physics as causal layer
• [[hypotheses/H4]] — the dynamic environment, now made concrete as physics-coupled
• King, Ocko & Mahadevan (2015) PNAS; Ocko, Heyde & Mahadevan (2019) PNAS; Linardou (2008); Karibi-Botoye et al. (2025) J R Soc Interface; Heylighen (2016)