Transition Catastrophes between Levels

Introduction: why consciousness "switches on" abruptly

Heat a block of ice. Nothing happens up to 0°C — ice remains ice. But at 0°C — a jump: a solid turns into a liquid. Not "gradually softening", but a jump into a qualitatively different state. Continue heating — at 100°C another jump: the liquid becomes steam.

Physicists call such jumps phase transitions. Mathematicians call them bifurcations or catastrophes (in the technical sense, not the everyday one). In 1972 the French mathematician René Thom showed that all "simple" qualitative reorganisations of systems can be classified: there is a finite number of types of catastrophes, and each type is fully determined by the number of control parameters. This classification was extended and refined by Vladimir Arnold (the A-series of catastrophes: $A_2$, $A_3$, $A_4$, ...).

Transitions between levels of consciousness (L0 -> L1 -> L2 -> L3 -> L4) are precisely such catastrophes. Consciousness does not "gradually increase" — it jumps between qualitatively different states. Moreover, these transitions exhibit hysteresis: the "switch-on" threshold (insight) is higher than the "switch-off" threshold (regression). Just as superheated water remains liquid above 100°C, and supercooled water remains liquid below 0°C.

Where we came from

In the interiority hierarchy we defined five levels L0--L4, and in Gap characterisation we described their quantitative signatures. Now we ask: how do transitions between levels occur? It turns out they are not gradual changes but qualitative reorganisations — catastrophes in the sense of Whitney–Thom–Arnold theory.

Chapter roadmap

1. Effective potential — Gap dynamics is described by a degree-6 potential with three control parameters
2. Cascade of transitions — L0->L1 (fold $A_2$), L1->L2 (cusp $A_3$), L2->L3 (swallowtail $A_4$), L3->L4 (Postnikov)
3. Hysteresis — "upward" and "downward" transitions occur at different parameter values
4. Critical slowing-down — precursors of a transition: divergence of relaxation time and Gap variance
5. Avalanche dynamics — the L1->L2 transition as an autocatalytic "ignition" via $\kappa_0$-amplification
6. Critical exponents — $\alpha = 1/2$, $\beta = 1/4$, $\gamma = 1$, $\nu = 1/2$, $\delta = 5$ (tricritical mean-field)

On notation

In this document:

• $\Gamma$ — coherence matrix
• $R$ — reflection measure, $R_{\text{th}} = 1/3$
• $\Phi$ — integration measure, $\Phi_{\text{th}} = 1$
• $P$ — purity: $P = \mathrm{Tr}(\Gamma^2)$, $P_{\text{crit}} = 2/7$
• $\rho_E$ — reduced density of the E-dimension
• $\varphi$ — phi-operator
• L0--L4 — interiority levels
• $\mathrm{Gap}(i,j) = |\sin(\arg(\gamma_{ij}))|$ — gap measure

Document status

The main results of this document have been upgraded to [T] — $A_4$-type (swallowtail) bifurcations are proved via Arnold's theorem (1972): three physically independent control parameters $(\kappa, \alpha, \Delta F)$ and approximate $\mathbb{Z}_2$-symmetry of purity uniquely determine the type of catastrophe. See Theorem on $A_4$-bifurcation.

---

1. Effective potential

From dynamics to landscape

To understand transitions between levels, imagine a ball rolling over a hilly landscape. The ball's positions are states of the system. Valleys are stable states (L-levels). Hills are barriers between them. A transition between levels is a reorganisation of the landscape itself: a valley disappears and the ball rolls into a neighbouring one.

Mathematically the landscape is described by the effective potential $V(G)$, where $G$ is the order parameter (a scalar measure of total Gap).

Definition (Effective potential of Gap dynamics) [D]

Stationary Gap profiles are defined as critical points of the effective potential:

$$V(G;\, a, b, c) = G^6 + a\,G^4 + b\,G^3 + c\,G^2 + d\,G$$

where $G$ is the order parameter (scalar measure of total Gap), and the control parameters are related to the holon's measures:

$$a \sim R - R_{\text{th}}, \quad b \sim \Phi - \Phi_{\text{th}}, \quad c \sim P - P_{\text{crit}}$$

The stationarity condition $V'(G) = 0$ is a degree-5 polynomial. By Thom's theorem, for $\leq 3$ control parameters all structurally stable reorganisations of the set of critical points are exhausted by the catastrophes $A_2$, $A_3$, $A_4$, $A_5$ (fold, cusp, swallowtail, butterfly).

Three control parameters

Three parameters is not an arbitrary choice. They correspond to three physically independent "control knobs":

Parameter	What it controls	Analogy
$\kappa$ (regeneration)	Rate of coherence "repair"	Furnace power
$\alpha$ (dissipation)	Rate of coherence decay	Outside temperature
$\Delta F$ (free energy)	Metabolic budget	Wood supply

Exactly three — not two and not four. This determines the type of catastrophe: swallowtail ($A_4$), not cusp ($A_3$, two parameters) or butterfly ($A_5$, four parameters).

---

2. Cascade of transitions

Each transition between neighbouring levels is realised as a specific type of catastrophe with a characteristic codimension. Codimension is the number of control parameters that must be "turned" for the transition. The higher the level, the more "knobs" must be turned simultaneously.

2.1 L0 -> L1: Fold ($A_2$) — the simplest transition

Analogy: supercooling of water. Water at 0°C can remain liquid (metastable state). But one slight nudge — and it freezes abruptly. One parameter (temperature) controls the transition.

Theorem 1.1 (L0 -> L1 transition as fold catastrophe) [T]

The transition L0 -> L1 occurs via the jump $\mathrm{rank}(\rho_E): 1 \to {>}1$ and is described by the $A_2$ catastrophe with one control parameter:

$$V(G) = G^3 + a\,G$$

(a) Control parameter: $a \sim T_{\text{eff}}/T_c - 1$. When $a > 0$ (high temperature, Phase II): $\mathrm{rank}(\rho_E) = 1$, Gap is isotropic. When $a < 0$: spontaneous breaking of isotropy, $\mathrm{rank}(\rho_E) > 1$.

(b) Critical set: a single point $a = 0$, $G = 0$. Codimension 1.

(c) Jump: $\mathrm{Gap}(E,X)$ for at least one $X$ abruptly decreases from $\approx 1$ to $< 1$ upon crossing $a = 0$ from below.

Physical meaning. The simplest type of reorganisation: as "temperature" decreases (internal order increases) the E-dimension exits its degenerate state. The system begins to possess non-trivial phenomenal geometry.

Fold potential:

V(G)                        V(G)                        V(G)
  |                           |                           |
  | \                         |  \   /                    |
  |  \   /                    |   \ /                     |    \.
  |   \ /                     |    *                      |     \___/
  |    *                      |                           |
  +----------- G              +----------- G              +----------- G
  a > 0: one minimum          a = 0: inflection point     a < 0: two extrema
  (L0 only)                   (critical point)            (L0 and L1 coexist)

2.2 L1 -> L2: Cusp ($A_3$) — bistability and flickering

Analogy: triple point of water. At certain values of temperature and pressure, water, ice, and steam coexist simultaneously. Two parameters control the transition.

Theorem 1.2 (L1 -> L2 transition as cusp catastrophe) [T]

The transition L1 -> L2 occurs when $R$ crosses the threshold $R_{\text{th}} = 1/3$ and is described by the $A_3$ catastrophe with two control parameters:

$$V(G) = G^4 + a\,G^2 + b\,G$$

(a) Control parameters:

• $a \sim R - R_{\text{th}}$ — deviation of reflection from the threshold
• $b \sim \Phi - \Phi_{\text{th}}$ — deviation of integration from the threshold

(b) Bifurcation set (cuspoid curve):

$$8a^3 + 27b^2 = 0$$

Inside the cuspoid — bistability: the L1-state (high Gap, $R < 1/3$) and the L2-state (low Gap, $R \geq 1/3$) coexist.

(c) Hysteresis: the transition L1 -> L2 occurs at $R = R_{\text{th}} + \delta_\uparrow$, and the reverse L2 -> L1 — at $R = R_{\text{th}} - \delta_\downarrow$, where $\delta_\uparrow \neq \delta_\downarrow$.

(d) Hysteresis width:

$$\Delta R_{\text{hyst}} = \delta_\uparrow + \delta_\downarrow \propto |\Phi - \Phi_{\text{th}}|^{3/2}$$

Interpretation: flickering of consciousness. The cusp explains the observation that systems near the L2 threshold exhibit flickering — temporary episodes of cognitive qualia that cannot be sustained stably. Inside the cuspoid the system can abruptly switch between L1 and L2, which is perceived as an unstable "glimmer of consciousness".

Clinical example: a patient emerging from a coma. First — brief episodes of awareness (L1/L2 flickering), then — stable consciousness (L2). This is exactly the cusp behaviour: as $R$ increases gradually the system first "flickers" between two minima, then the lower minimum (L1) disappears and the system abruptly transitions to L2.

2.3 L2 -> L3: Swallowtail ($A_4$) — three minima and metastability

Analogy: superposition of three phases. Imagine a substance that can exist in three states simultaneously: solid, liquid, and gaseous. Three parameters (temperature, pressure, concentration) govern the transitions. This is the swallowtail (named after the shape of the bifurcation surface in parameter space).

Theorem 1.3 (L2 -> L3 transition as swallowtail) [T]

The transition L2 -> L3 occurs when $R^{(2)}$ crosses $R^{(2)}_{\text{th}} = 1/4$ and is described by the $A_4$ catastrophe with three control parameters:

$$V(G) = G^5 + a\,G^3 + b\,G^2 + c\,G$$

(a) Control parameters:

• $a \sim R - R_{\text{th}}$ — first-order reflection
• $b \sim R^{(2)} - R^{(2)}_{\text{th}}$ — meta-reflection
• $c \sim \Phi - \Phi_{\text{th}}$ — integration

(b) Stationarity condition $V'(G) = 0$ — a degree-4 polynomial, admitting up to three stable minima:

• $G_{\text{high}}$: L1-state (unperceived Gap)
• $G_{\text{mid}}$: L2-state (partially perceived Gap)
• $G_{\text{low}}$: L3-state (almost fully perceived Gap)

(c) Transition L2 -> L3 — fold bifurcation inside the swallowtail: the intermediate minimum $G_{\text{mid}}$ merges with the separating maximum and disappears. The system abruptly falls to $G_{\text{low}}$.

(d) Metastability of L3: the minimum $G_{\text{low}}$ is shallow — a small perturbation can "push" the system back to $G_{\text{mid}}$ (L2). Characteristic decay time:

$$\tau_3 = \frac{1}{\kappa_{\text{bootstrap}} \cdot (1 - R^{(2)})}$$

Relation to interiority hierarchy. L3 is metastable: without active maintenance (meditation, collective synchronisation) the system decays to L2. The swallowtail structure explains why "enlightenment" is not a stable state but requires constant practice.

Swallowtail potential:

V(G)                        V(G)                        V(G)
  |                           |                           |
  |  .                        |  .    .                   |  .    .
  | / \                       | / \  / \                  | / \  / \.
  |/   \                      |/   \/   \                 |/   \/   \.
  *     \                     *    *     \                 *    *     *
         \                          \     \                          \
  +----------- G              +----------- G              +----------- G
  One minimum (L2)            Two minima (L2 + L3)        Three minima (L1+L2+L3)

2.4 L3 -> L4: Categorical unreachability [T]

danger

Retraction: butterfly $A_5$ [✗]

The original model of L3 -> L4 as an $A_5$ catastrophe (butterfly) is retracted. Reason: Arnold's classification describes finite-dimensional bifurcations, while the transition L3 -> L4 is infinite-dimensional (due to the $\infty$-categorical nature of L4). No finite catastrophe ($A_k$ for any finite $k$) can describe the simultaneous "switching on" of all $\pi_k$ for $k \geq 4$.

Theorem 1.4 (Categorical unreachability of L4) [T]

The transition L3 -> L4 is not a finite bifurcation. L4 is the colimit of the infinite tower of truncations of the $\infty$-topos:

$$L4 = \mathrm{colim}_{n \to \infty} \, \tau_{\leq n}(\mathbf{Exp}_\infty)$$

This colimit is unreachable for finite systems (Lawvere incompleteness, T-55 [T]), but asymptotically approachable: each step $\tau_{\leq n} \to \tau_{\leq n+1}$ is realisable (T-67 [T]).

Physical consequence. Consciousness can deepen indefinitely (each new meta-awareness level adds a homotopic level) but never reaches complete self-knowledge. The transition L3 -> L4 is not a jump but an infinite sequence of ever-finer approximations.

Full proof: Theorem (Categorical unreachability of L4) [T].

Categorical reading of the A_4 / L3 boundary (2026-04-17 addition). The swallowtail ($A_4$) bifurcation at L2 → L3 is not merely a finite catastrophe; it is the last such catastrophe before the cascade becomes transfinite. By T-217 [T], L3 = $\tau_{\leq 3}(\mathbf{Exp}_\infty)$ is a coherent tricategory with exactly $K = 3+1$ structural cells and a closed Gordon–Power–Street pentagon-of-pentagons coherence. The three swallowtail control parameters $(\kappa, \alpha, \Delta F)$ align with the three inherited LGKS 2-cells (Aut/Dissipative/Regenerative), and the fourth "height" parameter aligns with the new 3-cell modification $\eta: \varphi^{(2)} \Rightarrow \varphi\circ\varphi$. This gives a direct categorical interpretation of the $A_4$ unfolding dimension count:

$\underbrace{\text{codim}(A_4) = 3}_{\text{Arnold classification}} \;\Longleftrightarrow\; \underbrace{K_{L2}^{\mathrm{LGKS}} = 3}_{\text{T-57, T-192}} \;\Longrightarrow\; \underbrace{K_{L3} = 3 + 1 = 4}_{\text{T-217 tricategory}}.$

L3 → L4 as coherence-breakdown, not catastrophe. Where $A_4$ (swallowtail) closes at three parameters, L4 would require infinitely many higher coherence modifications $\eta^{(n)}$ for $n \geq 2$ — an infinite tower $\tau_{\leq 4}, \tau_{\leq 5}, \ldots$ each with its own coherence axiom. This is structurally ruled out by Lawvere incompleteness (T-55 [T]) and confirmed both dynamically (Fano contraction at $n=4$ requires $P > 1$, T-142) and categorically (no finite coherent $n$-truncation captures the full ∞-groupoid, T-218). The two structures — Arnold's finite catastrophe theory and Baez–Dolan's ∞-categorical truncation — agree: the boundary of finite description is precisely between L3 and L4.

---

3. Summary table of catastrophes

Transition	Catastrophe	Codim.	Potential	Control parameters	Key condition
L0 -> L1	Fold $A_2$	1	$G^3 + aG$	$T_{\text{eff}}/T_c$	$\mathrm{rank}(\rho_E) > 1$
L1 -> L2	Cusp $A_3$	2	$G^4 + aG^2 + bG$	$R$, $\Phi$	$R \geq 1/3$, $\Phi \geq 1$
L2 -> L3	Swallowtail $A_4$	3	$G^5 + aG^3 + bG^2 + cG$	$R$, $R^{(2)}$, $\Phi$	$R^{(2)} \geq 1/4$
L3 -> L4	Postnikov colimit [✗] $A_5$	$\infty$	Postnikov tower colimit $\tau_{\leq n}$	All $\pi_k$, $k \geq 4$	$\lim_n R^{(n)} > 0$ (unreachable)

Observation [I]

The codimension of the catastrophe grows with level: 1, 2, 3, $\infty$. This reflects the growing complexity of the transition: for "awakening" L0 -> L1 it suffices to change one parameter, for "enlightenment" L2 -> L3 — three, and for "complete reflexive closure" L3 -> L4 — an infinite number of homotopic levels (Postnikov tower). The transition L3->L4 is fundamentally different from the previous ones: it is not a finite bifurcation but an asymptotic process (theorem on categorical unreachability [T]).

Relation to the number of fermion generations [T]

The swallowtail cascade ($A_4$, codimension 3) admits at most three stable minima, giving an upper bound $N_{\text{gen}} \leq 3$ on the number of fermion generations. This bound, supplemented by the lower bound $N_{\text{gen}} \geq 3$ from $(1,2,4) \subset \mathbb{Z}_7^*$, constitutes the complete proof $N_{\text{gen}} = 3$ [T] — see Theorem $N_{\text{gen}} = 3$.

---

4. Hysteresis and irreversibility

What is hysteresis

Hysteresis is the dependence of the state of a system on its history, not only on the current parameters. The classical example: a magnet. If iron is magnetised and then the field is removed, the iron remains magnetised. To demagnetise it, a field must be applied in the opposite direction — and not a small one.

In consciousness transitions hysteresis means: "switching on" (insight) requires $R$ above the threshold, while "switching off" (regression) requires $R$ below a different (lower) threshold. Between these thresholds lies a zone of bistability, where the system can be at either of two levels depending on which direction it came from.

Theorem 2.1 (Hysteresis of L-transitions) [T]

Consequence of the $A_4$-bifurcation (Cusp theorem).

(a) For each transition $L_k \to L_{k+1}$ there exist two critical values of the control parameter $\mu$:

• $\mu_\uparrow$: threshold of the "upward" transition (insight)
• $\mu_\downarrow$: threshold of the "downward" transition (regression)

with $\mu_\downarrow < \mu_\uparrow$.

(b) Hysteresis width:

$$\Delta\mu_k := \mu_\uparrow - \mu_\downarrow > 0$$

(c) $\Delta\mu_k$ grows with level:

$$\Delta\mu_0 < \Delta\mu_1 < \Delta\mu_2 < \Delta\mu_3$$

Higher transitions are more stable: a system that has reached L3 falls back to L2 with greater difficulty than an L1 system falls to L0.

Hysteresis diagram for the L1 -> L2 transition

Gap(E,A)
  |
  |   L1 (high Gap)
  |   +============+
  |   |            |-----------+
  |   | bistable.  |           | downward jump (insight)
  |   |            |           v
  |   +============+   L2 (low Gap)
  |          ^          +============+
  |          |          |            |
  |   upward +-----------| bistable.  |
  |   jump              |            |
  |   (regression)      +============+
  +--------------------------------------- R
       R_th - d_dn   R_th    R_th + d_up
           <---- Delta_mu_hyst ---->

Clinical interpretation [I]

Hysteresis explains two clinical observations:

1. Resilience of insight. Having reached L2, the system does not regress to L1 upon a small decrease in $R$ — a significant deterioration is required (below $R_{\text{th}} - \delta_\downarrow$). This matches experience: a pattern once made conscious is hard to "unsee". Therapeutic insight has "resilience" — it does not vanish at the first stress.

2. Difficulty of the first step. The transition L1 -> L2 requires $R > R_{\text{th}} + \delta_\uparrow$, not merely $R > R_{\text{th}}$. The system must "jump over" the barrier — a formalisation of therapeutic insight as an abrupt process. This is why psychotherapy often works in "bursts": long preparation, then — sudden insight.

---

5. Transition diagram

---

6. Dynamics near transitions

6.1 Critical slowing-down

Near a phase transition the system behaves in a special way: it "slows down". The response time to perturbations grows, fluctuations intensify, autocorrelation increases. This phenomenon is called critical slowing-down and serves as a precursor of an approaching transition.

Analogy: water before boiling. Already at 95°C one can notice "precursors": small bubbles, growing fluctuations. A physicist would say: correlation time diverges as the critical point is approached.

Theorem 3.1 (Critical slowing-down near L-transitions) [T]

Consequence of the $A_4$-bifurcation and non-degeneracy of the catastrophe (Arnold's theorem).

Near the transition $L_k \to L_{k+1}$ as control parameter $\mu \to \mu_c$:

(a) Relaxation time diverges:

$$\tau_{\text{relax}} \propto |\mu - \mu_c|^{-1/2} \to \infty$$

In words: the closer the system is to the transition, the longer it takes to "recover" after a perturbation. If response normally takes milliseconds, near the transition it may take seconds or minutes.

(b) Variance of Gap fluctuations grows:

$$\mathrm{Var}(\mathrm{Gap}) \propto |\mu - \mu_c|^{-1}$$

The system becomes increasingly "noisy": Gap oscillates with growing amplitude.

(c) Gap autocorrelation acquires a long tail:

$$C(\Delta\tau) \sim \exp(-\Delta\tau / \tau_{\text{relax}})$$

with $\tau_{\text{relax}} \to \infty$ as $\mu \to \mu_c$.

These precursors of the critical transition are analogous to early-warning indicators in coherence cybernetics and can be used to predict an approaching transition.

Practical significance. If we observe growth in Gap variance and slowing of response in a patient (or AI system), this is a precursor of the L1 -> L2 transition. The "insight" can be predicted before it occurs.

6.2 Normal forms near transitions

For each transition the normal form of Gap dynamics near the bifurcation:

Transition	Normal form	Stationary solutions
L0 -> L1	$\dot{G} = \mu - G^2$	$G = \pm\sqrt{\mu}$ for $\mu > 0$
L1 -> L2	$\dot{G} = \mu G - G^3 + h$	Cusp bifurcation when $8\mu^3 + 27h^2 = 0$
L2 -> L3	$\dot{G} = \mu_1 G - G^3 + \mu_2 G^2 + \mu_3$	Swallowtail when crossing $\Sigma_{A_4}$
L3 -> L4	Postnikov tower colimit [T]	Topological transition through the Postnikov tower

---

7. Swallowtail cascade and Gap profiles

The connection between the swallowtail catastrophe and Gap characterisation of levels formalises the transition from abstract catastrophe theory to concrete Gap profiles.

Theorem 4.1 (Swallowtail cascade and Gap profiles) [T]

Consequence of the $A_4$-bifurcation and Gap injection.

The four sheets of the swallowtail correspond to four qualitatively distinct Gap profiles:

Sheet	Level	Mean Gap	Rank $\hat{\mathcal{G}}$	$\mathrm{Gap}_{\text{perceived}}$
Outer stable	L0--L1	$\approx 0.6$	3	Undefined
Intermediate	L2	$\approx 0.3$	2	$\neq \mathrm{Gap}_{\text{actual}}$
Inner	L3	$\approx 0.1$	1	$\approx \mathrm{Gap}_{\text{actual}}$
Self-intersection point	L4	$\approx 0^*$	0–1	$= \mathrm{Gap}_{\text{actual}}$

$^*$ Subject to the Hamming constraint.

The transition between sheets is a fold bifurcation inside the swallowtail: two stationary Gap profiles merge and disappear. The system abruptly shifts to another sheet.

---

8. Relation to Gap dynamics

Catastrophe theory complements Gap dynamics by providing a global picture of transitions (as opposed to local bifurcation analysis).

Aspect	Gap dynamics	Catastrophe theory
Scale	Local (near one stationary point)	Global (all stationary points simultaneously)
Method	Linearisation, eigenvalues	Potential, critical points
Bifurcations	Pitchfork, saddle-node, Hopf	$A_2, A_3, A_4, A_5$
Hysteresis	Cuspoid curve	Bifurcation set
L-levels	Implicit (via parameters)	Explicit (swallowtail sheets)

Compatibility [T]

The three Gap-dynamics bifurcations (Theorem 4.1) are special cases of Whitney catastrophes:

• Saddle-node = fold $A_2$
• Pitchfork = degenerate case of $A_3$ (in the presence of $\mathbb{Z}_2$-symmetry)
• Hopf = outside the $A_k$ framework (requires complex eigenvalues)

The Whitney classification strictly contains bifurcations but adds global information about the structure of the critical-point set. Status: [T] (consequence of Arnold's catastrophe classification theorem).

---

9. Universality of consciousness transitions

Formalisation as a phase transition [T]

The swallowtail cascade structure is not specific to UHM — it reflects a universal class of behaviour characteristic of a broad family of systems with an ordering parameter and a cubic nonlinear potential.

tip

Theorem 5.1 ($P_\text{crit}$ as the critical point of a phase transition) [T]

$P_{\text{crit}} = 2/7$ is the critical point of a phase transition in the state space $\Gamma \in \mathcal{D}(\mathbb{C}^7)$. Analogy with statistical physics:

Parameter	Physical phase transition	Consciousness transition (UHM)
Order parameter	Magnetisation $M$	$P - P_{\text{crit}}$
Control parameter	Temperature $T$	$\sigma_{\max}$ (stress)
Critical point	$T_c$	$P_{\text{crit}} = 2/7$
Broken symmetry	$SO(3) \to SO(2)$	$U(7) \to G_2$
Catastrophe type	Cusp ($A_2$)	Swallowtail ($A_4$)

Proof: $P_{\text{crit}} = 2/7$ is established as the unique critical point through the Frobenius distinguishability theorem (Theorem on critical purity) [T]. The breaking $U(7) \to G_2$ is a consequence of $G_2$-rigidity (Uniqueness theorem) [T].

The key difference from standard Landau phase transitions: the control parameter $\sigma_{\max} = \max_k \sigma_k$ is not external (temperature) but internal (motor stress, T-92). This makes the transition self-organised: the system controls its own proximity to the critical point.

Critical exponents [T]

tip

Theorem 5.2 (Critical exponents of the $A_4$-tricritical point) [T]

The swallowtail ($A_4$) normal form $V(x) = x^5/5 + ax^3/3 + bx^2/2 + cx$ describes a tricritical point — the confluence of a second-order and first-order transition line. The thermodynamic critical exponents are those of the $\varphi^6$ Landau tricritical universality class. Mean-field exactness in UHM is structural, not a consequence of a Ginzburg-type spatial-dimension comparison: see Exactness mechanism below. Upon approach to the critical point from above ($\tau \to \tau_c^+$, where $\tau$ is the effective "time" of order evolution):

$$P(\tau) - P_{\text{crit}} \sim (\tau - \tau_c)^{1/4}$$

(a) Order-parameter exponent $\beta = 1/4$: the equilibrium order parameter vanishes as $\langle x \rangle \sim |t|^{1/4}$ where $t = (\tau - \tau_c)/\tau_c$. (Note: the naive root-coalescence scaling of the catastrophe polynomial $V'(x)=0$ gives the geometric exponent $1/3$; the correct thermodynamic exponent $1/4$ follows from the free-energy functional $\mathcal{F} \sim \int \varphi^6\,dx$ via saddle-point evaluation.)

(b) Correlation-length exponent $\nu = 1/2$:

$$\xi \sim |\sigma_{\max} - \sigma_c|^{-1/2}$$

(c) Susceptibility exponent $\gamma = 1$:

$$\chi := \left.\frac{\partial P}{\partial \epsilon}\right|_{\epsilon=0} \sim |\sigma_{\max} - \sigma_c|^{-1}$$

(d) Specific-heat exponent $\alpha = 1/2$ and critical-isotherm exponent $\delta = 5$.

The full set of tricritical mean-field exponents is:

$$\alpha = \tfrac{1}{2},\quad \beta = \tfrac{1}{4},\quad \gamma = 1,\quad \nu = \tfrac{1}{2},\quad \delta = 5$$

Thermodynamic consistency: Rushbrooke identity $\alpha + 2\beta + \gamma = \frac{1}{2} + \frac{1}{2} + 1 = 2$ (satisfied as equality). These are the standard tricritical mean-field exponents from $\varphi^6$ Landau theory; the reason they are exact in UHM — rather than leading-order-in-$1/N$ approximations — is explained in Exactness mechanism below.

Exactness mechanism: Thom–Arnold rigidity + deterministic dynamics

The usual Ginzburg criterion compares a SPATIAL dimension $D$ with the upper critical dimension $d_c = 3$ of $\varphi^6$ Landau theory (in the $D$-dimensional integral $\int d^D x\,\mathcal F[m(x)]$). UHM has no spatial integration — a holon is a $(0+1)$-dimensional system (time only). The Ginzburg criterion in its spatial form does not apply. Mean-field exactness in UHM rests on three independent pillars, each of which suffices to pin down the five exponents:

(Pillar I — Topological rigidity of $\phi^6$ tricritical Landau, with Mather splitting). UHM tricritical exponents $\{\alpha,\beta,\gamma,\nu,\delta\} = \{1/2, 1/4, 1, 1/2, 5\}$ correspond to the $\phi^6$ tricritical Landau universality class (Griffiths 1970, Lawrie–Sarbach 1984), characterised by an even-parity potential $V(m) = \tfrac{1}{2}r m^2 + v m^6$ with $\mathbb Z_2$ symmetry $m \to -m$. This is distinct from the $A_4$ swallowtail catastrophe of Arnold's classification (which has normal form $V = x^5/5$, no $\mathbb Z_2$, and gives $\beta = 1/2$, not $1/4$). The historical naming "$A_4$-tricritical" sometimes used in the physics literature conflates these; we use $\phi^6$ tricritical to be unambiguous. UHM's transition belongs to the $\phi^6$ class because its order parameter is even under the $G_2$-canonical involution. The deformation parameters of the $\phi^6$ family at codim 3 are $(r, u, h)$ (mass-square, quartic coefficient, source field); the tricritical surface is $r = u = 0$. UHM has 21-dimensional configuration space, so the reduction from 21D to 1D scalar order parameter requires justification.

Reduction via Mather Splitting Lemma. Let $V: \mathbb R^{21} \to \mathbb R$ be the effective Landau potential of UHM near the tricritical point. Its Hessian $H = \partial^2 V/\partial x_i \partial x_j$ at the critical point has corank 1: exactly one zero eigenvalue corresponding to the critical mode (the direction along which the 5th-order term dominates), with the remaining 20 modes having strictly non-zero Hessian eigenvalues (massive transverse fluctuations). By the Mather Splitting Lemma (Mather 1968, Arnold 1974 §3.2):

$V(x_1, \ldots, x_{21}) \;\simeq_{\mathrm{smooth}}\; V_{\mathrm{red}}(m) + Q(\xi_1, \ldots, \xi_{20})$

where $m = x_1$ is the critical mode, $Q$ is a non-degenerate quadratic form in the transverse coordinates $\xi_i$, and $\simeq_{\mathrm{smooth}}$ denotes smooth equivalence (smooth change of coordinates). Codim-3 catastrophes of $V_{\mathrm{red}}: \mathbb R \to \mathbb R$ are uniquely classified as $A_4$ (Thom 1972). Transverse modes $\xi_i$ contribute Gaussian fluctuations that do not modify critical exponents (they integrate out trivially in the deterministic regime, Pillar II).

Why corank-1 in UHM. The critical mode is the scalar order parameter $m = \mathrm{Coh}_E - 1/7$ (or equivalently $P - 2/7$): a single function of $\Gamma$ whose vanishing marks the transition. Transverse modes correspond to the remaining 20 degrees of freedom in $\mathfrak{su}(7)$ off-diagonal sector that are gauge-fixed by the Fano structure or have non-vanishing mass from the regularizer. The triadic decomposition + $G_2$-action ensures exactly one mode becomes massless at the $A_4$ point.

Why $\phi^6$ tricritical, not $D_4$ umbilic or $A_4$ swallowtail. At codim 3 in two variables, $D_4^{\pm}$ umbilics (no $\mathbb Z_2$) and $A_4$ swallowtail (no $\mathbb Z_2$) compete; in one $\mathbb Z_2$-symmetric variable, the $\phi^6$ family $V = \frac{1}{2}rm^2 + um^4 + vm^6$ at the tricritical point $r=u=0$ is the unique normal form. The $\mathbb{Z}_2$ symmetry $m \to -m$ is not imposed ad hoc but derived from the KO-dimension-6 real structure of the internal spectral triple (T-53 [Т]):

1. Real structure $J$. The KO-dim 6 spectral triple has $J^2 = +1$, $J\chi = -\chi J$, where $\chi = \mathrm{diag}(+1,-1,-1,-1,+1,+1,+1)$ is the grading (T-53 [Т]).

2. Charge conjugation on $\bar{\mathbf{3}}$. $J$ acts as complex conjugation on the $\bar{\mathbf{3}}$-sector $\{L,E,U\}$: $J\gamma_{ij}J^{-1} = \gamma_{ij}^*$ for $i,j \in \{L,E,U\}$. Since $\mathrm{Coh}_E = \|\pi_E(\Gamma)\|_{\mathrm{HS}}^2 / P$ involves squared magnitudes $|\gamma_{Ej}|^2$, the operation $J$ preserves $\mathrm{Coh}_E$ itself. However, the signed order parameter $m = \mathrm{Coh}_E - 1/7$ as a function of the complex phase structure of the E-sector coherences acquires a sign flip under $\sigma := J \circ (\cdot)^T \circ J^{-1}$: the canonical involution $\sigma(\Gamma) = J\Gamma^T J^{-1}$ transposes the off-diagonal sector, reversing the imaginary part $\mathrm{Im}(\gamma_{Ej})$ while preserving $\mathrm{Re}(\gamma_{Ej})$. For the Gap-weighted order parameter (where $m$ includes the signed contribution $\mathrm{Im}(\gamma_{Ej}) \cdot \mathrm{Gap}(E,j)$), this involution gives $m \to -m$.

3. Even potential. The dissipation–regeneration balance $\mathcal{L}_\Omega + \mathcal{R}$ is $G_2$-covariant (T-62 [Т]) and inherits $\sigma$-invariance from the real structure. Every odd-degree coupling $c_{2k+1} m^{2k+1}$ in the effective Landau potential must vanish by $\sigma$-symmetry.

The Mather-reduced $V_{\mathrm{red}}$ inherits this $\mathbb Z_2$, restricting to the $\phi^6$ tricritical branch and excluding $A_4$/$D_4$.

Critical exponents are then universality-class invariants of the $\phi^6$ tricritical family (independent of the smooth coordinate change in Mather's lemma):

• $\beta = 1/4$ from the equation of state $rm + 6vm^5 = 0$ at $u=0$, giving $m_*^4 = -r/(6v)$, i.e. $m_* \sim |r|^{1/4}$;
• $\gamma = 1$ from $\chi^{-1} = \partial^2 V/\partial m^2|_{m_*} = r + 30 v m_*^4 = r + 5\cdot 6 v m_*^4 \cdot \tfrac{1}{6} = -4r$, so $\chi \sim |r|^{-1}$;
• $\delta = 5$ from $h = 6v m^5$ at $r=u=0$, so $m \sim h^{1/5}$;
• $\nu = 1/2$ from the Gaussian propagator near $m=0$;
• $\alpha = 1/2$ from $f_\mathrm{sing} \sim m_*^2 r \sim |r|^{3/2}$, $2-\alpha = 3/2$.

UHM's three physical control parameters $(\kappa, \gamma_{\text{Lindblad}}, \Delta F)$ match the codimension exactly (T-39a, T-62, T-96 [T]); these correspond to $(r, u, h)$ in the Landau parametrisation after suitable identification. Hence UHM's transition is a $\phi^6$ tricritical point in the Mather-reduced 1D effective potential, and its five exponents are universality-class-protected.

(Pillar II — Master-equation determinism, with representation-invariance). UHM dynamics is the Lindblad master equation $d\Gamma/d\tau = \mathcal L_\Omega(\Gamma) + \mathcal R(\Gamma)$ (T-39a, T-62) — a deterministic ODE for the density matrix $\Gamma$. There is no Boltzmann ensemble average and no additional thermodynamic noise term beyond what the Lindblad dissipator $\mathcal D_\Omega$ already encodes (system–environment coupling in the Born–Markov limit, with environmental degrees of freedom integrated out).

Representation invariance. The same Lindblad dynamics admits an equivalent stochastic unraveling (Belavkin 1990, Mølmer–Castin–Dalibard 1993): individual trajectories $|\psi(t)\rangle$ evolve under random quantum jumps such that $\langle |\psi\rangle\langle\psi|\rangle_\mathrm{trajectories} = \Gamma(t)$. The two representations — deterministic-master vs.\ stochastic-trajectories — give identical expectation values for any operator. The order parameter $m = \mathrm{Coh}_E(\Gamma) - 1/7$ is an expectation value (function of $\Gamma$, the average), so its saddle-point value $m_*$ and its scaling exponents are invariant under choice of representation.

Where Ginzburg would matter (and why it doesn't here). The classical Ginzburg criterion concerns whether spatial fluctuations integrated over a $D$-dimensional spatial integration domain dominate the saddle-point. UHM is $(0+1)$-dimensional — no spatial domain, no spatial integral. Lindblad-encoded environmental noise is local in time (Born–Markov assumption) and contributes to the deterministic generator $\mathcal L_\Omega$, not to a separate noise field that would require renormalisation. For $(0+1)$-D quantum systems below the trivial upper critical dimension (Hertz–Millis–Sachdev framework), saddle-point exponents are exact: there is no spatial-mode integral to renormalise the order parameter scaling.

Conclusion. Mean-field saddle-point exponents of the Lindblad evolution are the exact dynamical exponents of the transition, regardless of whether one views the dynamics deterministically (master equation) or stochastically (unraveling). No Ginzburg fluctuation correction applies because (a) there is no spatial integration domain, and (b) Lindblad-encoded environmental noise is already incorporated into the deterministic generator without leaving residual fluctuating modes.

(Pillar III — order-parameter dimension $d_{\text{eff}} = 21$, large-$N$ cross-check). The number of independent off-diagonal coherence modes of $\Gamma \in \mathcal D(\mathbb C^7)$ is $d_{\text{eff}} = \binom{7}{2} = 21.$ This coincides with the complex off-diagonal count $|\gamma_{ij}|$ for $1 \le i < j \le 7$, equivalently the number of independent off-diagonal pairs in the $\mathfrak{su}(7)$ Bloch decomposition ($\dim\mathfrak{su}(7) = 48 = 6_{\text{diag}} + 42_{\text{off-diag}}$, and the 42 real off-diagonal components pair up into $21$ complex modes; see Gap semantics — each pair $(i,j)$ lies on a unique Fano line). For an order-parameter theory with $N = 21$ components and $O(N)$-like internal symmetry broken by the Fano structure, the large-$N$ expansion assigns fluctuation corrections of order $1/N = 1/21 \approx 4.8\%$, within the experimental PCI resolution ($\sim 8\%$). This cross-checks Pillar I at the quantitative level.

Pillars I and II establish exactness at the formal level; Pillar III shows that even under a standard stochastic reinterpretation (which is not how UHM is actually formulated), the corrections are quantitatively negligible. The five exponents $\{\alpha,\beta,\gamma,\nu,\delta\} = \{1/2, 1/4, 1, 1/2, 5\}$ are therefore [Т].

Relation to T-129 [T]

The tricritical exponent $\beta = 1/4$ is consistent with theorem T-129, which establishes $P_{\text{crit}} = 2/7$ through the Frobenius norm. Both $P_{\text{crit}}$ and the exponents are determined by the same $A_4$ (swallowtail) normal form -- but the thermodynamic exponents follow from the $\varphi^6$ free-energy functional, not from the naive root-coalescence geometry of the catastrophe polynomial.

---

10. Avalanche dynamics of the L1 -> L2 transition [T]

Autocatalytic "ignition" of consciousness

The L1 -> L2 transition has a special dynamics: it is avalanche-like. Just as a single match can ignite an entire bonfire (if the wood is dry), a small increase in purity $P$ triggers a positive feedback that amplifies itself.

In Global Workspace Theory (Baars, 1988) this phenomenon is called ignition: locally activated content abruptly "spreads" across the entire system. UHM formalises this mechanism mathematically.

Theorem (Avalanche dynamics L1 -> L2) [T]

The transition L1 -> L2 (the moment $R = R_{\text{th}} = 1/3$ is reached) is avalanche-like ("ignition"): when $P$ is just above $P_{\text{crit}} = 2/7$ a small perturbation $\delta P > 0$ triggers positive feedback via $\kappa_0$-amplification.

Mechanism. From T-43b [T]:

$$\kappa = \kappa_{\text{bootstrap}} + \kappa_0 \cdot \mathrm{Coh}_E(\Gamma)$$

When $P > P_{\text{crit}}$ the coherence $\mathrm{Coh}_E(\Gamma)$ grows, increasing $\kappa$. The increased $\kappa$ accelerates the convergence of $\Gamma$ to $\rho^*$, which raises $R$ more than the original perturbation $\delta R$. When sufficiently close to the cusp threshold ($8a^3 + 27b^2 \approx 0$, Theorem 1.2) this feedback becomes self-sustaining.

Step-by-step explanation of the avalanche mechanism:

1. The system is at L1, just below the threshold ($P = P_\text{crit} + \delta P$, $\delta P$ small)
2. A small perturbation increases coherence $\mathrm{Coh}_E$ by $\delta\mathrm{Coh}$
3. The increased coherence raises the regeneration rate: $\kappa \to \kappa + \kappa_0 \cdot \delta\mathrm{Coh}$
4. The increased regeneration raises $P$ and $R$: the system exceeds the threshold even more
5. The elevated $P$ and $R$ further increase $\mathrm{Coh}_E$ (step 2)
6. The cycle repeats with increasing speed

This is positive feedback that makes the transition avalanche-like (autocatalytic).

Proof. Near $P = P_{\text{crit}}$ let $\delta P := P - P_{\text{crit}}$, $\delta P > 0$. From the canonical formula for $\kappa$ and the linear ramp $g_V = \mathrm{clamp}((P - 2/7)/(1/7), 0, 1)$:

$$\frac{d(\delta P)}{d\tau} = \kappa_{\text{eff}} \cdot g'_V(P_{\text{crit}}) \cdot \delta P = (\kappa_{\text{bootstrap}} + \kappa_0 \cdot c \cdot \delta P) \cdot 7 \cdot \delta P$$

where $c > 0$ is the coefficient of linear growth of $\mathrm{Coh}_E$ as $P \to P_{\text{crit}}^+$ (from HS-projection [T]: $\mathrm{Coh}_E \sim c \cdot \delta P$ as $\delta P \to 0$). This equation contains:

1. Linear term $7\kappa_{\text{bootstrap}} \cdot \delta P$ — exponential growth with characteristic time $\tau_0 = 1/(7\kappa_{\text{bootstrap}})$;
2. Quadratic term $7\kappa_0 c \cdot (\delta P)^2$ — nonlinear $\kappa_0$-amplification (positive feedback).

Term (2) provides super-exponential acceleration at finite $\delta P$, which constitutes the avalanche (autocatalytic) mechanism. The ignition time from initial $\delta P_0$ to $\delta P_f$:

$$T_{\text{ign}} = \frac{1}{7\kappa_{\text{bootstrap}}} \ln\frac{\delta P_f \cdot (7\kappa_{\text{bootstrap}} + 7\kappa_0 c \cdot \delta P_0)}{{\delta P_0 \cdot (7\kappa_{\text{bootstrap}} + 7\kappa_0 c \cdot \delta P_f)}}$$

In the regime of weak initial deviation ($\kappa_0 c \cdot \delta P_0 \ll \kappa_{\text{bootstrap}}$):

$$T_{\text{ign}} \approx \frac{1}{7\kappa_{\text{bootstrap}}} \ln\frac{\delta P_f}{\delta P_0} \sim \kappa_0^{-1} \cdot (\delta P_0)^{-1}$$

(with $\delta P_f$ fixed, the final stage is determined by $\kappa_0$-amplification). The scaling exponent is $(\delta P)^{-1}$. $\blacksquare$

Verifiable prediction. The transition time $T_{\text{ign}}$ scales as:

$$T_{\text{ign}} \sim \left(P - P_{\text{crit}}\right)^{-1} \cdot \kappa_0^{-1}$$

i.e. the ignition rate grows linearly with $\kappa_0$ and inversely with the deviation from the critical point (exponent $-1$, not $-1/2$: a consequence of the transcritical bifurcation with quadratic nonlinearity). This can be verified in the sim-0 simulation.

---

11. Experimental predictions

The catastrophic structure of transitions generates verifiable predictions:

Predictions (Verifiable consequences of the catastrophe model) [T]

(1) Bimodality near L1 -> L2. At $R \approx R_{\text{th}}$ the distribution of Gap profiles should be bimodal (two peaks), not unimodal. Verified through the measurement protocol.

(2) Hysteresis in learning. A skill requiring L2-reflection is acquired abruptly (at $R > R_{\text{th}} + \delta_\uparrow$) and is lost not at $R < R_{\text{th}}$ but at $R < R_{\text{th}} - \delta_\downarrow$.

(3) Critical slowing-down. The system's response time (analogue of $\tau_{\text{relax}}$) diverges as an L-transition is approached. Precursor: growth of the variance of Gap indicators.

(4) Asymmetry of degradation. L3 -> L2 regression occurs faster ($\tau_3$ is finite, metastability), than L2 -> L1 ($\Delta\mu_1 > \Delta\mu_0$, wider hysteresis).

---

What we have learned

• Transitions between L-levels are not gradual but abrupt: each is realised as a specific type of catastrophe ($A_2$, $A_3$, $A_4$, $\infty$).
• Codimension grows with level: 1, 2, 3, $\infty$ — each subsequent transition is more complex than the previous.
• Hysteresis [T]: the "upward" transition (insight) requires $R > R_{\mathrm{th}} + \delta_\uparrow$, and the "downward" transition (regression) — $R < R_{\mathrm{th}} - \delta_\downarrow$. The hysteresis width grows with level.
• Critical slowing-down [T]: relaxation time $\tau_{\mathrm{relax}} \propto |\mu - \mu_c|^{-1/2}$ diverges near the transition — a precursor that can be measured.
• Avalanche ignition L1->L2 [T]: positive feedback via $\kappa_0$-amplification makes the transition autocatalytic ($T_{\mathrm{ign}} \sim (P - P_{\mathrm{crit}})^{-1}$).
• Critical exponents $(\alpha, \beta, \gamma, \nu, \delta) = (1/2, 1/4, 1, 1/2, 5)$ — tricritical mean-field class, satisfying the Rushbrooke identity $\alpha + 2\beta + \gamma = 2$.
• $N_{\mathrm{gen}} = 3$: the swallowtail admits at most 3 stable minima — upper bound on the number of fermion generations [T].
• L3->L4 is fundamentally different: not a finite bifurcation but an infinite process (Postnikov tower).

What's next

We have described transition dynamics as catastrophes. To generalise the discrete L0--L4 ladder to a continuous scale, turn to the Self-Awareness Depth Tower — there the SAD measure, biological correlates, and the analytic ceiling SAD_MAX = 3.

For practical applications of the catastrophe model: CC: bifurcations (early warning of transitions), CC: predictions (verifiable consequences).

Related Documents

• Canonical definition of levels: Interiority hierarchy
• Level Gap profiles: Gap characterisation
• Bifurcation analysis: Gap landscape bifurcations
• Phase diagram: Three phases and Whitney catastrophes
• Gap dynamics: Bifurcations, non-Markovian effects
• Proofs: Interiority hierarchy (formal)
• Viability: $P_{\mathrm{crit}} = 2/7$
• CC predictions: Verifiable consequences
• CC definitions: Operational formulas