Prerequisites. This paper assumes the cause-plex framework as developed in Cause-Plex and Spacetime: the Lorentzian metric and Lorentz invariance are derived from causal structure (Malament 1977 + event counting); the metric signature has exactly one timelike dimension from P1. The present paper addresses what fixes the number of spatial dimensions at 3.

Positioning. This is an observer-selection argument, not a derivation. The distinction is important: the argument does not derive $n_s = 3$ from the cause-plex primitive; it explains why any observer capable of asking the question would find themselves in $n_s = 3$ . This is the honest and correct claim.

Abstract

We present the selective convergence argument for why the physical world has 3+1 spacetime dimensions in the cause-plex framework. The cause-plex primitive — a locally finite strict partial order of causal events — does not uniquely determine dimensionality. However, the subset of cause-plexes supporting observer-class entities (those with $\rho_{\mathrm{ac}} > 0$ and $\mathrm{CI} > \mathrm{CI}_{\min}$ ) is tightly constrained: $n_t = 1$ is forced by the causal ordering (P1 fails for $n_t \neq 1$ ); $n_s = 3$ is the unique value satisfying both Tangherlini's orbital stability requirement ( $n_s \leq 3$ ) and the knot-topology requirement for topologically diverse loop structures ( $n_s \geq 3$ ). The argument is an observer-selection conjecture — rigorously grounded but not a derivation from primitives. It is compared to string theory's landscape approach and Wolfram's singular convergence claim, and shown to be more honest and more precisely grounded than either.

1. The Problem

The cause-plex framework derives the Lorentzian metric structure of spacetime from causal order (Malament 1977) and event counting (number-volume conjecture). What it does not derive is the specific dimensionality of the manifold — the number of timelike and spacelike directions.

This is not a gap in the derivation; it is a feature. The cause-plex $(E, \prec)$ is an abstract partial order — it does not come pre-labeled with spatial dimensions. The dimensions emerge in the continuum limit, and what determines them is not the abstract structure but the physical content: specifically, whether the resulting manifold supports the kind of complex auto-causal loop structures that constitute observers.

The question is: why do we observe $(-,+,+,+)$ rather than $(-,+,+)$ or $(-,+,+,+,+)$ ?

2. Honest Status

The selective convergence argument is a structurally grounded version of the weak anthropic principle. It does not derive 3+1 dimensionality from the cause-plex primitive; it explains why observers would necessarily find themselves in 3+1 by showing that other dimensionalities fail to support the structural requirements for observer-class entities.

This is a correct and honest claim. The dimensionality entry in the spacetime paper's summary table is marked ⚠️ (observer-selection argument) accordingly — not ✅ (derived).

3. How Other Frameworks Handle Dimensionality

Framework	Why these dimensions?	Mechanism	Honest status
Standard Model / GR	Assumes 3+1	None	Empirical input
String theory	10D from Weyl anomaly; 4D via compactification	Internal consistency + landscape + anthropic selection	10D derived; 4D needs observer selection
Wolfram (singular convergence)	All rules converge for any complex observer	Claimed; unsupported by the ruliad structure	Overclaims
Epimechanics (selective convergence)	$n_t = 1$ from P1; $n_s = 3$ from stability filter on observer-class entities	Tangherlini + knot topology + observer selection	Observer-selection conjecture; rigorous and honest

String theory comparison. String theory requires 10 dimensions from the Weyl anomaly — an internal consistency condition. To recover 4D, the 6 extra dimensions are compactified on Calabi-Yau manifolds, of which there are approximately $10^{500}$ consistent choices (the string landscape). String theory cannot uniquely predict 4D and ultimately needs observer selection from the landscape. The cause-plex framework's observer-selection argument enters explicitly and upfront — which is more transparent.

Wolfram comparison. Wolfram claims singular convergence: all possible computational rules converge to the same physics for any sufficiently complex observer. This is stronger than the evidence supports. The cause-plex framework claims only selective convergence: those cause-plexes that support complex observers are restricted to a stable neighborhood of manifold space. This is a strictly weaker and more defensible claim.

4. The Selective Convergence Argument

Claim. The dimensionality of the physical world is fixed by a stability filter: cause-plexes that support complex auto-causal loop structures are restricted to a small stable region of manifold space. We observe 3+1 because we are on one of those stable manifolds — not because it is the only possible outcome, but because it is the only neighborhood that permits our kind of loop structure.

Formal statement:

We observe 3+1 dimensional spacetime not because it is uniquely derivable from the cause-plex primitive, but because 3+1 is the stable manifold neighborhood in which complex auto-causal loops — the structural requirement for observers — can form and persist.

5. Fixing $n_t = 1$ : From the Causal Partial Order

Claim. The cause-plex must have exactly one timelike dimension.

Argument. The cause-plex $(E, \prec)$ is a strict partial order — irreflexive, asymmetric, transitive. This is the definition of a single causal direction. In the continuum limit:

$n_t = 0$ : no causal ordering at all (elliptic PDEs). P1 fails — the partial order doesn't exist.
$n_t = 1$ : the causal partial order holds; PDEs are hyperbolic; deterministic forward propagation from initial data exists.
$n_t \geq 2$ : ultrahyperbolic PDEs. No well-posed initial value problem — state cannot be propagated forward deterministically. The causal partial order loses its structure: there is no clean "before/after," so P1 fails.

$n_t = 1$ is the unique value consistent with the cause-plex being a well-defined strict partial order with deterministic causal propagation. This step is rigorous — it follows directly from the definition of the cause-plex.

6. Fixing $n_s = 3$ : From Observer-Class Stability

Given $n_t = 1$ , we need to determine $n_s$ . Two independent arguments constrain it from above and below.

6.1 Upper bound: $n_s \leq 3$ from orbital stability

Tangherlini (1963): In $n_s > 3$ spatial dimensions, gravitational and electromagnetic potentials scale as $r^{-(n_s-2)}$ rather than $r^{-1}$ . Bertrand's theorem requires $1/r^2$ force laws for stable closed orbits. For $n_s > 3$ :

No stable closed orbits → no stable atoms
No stable atoms → no chemistry
No covalent bonds → no molecular structures
No bond stability ( $\sigma_b < k_BT$ ) → no persistent auto-causal loops
No persistent loops → $\rho_{\mathrm{ac}} \to 0$ , no observer-class entities

Therefore $n_s > 3$ is excluded for observer-class cause-plexes.

6.2 Lower bound: $n_s \geq 3$ from loop topology

Knot theory argument: In $n_s < 3$ , knot theory is trivial — all closed loops are isotopic (continuously deformable into each other). In $n_s = 3$ , loops can be topologically distinct (knots, links), creating topological diversity in bond structures.

Why this matters: Observer-class entities require $\mathrm{CI} > \mathrm{CI}_{\min}$ , where $\mathrm{CI}_{\min}$ is the minimum cause-plex index for topologically non-trivial loop structures.

Definition ( $\mathrm{CI}_{\min}$ ). The minimum CI for observer-class entities is the smallest CI value at which a cause-plex contains at least one topologically non-trivial loop — a closed causal loop not isotopically equivalent to a circle in 3D space. This threshold requires $n_s \geq 3$ , making the argument non-circular: $\mathrm{CI} > \mathrm{CI}_{\min}$ means specifically "entity with topologically non-trivial loop structure exists," not merely "observer exists."

In $n_s < 3$ : all loops are isotopic → no topological diversity → $\mathrm{CI} \leq \mathrm{CI}_{\min}$ → no observer-class entities.

Therefore $n_s < 3$ is excluded for observer-class cause-plexes.

6.3 Unique solution

Combining §6.1 and §6.2: $n_s \leq 3$ and $n_s \geq 3$ , so $n_s = 3$ .

7. Tegmark's Stability Table

Tegmark (1997) independently mapped the stability of physics across different $(n_s, n_t)$ signatures, confirming the argument:

Manifold $(n_s, n_t)$	PDE type	Stable atoms?	Sufficient loop richness?	Observers possible?
$(1,1)$ , $(2,1)$	Hyperbolic	Yes	No — trivial knot theory	No
$(3,1)$	Hyperbolic	Yes	Yes	Yes
$(n_s > 3, 1)$	Hyperbolic	No — Tangherlini	Moot	No
$(n_s, n_t \geq 2)$	Ultrahyperbolic	No	No	No
$(n_s, 0)$	Elliptic	No	No	No

$(3,1)$ is the unique manifold signature supporting both requirements simultaneously.

8. What Entities Would Look Like in Other Dimensions

For $n_t = 1$ (required) and varying $n_s$ :

Manifold	Bond stability	Loop topology	Max entity type	$\rho_{\mathrm{ac}}$ ceiling
$(2,1)$	Limited bond diversity; no 3D chirality	All loops isotopic	Low-CI composites; no stereochemistry	Low
$(3,1)$ — ours	Full bond diversity; stable atoms; rich knot topology	Topologically distinct loop types	Full range: cells, organisms, meta-entities	Observed maximum
$(4,1)$	Atoms unstable (Tangherlini)	Moot — no stable bonds	Dissipative only	Near zero

The $(3,1)$ manifold sits at a maximum of loop composition richness subject to bond stability. This is not a coincidence — it is selective convergence: the cause-plexes capable of asking the question occupy this maximum by definition.

Knot theory and biology. The connection between 3D topology and biological complexity is concrete. DNA supercoiling, protein folding, and enzyme active site geometry all depend on 3D knot topology — loops that cannot be continuously deformed into each other without breaking bonds. In $n_s < 3$ , all such loops are equivalent; the structural diversity underlying molecular biology vanishes. The cause-plex framework provides a structural explanation for why biology requires 3D space.

9. The Stable Observer Manifold Conjecture

Conjecture (Stable Observer Manifold). A cause-plex $\mathcal{C}^*$ supporting entities with $\rho_{\mathrm{ac}} > 0$ and cause-plex index $\mathrm{CI} > \mathrm{CI}_{\min}$ requires, in the continuum limit, a manifold with exactly $n_t = 1$ timelike dimension and $n_s = 3$ spacelike dimensions.

Proof sketch:

$n_t = 1$ : required by causal partial order (P1) — proved rigorously (Section 5).
$n_s \geq 3$ : required by $\mathrm{CI} > \mathrm{CI}_{\min}$ (knot topology, Section 6.2) — well-established.
$n_s \leq 3$ : required by bond stability (Tangherlini, Section 6.1) — well-established.
Therefore $n_s = 3$ uniquely. ∎

Status: Step 1 is rigorous. Steps 2 and 3 rely on Tangherlini's result and the knot theory argument — both are well-established in the literature. The conjunction (step 4) requires formalizing "sufficient loop richness" precisely in terms of CI and topological bond diversity. This is Open Problem 1 of this paper.

10. Open Problems

Open Problem 1: Formalize the Stable Observer Manifold Conjecture

What remains. Define "sufficient loop richness" precisely in terms of CI and topological bond diversity, and prove formally that $n_s < 3$ is insufficient for $\mathrm{CI} > \mathrm{CI}_{\min}$ . The Tangherlini result and knot theory argument are both established; their conjunction as a formal theorem about cause-plex CI thresholds requires writing up.

Approach. The CI threshold $\mathrm{CI}_{\min}$ is defined as the minimum CI for a cause-plex containing a topologically non-trivial loop. In 3D, this corresponds to the simplest non-trivial knot (the trefoil). The formal claim needed: in $n_s < 3$ , all loops in any locally finite cause-plex are isotopically trivial; in $n_s \geq 3$ , non-trivial loops exist. This is a statement about the embedding space and follows from the classification of knots in low dimensions — a known result in topology, but not yet stated in cause-plex terms.

Status: The mathematics exists; the translation to cause-plex language remains.

Open Problem 2: Number-Volume Conjecture

The conformal factor in the Lorentzian metric is fixed by equating causal event count with spacetime volume. This is well-supported numerically but not proved in full generality. Separate from dimensionality but required for the full metric derivation.

References

Tangherlini, F.R. (1963). Schwarzschild field in n dimensions and the dimensionality of space problem. Il Nuovo Cimento, 27, 636–651. DOI: 10.1007/BF02784569
Tegmark, M. (1997). On the dimensionality of spacetime. Classical and Quantum Gravity, 14(4), L69–L75. DOI: 10.1088/0264-9381/14/4/002
Malament, D.B. (1977). The class of continuous timelike curves determines the topology of spacetime. Journal of Mathematical Physics, 18(7), 1399–1404. DOI: 10.1063/1.523436
Bombelli, L., Lee, J., Meyer, D., & Sorkin, R.D. (1987). Space-time as a causal set. Physical Review Letters, 59(5), 521–524. DOI: 10.1103/PhysRevLett.59.521
Wolfram, S. (2020). A class of models with the potential to represent fundamental physics. Complex Systems, 29(2), 107–536.

Cause-Plex and Spacetime | Cause-Plex and Quantum Mechanics | Series Map

Why 3+1: Observer-Selection Constraints on Spacetime Dimensionality in the Cause-Plex Framework

Contribute to Epiphysics