Distinction-Lattice Framework: Eigenforms, Bisimulation, and the Realizability Boundary of Self-Reference

#Abstract

A conservative formal version of the Distinction-Lattice Framework (DLF): stable entities are bisimulation classes of observational behaviour, certified coinductively, with realizability separated from fixed-point existence. This version (v0.2.0, superseding v0.1.1) narrows novelty claims, corrects the JPCUB/Landauer bridge, and records the QNFO archive as not yet assigned.

#1. Scope and Status

This document is the formal statement of the Distinction-Lattice Framework (DLF), and it is deliberately conservative. It makes one structural proposal, one definitional commitment, and one conjecture with explicit falsification conditions. It does not claim a solved theorem where a conjecture stands, and it does not bridge to physical claims beyond what the referenced programs independently support.

The central proposal is this: distinction precedes number, behavior precedes proof, and the Lawvere diagonal is the invariant boundary of all self-referential notation. The framework's technical content is the formalization of that sentence.

#2. The Substrate: Distinctions and the Realizability Filter

Let the primitive act be not the element $x \in S$ but the distinction $\delta = (A \mid A^{\complement})$: a partition of a state space $S$ into what an observer can tell apart and what it cannot.

Definition (Distinction). Let $\mathcal{S}$ be a state space. A distinction on $\mathcal{S}$ is an equivalence relation $\sim$ on $\mathcal{S}$; the class of $A$ under $\sim$ is what the distinction counts as the same, and its complement is what the distinction counts as other.

Definition (Realizability filter). Let $R \subseteq \mathcal{P}(\mathcal{S})$ be a family of subsets that is closed upward (if $A \in R$ and $A \subseteq B$ then $B \in R$) and closed under finite intersection (if $A, B \in R$ then $A \cap B \in R$). Call $R$ a realizability filter. A distinction $\delta$ is $R$-realized when its realized side lies in $R$.

The filter is the framework's definitional commitment (axiom D5 in the program's internal numbering): a distinction is real if it can be the output of an observation. Closure under finite intersection is the formal shape of "observations compose"; upward closure is the shape of "if something is observable, anything it entails is at least as observable."

#3. Stable Entities as Bisimulation Classes

The DLF identifies entity with stable observational behavior.

Definition (Behavioral functor). Let $F$ be an endofunctor on a category of states and observations, with the intended reading that an $F$-coalgebra $\delta \to F\delta$ is one step of observed behavior from a distinction.

Definition (Realizability as behavioral fixed point). A distinction $\delta$ is realized, written $R(\delta)$, if and only if $\delta$ is a fixed point of the behavioral functor $F$ up to bisimulation:

Proposition (Lambek). If $F$ admits a final coalgebra, then the final coalgebra $\Omega$ satisfies $\Omega \cong F\Omega$ (Lambek's lemma). The space $\Omega$ is then the space of all observable behaviors; every realized distinction is, up to bisimulation, a point of $\Omega$.

Consequently, a stable entity is a bisimulation class of observational behavior — an equivalence class of states that no observation within the filter can separate. The framework's structural claim is that this is what "being the same thing" means operationally, and that set-theoretic identity is the special case in which the filter is total.

#4. Eigenforms: Objects as Fixed Points of Their Own Observation

Kauffman's eigenform formulation reads an object as a solution of

where $F$ is the operation of the object's own observation (Kauffman, arXiv:1102.2048). The DLF adopts this reading and places it in the coalgebraic setting of Section 3: an eigenform is a fixed point of the behavioral functor, and the final coalgebra $\Omega \cong F\Omega$ is the universal eigenform.

The framework does not claim priority over the eigenform concept; it claims only the specific reading that bisimulation up to the realizability filter is the correct notion of solution for self-referential equations. The novelty statement is correspondingly narrow: the contribution is the filter, not the fixed-point equation.

#5. The Self-Reference Boundary: The Lawvere Diagonal

Lawvere's diagonal argument shows, in categorical form, that a cartesian closed category with an object $A$ admitting a fixed point for every endomap of $A$ cannot contain a surjective map $A \to Y^A$ (Lawvere, "Diagonal arguments and cartesian closed categories"). Every self-referential construction — from Cantor's theorem to Goedel-type incompleteness to fixed-point combinators — is a special case of this diagonal.

The DLF's reading: self-referential notation always asks for a fixed point of an endomap built from the notation itself. The diagonal is therefore the invariant boundary of self-referential notation: where a notation can name itself, the Lawvere diagonal delimits what that naming can achieve.

#6. The Central Separation: Realizability Is Not Fixed-Point Existence

The framework's central conjecture separates two notions that are routinely conflated:

Conjecture (Realizability boundary). Fixed-point existence does not imply realizability. There exist distinctions $\delta$ such that the equation $\delta \cong F\delta$ has solutions in the ambient category, yet no solution lies in the realizability filter $R$ — and, conversely, there exist filters for which every realizable fixed point is a bisimulation class of the final coalgebra.

Falsification conditions (any one of these refutes the conjecture):

1. Exhibit a realizability filter $R$, a behavioral functor $F$ with a final coalgebra, and a fixed point $\delta \cong F\delta$ such that $\delta \in R$ but $\delta$ is not a bisimulation class of $\Omega$.
2. Show that for every filter closed upward and under finite intersection, every fixed point is realized — i.e., that the filter never cuts.
3. Show that the diagonal boundary collapses: construct a notation that names itself through the Lawvere diagonal without leaving $R$.

The conjecture is falsifiable, and the author does not currently know whether it holds in full generality; the framework's value is that it forces the question into a form where the answer is checkable.

#7. The JPCUB/Landauer Bridge (Corrected)

The program's thermodynamic bridge (internal reference QNFO.JPC.003) states a falsifiable connection between the realizability filter and the Landauer bound: the minimal cost of realizing a distinction is bounded below by the information the filter must discard. Version v0.1.1 corrected this bridge: it is not established that the filter's cost saturates the Landauer bound, and the bridge is a conjecture to be tested against the JPCUB program's measurements, not a derived result of this framework.

#8. Open Problems

1. Find a natural class of categories and functors for which Conjecture 6 holds with proof (the author's current candidate: accessible categories with finitary behavioral functors).
2. Characterize the realizability filters that correspond to physical observations (which upward-closed, intersection-closed families arise from actual measurement apparatus?).
3. Sharpen the JPCUB/Landauer bridge into an inequality with explicit constants.
4. Determine whether the realizability boundary is computable: given $F$, $R$, and $\delta$, is $R(\delta)$ decidable?

#9. References

• Kauffman, L. H. Eigenform. arXiv:1102.2048.
• Lambek, J. A fixpoint theorem for complete categories. Mathematische Zeitschrift 103 (1968) 151–161.
• Lawvere, F. W. Diagonal arguments and cartesian closed categories. In: Category Theory, Homology Theory and their Applications II, Lecture Notes in Mathematics 92 (1969), Springer, 134–145.
• QNFO.JPC.003 — internal program reference for the Landauer-bound bridge (conjectural, per Section 7).

Spotted a typo or a possible error? Select the passage and choose Report, or report it here.