motif
the non-linear utterance of neural: a recurring subgraph shape, meaning carried by topology rather than by order. where a sentence is a line, a motif is a form. motifs are the morphemes of the language — and the one primitive that carries meaning its links do not. the lineage is network motifs from graph theory; the full grounding is in frontier.
what a motif is
one shape, read by six convergent lenses (see frontier):
- a typed string diagram — a wiring of words with an interface (open wires await arguments, closed wires are a finished meaning)
- an operadic operation — the unit the motif algebra substitutes
- a higher-order cell — a triad is a filled 2-cell, an axon its filled face; it carries its own Laplacian and is promotable to a particle (homoiconic)
- a functional circuit — a shape that computes (see the table)
- a persistent homology class — a loop that is born and survives across scale, not an exact template
- a construction — a form bound to a meaning, the same substance as a word at lower schematicity
a motif is the fixed point where these coincide: a typed, composable, promotable, temporally-transitioning higher-order cell.
the shapes, and what they compute
each shape carries a function, not only a form:
| motif | shape | function |
|---|---|---|
| triad | A→B, B→C, A→C | closure — trust / relevance settled |
| star | one particle linked by many | a hub, a definition |
| chain | sequential links | inference (a sentence) |
| diamond | convergent–divergent | corroboration — many paths agree |
| cycle | A→…→A | self-reference, feedback, or contradiction |
| feed-forward | A→B, A→C, B→C | a persistence filter — passes meaning only if upstream support is sustained |
| co-citation | many neurons link one pair | consensus |
the motif algebra
motifs compose — this is how neural reasons without stated rules. formally, operad substitution:
- concatenation — two motifs side by side (the monoidal tensor)
- nesting — a motif plugged into a hole of another (substitution)
- intersection — a shared interface (pullback)
- complement — the under-represented, forbidden subgraph; an anti-motif is itself meaning
- morphism — a motif mapped onto a new frame. this is metaphor: TIME-IS-MONEY maps the COMMERCE star onto the TIME frame. the morphism operator is how a finite motif vocabulary covers unbounded meaning, and the most powerful thing a motif does
because the interpreting dialect is a functor, defining meaning on the sigil generators forces a meaning on every composite motif — composition for free.
the fingerprint
every motif carries a triple that survives renaming:
- persistence — whether the shape is real (born early, survives scaling) or noise; via zigzag, whether it is about to be born or to die
- curvature — positive (a settled, tightly-bound meaning) or negative (a fragile bridge, a polysemy fault line)
- cohomology (H¹) — zero (internally coherent, a stable morpheme) or nonzero (a built-in paradox, an odd cycle of opposite links)
(persistence, curvature, H¹) is a complete, computable signature of a motif's health and semantic role.
discovery, and why first-class
motifs are discovered, not designed: the tri-kernel in tru surfaces them (diffusion finds bridges, springs reveal stable shapes, heat weights by adoption), and the canonical motif set is the one that most compresses the cybergraph (minimum description length). a motif that recurs and holds focus becomes a unit a dialect can name.
motifs must be first-class because plain diffusion is structurally blind to them — message passing is bounded by the 1-Weisfeiler-Leman test and cannot count triangles or cycles. φ* sees the non-linear shape of meaning only when motifs are their own objects. and a motif's transitions P(motif' | motif) are a dialect's grammar of expectation — the engine that turns structure into foresight.