Persuasion theory (the rigorous spine)

The academic layer beneath the practitioner delivery-craft of communication: the tested frameworks of message design and persuasion. Replication status is flagged per item — this is theory, and some of it is oversold (the honest weak levers are marked, not hidden).

The practitioner corpus tells you how to say it; this tells you why it persuades and which move fits which audience. The organizing insight is the router (ELM — the elaboration-likelihood model):

The router — ELM decides everything downstream

Elaboration Likelihood Model (Petty & Cacioppo, well-established): attitude change runs on two routes. Central route engages when the audience is motivated (topic matters to them) AND able (knowledge, time, focus) — they scrutinize argument quality, and the resulting attitudes are durable and resistant. Peripheral route dominates under low motivation/ability — people lean on cues (credibility, likeability, number of arguments, emotion, fluency). DO THIS: diagnose involvement first, then match — high-involvement (someone evaluating your product/thesis) → lead with substance, evidence, structure (cues backfire); low-involvement (someone scrolling) → win on credibility signals, a likeable voice, one vivid claim (arguments won't be processed). Detail: the-two-routes. This is why the same message wins one room and slides off another — it's an involvement mismatch.

The branches

the-two-routes — ELM in full; the involvement × ability switch; what persuades on each route.
the-three-appeals — Aristotle: ethos/pathos/logos + kairos + the five canons; why logos-only slides off.
narrative-and-concreteness — why story out-persuades argument: transportation theory, Made to Stick (SUCCESs), the identifiable-victim effect, the curse of knowledge.
peripheral-cues — the low-involvement levers: Cialdini's 7, processing fluency / mere exposure / rhyme-as-reason, framing (with the honest weak-lever caveat), primacy/recency.
clarity-and-defense — Grice's maxims (the concision checklist + implicature), inoculation ("prebunking"), dual-coding / cognitive load.

The cross-cutting map (how they compose)

ELM routes: high-involvement → the central-route set (argument quality, logos, Grice + inoculation); low-involvement → the peripheral set (Cialdini + fluency) and story/emotion (narrative-and-concreteness). Over all of it: structure with primacy/recency, control salience with framing, and make everything concrete, fluent, single-protagonist. Almost all of this transfers to written and AI-generated content (it's message design, not live delivery) — the machine's generator can apply it directly; dual-coding is the only partial (needs a visual channel).

Honest ledger. Well-replicated: ELM, transportation, mere-exposure/fluency, primacy-recency, inoculation, identifiable-victim, dual-coding. Foundational-but-not-"tested": Aristotle, Grice (they're frameworks, not experiments). Oversold — hold loosely: message-level gain-vs-loss framing (meta-analyses find effects small and inconsistent, r ≈ −.04). The core framing/salience insight is robust; the gain/loss dial is not a rule.

Source: agent research 2026-07-16; per-branch citations in the child notes.