How does the first-order and second-order distinction unify classical and modern argument theory?
This explores Wagemans's claim that a single distinction — whether an argument reasons about the world (first-order) or about other arguments (second-order) — bridges the old Greek divide between internal and external proofs and the modern divide between sound and fallacious reasoning.
This explores Wagemans's claim that one structural cut — first-order versus second-order arguments — can do double duty, mapping onto both a classical and a modern way of carving up argument theory. The corpus has this as its anchor Do first-order and second-order arguments unify classical and modern divisions?. The classical side is Aristotle's split between *internal* topoi (reasoning drawn from the substance of the matter) and *external* ones (reasoning that leans on something outside it, like a witness or an authority). The modern side is the dialectical split between reasonable moves and fallacious ones. Wagemans's wager is that these two inherited distinctions are really the same line drawn twice: a first-order argument reasons directly about the thing under dispute, while a second-order argument reasons about the standing of a claim or its source. That reframing matters because it relocates fallacy from being a matter of conversational misbehavior to being something you can read off the *formal shape* of an argument — fallacy as structure, not just bad manners in a debate.
What makes this more than a tidy relabeling is where it leads: a closed, predictive classification of all arguments rather than an ever-growing list. The first-order/second-order axis is one of three orthogonal axes in Wagemans's "Periodic Table" of argument schemes Can three axes organize all possible argument schemes?, sitting alongside subject-predicate structure and the pairing of proposition types. Together they generate a finite coordinate space, the way the chemical periodic table generates elements — which is exactly the move from Walton's open-ended family of 60-plus schemes to a principled ordering Can argument schemes be organized by formal principles instead of lists?. So the unification isn't decorative; it's the thing that lets you predict scheme types nobody has catalogued yet, the same way gaps in the periodic table predicted undiscovered elements.
Here's the turn a curious reader might not expect: that second-order category — arguments about a claim's authority rather than its content — is precisely where today's AI systems break down. When an argument's force comes from *who* is making it (reputation, track record, standing), a language model that only sees text loses the social world that gives expert claims their weight Can language models distinguish expert arguments from common assumptions?. In Wagemans's terms, models struggle with the second-order layer because that layer points outward to a world the model can't access. And classifying schemes at all turns out to be unusually hard for LLMs — they plateau where other language tasks soar, because recognizing an inferential pattern means integrating cues scattered across the whole text rather than reading local surface features Why does argument scheme classification stumble where other NLP tasks succeed?, with even the strongest models barely clearing the bar and only with worked examples in the prompt Can large language models classify argument schemes reliably?.
The deeper payoff of a formal classification is that it makes arguments *contestable* in a way prose never is. Once an argument has a known structure — premises, warrant, the type of inferential move — you can point at the exact joint you reject, which is what structured argumentation frameworks give you and raw model output cannot Can formal argumentation make AI decisions truly contestable?. That same logic is why feeding a scheme's critical questions back to a model as explicit prompting steps sharpens its reasoning: it forces the implicit warrant into the open Can structured argument prompts make LLM reasoning more rigorous?. The first-order/second-order distinction, then, is the small hinge on which a much larger door swings — from a list to a system, and from arguments you can only feel are wrong to arguments you can show are wrong.
Sources 8 notes
Wagemans proposes that the first-order vs second-order argument distinction reflects both the classical internal-external topoi divide and the modern reasonable-fallacious distinction. This suggests fallacy theory operates through specifiable formal-linguistic structure rather than purely dialectical criteria.
Wagemans's Periodic Table maps all argument schemes onto coordinates across three axes: subject-predicate structure, first-order versus second-order reasoning, and proposition-type pairings. This combinatorial approach replaces Walton's open-ended list with a closed, systematic space enabling computational analysis and discovery of unstudied scheme types.
Wagemans shows that three orthogonal axes generate a closed, finite classification space for all argument types, replacing the family-resemblance logic behind Walton's 60+ schemes. This mirrors the chemical periodic table's shift from contingent lists to predictive structure.
LLMs lose the social context that gives expert claims their force—reputation, track record, and standing—because they process only text, not the social world where expertise is built and evaluated.
Scheme classification requires recognizing inferential patterns across distributed text spans, not local surface features. Models plateau at F1 0.55–0.65 while the same systems exceed 0.80 on component tagging and stance, suggesting the integrative reasoning demand is fundamentally different.
Zero-shot prompting fails uniformly across models. Few-shot with scheme descriptions helps, but only larger models exceed F1 0.55, with Claude reaching 0.65. Smaller models plateau around 0.53, suggesting a representational capacity threshold.
Dung-style argumentation structures AI outputs as traversable attack/defense graphs, allowing users to identify and contest specific premises. Standard LLM outputs lack this structure, making it impossible to pinpoint which claims users actually reject.
Applying Toulmin's argument model as explicit prompting steps (CQoT) improves LLM reasoning by forcing models to identify warrants and backing rather than skipping implicit premises. The method catches failures that standard chain-of-thought prompting allows.