Can unfilled cells in the periodic table represent undiscovered argument schemes?
This explores whether the 'periodic table' metaphor for argument schemes is literal enough that empty coordinates predict real, not-yet-cataloged argument types — the way Mendeleev's gaps predicted undiscovered elements.
This explores whether the periodic-table metaphor for arguments does real predictive work — whether a blank cell forecasts a discoverable argument scheme the way a gap in Mendeleev's table forecast gallium. The corpus says yes, and the move is more than analogy. Wagemans's table maps every argument scheme onto three orthogonal axes: whether the claim is about a subject or a predicate, whether the reasoning is first-order or second-order, and which proposition types are paired Can three axes organize all possible argument schemes?. Because those axes are combinatorial, the space they define is *closed* — every coordinate either names a scheme people already use or names one nobody has studied yet. That second category is exactly the unfilled cell: a structurally valid combination awaiting an example. This replaces Walton's open-ended, ever-growing list with a finite grid, and a finite grid has gaps you can point at.
The deeper reason the metaphor holds is that it rhymes with how other reasoning structures in this collection turn out to be combinatorial rather than list-like. Long chain-of-thought reasoning, it turns out, has a 'molecular bond' structure — deep reasoning behaves like covalent bonds, self-reflection like hydrogen bonds, self-exploration like weak van der Waals forces — and only certain combinations form stable wholes Does long chain of thought reasoning follow molecular bond patterns?. Neural networks, similarly, decompose tasks into modular, recombinable subnetworks rather than one monolithic blob Do neural networks naturally learn modular compositional structure?. The shared lesson: when a domain is built from a small set of primitives combined along independent axes, the empty combinations aren't noise — they're predictions. That's what makes a periodic table a periodic table and not just a sorted list.
But a predicted cell is only useful if you can actually go find what belongs there, and here the corpus adds a sharp caveat. Machines are still mediocre at recognizing the schemes that *do* exist: LLMs classify argument schemes acceptably only with few-shot examples plus written descriptions, and even then the best model tops out around F1 0.65 while smaller ones plateau near 0.53, hinting at a representational-capacity floor Can large language models classify argument schemes reliably?. So the table can *say* a cell should be occupied, but our automated tools can't yet reliably confirm a candidate fits it. Discovery of the unfilled cell remains, for now, a human job that the grid scaffolds rather than automates.
The genuinely interesting twist comes from connecting this to how creativity itself gets carved up. One line of work splits creative reasoning into three modes — combinational, exploratory, and transformational Can LLMs reason creatively beyond conventional problem-solving?. A periodic table of arguments is a machine for the first two: combinational discovery (mix axis values you haven't mixed before) and exploratory discovery (walk the defined space looking for the empty seats). What it cannot do is the third — transformational discovery, where you add a *new axis* and redraw the table entirely. So the honest answer is layered: unfilled cells absolutely can represent undiscovered argument schemes *within the three-axis world Wagemans drew*. The schemes that would force a fourth axis are precisely the ones no grid can predict — which is the same boundary every closed combinatorial system in this collection runs into.
Sources 5 notes
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.
Deep-Reasoning (covalent), Self-Reflection (hydrogen bonds), and Self-Exploration (van der Waals forces) form stable distributions in effective Long CoT. Mixing these stable structures from different teachers destabilizes learning despite matched performance metrics.
Pruning experiments reveal that neural networks implement compositional subroutines in isolated subnetworks, with ablations affecting only their corresponding function. Pretraining substantially increases the consistency and reliability of this modular structure across architectures and domains.
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.
Research identifies combinational, exploratory, and transformational reasoning as distinct creative modes grounded in cognitive science. Existing LLM reasoning methods address only conventional problem-solving, leaving creative paradigms unaddressed and potentially explaining diversity collapse in ideation.