Logical Realism and Conventionalism

The Problem of Logical Authority

Logical inquiry asks what follows from what, why the inference is correct, and what entitlement we acquire by recognizing that correctness. From the premises that every mammal is an animal and that an individual is a mammal, the conclusion that the individual is an animal follows. The biological predicates supply content, while the arrangement of quantification and predication allows a conclusion to be derived independently of the particular identity of the individual. This elementary case invites a philosophical question more difficult than checking the inference: is its necessity a fact discovered, an artifact of the meanings of its expressions, or an obligation instituted by conventions that govern a language?

The answer matters because the authority of logic extends beyond an individual's confidence. A person can believe that an argument is valid and be mistaken. An entire intellectual community can employ rules that turn out to be unsound for its intended interpretation. Mathematical logicians also design multiple formal calculi and prove different metatheorems about them. To say that a proof depends on rules chosen by human beings leaves open whether the rules capture objective relations, what makes them suitable, and why applications of a rule can be right or wrong.

Four inquiries must be distinguished at the outset. The metaphysical inquiry asks what makes logical facts or consequences hold. The semantic inquiry asks how a logical vocabulary gets its meaning and what remains invariant across reinterpretations. The epistemic inquiry asks how logical principles and consequences can be known. The normative inquiry asks why they provide reasons for assertion, belief, inference, or revision. A fifth, methodological inquiry asks when a language or calculus is well chosen for the purposes of a science. These questions interact, but an answer to any one does not by itself settle the remaining four (Frege, 1918/1956; Carnap, 1937; Quine, 1936; Tarski, 1936/1983).

The guiding claim is that an adequate account of logical authority must differentiate specification from correctness and formal correctness from justified application. This distinction neither assumes an eternal realm of logical objects nor treats rigorously established consequence as an arbitrary human decision. It requires every philosophical position to say at which level its appeal to independence, convention, truth, meaning, or proof is supposed to operate.

Realism and Conventionalism as Families of Positions

Logical realism assigns objective standing to some logical laws, truths, consequences, structures, or properties. One version regards logical relations as relations among propositions that exist independently of anyone's thinking. Another takes logical truth to be a mind-independent feature of what meanings or logical structures determine. A third defends the objectivity of consequence without making an additional ontological commitment to a distinct realm of propositions. A fourth gives priority to the rational or normative correctness of inferential practice. These positions can overlap while differing sharply in metaphysical strength. Logical realism must also be separated from realism about material objects, from mathematical Platonism, and from semantic realism in particular disputed discourses.

Conventionalism is similarly diverse. Notational conventionalism concerns the choice of signs and symbols. Syntactic conventionalism concerns the selection of formation and transformation rules. Semantic conventionalism concerns the role of rules in fixing meaning or truth conditions. Framework conventionalism concerns the adoption of formal languages or conceptual schemes. An epistemic form of conventionalism claims that accepting or understanding such rules can justify certain logical judgments. A far stronger metaphysical doctrine would claim that agreement itself makes independently understood propositions true. Moving from any modest thesis to the last requires a separate argument (Boghossian, 1996; Ebbs, 2011).

A formal language can be introduced by decision without its theorems being freely alterable afterward. If conjunction means that both conjuncts hold, then a structure in which A and B are true makes A true. Once the relevant definitions are fixed, this is a determinate mathematical statement. Conversely, the fact that a formula is derivable in a stipulated calculus does not prove that the calculus is a good representation of every possible form of reasoning. The correctness of implementation also depends on whether the stipulated symbol actually represents the relation intended in the application.

A further alternative is logical pluralism, the thesis that more than one logic can be correct. Pluralism is independent of conventionalism in principle. One may hold that multiple objective consequence relations exist and are suited to different intellectual settings, or that several systems are useful conventions while only one fully captures some unrestricted standard. Similarly, monism can coexist with the ordinary recognition that alternative symbolizations are possible (Beall & Restall, 2006; Shapiro, 2014).

The historical dispute is best reconstructed as a series of arguments about these commitments. Its genuinely philosophical content appears whenever the validity of an inference, the choice of a logical system, the meaning of an expression, and the justification of a theory are treated as distinguishable problems whose relations require reasons.

Aristotle: Necessary Consequence and Demonstration

Aristotle's Prior Analytics offers a systematic account of the syllogism as an argument in which something different from what was laid down follows of necessity because of what was laid down (Prior Analytics, I.1, 24b18–20). This formulation makes necessity central to correct deduction. In the traditional first-figure pattern later called Barbara, the universal relation between two classes and the subsumption of another under the first warrant a further universal conclusion. A person can draw the inference or fail to draw it; what the premises necessitate is a distinct question.

Aristotelian logic belongs to a framework of terms, predication, and forms of scientific knowledge rather than to contemporary first-order model theory. Its historical contribution should accordingly be reconstructed from the original architecture. The treatment of syllogisms proceeds by investigating valid moods and figures, conversions, and proofs. The Posterior Analytics places these patterns in a theory of demonstration that requires explanatory premises. An argument may be formally correct and yet fail to explain the cause of the conclusion. Formal consequence and scientific understanding already answer different questions.

The Stoics pursued a partially different logical organization, developing arguments around propositional relationships involving conditional, negation, and disjunction. Our textual evidence is mediated through ancient reports and surviving fragments rather than through a complete work by Chrysippus. The historical existence of distinct logical traditions establishes plurality of logical projects. It does not demonstrate that all proposed inferential schemes are equally correct, because these traditions investigated validity by distinguishing permissible inferences from counterexamples.

The ancient material also limits a retrospective reading of the modern debate. Aristotle did not argue against a twentieth-century semantic conventionalism, and the Stoics did not formulate the modern problem of selecting a formal language. What endures is the conceptual difference between an act of reasoning and the relation that makes it a correct deduction. A contemporary realist may treat this as evidence for objective relations, while a conventionalist can maintain that formalizing the relation requires selective interpretive decisions. Ancient logic supplies the foundational question without deciding the later opposition.

Kant: General Logic, Apriority, and Judgment

Kant distinguishes general logic, which abstracts from the content of cognition and examines its formal employment, from transcendental logic, which investigates a priori conditions of cognition insofar as it relates to objects. The contrast prevents a simple identification of logical form with all the conditions of objective knowledge. Kant's account of analytic and synthetic judgments in the Critique of Pure Reason also influenced later discussions of truths grounded in concepts, although his formulation and philosophical aims differ from Carnap's theory of constructed formal languages (Kant, 1781/1787, Introduction, IV, A6–7/B10–11).

Analyticity, apriority, and necessity are distinguishable categories. Analyticity concerns a supposed dependence on conceptual containment or meaning; apriority concerns the source of epistemic justification; necessity concerns modal status. Kant's thesis that mathematical judgments are synthetic a priori directly resists the idea that mathematical knowledge merely unpacks prior meanings. This is one historical reason why an appeal to stipulated definitions does not exhaust the philosophical problem of mathematical and logical cognition.

The authority of a rule also depends on how it is applied. If a speaker adopts an inference rule but cannot reliably distinguish instances of it, the stipulation has not by itself supplied a complete epistemology of correct reasoning. Likewise, an argument that conventions create their own legitimacy must explain how the conventions can be instituted and used without presupposing some capacity to infer. This challenge was sharpened in the twentieth century by Quine's criticism of truth by convention, though its conceptual antecedent lies in a much wider rationalist concern about conditions of justification.

Kant's contribution to this investigation therefore concerns the roles of form, judgment, and warrant. It would be historically misleading to treat his logic as a direct formal predecessor of contemporary semantic frameworks. The philosophical continuity lies in the demand to account for the difference between the rules of correct thinking and the grounds of knowledge of objects.

Bolzano: Propositions in Themselves and Logical Variation

Bernard Bolzano's Wissenschaftslehre (1837) develops the notion of Sätze an sich, propositions in themselves, whose identity and truth are independent of the contingent psychological acts through which they may be grasped. A proposition may be believed, disbelieved, or never entertained while retaining its status as a possible bearer of truth. This account offers a powerful argument against psychologism: the content of a logical relation cannot simply be the episode of a thinker judging it, since the same proposition may be considered on different occasions by different persons (Bolzano, 1837; Rusnock, 2000).

Bolzano's account of deducibility employs systematic variation of selected constituent ideas. The method anticipates later attempts to identify consequence through the substitution or reinterpretation of nonlogical constituents. Yet choosing which ideas vary is itself philosophically consequential. Fixing too much content can make empirical connections appear logically invariant; varying too much can eliminate connections one had intended to preserve. A condition of formal constancy must therefore accompany the method of variation.

The realist achievement is clear: Bolzano offers a theory of shared objective contents on which logically correct and incorrect relations may depend. The philosophical challenge remains explaining how those contents are individuated and which among their relations are specifically logical. The mere postulation of objective propositions does not determine the boundary of logic.

A further conceptual lesson emerges from the distinction between the objectivity of a proposition and the objectivity of a choice of logical vocabulary. The former concerns the independence of the content from mental acts. The latter concerns why some parts of that content are held fixed under reinterpretation. Bolzano's system shows that the dispute between logical realism and conventionalism persists even after psychologism is rejected.

Frege: Logic as the Science of Truth

Frege's Begriffsschrift (1879) introduced a formal language capable of analyzing relations that escaped traditional term logic. His Grundlagen der Arithmetik (1884) and Grundgesetze der Arithmetik (1893–1903) pursued a logicist account of arithmetic. The mathematical ambition was sustained by a philosophical distinction between objective truth and the subjective process of thinking. In “Der Gedanke” (“The Thought,” 1918–1919), Frege distinguishes laws of truth from psychological laws describing how people actually think (Frege, 1918/1956).

The argument for logical objectivity proceeds from the possibility of shared judgment, genuine error, and correction. If truth depended on someone's assent, disagreement would concern separate personal states rather than a common issue. A shared thought can be entertained by many people while preserving its identity and truth value. Logical laws therefore concern objective relations to truth; rules for judging and asserting derive their authority from those relations rather than from the statistical regularity of assent.

The metaphysical and epistemic costs of this position must be distinguished. The concept of objective thought gives an account of the common subject of reasoning, but the theory must still explain how finite thinkers grasp such thoughts and identify valid relations among them. Objective logical laws cannot serve as an explanation of successful reasoning merely by being asserted. Their recognition requires an epistemology of proof, understanding, and correction.

Russell's identification of a contradiction in Frege's unrestricted extension principles undermined the consistency of a central component of the Grundgesetze. The historical result demonstrates that even a philosophical commitment to objective logical truth cannot guarantee the adequacy of a proposed formalization. A realist must be able to recognize that a system designed to exhibit truth has formal defects. The ideal of objectivity is itself a reason to subject axioms and rules to criticism.

Frege's position remains central because the correctness of reasoning is separated from its sociological acceptance. Conventionalist accounts of adopted languages need an explanation of how communal mistakes remain possible. Frege's insistence on shared objective content supplies that demand in especially clear form.

Poincaré: Convention, Geometry, and Scientific Structure

Henri Poincaré's La Science et l'Hypothèse (Science and Hypothesis, 1902) develops an account of geometrical conventions and the organization of physical science. Alternative geometric frameworks may be employed in conjunction with appropriate auxiliary assumptions; which is preferable can depend on simplicity, convenience, and the organization of empirical results. These choices are constrained by scientific inquiry. They are not arbitrary ways to declare every empirical proposition true (Poincaré, 1902; Stanford Encyclopedia of Philosophy, “Henri Poincaré”).

Poincaré also distinguished geometrical conventions from mathematical practices that involve induction and constructive insight. His criticism of wholesale logicist reduction therefore complicates any inference from conventional geometry to the claim that all mathematics and logic consist of freely selected definitions. The historical conventionalist tradition is internally differentiated.

His philosophical contribution may be understood as an analysis of theory choice. One can select a formal framework while retaining substantive criteria for its usefulness. Changing geometry together with adjustments to physical assumptions need not change every observational prediction. But the joint theoretical system remains accountable to experience. Similarly, the choice of a logical calculus can be influenced by practical and representational aims without making the derivations within it arbitrary.

Poincaré's conventionalism is an important precursor of twentieth-century linguistic conventionalism, though the later development introduced different objects of analysis: formal syntax, semantic consequence, analyticity, and language frameworks. Their genealogy is a sequence of intellectual transformations, not a single unchanging doctrine. The distinction is essential for historical fidelity.

Logicism, Tautology, and the Limits of Stipulation

Russell and Whitehead's Principia Mathematica (1910–1913) sought to reconstruct mathematics through a sophisticated logical apparatus. Its use of axiomatic conventions and formal notation does not entail that mathematical truths are created by arbitrary definitions. Logicism is a proposed reduction of the justification and derivability of mathematical theorems to an underlying logic. Its success depends on the soundness, consistency, and philosophical status of that underlying apparatus.

Wittgenstein's Tractatus Logico-Philosophicus (German edition 1921; bilingual English edition 1922) develops a different conception of logical propositions as tautologies (6.1–6.13). Under the Tractarian picture, a tautology can reveal the logical structure of representation without depicting a contingent state of affairs. A truth-functional display may establish that a given propositional schema is true under all valuations, yet this fact does not imply that a human decision created the structure of possibility to which the schema belongs (Wittgenstein, 1922).

The distinction between the truth-functional part of logic and broader logical consequence matters here. Propositional tautologies can be checked with finite truth tables, whereas first-order quantification demands treatment of domains and interpretations and does not possess a general truth-table decision procedure. Calling a proposition tautological cannot settle the status of every kind of logical consequence without specifying the logical system concerned.

Wittgenstein influenced Carnap's turn to formal language, but Carnap transformed the issue by treating logical syntax as an object of explicit construction and comparative evaluation. The two theories should not be identified simply because both separate logical from empirical propositions. Their accounts of language, the philosophical work of logical syntax, and permissible alternative forms of logic differ substantially.

The broader lesson from logicism and tautological theories is that a formal proof has both a system-relative character and a substantive philosophical interpretation. Determining the consequences of an adopted calculus is one task; justifying its philosophical role is another.

Carnap's Principle of Tolerance and Logical Syntax

Rudolf Carnap's Logische Syntax der Sprache appeared in 1934, followed by the revised English translation The Logical Syntax of Language in 1937, translated by Amethe Smeaton. The book proposed an explicitly metalinguistic method of examining formal languages, their rules for forming expressions, and their transformation rules for inference. The methodological premise was that a rigorous philosophical investigation of mathematics and science could compare linguistic forms rather than search for a uniquely legitimate ordinary language (Carnap, 1934/1937; Stanford Encyclopedia of Philosophy, “Rudolf Carnap,” “Logical Syntax of Language”).

Carnap's principle of tolerance licenses the adoption of different forms of language for different tasks, provided that the relevant rules are stated clearly. This freedom is often misunderstood as the thesis that any claim can become an objective truth simply because a group agrees to accept it. In the more precise reconstruction, one can institute a calculus and then distinguish correctly derived from incorrectly derived formulas relative to that calculus. One can also ask which calculus is suitable for the work at hand. The method gains its intellectual power from the combination of freedom of specification with rigorous constraint after specification.

The contrast between Carnap's Languages I and II exemplifies the program. Language I supports a more restricted constructive orientation; Language II provides stronger classical resources. Their mathematical consequences and powers of expression differ. These differences can be studied precisely without deciding in advance that exactly one language alone represents all legitimate mathematical reasoning. Carnap's plural options serve an analytical purpose: they turn claims about what mathematics requires into determinate questions about the explicit rules of a particular system.

The philosophical difficulty concerns the metalanguage. Stating that a formula follows from certain transformation rules already requires recognizing a finite sequence, an application of a rule, identity of formulas, and the relation of a proof step to its predecessors. If the authority of all inference is supposedly created by a rule stipulated in the language under investigation, the account risks presupposing inferential legitimacy in the act of identifying a valid rule application. Carnap can respond that his project is an explication of formal language rather than an attempt to obtain a noncircular justification of every possible rule of reasoning.

This distinction is decisive. The claim that syntactic methods clarify a philosophical problem is compatible with the claim that all adequate clarification employs some background logical resources. The stronger claim that syntax alone metaphysically creates every logical truth demands further support. Reconstructing Carnap through the former rather than automatically attributing the latter produces a historically sounder understanding of his method.

Carnap's Semantic Turn and Framework Questions

Carnap's later work broadened the investigation from syntax to semantics. Introduction to Semantics (1942), Meaning and Necessity (1947), and “Empiricism, Semantics, and Ontology” (1950) treated meaning, interpretation, and linguistic frameworks as central philosophical concerns. The development reflects the inadequacy of treating a purely syntactic description as a comprehensive account of all meaningful distinctions. Logical truth can be examined relative to an interpreted language and its semantic rules, not merely by inspecting the visual form of written symbols (Carnap, 1942, 1947, 1950).

In the 1950 essay, Carnap distinguishes internal questions formulated within a framework from external practical questions about adopting that framework. Within a number language, the assertion that there are prime numbers greater than ten is a determinate mathematical claim. Whether to adopt the numerical framework for some scientific purpose is a question concerning its utility, clarity, and role in inquiry. The distinction is designed to prevent a category of apparently metaphysical disputes from being confused with disputes over how best to conduct research.

The independence of internal correctness from individual preference is entirely compatible with the practical status of the adoption decision. Mathematical results cannot be changed by wishing them different once the pertinent formal commitments are fixed. Yet practical decision criteria may guide the selection among alternative languages. In this sense the conventionalist's freedom is neither unlimited nor wholly illusory.

A philosophical objection remains: the line between internal and external questions can be difficult to locate, especially where a framework change alters the objects, inferential practices, or empirical expectations of a theory. If the very standards for evaluating utility are themselves expressed within some language, the pragmatic evaluation may also require philosophical analysis of rationality. This is one reason the Carnap–Quine dispute extends far beyond the simple recognition that formal systems are stipulative.

The later Carnap cannot be treated as an unmodified restatement of his 1934 syntax project. His development shows that a philosophical position may preserve a commitment to clarity, tolerance, and explication while changing its formal and conceptual resources.

Quine's Challenge in Truth by Convention

Quine's “Truth by Convention” (1936) examined whether logical truths can be explained by explicit linguistic decisions. Its most durable objection concerns the attempt to justify inference by appeal to conventions whose application already requires inference. If a convention is introduced by a definition, one must still understand how the definition may be substituted in a proof. If a general convention licenses infinitely many instances, some method must identify those instances. Merely adopting the convention cannot explain the legitimacy of every logical operation used to apply it (Quine, 1936).

The challenge can be reconstructed in four steps. First, an ambitious conventionalist wishes to explain logical truths as true because of conventions. Second, conventions must be formulated as general rules, since the space of logical truths cannot feasibly be settled through the independent enumeration of all its members. Third, deriving a particular logical truth from a general rule requires at least some capacity for valid inference or rule application. Fourth, if that capacity itself is explained by a further convention, the same difficulty threatens to recur. Hence unrestricted truth by convention does not provide an assumption-free foundation for the validity of reasoning.

This argument has limited scope. It does not show that symbols cannot be assigned meanings by stipulation. It does not show that theorems of an axiomatic theory cannot be derived from chosen axioms. It does not establish that every system of logic must be discovered as a timeless metaphysical object. Its target is a circular explanatory program that purports to create the authority of all logical inference by conventions alone.

The historical interpretation must also avoid a familiar oversimplification. Gary Ebbs (2011) argues that Quine's 1936 criticism did not refute a metaphysical doctrine of truth creation that Carnap in fact held. On Ebbs's reconstruction, Quine's discussion belongs to his developing naturalistic perspective and even permits restricted ways of speaking of adopted postulates as conventional. An adequate history must distinguish the reception of a purported doctrine from what the primary philosophical project actually requires.

The central philosophical question left by Quine is whether a rule can have constitutive force without possessing foundational force. A rule may define what counts as a derivation within a calculus. This does not explain why the calculus should be trusted as a representation of an intended objective relation. That further trust requires reasons concerning soundness, meaning, applicability, and inferential goals.

Analyticity and Quine's Second Challenge

In “Two Dogmas of Empiricism” (1951), Quine criticized the analytic–synthetic distinction and the theory that each meaningful statement can be associated with a determinate isolated empirical confirmation condition. His examination of analyticity passes through appeals to meaning, synonymy, definition, interchangeability, and semantic rules. The objection is that these proposed explanations frequently rely on concepts that seem to require independent clarification of precisely the disputed distinction (Quine, 1951, §§1–6).

The critique bears differently on different conventionalist views. If a view simply states that a finite formal calculus has specified axioms and rules, the existence of such a calculus remains unchallenged. If it claims that an unrevisable class of ordinary sentences can be identified as true entirely by virtue of meaning, Quine demands a substantive criterion and justification. If it claims that logical laws are exempt from revision regardless of their contribution to scientific inquiry, Quine's confirmation holism provides a contrary picture in which statements acquire significance as part of larger systems.

Holism does not entail that any conclusion whatever becomes logically valid, or that logic has no important role in science. Rather, it challenges an epistemological picture that separates logical principles from all of the broader intellectual considerations under which a scientific theory may be revised. The precise distinction is between the practice of employing a logic within inquiry and the philosophical claim that its authority is grounded in a sharply bounded class of meanings independent of that inquiry.

Carnapian explication remains a possible reply. A rigorously specified notion of analytic truth for a constructed language need not reproduce every ambiguity of natural-language meaning. A formal explication can be valuable because it resolves an imprecise problem into tractable questions. What remains contested is the philosophical inference from formal explication to an unrestricted theory of apriority or necessity.

Paul Boghossian (1996) accordingly distinguishes metaphysical from epistemic forms of analyticity. The first may suggest that meaning somehow makes propositions true, whereas the second concerns whether grasp of meanings can confer justificatory entitlement. This distinction returns the dispute to a more productive question: which kinds of conceptual or inferential competence supply genuine grounds for knowledge, and why? Convention alone may be insufficient, but the epistemic role of understanding remains to be investigated.

Tarski's Model-Theoretic Conception of Consequence

Alfred Tarski's 1936 paper “On the Concept of Logical Consequence,” originally published in Polish and German, gave a rigorous account of consequence in terms of truth preservation under admissible reinterpretation. Its modern model-theoretic formulation may be written as follows: Γ ⊨ φ just in case every admissible structure in which each member of Γ is true also makes φ true. This captures the impossibility of a counterexample relative to the chosen language, interpretation rules, and range of structures (Tarski, 1936/1983; Stanford Encyclopedia of Philosophy, “Alfred Tarski”).

The definition has two major philosophical virtues. It is semantically transparent, because the consequence relation is articulated by truth conditions in a model. It is mathematically systematic, because a theory of models can be used to compare many different formal languages, including those with complex quantification and relational structure. A conclusion can be shown to be a consequence without requiring that every individual investigator happens to recognize that fact.

The familiar inference from “Every F is G” and “a is F” to “a is G” illustrates the method. In any interpretation on the usual first-order semantics where all Fs belong to the extension of G and a belongs to F, a also belongs to G. No admissible countermodel exists. The proof of this condition is a mathematical matter once the semantics are fixed. Nothing in the example depends on whether the interpreter likes the conclusion.

Yet a central philosophical question remains. Model-theoretic validity depends on which vocabulary is held fixed as logical, which interpretations are permissible, and whether mathematical structures represent all relevant possibilities. If those specifications are selective, the definition's exactness does not automatically establish that it coincides with an antecedently unique concept of necessary consequence.

John Etchemendy's The Concept of Logical Consequence (1990) emphasizes difficulties with understanding model-theoretic definitions as explanations of necessary and formal truth preservation. A model is a mathematical structure, and its relation to metaphysically possible situations cannot be read from the bare set-theoretic definition. Subsequent discussions investigate the representational and interpretational roles of models and the extent to which both necessity and formality can be preserved (Etchemendy, 1990; Shapiro, 2014; Stanford Encyclopedia of Philosophy, “Logical Consequence”).

Tarski's achievement therefore sharpens the realism–conventionalism dispute. He provides an exceptionally precise mathematical explication. The realist must explain the objective significance of its truth-preserving relation, while the conventionalist must explain why specifications of admissible models and logical vocabulary are accountable to the intended subject matter. Mathematical precision brings philosophical alternatives into better focus; it does not by itself remove them.

What Counts as a Logical Constant?

The definition of consequence ordinarily holds some vocabulary fixed and permits the interpretation of other vocabulary to vary. Logical operators such as negation, conjunction, quantification, and identity in many standard systems occupy the fixed category. Predicates denoting specific objects or properties generally vary. Why should these divisions be made in one way rather than another? The answer shapes which patterns count as valid independently of content.

An invariance approach argues that logical notions should remain unchanged under permissible permutations of the objects of a domain. The idea is that logic expresses maximally general structural operations, independent of the identity of particular objects. Tarski's later lecture “What Are Logical Notions?” (published in 1986) and the subsequent literature give this approach a formal shape. But permutation invariance is itself subject to philosophical evaluation, because quite substantial mathematical operations can exhibit the relevant generality (Stanford Encyclopedia of Philosophy, “Logical Constants”).

If logical vocabulary is identified solely through a decision to call certain signs logical, the distinction may become too easy: any predicate could be privileged without explanation. If it is identified through a robust structural criterion, philosophers must explain why that criterion captures the special normative and epistemic role of logic. The latter task is neither a purely lexical exercise nor a consequence of the existence of a formal system.

The example of identity clarifies the difficulty. A first-order theory with identity treated as a logical primitive licenses certain inferences under every admissible interpretation. An otherwise related theory that represents identity through a designated binary predicate must include explicit axioms to secure comparable inferential behavior. The resulting overall mathematical consequences may be similar in many settings, yet the division between logical and nonlogical consequence has changed.

The choice of logical constants thus links formal objectivity with questions of meaning, semantic variation, and the purpose of formalization. A realist may defend a principled boundary; a conventionalist may defend a pragmatic boundary. Both must address the consequences of the classification and justify its claimed scope.

Proof-Theoretic Semantics and the Legitimacy of Rules

Semantic consequence concerns what holds across interpretations. Proof theory begins from the operations by which conclusions are derived. Natural-deduction systems articulate introduction and elimination rules for logical constants. A conjunction-introduction rule derives A ∧ B from A and B, while elimination derives each component from A ∧ B. The inferential relationships can be used to explain the meaning of conjunction through the role the connective plays in proof.

This approach offers a philosophical explanation closer to a reasoner's actual mastery of inference. Understanding a connective can be linked to grasping which proofs authorize assertions involving it. Proof-theoretic semantics accordingly provides a sophisticated alternative to the view that meaning is established wholly through independent semantic conditions concerning models or abstract truth values.

But the permission to stipulate inferential rules cannot be unlimited. Arthur Prior's “tonk” example (1960) combines rules that appear to permit the introduction of a composite expression from one statement and the extraction of an arbitrary further statement. The system threatens inferential triviality. The philosophical point is that introducing and eliminating a sign must be subject to constraints of harmony or some related proof-theoretic coherence if the rules are to articulate a legitimate connective.

A conventionalist may defend such constraints as pragmatic criteria for adopting rules that serve an intellectual purpose. A realist may interpret them as evidence of more fundamental inferential relations. Either way, the example shows that a bare stipulation is an insufficient explanation of rational acceptability. Meaning-conferring rules themselves require criticism.

In classical first-order logic, soundness and completeness results connect syntactic derivability with model-theoretic validity. Under the standard assumptions and definitions, Γ ⊢ φ holds if and only if Γ ⊨ φ. The equivalence is a substantive mathematical theorem linking proof rules and model-theoretic interpretations. It does not show that proof and model theory supply identical philosophical explanations, that all logics are complete, or that the validity problem for unrestricted first-order logic is decidable.

This distinction returns us to two irreducible tasks. A proof system makes the steps of correct inference explicit and can offer a public criterion for checking a proposed derivation. Its interpretation and application still need to be assessed. The rigorous collaboration of proof and model theory is a major achievement of twentieth-century logic, but a philosophical foundation must also explain the nature of the agreement and the limits of the formal representation.

Logical Realism, Semantic Realism, and Bivalence

The relationship between logical realism and realism about a disputed subject matter was transformed by twentieth-century theories of meaning. Michael Dummett's discussions of realism connect the meaning of statements with the conditions under which they are true or justifiably assertible. The unrestricted application of bivalence is controversial when a discourse contains statements whose truth conditions may outrun the available means of determining them. Logical principles thereby become implicated in theories of understanding (Dummett, 1991).

The familiar contrast concerns the law of excluded middle. Classical logic validates A ∨ ¬A under its standard truth-conditional semantics. Intuitionistic logic generally requires an appropriate proof of one disjunct before a disjunction can be asserted on constructive grounds. This disagreement becomes intelligible only when one examines the interpretations of negation, disjunction, proof, and truth. To say that the systems use different rules reports a formal fact; to explain why either is an adequate theory of a domain is to take a further philosophical step.

The classical logician can argue that a proposition is determinately true or false whether its truth is known. An intuitionist may argue that the mathematical content of a proposition is tied to the possibility of a construction or proof. The disagreement is deeper than a preference for symbols because it concerns what it is for a mathematical claim to possess determinate content.

At the same time, the positions should not be collapsed. Logical realism does not imply that every empirical or mathematical claim is mind-independent in the same respect. Anti-realism about a particular discourse does not uniquely prescribe one revision of classical logic. Each purported connection requires a bridge principle from a theory of meaning to the semantic or proof-theoretic treatment of the logical operators.

The historical significance of the Dummettian project lies in relocating the dispute from an abstract choice between timeless laws and arbitrary conventions into the conditions under which logical reasoning can be understood. The project also tests conventionalism: a language's rules may be chosen, but coherent grasp of those rules and their appropriate use may still be subject to substantive constraints.

Logical Pluralism and the Question of a Unique Logic

Logical pluralism holds, in its substantive forms, that multiple logics can be correct. The contemporary discussion was advanced by J. C. Beall and Greg Restall, whose case-based pluralism permits different admissible understandings of the cases over which validity is assessed. One system may consider classical models, another constructive constructions, and another a different family of semantic situations. The plurality is intended to concern correct consequence relations rather than the banal existence of several formal calculi (Beall & Restall, 2006; Stanford Encyclopedia of Philosophy, “Logical Pluralism”).

The monist's objection is immediate. If one conclusion follows from premises while another allegedly correct logic denies the inference, can both be correct about a single relation of consequence? One possible response is that the disputants employ different concepts or different interpretations of the logical constants. Another is that the concept of validity admits several correct precisifications. Still another is that logics are appraised in relation to inferential purposes, including truth preservation, information sensitivity, or preservation of warrant.

Each response must avoid trivializing the dispute. The mere existence of alternative interpretations does not establish that more than one interpretation is correct for an intended argument. Likewise, a formal language designed for inconsistent databases may be useful without being a complete theory of every inference whatsoever. The pluralist must explain which features confer legitimacy on each candidate and whether a disagreement remains after vocabulary, domain, and aim have been aligned.

Stewart Shapiro's Varieties of Logic (2014) examines the question through logical consequence, mathematical practice, and the philosophical plurality of structures and inferential purposes. Field's account of logical normativity also informs pluralism: several inferential norms may be defensible in relation to different epistemic goals, with no uniquely best choice guaranteed. Such proposals provide alternatives to a metaphysical monism that assumes one calculus must govern all inquiry (Field, 2009; Shapiro, 2014).

Logical pluralism is related to conventionalism without following from it. A realist could affirm that several objective structural consequence relations exist. A conventionalist could argue that some particular calculus is uniquely advantageous in a given domain. The philosophical question is how standards of objective correctness apply when languages and inferential tasks vary.

The Normative Authority of Inference

Logic appears normative insofar as the recognition of a valid consequence bears on how a reasoner should manage assertions, commitments, and beliefs. Frege's connection between the laws of truth and the norms of thinking supplies one classical realist formulation. However, the simple command to believe every consequence of one's existing beliefs cannot function as a general rule of rational conduct. Ordinary agents cannot enumerate all their logical consequences, and some premises should be abandoned when their consequences reveal contradictions or conflicts with better evidence.

The difference between the validity of an argument and the rationality of believing its conclusion is therefore critical. An inference from a false premise can be formally valid but provide no sound basis for believing the conclusion. A person may reasonably postpone a conclusion while checking a complex proof, or withdraw a premise upon discovering an unwelcome implication. Formal consequence constrains possibilities for jointly retaining commitments; it does not automatically instruct an agent to add every entailed proposition to a finite belief set.

Hartry Field (2009) investigates the normative role of logic and the possibility of assessing inferential systems against epistemic purposes. This approach does not dissolve rational evaluation into personal taste. An inferential norm can perform better or worse in achieving a stated aim, and the significance of that aim can be publicly argued. The difference from strict monism is that correctness need not be expressed as a unique purpose-independent calculus.

A conventionalist explanation can invoke conditional commitment: participants in a formal practice undertake to use adopted rules, so applying a rule correctly is part of what it means to carry out that practice. Yet conditional compliance is weaker than a vindication of the practice itself. A fallaciously designed inference system can be used correctly according to its own rules and still be an unsuitable instrument for preserving truth.

An objective logical relation and a rational obligation to reason in a particular way are related through bridge principles concerning an agent's capacities, commitments, goals, and evidence. Their relation cannot be reduced to the bare identity of validity and belief. The distinction provides a way to understand why logical laws can be general while rational inquiry remains responsive to cognitive limitations and changing information.

Case Study: Excluded Middle and Constructive Meaning

The classical validity of excluded middle can be expressed by the schema A ∨ ¬A. Classical two-valued semantics validates the schema under every assignment of truth values to A. A constructive account of disjunction instead connects asserting A ∨ B with possessing a proof of one of the disjuncts; unrestricted recognition of A ∨ ¬A is not licensed merely from the grammatical form of the sentence.

The dispute has two possible interpretations, which must be examined rather than presumed. Under a semantic-disagreement reading, the operators denote different concepts, so the parties have presented different formal claims using visually similar signs. Under a shared-subject-matter reading, the parties disagree about the right account of inference in mathematics or another domain. Either interpretation needs justification from the intended use and meaning of the expressions.

The case demonstrates the insufficiency of an unrestricted conventionalist response. Saying that speakers chose different rules provides a description of the historical situation but does not determine whether the same underlying issue was at stake. It also demonstrates a limit on simple realism. Asserting that one logic states an independently valid law does not settle what meanings that law has or how its application to a contested discourse is defended.

A substantive resolution examines truth conditions, constructions, proof requirements, and the purpose of mathematical assertion. It asks whether classical reasoning is conservative over a constructive fragment for the relevant problem, what additional principles are involved, and whether those principles have independently acceptable justification. This is a philosophical assessment informed by technical logic.

Case Study: Contradiction, Explosion, and Inconsistent Information

Classical logic validates the inference from A and ¬A to an arbitrary B. This is commonly called explosion or ex falso quodlibet. Certain paraconsistent and relevance logics reject explosion while preserving other useful consequence patterns. Their development reflects questions about inconsistent theories, relevance of premises to conclusions, and systems in which local contradictions should not entail everything (Priest, 2006; Stanford Encyclopedia of Philosophy, “Logical Pluralism”).

A strict classical realist may regard explosion as a genuine consequence of the accepted meanings of negation and implication. A paraconsistent theorist may insist that the same inferential pattern fails for the intended treatment of contradiction or information. The nature of the dispute depends upon whether the logical connectives have been assigned common meanings and whether a unique relation of consequence has been fixed.

A database containing contradictory reports need not be treated as supporting every arbitrary conclusion. A paraconsistent calculus can make this methodological constraint explicit. Yet its practical utility does not by itself demonstrate that classical logic is false for every purpose; nor does the formal proof of explosion show that the classical calculus should always govern the management of inconsistent information.

This case supports a disciplined realism about mathematical properties of each calculus, joined to critical assessment of their intended applications. The dispute is substantive precisely because the logics behave differently under well-specified tests. Rules can be devised, but their adequacy to a target practice remains open to reasoned evaluation.

Case Study: Theory Dependence and Logical Vocabulary

One may distinguish logical consequence from consequence relative to a mathematical or scientific theory. Let Γ contain ordinary premises, and let T contain special assumptions about a domain. That Γ together with T entails φ does not mean that Γ alone logically entails φ. This distinction is elementary in formal semantics but indispensable for evaluating claims about convention.

Suppose identity is built into a logic as a distinguished symbol with its usual semantic treatment. Certain inferences are then logically valid across all admissible interpretations. If an apparently comparable symbol is treated as a nonlogical binary predicate, its reflexivity, symmetry, and transitivity may have to be supplied through explicit axioms. Results can sometimes be recovered by adjoining these axioms, yet the boundary between logical and theory-dependent consequence has moved.

This fact is a point for the conventionalist: classifying vocabulary shapes the system. It also marks a limit to unrestricted conventionalism: once the system is specified, there is a determinate fact about which consequences follow. The criterion for preferring one classification remains philosophical and mathematical, involving generality, semantic uniformity, elegance, and intended use.

The lesson extends to modal, temporal, and higher-order operators. Which assumptions are built into the logic and which are relegated to an auxiliary theory will affect the classification of validities. A comprehensive account of logical objectivity therefore requires an account of the logical/nonlogical distinction rather than a mere list of the connectives familiar from introductory textbooks.

Logic in Mathematics, Science, and Computation

Mathematical logic is both a tool for proving results and an object of foundational study. Gödel's incompleteness theorems establish, under precisely stated assumptions, limitations on effective systems strong enough to represent suitable fragments of arithmetic. These results concern specified formal systems and the relations among provability, truth, and consistency. They cannot reasonably be read as evidence that every logical truth is beyond knowledge or that any adopted rule is as good as any other.

Scientific practices employ formal languages, mathematical modeling, and inferential transformations whose adequacy depends on more than formal derivability. An argument may be impeccable relative to an accepted model and fail to establish a substantive conclusion because the translation from empirical claims to formal premises is inaccurate. Formal validity has a vital critical role: it allows one to isolate whether a failure is in the premises, the translation, or the inferential step.

Automated theorem provers and proof assistants make these layers particularly visible. A proof checker can evaluate whether a derivation satisfies a fixed kernel of rules. Such checking can be highly reproducible while the system remains dependent on sound implementation, well-chosen axioms, and faithful formalization of the intended theorem. A proof that passes a checker is a result about formal derivability; determining what has been established outside the calculus requires additional analysis.

Formal methods in computer science also illuminate the difference between semantic and epistemic consequence. A machine may efficiently produce a derivation that is hard for humans to survey. A human may recognize a consequence without representing its entire derivation. The objective status of derivability in a formal system is conceptually distinct from the biological or technological mechanism that discovers or verifies the proof.

The historical movement from Homo to Artificial raises questions about public inference and the portability of formal norms across different kinds of reasoning agents. Those questions can be investigated by drawing on the philosophical distinctions established here. They are applications of the analysis, rather than modifications of any canonical Aisentica definition or a claim that the historical debate has already established an ontology of Artificial.

Open Questions in the Foundations of Logic

A first outstanding question concerns the criterion of logicality. Topic neutrality, invariance under permutation, formal generality, inferential role, and natural-language usage provide different constraints on what should count as a logical constant. The differences may affect the extension of logical consequence. A complete account must justify its choice of criterion and explain which features of a discourse the criterion preserves.

A second question concerns the epistemic grounding of fundamental inferential rules. An attempt to justify every rule by employing more primitive rules threatens regress; an attempt to justify a rule by using the rule threatens circularity. Such circularity may be destructive when a rule's legitimacy is simply assumed as its own proof, while more sophisticated reflective justification may appeal to coherence, epistemic success, and mutual support among practices. The distinction calls for further investigation.

A third question concerns whether all correct logical consequence can be represented by one relation. The development of intuitionistic, relevant, paraconsistent, modal, and other systems raises substantive issues of meaning and domain. The possibility of multiple mathematical definitions is clear; whether multiple logics are objectively correct of the same subject matter remains contested.

A fourth question concerns the relation between semantic meaning and inferential entitlement. If understanding a logical operator gives some epistemic access to its use, the account must avoid explaining correct reasoning entirely through antecedent reliance on that reasoning. If meaning is fixed externally by truth conditions, the theory must explain how a competent agent gains justified access to those conditions.

A fifth question concerns rational revision of logic. Quinean holism allows the possibility of revising even deeply entrenched principles in response to explanatory and scientific pressures. A theory of such revision must distinguish rational criticism of a principle from mere alteration of vocabulary. Otherwise disagreement becomes invisible whenever one party changes a formal rule.

A sixth question concerns the relation of objectivity to convention. Rules may constitute a calculus while failing to explain why its outputs should be trusted outside the calculus. Conversely, objective relations may be discovered through several formal systems without being exhausted by any one representation. Clarifying this relation is essential to any comprehensive philosophical theory of logical authority.

Philosophical Synthesis

The historical record yields a sequence of increasingly precise distinctions. Aristotle analyzes necessity in deduction. Kant separates formal logic from conditions of knowledge of objects. Bolzano and Frege articulate the independence of logical content and truth from psychological acts. Poincaré shows the intellectual legitimacy of selecting theoretical frameworks under practical constraints. Wittgenstein and the logicists distinguish the formal structure of reasoning from empirical assertion. Carnap constructs formal languages and proposes tolerance among their forms. Quine exposes the regress threatening an unrestricted explanation of logical authority through convention. Tarski supplies an exact model-theoretic explication, while proof theorists, semantic anti-realists, and pluralists test its conceptual limits.

These contributions support a four-level analysis. Formal specification consists in choosing vocabulary, rules, interpretations, and the permitted forms of proof. Formal determination concerns what is derivable or true in all models once the specification is fixed. Applicative adequacy concerns whether the calculus faithfully represents the domain or practice that it claims to illuminate. Philosophical grounding concerns why any of these relations express truth-preservation, meaning, or rational normativity. Each level involves different standards of explanation.

An argument that conventions matter at the first level cannot establish that all correctness is arbitrary at the second. The mathematical consequences of precisely stated rules are publicly testable and independent of a particular individual's preference. Similarly, a mathematical proof at the second level cannot establish that the adopted representation is adequate to every philosophical or empirical domain at the third. The fourth level remains a site of metaphysical, semantic, and epistemological contestation.

The strongest realist thesis recognizes standards of correctness that transcend individual or collective assent. Its challenge is to explain how those standards are identified, known, and applied. The strongest conventionalist thesis demonstrates the legitimacy and importance of choosing explicit frameworks; its challenge is to articulate rational limits on those choices. Both approaches improve when they distinguish objectivity within formal practice from claims of universal metaphysical privilege.

This study accordingly establishes a qualified thesis of conditional formal objectivity and public justificatory accountability. Within a fully specified interpreted calculus, the distinction between valid and invalid consequence is objective in the mathematical sense. The choice and application of the calculus are accountable to arguments about meaning, soundness, purpose, and domain. Whether there is one uniquely correct language-independent consequence relation remains an independent question that neither a formal stipulation nor a general assertion of logical realism can resolve by itself.

Conclusion

Logical realism and conventionalism investigate the sources of authority in logical reasoning. Their history demonstrates that the adoption of formal rules and notations is an intellectual activity, that the mathematical consequences of those rules are determinate once an appropriate system is fixed, and that the adequacy of the system requires argument beyond its internal derivations. The controversies concern truth, meaning, epistemic entitlement, rational normativity, and the scope of logical consequence.

Frege's defense of objective thoughts explains why correctness cannot be equated with acceptance. Carnap's tolerance exposes the significance of alternative frameworks and exact formal specification. Quine's criticisms show why general conventions cannot be invoked as an assumptionless foundation for the validity of their own application. Tarski's theory makes consequence mathematically precise while drawing attention to admissible interpretations and logical vocabulary. Pluralist and proof-theoretic approaches clarify the variety of possible inferential systems and the constraints on their legitimate use.

The resulting philosophical conclusion separates four questions: what system has been specified, what follows inside it, how well it represents its intended subject matter, and what warrants treating its consequences as authoritative. The separation is the basis of a more rigorous philosophy of logic. It preserves the objective assessment of formal results, identifies the genuine work performed by convention, and leaves the metaphysical and epistemic grounding of logical necessity open to systematic investigation.