Who is associated with inferentialism?

The term inferentialism is associated with the philosopher Robert Brandom. Its roots lie in Wilfrid Sellars and the later Wittgenstein's idea of meaning as use, and in the logical work of Gerhard Gentzen, Michael Dummett, and Dag Prawitz, who held that the introduction rules of a connective fix its meaning. Proof-theoretic semantics is the formal development of these ideas.

Is inferentialism the same as proof-theoretic semantics?

They are closely related but not identical. Inferentialism is a broad philosophical position about the nature of meaning. Proof-theoretic semantics is its precise mathematical development within formal logic, providing definitions and theorems. One can hold inferentialism as a philosophical view without working in proof-theoretic semantics, and vice versa.

Does inferentialism reject truth-conditional semantics?

Inferentialism rejects the claim that truth conditions are the fundamental notion of meaning. It holds that meaning is constituted by inferential role, and that truth is to be understood through inference rather than the other way around. It need not deny that sentences have truth conditions, only that those conditions are what meaning ultimately consists in.

How does inferentialism relate to artificial intelligence?

Inferentialism makes the structure of reasoning the seat of meaning, which bears directly on whether an AI system genuinely reasons. A system that can trace its conclusions to explicit inferential steps has, on the inferentialist picture, a grip on the content of what it asserts; one that only predicts statistically does not. This gives a principled way to distinguish reasoning from approximation.

What Is Inferentialism? — Alexander V. Gheorghiu

Q: What is inferentialism?

Inferentialism is the view that the meaning of an expression is constituted by its inferential role — the inferences it licenses and the inferences that license it — rather than by an object it denotes or the conditions under which it is true. In logic, the meaning of a connective such as 'and' or 'if…then' is given by the inference rules governing its use.

Summary

Inferentialism is the view, in the philosophy of language and logic, that the meaning of an expression is constituted by its inferential role — the inferences it licenses and the inferences that license it — rather than by an object it denotes or the conditions under which it is true. The term is associated with Robert Brandom; the idea has roots in Wittgenstein, Sellars, and the logical work of Gentzen, Dummett, and Prawitz. Proof-theoretic semantics is its formal development for the logical constants.

The core idea

The dominant tradition in the philosophy of language, following Frege and Tarski, treats reference and truth as fundamental: to grasp a sentence is to know the conditions under which it would be true. Inferentialism reverses the order of explanation. What you understand, when you understand an expression, is how to use it — which inferences it entitles you to make, and which entitle you to it.

The point is familiar outside logic. Asked what “Tammy is a vixen” means, it is unhelpful to say that “Tammy” denotes an object falling under the category “vixen”. It is far more illuminating to give the inferential connections: that it follows from “Tammy is a fox” and “Tammy is female”, and that each of these follows from it. To master those inferences is to grasp the meaning.

Applied to logic, this yields logical inferentialism: the meaning of a connective is fixed by its inference rules. Following Gentzen, the introduction rules act as definitions, and the elimination rules are justified as their consequences. Inference, not truth, is conceptually prior.

Key concepts

Inferential role

The meaning of an expression is its place in a web of inferences — what it follows from and what follows from it. To know the meaning is to have mastered that role, not to have identified a referent.

Rules as definitions

On Gentzen’s view, the introduction rules of a connective define it: they say what it takes to assert it. The eliminations are then read off as what that definition entitles us to infer.

Harmony

For the rules to confer a coherent meaning they must be in harmony: the eliminations should extract exactly what the introductions put in. Harmony is the inferentialist’s criterion for a well-formed concept.

Logical vs semantic inferentialism

Brandom’s semantic inferentialism is a general theory of conceptual content; logical inferentialism is the narrower, formally precise thesis about the logical constants that proof-theoretic semantics makes rigorous.

Meaning from rules

Inferentialism invites a constructive question: given only rules, can one recover a logic’s meanings — and even derive which logic results? This is the thread that runs through the research below.

How this research develops inferentialism

Gheorghiu’s research gives logical inferentialism precise mathematical form and pushes it into new territory:

Inferentialist semantics for intuitionistic logic. An Inferential Semantics for Intuitionistic Sentential Logic relates the base-extension semantics of intuitionistic logic to Mints’ resolution method, recovering soundness and completeness as corollaries.

Inferentialism and resources. Inferentialist Resource Semantics extends the inferentialist programme to resource semantics, connecting meaning-as-use with reasoning in which assumptions are consumed, as in the logic of bunched implications.

Deriving logic from structure. Defining Logical Systems via Algebraic Constraints on Proofs and On the Logical Content of Knowledge Bases ask what logic emerges when one starts from constraints on proofs, or from the structure of a body of knowledge, rather than assuming the connectives in advance.

For the broader setting, see A Survey of Proof-theoretic Semantics.

Why it matters

Philosophy of language and mathematics. Inferentialism offers an account of meaning that does not rest on reference to abstract objects, making it congenial to those sceptical of platonism — and it explains how speakers can learn meanings by mastering use.

Meaning without bivalence. Because it grounds meaning in inference, the view can make sense of logical vocabulary — even classical vocabulary — without assuming that every sentence is determinately true or false. Its formal counterpart is base-extension semantics.

Artificial intelligence. If meaning lives in inferential role, then a system that cannot trace its outputs to inferential steps lacks a grip on what it asserts. This gives a principled handle on the difference between reasoning and statistical prediction — see How does logic relate to AI?

Common Questions

Frequently asked

Is inferentialism the same as the “meaning as use” idea?

It is a precise version of it. Wittgenstein’s slogan that meaning is use leaves open what counts as the relevant use. Inferentialism answers: the use that matters is inferential — the role an expression plays in reasoning. Logical inferentialism makes this exact for the logical constants.

What is the difference between Brandom’s inferentialism and logical inferentialism?

Brandom’s inferentialism is a sweeping account of conceptual content across language as a whole. Logical inferentialism restricts attention to the logical constants and gives the thesis mathematical content, through proof theory and proof-theoretic semantics. The latter is where formal results — soundness, completeness, harmony — can be proved.

Does inferentialism favour intuitionistic logic?

Historically it was thought to, because classical connectives resisted harmonious formulation, leading Dummett towards intuitionism. More recent work — including Gheorghiu’s — develops inferentialist and base-extension semantics for classical logic too, so the commitment to inferentialism need not force a revision of which logic is correct.

From This Site

Inferentialism’s formal development is proof-theoretic semantics, and its completeness-bearing technical form is base-extension semantics.

For definitions of the key terms, see the Glossary; for the relevance to AI, see How does logic relate to AI?; and for the full body of research, the Publications page.