Ryan Simonelli

Do large language models (LLMs) possess concepts, such that they can be counted as genuinely understanding what they're saying? In this paper, I approach this question through an inferentialist account of concept possession, according to which one's possession of a concept is understood in terms of one's mastery of the inferential role of a linguistic expression. I suggest that training on linguistic data is in principle sufficient for mastery of inferential role, and thus, LLMs trained on nothing but linguistic data could in principle possess all concepts and thus genuinely understand what they're saying, even when speaking about such things as colors and tastes, guilt and folly, life and death. This doesn't mean, however, that they are conscious. I draw a classical distinction between sentience (conscious awareness) and sapience (conceptual understanding) and argue that we might think of LLMs as genuinely possessing the latter without even a shred of the former. In defending this claim, I argue that attributing conceptual understanding to a system is not a matter of describing some specific empirical property that the system shares with us but, rather, as Wilfrid Sellars says, "placing it in the logical space of reasons,'' treating it as answerable to calls for reasons, clarifications, corrections, and so on. I claim that we may aptly adopt this attitude towards sufficiently capable LLMs without thereby treating them as conscious subjects.

A modest theory of meaning does not attempt to give an account of the concepts expressed by the primitive expressions of a language. Throughout his career, John McDowell has argued that our semantic aspirations ought to be modest in this sense. By contrast, inferentialism, as developed by Wilfrid Sellars and Robert Brandom, is clearly an immodest theory of meaning: it aims to account for the concepts expressed by linguistic expressions in terms of the inferential rules governing their use. In this paper, I defend inferentialism against all of McDowell’s criticisms, arguing that they either fail to apply to the view, or beg the question against it.

When faced with the question of whether to assert or deny a paradoxical sentence such as the liar, it seems that there are two plausible responses: neither asserting it nor denying it, or both asserting it and denying it. In this paper, I make this thought concrete by formulating bilateral proof systems (of both the natural deduction and sequent calculus variety) for the logics in the FDE family: K3, LP, and FDE. The different logics are simply the result of different choices of “coordination principles,” bilateral structural rules which coordinate the opposite speech acts of assertion and denial. I show that conceiving of these logics in bilateral terms has important philosophical consequences, most notably for the debate between “subclassical” and “substructural” approaches to paradox. In particular, I show how adopting Bilateral K3 enables one to endorse the “non-transitive” solution to paradox, as developed by Ripley, while maintaining that logical consequence (understood bilaterally) is transitive.

When faced with the question of whether to assert or deny a paradoxical sentence such as the liar, it seems that there are two plausible responses: neither asserting it nor denying it, or both asserting it and denying it. In this paper, I make this thought concrete by formulating bilateral proof systems (of both the natural deduction and sequent calculus variety) for the logics in the FDE family: K3, LP, and FDE. The different logics are simply the result of different choices of "coordination principles," bilateral structural rules which coordinate the opposite speech acts of assertion and denial. I show that conceiving of these logics in bilateral terms has important philosophical consequences, most notably for the debate between "subclassical'' and "substructural" approaches to paradox. In particular, I show how adopting Bilateral K3 enables one to endorse the ``non-transitive'' solution to paradox, as developed by Ripley, while maintaining that logical consequence (understood bilaterally) is transitive.

According to an act-based conception of propositions, propositions are types of cognitive or linguistic acts. Such accounts are advertised as having major metaphysical and epistemological advantages over traditional platonic accounts. However, existing versions of such accounts appeal to platonic properties and relations in order to account for the contents expressed by predicates, reintroducing many of the problems they aim to solve. Characterizing both *that a is F* and *that it’s F* as different types of “assertibles” (the former can be asserted full-stop and the latter can be asserted of things), the issue can be seen as a limitation of existing act-based approaches: they apply only to a restricted class of assertibles. In this paper, I show how adopting a normative functionalist approach to linguistic meaning enables one to generalize the act-based approach to all assertibles such that no appeal to extrinsic properties and relations is needed. I show, further, how this radicalized act-based account provides the resources for a satisfactory account of our knowledge of objective states of affairs, properties, and relations (which I dub “instantiables”) in terms of our mastery of linguistic norms.

Jc Beall argues that if FDE is logic proper, then there is no logical negation. This claim is largely based on the fact that, in standard proof systems for FDE, there are no stand-alone negation rules that suffice to capture the behavior of negation. In this paper, I show that by adopting a bilateral proof system for FDE, one can maintain that there is a logical negation, it is the very same logical negation that belongs to classical logic, and its basic function is to flip-flop between assertion and denial. After laying out the bilateral proof systems on which this claim is technically based, I develop the conception of assertion and denial on which this claim is philosophically based, responding to a number of objections. I conclude by considering the possibility of a different, so-called "Boolean" negation, in the context of this bilateral framework.

Bilateralism in proof-theoretic semantics, according to which rules are provided for both assertions and denials, has been central to the development of inferentialist theories of meaning over the past few decades. Bilateralists are split, however, as to what this position amounts to, with two different formal approaches to bilateralism corresponding to fundamentally different ways of thinking about the basic notion in terms of which an inferentialist theory is to be cast: normative incoherence or committive consequence. In this paper, I show how these two forms of bilateralism can be brought together in a single framework that has the virtues of both. The resulting framework is uniquely capable of serving as a proof-theoretic basis to carry out an inferentialist semantic theory.

In a recent paper, Nils Kürbis argues that bilateral natural deduction systems in which assertions and denials figure as hypothetical assumptions are unintelligible. In this paper, I respond to this claim on two counts. First, I argue that, if we think of bilateralism as a tool for articulating discursive norms, then supposition of assertions and denials in the context of bilateral natural deduction systems is perfectly intelligible. Second, I show that, by transposing such systems into sequent notation, one can make perfect sense of them without talking about supposition at all, just talking in terms of relations of committive consequence. I conclude by providing some motivation for adopting this normative interpretation of bilateralism on which this response to Kürbis’s argument is based.

Bilateral proof systems, which provide rules for both affirming and denying sentences, have been prominent in the development of proof-theoretic semantics for classical logic in recent years. However, such systems provide a substantial amount of freedom in the formulation of the rules, and, as a result, a number of different sets of rules have been put forward as definitive of the meanings of the classical connectives. In this paper, I argue that a single general schema for bilateral proof rules has a reasonable claim to inferentially articulating the core meaning of all of the classical connectives. I propose this schema in the context of a bilateral sequent calculus in which each connective is given exactly two rules: a rule for affirmation and a rule for denial. Positive and negative rules for all of the classical connectives are given by a single rule schema, harmony between these positive and negative rules is established at the schematic level by a pair of elimination theorems, and the truth-conditions for all of the classical connectives are read off at once from the schema itself.

An “inferentialist” semantic theory for some language L aims to account for the meanings of the sentences of L solely in terms of the inferential rules governing their use. A “hyper-inferentialist” theory admits into the semantics only “narrowly inferential” rules that normatively relate sentences of L to other sentences of L. A “strong inferentialist” theory also admits into the semantics “broadly inferential” rules that normatively relate perceptual states to sentences of L or sentences of L to intentional actions. It is widely thought that only the latter of these two sorts of semantic theories is at all viable. I argue here that the opposite is so. Negatively, strong inferentialism is viciously circular: including rules into the semantic theory that relate perceptual states to sentences of the language requires us to appeal, in individuating those perceptual states, to the very meanings for which we are supposed to be inferentially accounting. Hyper-inferentialism does not face this problem because it does not appeal to any nonlinguistic states. Positively, though hyper-inferentialism is widely thought to be a theoretical non-starter, I argue here that it is a genuine theoretical possibility insofar as it essentially includes cross-perspectival inferences, inferences along the lines of the one from the sentences “The ball is in front of n,” “The ball is red,” and “The lighting is good” to the sentence “n sees that the ball is red.” I make the further exegetical claim that Robert Brandom, though unanimously taken to be a strong inferentialist, is in fact a hyper-inferentialist.

Why must incompatibility be symmetric? An odd question, but recent work in the semantics of non-classical logic, which appeals to the notion of incompatibility as a primitive and defines negation in terms of it, has brought this question to the fore. Francesco Berto proposes such a semantics for negation argues that, since incompatibility must be symmetric, double negation introduction must be a law of negation. However, he offers no argument for the claim that incompatibility really must be symmetric. Here, I provide such an argument, showing that, insofar as we think of incompatibility in normative pragmatic terms, it can play its basic pragmatic function only if it is symmetric. The upshot is that we can vindicate Berto’s claim about the symmetry of incompatibility but only if we, pace Berto, think about incompatibility, in the first instance, as a pragmatic relation between acts rather than a semantic relation between contents.

I motivate and develop a use-based semantic theory in opposition to the dominant paradigm in philosophical and linguistic semantics. Drawing inspiration from Wilfrid Sellars, I argue that contemporary semantic theories are faced with a basic problem of explanatory circularity. These theories universally presuppose that worldly knowledge of such things as properties or sets of possible worlds precedes and underlies knowledge of meaning. However, I argue that it is only through learning a language--mastering the rules governing the use of the expressions belonging to that language--that speakers can possibly have knowledge of the worldly entities appealed to by semanticists at the base level of their semantic theories. In response to this fundamental problem, I develop a formal semantic framework in which the meaning of a sentence is understood directly in terms of its role in discourse. In contrast to existing semantic frameworks, this framework does not presuppose speakers' worldly knowledge, and so is actually able to explain it. The result is not just a new kind of semantic theory, but a new conception of the relation between meaning and the world.

According to existing accounts of indicative conditionals, any argument of the following form is valid: ϕ → ψ, ( ϕ ∧ ψ ) → χ ∴ ϕ → χ. Here, I present a set of counterexamples to show that there exist invalid arguments of this form. I argue that this data poses serious problems to variably strict accounts of conditionals, as such accounts are structurally unable to accommodate it. Dynamic strict accounts, however, are a different story. While existing dynamic strict accounts do not accommodate the data, they are in principle able to, and I propose a modified dynamic strict account, drawing from von Fintel, that does. The key modification is this: whereas existing dynamic strict accounts take into account only the effects of conditional antecedents in changing the semantic context, the data shows that we must also take into account the effects of conditional consequents in changing the semantic context.

In recent work, Robert Brandom has articulated important connections between the deontic normative statuses of entitlement and commitment and the alethic modal statuses of possibility and necessity. In this paper, I articulate an until now unexplored connection between Brandom’s core normative statuses of entitlement and commitment and the agentive modal statuses of ability and compulsion. These modals have application not only in action, but also in perception and inference, and, in both of these cases, there is a direct mapping between the normative statuses that one bears towards various claims, articulated from the perspective of the attributor of commitments and entitlements, and the agentive modal statuses that one bears towards various judgments, articulated from the perspective of the undertaker of commitments. I will highlight this correspondence, focusing on the case of perception, and show how it sheds light on the account of mindedness that emerges from Brandom’s theory of discursive practice.

Wilfrid Sellars is widely known for two positions that he calls “nominalism.” On the one hand, there is his “psychological nominalism,” according to which any awareness one might have of abstract entities—be they properties, relations, or facts—is a thoroughly linguistic affair, and so cannot be presupposed in thinking about the process of learning a (first) language. On the other hand, there is his ontological nominalism, according to which the world, as it is in itself, is fundamentally a world of concrete particulars and so does not ultimately contain such things as properties, relations, or propositionally-structured facts. Sellars clearly takes these two sorts of “nominalism” to go together. However, one of the most influential inheritors of Sellars's philosophy, Robert Brandom, thinks that they do not. Brandom endorses the former, but denies that the latter “is in the end so much as intelligible.” In this paper, I articulate the connection between Sellars's psychological and ontological nominalism and draw on Brandom's own development of Sellars's functional role semantics to argue, against Brandom, that Sellars's ontological nominalism not only harmonizes with the rest of his philosophical commitments, but is actually made fully intelligible by the very aspect of Sellars's theory that Brandom himself develops.

Donald Davidson argues that the very nature of belief ensures that, if we have any beliefs at all, most of them must be true. He takes this to show that Cartesian skepticism is fundamentally mistaken. Many commentators, however, find this response to skepticism to be lacking. In this paper, I draw from recent work by Rebecca Kukla and Mark Lance and attempt to give Davidson’s argument a newfound force by applying it to our acts of ostension, of pointing others to features in our shared environment. The sort of ostensive argument that I extract from Davidson remains largely unexplored in skepticism’s vast literature, and I argue that this is the strongest way to read Davidson’s epistemological work, providing a response to some influential objections to his arguments.

In Evidence and Inquiry, Susan Haack presents a dual-aspect account of evidence in which both casual and logical relations play a necessary factor. In this paper, I reformulates how these two aspects fit together to form a comprehensive picture of discursive justification. Drawing from Quine’s work on the “observation sentence,” I show how we can move from causal justifications to inferential justifications. Conversely, I also attempt to show how we can correct and improve our causally justified, noninferential beliefs by challenging and inferentially justifying them. The resulting account should be one in which our beliefs are connected to the world, but one which is non-foundationalist and allows any belief to be called into question.