Inference

In a series of papers over the past twenty years, and in a new book, Igor Douven has argued that Bayesians are too quick to reject versions of inference to the best explanation that cannot be accommodated within their framework. In this paper, I survey their worries and attempt to answer them using a series of pragmatic and purely epistemic arguments that I take to show that Bayes’ Rule really is the only rational way to respond to your evidence.

“Is Professor Crispin Wright on Quine’s indeterminacy of translation thesis worth reading?” you ask. “What does it say? What is the main point? And what is Alex Miller’s response?” Beyond those who doubt the value of the tradition of analytic philosophy in general, there are probably American analytic philosophers with questions along these lines, who doubt the value of much British philosophy probably. In this paper, I offer a light substitute which bears some loose resemblance to Wright’s paper, for those (...) wanting a quick feel of what Wright’s paper is like. I argue that Quine’s argument from above cannot succeed, given the kind of underdetermination to which he is committed. But his argument from below is less susceptible to being undermined by considerations of overall coherence than one might think. (shrink)

Professor Timothy Williamson has written a paper addressing the world of jurisprudence as well, entitled "The content of legal evidence." But what should he be saying to jurisprudence, which we can here understand as legal philosophy. Perhaps "saying" is the wrong word. We should speak of checking! He should check that his audiences have a reliable grasp of the premise-inference-conclusion distinction. Let's help any audience members who are not too sure. Here is an example of an argument, with two premises (...) and an inference and a conclusion - prompted by words I actually heard. "(1) If my young children have been taken away to Pakistan without my consent, then my rights have been violated. (2) My young children have been taken away to Pakistan without my consent. Therefore: (3) My rights have been violated." (1) is a premise. (2) is also a premise. The statement above of it ends with "without my consent." The word "therefore" is an inference from these two premises. The conclusion is (3). Given these distinctions, the following representation of the argument is a mistake: "(1) If my young children have been taken away to Pakistan without my consent, then my rights have been violated. (2) My young children have been taken away to Pakistan without my consent. (3) Therefore, my rights have been violated." (3) is supposed to be the conclusion, but here it also includes the inference - the therefore! But this mistake is not uncommon in various circles (I have not been collecting published examples, sorry) and suggests to me a poor “grasp” of the distinctions between premises and inference and conclusion. Probably there is a paper in the nineteenth century regarding what disasters arise by not separating out the inference. But probably also you intuitively grasp the distinction, even if you are not disposed to write in line with it. If I ask you, what follows from premises (1) and (2), you will say (or at least think): my rights have been violated. If I ask you, what conclusion can you draw from premises (1) and (2) - assuming they are true - you will say (or at least think): my rights have been violated. If I ask you, what can you infer from premises (1) and (2) - assuming they are true - you will say (or at least think): my rights have been violated. In answering in this way, you don't include the inference in your answer: the answer being your specification of the conclusion. Including it is like saying "Therefore therefore..." If I draw attention to these two premises and say, "Therefore" you just think the conclusion as what follows. Hopefully! (The title I have given to this contribution quotes a poem by Laura Riding entitled "The World and I."). (shrink)

This paper presents Anca Gheaus’s attempt to reconstruct Harry Brighouse and Adam Swift’s defence of the family against abolitionists. I propose that the reconstruction suffers from at least two problems: it is missing a premise given the aim of not attributing an invalid argument unnecessarily; and it fails to distinguish the inference to a conclusion from the conclusion itself.

Philosophy of mind, cognitive science, and AI lack a unified model to explain how systems form meaningful propositional attitudes—states like belief, desire, and intention. This work introduces and defends the Trialectics framework as a response to this theoretical gap. We propose that any system capable of forming meaningful propositional attitudes—that is, internal states with truth-evaluable, compositionally structured content that can be rationally evaluated and guide goal-directed behavior—must instantiate three interdependent functional capacities: Reference (R), the ability to establish semantic links; Inference (...) (I), the capacity to process and transform content systematically; and Preference (P), the function of internal evaluation and goal-orientation. The central thesis is that these capacities are not modular but mutually conditioning: meaningful content arises only from their dynamic entanglement, where reference shapes and is shaped by inference and preference, and so on. By embedding evaluation (P) within the cognitive loop, Trialectics offers a naturalized solution to the problem of normativity in reasoning and reconciles representational requirements with enactivist critiques by defining reference functionally rather than metaphysically. The framework’s utility is demonstrated by showing how it implicitly structures existing cognitive architectures (e.g., BDI, Active Inference) and provides a powerful diagnostic and design tool for AI, particularly concerning alignment, generalization, and interpretability. Ultimately, Trialectics posits a fundamental architectural condition for any system—biological or artificial—to achieve the flexible, offline, and rationally evaluable propositional attitudes characteristic of sophisticated intelligence. (shrink)

Counterfactual reasoning has always played a role in human life. We ask questions like, “Could it have been different?”, “Under which conditions might/would it have been different?”, “What would have happened if…?” If we don’t find an answer, i.e. what we accept as an answer, we may start reasoning. Reasoning means introducing still new information/assumptions, new questions, new answers to new questions etc. From a formal point of view, it may be compared with stepwise moving towards a destination in a (...) path-system, in which you never fully have an overview. In this way, reasoning is an activity, with its own rationale, which will be studied from the agent’s own perspective. Questions include: What are the conditions where asking that specific question, or introducing this information/assumption, etc. will count as a reasonable step or progress towards the answer of the initial question? What makes this step more reasonable than another? (shrink)

Different definitions of deduction are offered. The first is that deduction is an inference type that is both possibly valid and possibly invalid. No inference can satisfy this definition, because valid inferences are not possibly invalid and invalid inferences are not possibly valid. In the second definition, deductive inferences are defined as being possibly valid or possibly invalid, but not both. This hypothesis does not work, since it implies that inductive inferences are deductive. To avoid this objection, deduction can be (...) understood as an inference that aims for validity. This definition also has unwanted consequences, including the fact that invalid inferences are only deductive when they are thought to be possibly valid. If the only genuine deductions are the valid ones, then our talk about deduction is an indirect and thoughtless way of referring to valid inferences rather than to an inference type. There is simply no deduction to speak of. Only valid inferences. (shrink)

This article provides a defense of deductivism, the notion that all inferences are deductive. It’s argued that all inferences are deductive according to the reasoner’s intended domain of quantification. This implicit range might include different modalities (logical, nomological, metaphysical), different numbers of intended worlds (all, some, twenty, one...), and different sets of objects referred to. These restrictions require a different conception of validity where inferences can be truth preserving in different ranges. The main alternative to deductivism is inductivism, the view (...) that some inferences are genuinely inductive. Section 2 presents the problems of inductivism, highlighting its conceptual deficiencies and advancing deductivism as an alternative. In section 3, it’s argued that the epistemic commitments of inductivism and deductivism are fallibilism and infallibilism, respectively. Infallibilism is defended as a matter of coherence. Section 4 considers the metaphysical commitments of deductivism. It’s argued that deductivism implies determinism, which on its turn requires only deductive explanations. Section 5 concludes. Throughout the paper other conceptual innovations are offered such as modal indexes, a modal interpretation of probability and a distinction between type and token propositions. (shrink)

Although inference is applaudable while fallacy is assumed culpable, both are the same cognitive faculty. The different appearance is a matter of epistemic l{a,u}ck of agency, in tight connection with a convenient arbitrariness of perceived reality. Conversely, when that arbitrariness collapses, strictly useful knowledge promises both salvation and coercion.

When entering an environment, animals – including humans – tend to consult their memories to determine what they know about the place. This information is useful to determine: is this place safe? And what happens next? In this chapter, we argue on both empirical and conceptual grounds that memory is largely organized by space. Spatial relations determine what is recalled and which experiences are combined in generalizations. Time does not play an analogous role. We show that space and time in (...) memory are thus deeply asymmetrical. We conclude with a consequence of our view: spatial organization entails a basic form of abstraction in a way temporal organization does not, and so the spatial organization of memory reveals that memory involves generalization from the very beginning. (shrink)

Recent epistemological debates have increasingly focused on the contentious Counter-Closure principle, which holds that, necessarily, if an agent S believes that q based solely on a competent inference from p, and knows q, then S also knows p. This principle has drawn attention due to various challenges, particularly the issue of inferential knowledge derived from false premises. In this article, we pursue two objectives. First, we argue that the Counter-Closure principle is untenable but for reasons that depart from traditional critiques. (...) Specifically, we will present a novel argument against the internalist approach that supports the cases of knowledge from falsehoods. Second, we show that the Counter-Closure principle’s failure can be better addressed within an externalist framework by exploring novel theories of defeaters and the relationship between doxastic and propositional warrant. (shrink)

Having justification for inferring a conclusion from some premises is independent of seeing how to make that inference. But if you do make the inference, then your belief in the conclusion is justified based on those premises. These natural thoughts about inference and justification generate two puzzles. One concerns logical omniscience, and the other concerns inferences to correct conclusions via fallacious rules. In this paper I introduce and develop what I call the Inferential Parsing Hypothesis as a way of resolving (...) these two puzzles. The hypothesis recruits an analogy between inferential processes and syntactic parsing processes to motivate novel epistemic theses about the nature of inferential basing. (shrink)

This paper lays the groundwork for expanding on past work on the connection between logic and games, which has largely focused on formal models of dialogical games or mathematical theories of games and decision. I lay out a new approach by arguing that use of formal, deductive logic can be fruitfully analysed as a Suitsian game. To do so, I outline an account of use of deductive logic in reasoning and motivate the analysis by appealing to Suits’s classic account of (...) gameplay, his work on detective fiction and a discussion of puzzle games. I then respond to worries that use of logic is not less efficient and that one does not take on a lusory attitude in using a logic. I suggest that in my responses one can see where the analysis bears fruit, both with respect to debates in the philosophy of games and in the philosophy of logic. (shrink)

Recently, Dardashti, Hartmann, Thébault, and Winsberg (2019) proposed a Bayesian model for establishing Hawking radiation by analogical inference. In this paper we investigate whether their model would work as a general model for analogical inference. We study how it performs when varying the believed degree of similarity between the source and the target system. We show that there are circumstances in which the degree of confirmation for the hypothesis about the target system obtained by collecting evidence from the source system (...) goes down when increasing the believed degree of similarity between the two systems. We then develop an alternative model in which the direction of the variation of the degree of confirmation always coincides with the direction of the believed degree of similarity. Finally, we argue that the two models capture different types of analogical inference. (shrink)

Do large language models (“LLMs”) think? Daniel Stoljar and Zhihe Vincent Zhang have recently developed an argument from rationality for the claim that LLMs do not think. We contend, however, that the argument from rationality not only falters, but leaves open an intriguing possibility: that LLMs engage only in arational, associative forms of thinking, and have purely associative minds. Our positive claim is that if LLMs think at all, they likely think precisely in this manner.

The paper argues that Wright’s account of transmission failure cannot be correct as it stands. For, it requires that reasoners are warranted to believe all the authenticity conditions for their evidence relative to the premises of their inferences. There are, however, cases in which there are pairs of such authenticity conditions such that one cannot be warranted to believe both.

I take issue with recent enthusiasm for the alleged epistemic powers of imagination and argue that imagination cannot generate knowledge and justification. I distinguish two conceptions of imagination that have emerged from recent debates: primitivism, according to which imagination is sui generis and essentially sensory or imagistic, vs. reductionism, according to which imagination is akin to other psychological states and “propositional” or non-necessarily imagistic. Both conceptions are problematic. Sensory imagination has been shown to be largely unreliable; besides, it doesn’t play (...) either a necessary or sufficient role for knowledge. Propositional imagination, on the other hand, seems virtually indistinguishable from inferential reasoning, hence it’s unclear that it can yield distinctively imaginative justification. The issue reflects a broader dilemma for candidate accounts of the imagination and its epistemic powers: either imagination is primitive and sui generis but epistemically suspect; or it's reducible to more familiar inferential processes but lacks independent epistemic power. (shrink)

We explore the view that knowledge of grounding is based on knowledge of essence. We assess different existing accounts of the relation between essence and grounding and identify some of their shortcomings. In response, we propose a novel account that we argue is better suited to explain this relation and show how this can further explain knowledge of grounding. Finally, we examine how one can transition from knowledge of essence to knowledge of grounding. We maintain that, at least in some (...) cases, it is possible to know what grounds what—non-circularly and inferentially—on the basis of knowledge of essence. (shrink)

Is more information always better? Or are there some situations in which more information can make us worse off? Good (1967) argues that expected utility maximizers should always accept more information if the information is cost-free and relevant. But Good's argument presupposes that you are certain you will update by conditionalization. If we relax this assumption and allow agents to be uncertain about updating, these agents can be rationally required to reject free and relevant information. Since there are good reasons (...) to be uncertain about updating, rationality can require you to prefer ignorance. (shrink)

To inquire into some question Q is to try to answer Q. To understand inquiry, we must understand what constitutes success in this endeavor. What it is to answer Q? The issue has been systematically neglected in philosophical work on inquiry. It raises a real puzzle. A judgment with a content p that actually settles Q doesn't necessarily constitute answering Q. To answer Q, a judgment must have additional significance connecting p with Q. It's not clear what gives a judgment (...) such significance in inquiry. This article introduces this unsolved “problem of answering,” explains its difficulty, and draws a comparison with the problem of inference. (shrink)

Epistemologists typically discuss defeat as applying to beliefs and other cognitive states. But many other kinds of mental states can also be defeated, including emotions. This paper argues that there are strong structural analogies between the rational defeat of beliefs and the rational defeat of other kinds of mental states, analogies that call for a unified explanation. In the case of belief, there are at least three ways evidence can have a defeating effect: it can act as a rebutter, an (...) undercutter, or a higher-order defeater. Many other kinds of mental states can also be defeated in each of these three ways. This suggests that there is a general account of rational defeat that applies across mental states. This paper develops such an account, according to which defeat concerns the fittingness of mental states. (shrink)

This paper does four things. First it lays out an orthodox position on reasons and defeaters. Then it argues that the position just laid out is mistaken about “undercutting” defeaters. Then the paper explains an unpublished thought experiment by Dorothy Edgington. And then it uses that thought experiment to motivate a new approach to undercutting defeaters.

It’s generally thought that we finite beings can’t follow infinitary rules of inference like the omega rule. But this consensus has recently been challenged, most notably by Jared Warren, whose account of the determinacy of our arithmetical language depends on the claim that we do follow the omega rule. Warren argues for that claim by exhibiting a case in which it’s purportedly clear that we’re actually disposed to use that rule to infer a generalization from infinitely many instances. Here I (...) respond to this argument: I show that, in the case in question, there’s conclusive reason to think we’re inferring from the generalization to the instances rather than the other way around and so aren’t engaging in infinitary reasoning, and I argue, more generally, that finite beings like us can never use infinitary reasoning to come to accept new generalizations, as the omega rule would require us to do. (shrink)

Timothy Williamson has defended two hypotheses concerning counterfactual conditionals: that necessity can be defined in counterfactual terms; and that we follow a heuristic to the effect that a counterfactual is assessed by assessing the consequent while counterfactually supposing the antecedent. The two hypotheses form the bedrock for a program aiming to reduce the epistemology of modality to the epistemology of counterfactual thinking. This paper argues that the pair of theses, if construed as Williamson intends it, has the unwanted consequence of (...) trivializing our judgements about necessity and possibility, thus threatening the reductionist program. Trivialization can be avoided if we suitably weaken a background hypothesis concerning the way we assess pairs of mutually inconsistent statements. One important corollary of the proposed solution is that the suppositional heuristic need not yield incorrect judgements about counterpossibles, pace Williamson. Moreover, the proposal remains compatible with vacuism, the view that all counterpossibles are true. (shrink)

Nozick has an account of inferential knowledge which has rarely been discussed. According to this account, in order to know q via competent inference from p, S’s belief in q should track the truth of p in the right way. In detail, S knows via competent inference from p that q iff 1*. S knows that p. 2*. q is true, and S infers q from p. 3*. If q were false, S wouldn’t believe that p. 4*. If q were (...) true, S would believe that p. In this paper, it is argued that the account is problematic. It is both too restrictive and too inclusive. It is too restrictive because condition 4* is not satisfied in some cases of inferential knowledge. It is too inclusive because condition 3*, together with other conditions, could be easily satisfied in some cases where inferential knowledge does not exist. (shrink)

By a reasoned change in logic I mean a change in the logic with which you make inferences that is based on your evidence. An argument sourced in recently published material Kripke lectured on in the 1970s, and dubbed the Adoption Problem by Birman (then Padró) in her 2015 dissertation, challenges the possibility of reasoned changes in logic. I explain why evidentialists should be alarmed by this challenge, and then I go on to dispel it. The Adoption Problem rests on (...) a failure to distinguish between logical principles such as Universal Instantiation and Modus Ponens which might or might not govern your inferences with superficially similar laws which must govern your cognitive architecture. (shrink)

According to a widely held view in the philosophy of mathematics, direct inferential justification for mathematical propositions (that are not axioms) requires proof. I challenge this view while accepting that mathematical justification requires arguments that are put forward as proofs. I argue that certain fallacious putative proofs considered by the relevant subjects to be correct can confer mathematical justification. But mathematical justification doesn’t come for cheap: not just any argument will do. I suggest that to successfully transmit justification an argument (...) must satisfy specific standards, some of which are social. I contrast my view with Huemer’s inferential conservatism, which makes mathematical justification too easy to get. Although in this article I focus on mathematical inferential beliefs, the view on offer generalizes to other inferential beliefs. (shrink)

The aim of this paper is to motivate and defend Evidential Knowledge First, a view that combines the best of both knowledge first and evidentialist approaches to epistemology. To do so, I argue that knowledge is not the most general factive mental state. However, non-inferential knowledge is. After developing the core commitments of the view, I investigate what else the Evidential Knowledge Firster should say. A similar view is considered and rejected by Alexander Bird (2025). I conclude by addressing his (...) objections. (shrink)

In reasoning, we draw conclusions from multiple premises. But thinkers can be fragmented. And if there is no single fragment of the agent that thinks all of the premises, then the agent cannot draw any conclusions from them. It follows that reasoning from multiple premises depends on their being thought together. But what is it to think premises together? What is the condition that contrasts with fragmentation? This paper provides an answer to this question that is simple but compelling: to (...) think premises together is to attribute those very thoughts to one and the same self. The ability to reason depends on the ability to attribute the thoughts from which one reasons to oneself. Reasoning—at least the sort of reasoning that we do—requires de se thought. (shrink)

This paper argues that coherent reasoning is best understood as implication-sensitive branch selection rather than as local output emission or mere retention of historically successful patterns. Reasoning begins with a compressed claim whose implications are not yet fully explicit. What is commonly called chain-of-thought is better understood here as implication-unfolding reasoning: the progressive disclosure of the continuations, conflicts, and admissible developments already latent in an initial claim. -/- This unfolding reveals a structural divide between retention-principles, which explain persistence, origin, and (...) maintenance, and closure-principles, which explain why certain branches are valid, binding, or worth carrying forward independently of local preservation. Existing traditions already provide key ingredients for this picture: reasoning as model-based management of possibilities, communication as inferential coordination, and self-organizing cognition as uncertainty-sensitive regulation. -/- The contribution of this paper is to show that, when followed far enough, implication-unfolding reasoning exposes a non-reducible gap between persistence and purposive validity. Frameworks centered on self-maintenance or historical selection can therefore explain continuity without thereby explaining non-indexical bindingness or purposive resolution. The paper concludes that any complete theory of reasoning, whether biological or artificial, must supply an account of closure in addition to retention. (shrink)

Building on Nozick’s (2001) work, I provided a novel invariantist account of scientific objectivity: the Counterfactual Independence Account (Reutlinger 2024). Despite its virtues, this account needs to be articulated in a more nuanced manner – regarding its scope, content, and consequences. My goal in this paper is to develop such a more articulated and improved version of the Counterfactual Independence Account in two steps. First, I will discuss two instructive dissonances between Nozick’s version of invariantism and the Counterfactual Independence Account (...) – dissonances concerning what concept of objectivity is addressed and how the key notion of invariance is characterized. Second, I will develop an argument for the claim that (epistemic) objectivity plays a productive role in scientific inference or, more precisely, that objective evidence is, in a sense to be specified, more evidence. (shrink)

Testimony can be a source of knowledge. This paper examines how misinterpretation, or the risk of it, can prevent a hearer from acquiring testimonial knowledge. Because unreliability in interpretation can arise in many ways, section 2 considers a variety of such cases. Section 3 sketches some desiderata for a successful account of the role of interpretation in testimony, by analogous consideration of inference. On our account, interpretation needn’t proceed inferentially through knowledgeable belief about what is said. Finally, section 4 offers (...) a safety-theoretic account of reliable interpretation which explains how and when misinterpretation prevents knowledge. (shrink)

We argue that stereotypes associated with concepts like he-said–she-said, conspiracy theory, sexual harassment, and those expressed by paradigmatic slurs provide “normative inference tickets”: conceptual permissions to automatic, largely unreflective normative conclusions. These “mental shortcuts” are underwritten by associated stereotypes. Because stereotypes admit of exceptions, normative inference tickets are highly flexible and productive, but also liable to create serious epistemic and moral harms. Epistemically, many are unreliable, yielding false beliefs which resist counterexample; morally, many perpetuate bigotry and oppression. Still, some normative (...) inference tickets, like some activated by sexual harassment, constitute genuine moral and hermeneutical advances. For example, our framework helps explain Miranda Fricker’s notion of “hermeneutical lacunae”: what early victims of “sexual harassment” — as well as their harassers — lacked before the term was coined was a communal normative inference ticket — one that could take us, collectively, from “this is happening” to “this is wrong.”. (shrink)

Some philosophers have claimed that there is knowledge from falsehood (KFF, in short), i.e., inferential knowledge that involves a relevant false premise. The main thesis of this paper is that there are no standard cases of KFF. By ‘standard cases’ we mean cases in which the subject employs a measurement procedure in order to determine the value of some quantity, such as the time or the number of people present in a room. If knowledge is attained at all, it is (...) attained by inference not from a false premise but, instead, from two pieces of reflective knowledge. A brief look at putative non-standard cases of KFF reveals that they can be explained away in a different way, albeit a way that shares a common pattern with standard cases. (shrink)

For the first time in decades, our public conversation around technology feels fundamentally unstable. The surface narrative is familiar, rapid progress, disruptive potential, and staggering valuations, but just beneath lies a pervasive anxiety that seems to grow with every press release. Much of this unease is expressed through our usual tropes, the fear of AI replacing workers, frustration with corporate consolidation, questions about its ecological impact, concerns about bias and personality, and confusion about how models seem simultaneously brilliant and dumb. (...) These visible tensions are symptoms, not causes. The root of our problem is structural. (shrink)

Modern AI is driven by five dominant techniques, gradient descent, Bayesian inference, variational free-energy minimization, Markov models, and reinforcement learning. Although historically independent, these methods share a structural assumption; each optimizes a scalar objective inside a fixed representational space. Their successes arise when the world remains compatible with these frozen ontologies; their failures emerge whenever structural novelty is required. This paper reframes these techniques as constrained projections of a deeper coherence dynamic. When the representational space Ω is held rigid, coherence (...) ascent collapses into the familiar behaviors of parameter tuning, posterior updating, or policy optimization. When Ω is allowed to evolve, the full Coherence Engine becomes visible. Adaptive model expansion, structural pruning, and curvature-driven transitions restore the system’s ability to generate new abstractions, causal structures, and strategies. The future of AI depends not on improved optimization within predetermined models but on principled mechanisms for the evolution of models themselves. Coherence dynamics provides that mechanism, unifying inference, prediction, control, and representation under a single meta-law. (shrink)

The Coherence Engine presents a unified operational framework for understanding how physical, biological, cognitive, and social systems persist and adapt. It identifies coherence, recursive structural reinforcement, as the single invariant governing system viability across scales. Using minimal state representation and a universal update law, the framework models system evolution as the joint ascent of coherence gradients and descent of entropy gradients while preserving future adaptive capacity. The Spectrum of Possibility defines feasible trajectories, and Recursive Choice selects those maximizing coherence curvature (...) under constraints of noise, leakage, and dimensional change. Diverse scientific domains emerge as instantiations of the same kernel with quantum processes as coherence-weighted transitions, information as coherence-weighted entropy, cognition as attractor navigation, evolution as coherence selection, and societal dynamics as shifts in multi-scale structural alignment. By making coherence computable and measurable, the Coherence Engine enables simulation, empirical testing, and principled design of systems that must remain viable under uncertainty. (shrink)

An influential objection to the epistemic power of the imagination holds that it is uninformative. You cannot get more out of the imagination than you put into it, and therefore learning from the imagination is impossible. This paper argues, against this view, that the imagination is robustly informative. Moreover, it defends a novel account of how the imagination informs, according to which the imagination is informative in virtue of its analog representational format. The core idea is that analog representations represent (...) relations ‘for free,’ and this explains how the imagination can contain more information than is put into it. This account makes important contributions to both philosophy of mind, by showing how the imagination can generate new content that is not represented by a subject’s antecedent mental states, and epistemology, by showing how the imagination can generate new justification that is not conferred by a subject’s antecedent evidence. (shrink)

This paper argues that inference rule adoption is a diachronic process during which agents are inferentially guided by a statement of the rule they are adopting, but during which they do not use that rule. Rather, the ability to use the rule is the outcome at the end of the process. This account avoids a regress objection to inferentially guided adoption recently posed by Boghossian and Wright. Adoption, on this model, involves the use of six privileged inference rules, including universal (...) instantiation, modus ponens, and the naive truth rules. However, though these rules play a special role in adoption, they are not indispensable to the process in their full generality: so long as one has reasonably robust restrictions of them, the adoption process is unimpeded. Furthermore, there are no general forms that constitute the minimal restrictions of these rules required for adoption. It follows that these rules, in their fully generality, are themselves adoptable, so long as one begins with reasonable restrictions of them. It is argued that this is enough to overcome the ‘adoption problem’ as a challenge to anti-exceptionalism about logic and to the normative significance of proposals for alternative logics. (shrink)

Hilary Kornblith is one of the world’s leading epistemologists, a champion of an innovative philosophical research program that is at once traditional and revisionary. In viewing the study of knowledge as inseparable from the empirical study of the mind, Kornblith aligns himself closely with the approach of the traditional empiricists of the 17th and 18th centuries. Yet in taking contemporary empirical work seriously, Kornblith has developed views and arguments that shift the epistemological focus away from what is available first-personally _within_ (...) the mind towards what is revealed third-personally _of_ the mind instead. Indeed, anyone looking seriously at the history of 20th century epistemology can draw a straight line from W.V.O. Quine through Alvin Goldman to Hilary Kornblith – all three, in their ways, champions of a science-centered epistemology. On the innovative leading edge of that tradition, however, Kornblith has developed a systematic rejection of the traditional philosophical method of conceptual analysis (_pace_ Goldman) while establishing a thoroughly empirical and yet unambiguously philosophical approach to the study of knowledge (_pace_ Quine). As such, Kornblith’s work is required reading for anyone interested in the structure and nature of knowledge, the structure and grounds of justification, the sources of epistemic normativity, the history and the prospects for a naturalized epistemology, the legitimacy of intuition-dependent conceptual analysis as a philosophical method, and the significance and reliability of reflection and reasoning. This volume collects 16 original essays that advance the state of play by engaging critically and substantively with Kornblith’s views on these and other topics, along with an essay by Kornblith himself replying to his critics. (shrink)

It's often thought that when we reason to new judgments in inference, we aim at believing the truth, and that this aim of ours can explain important psychological and normative features of belief. I reject this picture: the structure of aimed activity shows that inference is not guided by a truth‐aim. This finding clears the way for a positive understanding of how epistemic goods feature in our doxastic lives. We can indeed make sense of many of our inquisitive and deliberative (...) activities as undertaken in pursuit of such goods; but the evidence‐guided inferences in which those activities culminate will require a different theoretical approach. (shrink)

Inferences are familiar movements of thought, but despite important recent work on the topic, we do not yet have a fully satisfying theory of inference. Here I provide a functionalist theory of inference. I argue that the functionalist framework allows us the flexibility to meet various demands on a theory of inference that have been proposed (such as that it must explain inferential Moorean phenomena and epistemological ‘taking’). While also allowing us to compare, contrast, adapt, and combine features of extant (...) theories of inference into one unified theory. In fleshing out the inference role, I also criticize the common assumption that inference requires rule-following. (shrink)

In the epistemology of self-knowledge, Inferentialism is the view that one’s current mental states are normally known to one through inferences from evidence. This view is often taken to conflict with widespread claims about normally-acquired self-knowledge, namely that it is privileged (essentially more secure than knowledge of others’ minds) and peculiar (obtained in a way that fundamentally differs from how others know your mind). In this paper I argue that Inferentialism can be reconceived so as to no longer conflict with (...) these claims. On a positive framing, Inferentialism earns a seat at the table alongside other theories of self-knowledge that take its potentially privileged and peculiar status seriously. On a less positive framing, Inferentialism’s plausibility now hinges on the plausibility of viewing self-knowledge as at least peculiar, and perhaps as privileged as well. (shrink)

Rodrigo Borges (2020, 2017) argumenta que os supostos casos de conhecimento via falsidade não são casos nos quais o status epistêmico da crença-alvo p dependa de uma crença falsa e, ainda assim, concede que a crença em p é uma instância de conhecimento. O objetivo deste artigo é analisar essa alegação. Para isso, iniciarei explicando o que é o problema do conhecimento via falsidade e, na sequência apresentarei a proposta concessiva de Rodrigo Borges. Por fim, argumentarei que essa proposta não (...) é adequada por não acomodar todos os supostos casos de conhecimento via falsidade. (shrink)

The perennial question of the nature of natural-language understanding has received renewed attention in recent years. Two kinds of natural-language understanding, in particular, have captivated the interest of philosophers: linguistic understanding and utterance understanding. While the literature is rife with discussions of linguistic understanding and utterance understanding, the question of how the two types of understanding explanatorily depend on each other has received relatively scant attention. Exceptions include the linguistic ability/know-how views of linguistic understanding proposed by Dean Pettit and Brendan (...) Balcerak Jackson. On these views, to tacitly linguistically understand a sentence just is to possess the linguistic ability/knowledge-how needed to derive/infer what is said by different utterances of the sentence. Despite their focus on linguistic understanding, both views can straightforwardly explain utterance understanding as the output of a derivation/inference from a representation of the sentence uttered. Here, I take issue with these approaches to utterance understanding and then develop an alternative. More specifically, I distinguish two kinds of utterance understanding, experiential and doxastic, and then argue that experiences of what is said by utterances play distinct rational roles in the two kinds of utterance understanding. I conclude by addressing a recent challenge to my proposal. (shrink)

There is a felt difference between following an argument to its conclusion and keeping up with an argument in your judgments while failing to see how its conclusion follows from its premises. In the first case there’s what I’m calling an inferential seeming, in the second case there isn’t. Inferential seemings exhibit a cluster of functional and normative characteristics whose integration in one mental state is puzzling. Several recent accounts of inferring suggest inferential seemings play a significant role in the (...) process, but none provides a fully satisfactory understanding of inferential seemings themselves. In this paper I critically examine theoretical options on offer in the existing literature, then develop an alternative view. I discuss implications for recent debates about general principles governing inference, such as Fumerton’s Principle of Inferential Justification and Boghossian’s Taking Condition. (shrink)

This paper discusses the account of alethic modality as presented by Wilfrid Sellars in his earlier work from 1947 to 1958. Its aim is twofold. First, I discuss Sellars' analysis by exploring its historical relationship to Carnap's account of modality. I argue that Carnap's early syntactic treatment of modality profoundly influenced Sellars' own so-called ‘regulist' account of modality in terms of rules of inference. Furthermore, it is suggested that Sellars' lesser-known possible worlds analysis was influenced by Carnap's later semantic account (...) of modality. The second aim of the paper is a critical one. I raise a number of objections to Sellars’ account of modality and argue that his account confronts some serious limitations. I also argue that Sellars does not provide a satisfying explanation of the relationship between his regulist account and his possible worlds analysis. I offer such an explanation by arguing that his regulist account should be understood as a pragmatic account of what is conveyed by the use of modal sentences, whereas his possible worlds analysis is a semantic account of the asserted content of modal statements. That is, the regulist account and the possible worlds analysis have different explanatory targets. (shrink)

The empirical success of particle physics rests largely on an approximation method: perturbation theory. Yet even within perturbative quantum field theory, there are a variety of different formulations. This variety teaches us that reformulating approximation methods can provide a tremendous source of progress in science. Along with enabling the solution of otherwise intractable problems, reformulations clarify what we need to know to obtain solutions, which can in turn make previously hidden properties manifest. To develop these lessons, I compare and contrast (...) three compatible formulations of perturbative QFT: (i) elementary perturbation theory, (ii) the method of Feynman diagrams, and (iii) a recent reformulation known as on-shell recursion. I propose and defend a novel account of what it means to ‘make a property manifest,’ based on the inferences that a formulation warrants. (shrink)

This paper discusses a novel response to two closely related regress arguments from Bolzano’s Theory of Science and Carroll’s What the Tortoise Said to Achilles. Bolzano’s argument aims to refute the thesis that full grounds must include propositions involving notions such as entailment, grounding or lawhood which link the respective grounds to their groundee. This thesis is motivated, Bolzano’s argument is reconstructed, and a response based on self-referential linking propositions is developed and defended against objections concerning self-reference and Curry’s paradox. (...) Finally, the idea is applied to a reading of Carroll’s dialogue and a corresponding solution to the so-called infinite regress problem of inference is proposed. (shrink)