[This is an edited version of a keynote talk at the conference on Neural Information Processing Systems (NeurIPS) on November 28, 2022, with some minor additions and subtractions.] -/- There has recently been widespread discussion of whether large language models might be sentient or conscious. Should we take this idea seriously? I will break down the strongest reasons for and against. Given mainstream assumptions in the science of consciousness, there are significant obstacles to consciousness in current models: for example, their (...) lack of recurrent processing, a global workspace, and unified agency. At the same time, it is quite possible that these obstacles will be overcome in the next decade or so. I conclude that while it is somewhat unlikely that current large language models are conscious, we should take seriously the possibility that successors to large language models may be conscious in the not-too-distant future. (shrink)

In recent years, metaphysics has undergone what some describe as a revolution: it has become standard to understand a vast array of questions as questions about grounding, a metaphysical notion of determination. Why should we believe in grounding, though? Supporters of the revolution often gesture at what I call the Argument from Explanatoriness: the notion of grounding is somehow indispensable to a metaphysical type of explanation. I challenge this argument and along the way develop a “reactionary” view, according to which (...) there is no interesting sense in which the notion of grounding is explanatorily indispensable. I begin with a distinction between two conceptions of grounding, a distinction which extant critiques of the revolution have usually failed to take into consideration: grounding qua that which underlies metaphysical explanation and grounding qua metaphysical explanation itself. Accordingly, I distinguish between two versions of the Argument from Explanatoriness: the Unexplained Explanations Version for the first conception of grounding, and the Expressive Power Version for the second. The paper’s conclusion is that no version of the Argument from Explanatoriness is successful. (shrink)

Conciliationism holds that disagreement of apparent epistemic peers often substantially undermines rational confidence in our opinions. Uniqueness principles say that there is at most one maximally rational doxastic response to any given batch of total evidence. The two views are often thought to be tightly connected. This paper distinguishes two ways of motivating conciliationism, and two ways that conciliationism may be undermined by permissive accounts of rationality. It shows how conciliationism can flourish under certain strongly permissive accounts of rationality. This (...) occurs when the motivation for conciliationism does not come from taking disagreement as evidence of one's own rational failings. However, divorcing the motive for conciliating from worries about rationality does not remove a feature of conciliationism that some find troubling: that conciliationism can lead to cases of “rational toxicity,” in which the most rational response to one's evidence involves violating some rational ideal. (shrink)

How do we determine whether some candidate causal factor is an actual cause of some particular outcome? Many philosophers have wanted a view of actual causation which fits with folk intuitions of actual causation and those who wish to depart from folk intuitions of actual causation are often charged with the task of providing a plausible account of just how and where the folk have gone wrong. In this paper, I provide a range of empirical evidence aimed at showing just (...) how and where the folk go wrong in determining whether an actual causal relation obtains. The evidence suggests that folk intuitions of actual causation are generated by two epistemically defective processes. I situate the empirical evidence within a background discussion of debunking, arguing for a two-pronged debunking explanation of folk intuitions of actual causation. I conclude that those who wish to depart from folk intuitions of actual causation should not be compelled to square their account of actual causation with the verdicts of the folk. In the dispute over actual causation, folk intuitions deserve to be rejected. (shrink)

Does consciousness collapse the quantum wave function? This idea was taken seriously by John von Neumann and Eugene Wigner but is now widely dismissed. We develop the idea by combining a mathematical theory of consciousness (integrated information theory) with an account of quantum collapse dynamics (continuous spontaneous localization). Simple versions of the theory are falsified by the quantum Zeno effect, but more complex versions remain compatible with empirical evidence. In principle, versions of the theory can be tested by experiments with (...) quantum computers. The upshot is not that consciousness-collapse interpretations are clearly correct, but that there is a research program here worth exploring. (shrink)

This paper defends the view that perceptual ascriptions such as “Jones sees a cat” are sometimes intensional. I offer a range of examples of intensional perceptual ascriptions, respond to objections to intensional readings of perceptual ascriptions, and show how widely accepted semantic accounts of intensionality can explain the key features of intensional perceptual ascriptions.

There is a difference between merely thinking that P and really grasping that P. For example, Jackson's (1982) black-and-white Mary cannot (before leaving her black-and-white room) fully grasp what it means to say that fire engines are red, but she can perfectly well entertain the thought that fire engines are red. The contrast between merely thinking and grasping is especially salient in the context of certain moral decisions. For example, an individual who grasps the plight of starving children thanks to (...) evocative images seems much more likely to donate for the children's benefit than someone who is merely told about their plight. This paper offers a phenomenal account of grasping according to which grasping a content (whether a proposition or concept) consists in one's thoughts about the content being constituted by phenomenal experiences of constituents of the content. This account improves on earlier phenomenal accounts of grasping (Bourget 2017a, 2018) by allowing for partial grasping and generalizing to concepts. The paper also addresses objections to phenomenal accounts of grasping, including the objection that recent developments in AI technology show that consciousness is irrelevant to grasping and understanding. (shrink)

The paper considers contemporary models of presumption in terms of their ability to contribute to a working theory of presumption for argumentation. Beginning with the Whatelian model, we consider its contemporary developments and alternatives, as proposed by Sidgwick, Kauffeld, Cronkhite, Rescher, Walton, Freeman, Ullmann-Margalit, and Hansen. Based on these accounts, we present a picture of presumptions characterized by their nature, function, foundation and force. On our account, presumption is a modal status that is attached to a claim and has the (...) effect of shifting, in a dialogue, a burden of proof set at a local level. Presumptions can be analysed and evaluated inferentially as components of rule-based structures. Presumptions are defeasible, and the force of a presumption is a function of its normative foundation. This picture seeks to provide a framework to guide the development of specific theories of presumption. (shrink)

I argue that some important elements of the current cosmological model are 'conventionalist’ in the sense defined by Karl Popper. These elements include dark matter and dark energy; both are auxiliary hypotheses that were invoked in response to observations that falsified the standard model as it existed at the time.

How should one react when one has a belief, but knows that other people—who have roughly the same evidence as one has, and seem roughly as likely to react to it correctly—disagree? This paper argues that the disagreement of other competent inquirers often requires one to be much less confident in one’s opinions than one would otherwise be.

Presentists, who believe that only present objects exist, should accept a thisness ontology, since it can do considerable work in defence of presentism. In this paper, I propose a version of presentism that involves thisnesses of past and present entities and I argue this view solves important problems facing standard versions of presentism.

Are the special sciences autonomous from physics? Those who say they are need to explain how dependent special science properties could feature in irreducible causal explanations, but that’s no easy task. The demands of a broadly physicalist worldview require that such properties are not only dependent on the physical, but also physically realized. Realized properties are derivative, so it’s natural to suppose that they have derivative causal powers. Correspondingly, philosophical orthodoxy has it that if we want special science properties to (...) bestow genuinely new causal powers, we must reject physical realization and embrace a form of emergentism, in which such properties arise from the physical by mysterious brute determination. In this paper, I argue that contrary to this orthodoxy, there are physically realized properties that bestow new causal powers in relation to their realizers. The key to my proposal is to reject causal-functional accounts of realization and embrace a broader account that allows for the realization of shapes and patterns. Unlike functional properties, such properties are defined by qualitative, non-causal specifications, so realizing them does not consist in bestowing causal powers. This, I argue, allows for causal novelty of the strongest kind. I argue that the molecular geometry of H2O—a qualitative, multiply realizable property—plays an irreducible role in explaining its dipole moment, and thereby bestows novel powers. On my proposal, special science properties can have the kind of causal novelty traditionally associated with strong emergence, without any of the mystery. (shrink)

Conceptual engineering is the design, implementation, and evaluation of concepts. Conceptual engineering includes or should include de novo conceptual engineering (designing a new concept) as well as conceptual re-engineering (fixing an old concept). It should also include heteronymous (different-word) as well as homonymous (same-word) conceptual engineering. I discuss the importance and the difficulty of these sorts of conceptual engineering in philosophy and elsewhere.

How should one react when one has a belief, but knows that other people—who have roughly the same evidence as one has, and seem roughly as likely to react to it correctly—disagree? This paper argues that the disagreement of other competent inquirers often requires one to be much less confident in one’s opinions than one would otherwise be.

Cartesian arguments for global skepticism about the external world start from the premise that we cannot know that we are not in a Cartesian scenario such as an evil-demon scenario, and infer that because most of our empirical beliefs are false in such a scenario, these beliefs do not constitute knowledge. Veridicalist responses to global skepticism respond that arguments fail because in Cartesian scenarios, many or most of our empirical beliefs are true. Some veridicalist responses have been motivated using verificationism, (...) externalism, and coherentism. I argue that a more powerful veridicalist response to global skepticism can be motivated by structuralism, on which physical entities are understood as those that play a certain structural role. I develop the structuralist response and address objections. (shrink)

Recent years have seen a surge in online collaboration between experts and amateurs on scientific research. In this article, we analyse the epistemological implications of these crowdsourced projects, with a focus on Zooniverse, the world’s largest citizen science web portal. We use quantitative methods to evaluate the platform’s success in producing large volumes of observation statements and high impact scientific discoveries relative to more conventional means of data processing. Through empirical evidence, Bayesian reasoning, and conceptual analysis, we show how information (...) and communication technologies enhance the reliability, scalability, and connectivity of crowdsourced e-research, giving online citizen science projects powerful epistemic advantages over more traditional modes of scientific investigation. These results highlight the essential role played by technologically mediated social interaction in contemporary knowledge production. We conclude by calling for an explicitly sociotechnical turn in the philosophy of science that combines insights from statistics and logic to analyse the latest developments in scientific research. (shrink)

Working within the broad lines of general consensus that mark out the core features of John Stuart Mill’s (1806–1873) logic, as set forth in his A System of Logic (1843–1872), this chapter provides an introduction to Mill’s logical theory by reviewing his position on the relationship between induction and deduction, and the role of general premises and principles in reasoning. Locating induction, understood as a kind of analogical reasoning from particulars to particulars, as the basic form of inference that is (...) both free-standing and the sole load-bearing structure in Mill’s logic, the foundations of Mill’s logical system are briefly inspected. Several naturalistic features are identified, including its subject matter, human reasoning, its empiricism, which requires that only particular, experiential claims can function as basic reasons, and its ultimate foundations in ‘spontaneous’ inference. The chapter concludes by comparing Mill’s naturalized logic to Russell’s (1907) regressive method for identifying the premises of mathematics. (shrink)

Bernard Williams’ discussion of immortality in “The Makropulos Case: Reflections on the Tedium of Immortality” has spawned an entire philosophical literature. This literature tends to focus on one of Williams’ central claims: if we were to relinquish our mortality, we would necessarily become alienated from our existence and environment—“bored,” in his terms. Many theorists have defended this claim; many others have challenged it. Even if this claim is false, though, it still isn’t obvious that we should choose to relinquish our (...) mortality, given the option. And this is puzzling. Death is generally regarded as an evil, meriting anxiety, dread, and avoidance. What is it about our mortality that makes us hesitant to relinquish it? In this paper, I aim to explain this hesitancy and to argue in its favor: we should be hesitant to relinquish our mortality. Further, I take my discussion to suggest that most of us probably shouldn’t take this option if we were given it. To make these points, I draw on Williams’ discussion of immortality, developing and extending his thought in order to better understand how mortality shapes the way we relate to our lives, particularly the commitments and values that make them meaningful. Our mortality, it seems, plays a crucial role in our lives standing for something. Relinquishing it thus means risking the loss of this value. (shrink)

This book concerns the foundations of epistemic modality and hyperintensionality and their applications to the philosophy of mathematics. I examine the nature of epistemic modality, when the modal operator is interpreted as concerning both apriority and conceivability, as well as states of knowledge and belief. The book demonstrates how epistemic modality and hyperintensionality relate to the computational theory of mind; metaphysical modality and hyperintensionality; the types of mathematical modality and hyperintensionality; to the epistemic status of large cardinal axioms, undecidable propositions, (...) and abstraction principles in the philosophy of mathematics; to the modal and hyperintensional profiles of the logic of rational intuition; and to the types of intention, when the latter is interpreted as a hyperintensional mental state. Chapter \textbf{2} argues for a novel type of expressivism based on the duality between the categories of coalgebras and algebras, and argues that the duality permits of the reconciliation between modal and hyperintensional cognitivism and modal and hyperintensional expressivism. I also develop a novel topic-sensitive truthmaker semantics for dynamic epistemic logic, and develop a novel dynamic two-dimensional semantics grounded in two-dimensional hyperintensional Turing machines. Chapter \textbf{3} provides an abstraction principle for epistemic (hyper-)intensions. Chapter \textbf{4} advances a topic-sensitive two-dimensional truthmaker semantics, and provides three novel interpretations of the framework along with the epistemic and metasemantic. Chapter \textbf{5} applies the fixed points of the modal $\mu$-calculus in order to account for the iteration of epistemic states in a single agent, by contrast to availing of modal axiom 4 (i.e. the KK principle). The fixed point operators in the modal $\mu$-calculus are rendered hyperintensional, which yields the first hyperintensional construal of the modal $\mu$-calculus in the literature and the first application of the calculus to the iteration of epistemic states in a single agent instead of the common knowledge of a group of agents. Chapter \textbf{6} advances a solution to the Julius Caesar problem based on Fine's `criterial' identity conditions which incorporate conditions on essentiality and grounding. Chapter \textbf{7} provides a ground-theoretic regimentation of the proposals in the metaphysics of consciousness and examines its bearing on the two-dimensional conceivability argument against physicalism. The topic-sensitive epistemic two-dimensional truthmaker semantics developed in chapters \textbf{2} and \textbf{4} are availed of in order for epistemic states to be a guide to metaphysical states in the hyperintensional setting. -/- Chapters \textbf{8-12} provide cases demonstrating how the two-dimensional hyperintensions of hyperintensional, i.e. topic-sensitive epistemic two-dimensional truthmaker, semantics, solve the access problem in the epistemology of mathematics. Chapter \textbf{8} examines the interaction between my hyperintensional semantics and the axioms of epistemic set theory, large cardinal axioms, the Epistemic Church-Turing Thesis, the modal axioms governing the modal profile of $\Omega$-logic, Orey sentences such as the Generalized Continuum Hypothesis, and absolute decidability. These results yield inter alia the first hyperintensional Epistemic Church-Turing Thesis and hyperintensional epistemic set theories in the literature. Chapter \textbf{9} examines the modal and hyperintensional commitments of abstractionism, in particular necessitism, and epistemic hyperintensionality, epistemic utility theory, and the epistemology of abstraction. I countenance a hyperintensional semantics for novel epistemic abstractionist modalities. I suggest, too, that observational type theory can be applied to first-order abstraction principles in order to make first-order abstraction principles recursively enumerable, i.e. Turing machine computable, and that the truth of the first-order abstraction principle for hyperintensions is grounded in its being possibly recursively enumerable and the machine being physically implementable. Chapter \textbf{10} examines the philosophical significance of hyperintensional $\Omega$-logic in set theory and discusses the hyperintensionality of metamathematics. Chapter \textbf{11} provides a modal logic for rational intuition and provides a hyperintensional semantics. Chapter \textbf{12} avails of modal coalgebras to interpret the defining properties of indefinite extensibility, and avails of hyperintensional epistemic two-dimensional semantics in order to account for the interaction between interpretational and objective modalities and the truthmakers thereof. This yields the first hyperintensional category theory in the literature. I invent a new mathematical trick in which first-order structures are treated as categories, and Vopenka's principle can be satisfied because of the elementary embeddings between the categories and generate Vopenka cardinals in the category of Set in category theory. Chapter \textbf{13} examines modal responses to the alethic paradoxes. I provide a counter-example to epistemic closure for logical deduction. Chapter \textbf{14} examines, finally, the modal and hyperintensional semantics for the different types of intention and the relation of the latter to evidential decision theory. (shrink)

I insist that I was able to raise my hand, and I acknowledge that a law would have been broken had I done so, but I deny that I am therefore able to break a law. To uphold my instance of soft determinism, I need not claim any incredible powers. To uphold the compatibilism that I actually believe, I need not claim that such powers are even possible. My incompatibilist opponent is a creature of fiction, but he has his prototypes (...) in real life. He is modeled partly after Peter van Inwagen and partly on myself when I first worried about van Inwagen's argument against compatibilism. (shrink)

Responding rationally to the information that others disagree with one’s beliefs requires assessing the epistemic credentials of the opposing beliefs. Conciliatory accounts of disagreement flow in part from holding that these assessments must be independent from one’s own initial reasoning on the disputed matter. I argue that this claim, properly understood, does not have the untoward consequences some have worried about. Moreover, some of the difficulties it does engender must be faced by many less conciliatory accounts of disagreement.

We present data and argument to show that in Tetris - a real-time interactive video game - certain cognitive and perceptual problems are more quickly, easily, and reliably solved by performing actions in the world rather than by performing computational actions in the head alone. We have found that some translations and rotations are best understood as using the world to improve cognition. These actions are not used to implement a plan, or to implement a reaction; they are used to (...) change the world in order to simplify the problem-solving task. Thus, we distinguish pragmatic actions ñ actions performed to bring one physically closer to a goal - from epistemic actions - actions performed to uncover information that is hidden or hard to compute mentally. To illustrate the need for epistemic actions, we first develop a standard information-processing model of Tetris-cognition, and show that it cannot explain performance data from human players of the game - even when we relax the assumption of fully sequential processing. Standard models disregard many actions taken by players because they appear unmotivated or superfluous. However, we describe many such actions that are actually taken by players that are far from superfluous, and that play valuable roles in improving human performance. We argue that traditional accounts are limited because they regard action as having a single function: to change the world. By recognizing a second function of action - an epistemic function - we can explain many of the actions that a traditional model cannot. Although, our argument is supported by numerous examples specifically from Tetris, we outline how the one category of epistemic action can be incorporated into theories of action more generally. (shrink)

Humour is worthy of serious ethical consideration. However, it is often taken far too seriously. In this paper, it is argued that while humour is sometimes unethical, it is wrong much less often than many people think. Non-contextual criticisms, which claim that certain kinds of humour are always wrong, are rejected. Contextual criticisms, which take issue with particular instances of humour rather than types of humour, are more promising. However, it is common to overstate the number of contexts in which (...) humour is wrong. Various mistakes of this kind are highlighted and cautioned against. (shrink)

This is a collective study of the epistemic significance of disagreement: twelve contributors explore rival responses to the problems that it raises for philosophy. They develop our understanding of epistemic phenomena that are central to any thoughtful engagement with others' beliefs.

In the pure powers ontology (PPO), basic physical properties have wholly dispositional essences. PPO has clear advantages over categoricalist ontologies, which suffer from familiar epistemological and metaphysical problems. However, opponents argue that because it contains no qualitative properties, PPO lacks the resources to individuate powers, and generates a regress. The challenge for those who take such arguments seriously is to introduce qualitative properties without reintroducing the problems that PPO was meant to solve. In this paper, I distinguish the core claim (...) of PPO: (i) basic physical properties have dispositional essences, from a hitherto unnoticed assumption: (ii) the dispositional essences of basic physical properties exclusively involve type-causal relations to other basic physical properties. I reject (ii), making room for a structuralist ontology in which all basic physical properties are pure powers, individuated by their places in a causal structure that includes not only other powers, but also physically realized qualitative properties such as shapes, patterns and structures. Such qualities individuate pure powers in the way that non-mental input and output properties individuate realized mental properties in functionalist theories of mind, except that here it is basic physical powers that are individuated by relations to realized non-powers. I distinguish one Platonic and two Aristotelian version of this theory, and argue that the Aristotelian versions require that grounding is not always a relative fundamentality relation, because the powers ground the qualities that individuate them. By considering ontic structural realism, I argue that symmetric grounding is the best way to make sense of relational individuation in structuralist ontologies, and is therefore no additional commitment of the one proposed here. (shrink)

Several different quantum gravity research programmes suggest, for various reasons, that spacetime is not part of the fundamental ontology of physics. This gives rise to the problem of empirical coherence: if fundamental physical entities do not occupy spacetime or instantiate spatiotemporal properties, how can fundamental theories concerning those entities be justified by observation of spatiotemporally located things like meters, pointers and dials? I frame the problem of empirical coherence in terms of entailment: how could a non-spatiotemporal fundamental theory entail spatiotemporal (...) evidence propositions? Solutions to this puzzle can be classified as realist or antirealist, depending on whether or not they posit a non-fundamental spacetime structure grounded in or caused by the fundamental structure. These approaches place different constraints on our everyday concepts of space and time. Applying lessons from the philosophy of mind, I argue that only realism is both conceptually plausible and suitable for addressing the problem at hand. I suggest a role functionalist version of realism, which is consistent with both grounding and causation, and according to which our everyday concepts reveal something of the true nature of emergent spacetime. (shrink)

In the course of daily life we solve problems often enough that there is a special term to characterize the activity and the right to expect a scientific theory to explain its dynamics. The classical view in psychology is that to solve a problem a subject must frame it by creating an internal representation of the problem’s structure, usually called a problem space. This space is an internally generable representation that is mathematically identical to a graph structure with nodes and (...) links. The nodes can be annotated with useful information, and the whole representation can be distributed over internal and external structures such as symbolic notations on paper or diagrams. If the representation is distributed across internal and external structures the subject must be able to keep track of activity in the distributed structure. Problem solving proceeds as the subject works from an initial state in mentally supported space, actively constructing possible solution paths, evaluating them and heuristically choosing the best. Control of this exploratory process is not well understood, as it is not always systematic, but various heuristic search algorithms have been proposed and some experimental support has been provided for them. (shrink)

Using as a starting point recent and apparently incompatible conclusions by Saunders and Knox, I revisit the question of the correct spacetime setting for Newtonian physics. I argue that understood correctly, these two versions of Newtonian physics make the same claims both about the background geometry required to define the theory, and about the inertial structure of the theory. In doing so I illustrate and explore in detail the view—espoused by Knox, and also by Brown —that inertial structure is defined (...) by the dynamics governing subsystems of a larger system. This clarifies some interesting features of Newtonian physics, notably the distinction between using the theory to model subsystems of a larger whole and using it to model complete universes, and the scale-relativity of spacetime structure. _1_ Introduction _2_ Newtonian Mechanics and Galilean Spacetime _3_ Vector Relationism and Maxwellian Spacetime _4_ Recovering the Galilei Group: Dynamics of Subsystems _5_ Knox on Inertial Structure _6_ Connections on Maxwellian Spacetime _7_ Knox on Newtonian Gravity _8_ Vector Relationism and Newton–Cartan Theory _9_ Inertial Structure in Newton–Cartan Gravity _10_ Reconciling Knox and Saunders _11_ Conclusions. (shrink)

In the first half of this paper, I argue that essential properties are intrinsic and that this permits a modal analysis of essence that is immune the sort of objections raised by Fine. In the second half, I argue that intrinsic properties collectively have a certain structure and that this accounts for some observations about essences: that things are essentially determinate; that things often have properties within a certain range essentially; and that the essential properties of things are their core (...) properties. (shrink)

A critical examination of the Reflection principle in Bayesian epistemology, and of the diachronic Dutch-book-style arguments that have been invoked to support Reflection and Conditionalization.

Why scientists reach an agreement on new experimental methods when there are conflicts of interest about the evidence they yield? I argue that debiasing methods play a crucial role in this consensus, providing a warrant about the impartiality of the outcome regarding the preferences of different parties involved in the experiment. From a contractarian perspective, I contend that an epistemic pre-requisite for scientists to agree on an experimental method is that this latter is neutral regarding their competing interests. I present (...) two medical experiments (on smallpox inoculation and Mesmerism) in which debiasing procedures such as blinding and data tabulation provided warrants of impartiality that made people agree on the experimental design even if they disagreed on the outcome. (shrink)

Lipsey and Lancaster's "general theory of second best" is widely thought to have significant implications for applied theorizing about the institutions and policies that most effectively implement abstract normative principles. It is also widely thought to have little significance for theorizing about which abstract normative principles we ought to implement. Contrary to this conventional wisdom, I show how the second-best theorem can be extended to myriad domains beyond applied normative theorizing, and in particular to more abstract theorizing about the normative (...) principles we should aim to implement. I start by separating the mathematical model used to prove the second-best theorem from its familiar economic interpretation. I then develop an alternative normative-theoretic interpretation of the model, which yields a novel second best theorem for idealistic normative theory. My method for developing this interpretation provides a template for developing additional interpretations that can extend the reach of the second-best theorem beyond normative theoretical domains. I also show how, within any domain, the implications of the second-best theorem are more specific than is typically thought. I conclude with some brief remarks on the value of mathematical models for conceptual exploration. (shrink)

Abstract: Theories of epistemic rationality that take disagreement (or other higher-order evidence) seriously tend to be “modest” in a certain sense: they say that there are circumstances in which it is rational to doubt their correctness. Modest views have been criticized on the grounds that they undermine themselves—they’re self-defeating. The standard Self-Defeat Objections depend on principles forbidding epistemically akratic beliefs; but there are good reasons to doubt these principles—even New Rational Reflection, which was designed to allow for certain special cases (...) that are intuitively akratic. On the other hand, if we construct a Self-Defeat Objection without relying on anti-akratic principles, modest principles turn out not to undermine themselves. In the end, modesty should not be seen as a defect in a theory of rational belief. (shrink)

Hylomorphists hold that substances are, in some sense, composites of matter and form. The form of a substance is typically taken to play a fundamental role in determining the unity or identity of the whole. Staunch hylomorphists think that this role is of a kind that precludes the ontological reduction of form to the physical and thus take their position to be inconsistent with physicalism. Forms, according to staunch hylomorphism, play a fundamental role in grounding their bearers’ proper parts and (...) that, it seems, rules out the physical grounding of form itself. I shall develop a physicalist version of hylomorphism that treats form as geometric structure and which, I shall argue, entails many of the central theses endorsed by staunch hylomorphists. Based on Shoemaker’s notion of conditional powers, I shall argue that the geometric structures of complex wholes are the conditions on at least some of the conditional powers of their bearers’ proper parts, transforming those powers into powers simpliciter. Thus, forms play a fundamental role in the dynamical evolution of the physical world, but without bestowing causal powers themselves and hence without violating the causal closure of the physical domain. If a substance’s proper parts are taken to be individuated by their powers simpliciter, then they derive their identities from their places in the substantial whole, which thus determines the identities of its proper parts without changing their intrinsic natures. (shrink)

The old evidence problem affects any probabilistic confirmation measure based on comparing pr(H/E) and pr(H). The article argues for the following points: (1) measures based on likelihood ratios also suffer old evidence difficulties; (2) the less-discussed synchronic old evidence problem is, in an important sense, the most acute; (3) prominent attempts to solve or dissolve the synchronic problem fail; (4) a little-discussed variant of the standard measure avoids the problem, in an appealing way; and (5) this measure nevertheless reveals a (...) different problem for probabilistic confirmation measures, a problem that is unlikely to lend itself to formal solution. (shrink)

It is conventional wisdom among political philosophers that ideal principles of justice must guide our attempts to design institutions to avert actual injustice. Call this the ideal guidance approach. I argue that this view is misguided— ideal principles of justice are not appropriate "guiding principles" that actual institutions must aim to realize, even if only approximately. Fortunately, the conventional wisdom is also avoidable. In this paper, I develop an alternative approach to institutional design, which I call institutional failure analysis. The (...) basic intuition of this approach is that our moral assessment of institutional proposals is most effective when we proceed from a detailed understanding of the causal processes generating problematic social outcomes. Failure analysis takes the institutional primary design task to be obviating or averting institutional failures. Consequently, failure analysis enables theorists to prescribe more effective solutions to actual injustice because its focuses on understanding the injustice, rather than specifying an ideal of justice. (shrink)

The causal closure of the physical poses a familiar causal exclusion problem for the special sciences that stems from the idea that if closure is true, then fundamental physical properties do all the causal work involved in bringing about physical effects. In this paper I aim to show that the strongest causal closure principle that is not ruled out by some simple physics in fact allows for a certain kind of downward causation, which in turn makes room for robust special (...) science autonomy. I focus on the case of vector composition, arguing that it involves irreducibly multilevel causation. To give a complete causal explanation of what goes on in cases that involve the effects of multiple forces, we need to appeal not only to the fundamental properties that generate those forces, but also to geometric properties that determine how multiple vectors compose. However, if all the causal work involved in such cases were due to fundamental physical properties, there would be a one-level explanation available. Hence, I argue, it cannot be the case that fundamental physical properties do all the causal work. The extra causal work, I suggest, comes from the fact that the causal powers bestowed by fundamental physical properties have irreducibly geometric manifestation conditions. I defend the resulting form of downward causation as a solution to the problem of special science autonomy, and discuss two potential accounts of its source, in terms of the debate between kinematic and dynamic theories of the origins of spacetime symmetries. (shrink)

We propose a formal framework for interpretable machine learning. Combining elements from statistical learning, causal interventionism, and decision theory, we design an idealised explanation game in which players collaborate to find the best explanation for a given algorithmic prediction. Through an iterative procedure of questions and answers, the players establish a three-dimensional Pareto frontier that describes the optimal trade-offs between explanatory accuracy, simplicity, and relevance. Multiple rounds are played at different levels of abstraction, allowing the players to explore overlapping causal (...) patterns of variable granularity and scope. We characterise the conditions under which such a game is almost surely guaranteed to converge on a optimal explanation surface in polynomial time, and highlight obstacles that will tend to prevent the players from advancing beyond certain explanatory thresholds. The game serves a descriptive and a normative function, establishing a conceptual space in which to analyse and compare existing proposals, as well as design new and improved solutions. (shrink)

Machine learning algorithms may radically improve our ability to diagnose and treat disease. For moral, legal, and scientific reasons, it is essential that doctors and patients be able to understand and explain the predictions of these models. Scalable, customisable, and ethical solutions can be achieved by working together with relevant stakeholders, including patients, data scientists, and policy makers.

In this paper, we develop an account of the justification thinkers have for employing certain basic belief-forming methods. The guiding idea is inspired by Reichenbach's work on induction. There are certain projects in which thinkers are rationally required to engage. Thinkers are epistemically justified in employing any belief-forming method such that "if it doesn't work, nothing will" for successfully engaging in such a project. We present a detailed account based on this intuitive thought and address objections to it. We conclude (...) by commenting on the implications that our account may have for other important epistemological issues and debates. (shrink)

Artificial intelligence has historically been conceptualized in anthropomorphic terms. Some algorithms deploy biomimetic designs in a deliberate attempt to effect a sort of digital isomorphism of the human brain. Others leverage more general learning strategies that happen to coincide with popular theories of cognitive science and social epistemology. In this paper, I challenge the anthropomorphic credentials of the neural network algorithm, whose similarities to human cognition I argue are vastly overstated and narrowly construed. I submit that three alternative supervised learning (...) methods—namely lasso penalties, bagging, and boosting—offer subtler, more interesting analogies to human reasoning as both an individual and a social phenomenon. Despite the temptation to fall back on anthropomorphic tropes when discussing AI, however, I conclude that such rhetoric is at best misleading and at worst downright dangerous. The impulse to humanize algorithms is an obstacle to properly conceptualizing the ethical challenges posed by emerging technologies. (shrink)

We often get evidence that bears on the reliability of some of our own first-order reasoning. The rational response to such “higher-order” evidence would seem to depend on a rational assessment of how reliable we can expect that reasoning to be, in light of the higher-order evidence. “Independence” principles are intended to constrain this reliability-assessment, so as to prevent question-begging reliance on the very reasoning being assessed. However, extant formulations of Independence principles tend to be vague or ambiguous, and coming (...) up with a tolerably precise formulation turns out to be tricky. This paper aims to make some progress toward developing a more precise Independence principle, and also to highlight some of the difficulties the project entails. The hope is to take a step toward understanding how rational belief is informed by evidence bearing on agents’ cognitive reliability. (shrink)

Computation is a process of making explicit, information that was implicit. In computing 5 as the solution to ∛125, for example, we move from a description that is not explicitly about 5 to one that is. We are drawing out numerical consequences to the description ∛125. We are extracting information implicit in the problem statement. Can we precisely state the difference between information thati s implicit in a state, structure or process and information that is explicit?