PHILOSOPHICAL STUDIES最新文献

A standard form of skeptical scenario, in the tradition of Descartes’ evil demon, raises the prospect that our sensory experiences are deceptive. A less familiar and importantly different kind of skeptical scenario raises the prospect that our beliefs have been debased (Schaffer, 2010). This paper provides a new and improved way of resisting this latter kind of debasing skepticism. Along the way, I explore how the debasing demon scenario connects with some potentially controversial epistemological principles and clear up various neglected or misunderstood points concerning debasing skepticism.

The field of AI safety considers whether and how AI development can be safe and beneficial for humans and other animals, and the field of AI welfare considers whether and how AI development can be safe and beneficial for AI systems. There is a prima facie tension between these projects, since some measures in AI safety, if deployed against humans and other animals, would raise questions about the ethics of constraint, deception, surveillance, alteration, suffering, death, disenfranchisement, and more. Is there in fact a tension between these projects? We argue that, considering all relevant factors, there is indeed a moderately strong tension—and it deserves more examination. In particular, we should devise interventions that can promote both safety and welfare where possible, and prepare frameworks for navigating any remaining tensions thoughtfully.

According to qualitativism, thisness is not a fundamental feature of reality; facts about particular things are metaphysically second-rate. In this paper, I advance an argument for qualitativism from ideological parsimony. Supposing that reality fundamentally contains an array of propertied things, non-qualitativists employ a distinct name (or constant) for each fundamental thing. I argue that these names encode a type of worldly structure (thisness structure) that offends against parsimony and that qualitativists can eliminate without incurring a comparable parsimony-offense.

This paper concerns the proxy problem: often machine learning programs utilize seemingly innocuous features as proxies for socially-sensitive attributes, posing various challenges for the creation of ethical algorithms. I argue that to address this problem, we must first settle a prior question of what it means for an algorithm that only has access to seemingly neutral features to be using those features as “proxies” for, and so to be making decisions on the basis of, protected-class features. Borrowing resources from philosophy of mind and language, I argue that the answer depends on whether discrimination against those protected classes explains the algorithm’s selection of individuals. This approach rules out standard theories of proxy discrimination in law and computer science that rely on overly intellectual views of agent intentions or on overly deflationary views that reduce proxy use to statistical correlation. Instead, my theory highlights two distinct ways an algorithm can reason using proxies: either the proxies themselves are meaningfully about the protected classes, highlighting a new kind of intentional content for philosophical theories in mind and language; or the algorithm explicitly represents the protected-class features themselves, and proxy discrimination becomes regular, old, run-of-the-mill discrimination.

Many varieties of understanding subsist in a thinker’s having the right kind of mental connection to a certain body of fact (or putative fact), a connection often called “grasp”. The use of a single term suggests a single connection that does the job in every kind of understanding. Then again, “grasp” might be an umbrella term covering a diverse plurality of understanding-granting mind-world relations. This paper argues for the former, unified view of grasp in two ways. First, it advances a broad, ability-based construal of grasp, along with a test for lack of grasp, that suggests that a certain specific connection plays an essential role in many varieties of understanding. Second, the paper considers a number of challenges to the thesis of unity that arise in a range of different kinds of understanding (scientific, moral, objectual, humanistic), and seeks to disarm them.

Sosa holds one may rationally want to understand how the specific features of a particular artwork ground its aesthetic value, and that this understanding cannot be gained at second-hand. Such understanding requires one to have insight into the link between grounding features and that value, and this can only be gained through first-hand engagement with the artwork. I distinguish two senses of second-hand. In one sense Sosa is correct that one cannot understand why P at second-hand: one must have insight oneself into the link between explanans and explanandum, and this is an exercise of one’s own mental power. But this allows another sense in which understanding may be gained at second-hand, via a description of an artwork provided to one through testimony. I argue that an expert can attain understanding of how key features of an artwork ground its aesthetic value from a suitably rich description of it. Sosa has misidentified the epistemic good that can only be obtained from first-hand engagement with an artwork. This is not understanding of what makes it good, but an enjoyable episode of aesthetic appreciation of it.

In this paper I outline a challenge for experiential passage realism, the view that we veridically perceptually experience the robust passage of time. The challenge lies in accommodating recent empirical data, according to which ~ 35% of people do not report that it seems as though time robustly passes, and ~ 65% report that it does. I argue that offering a plausible explanation for this data is especially challenging for the experiential passage realist. This gives us reason to reject experiential passage realism either by adopting a form of passage realism according to which although time robustly passes, we do not experience its passing, or by adopting deflationism, the view that time does not robustly pass, and we have veridical experiences of a passageless world.

In this paper, I present and defend a novel way for non-naturalists to account for the sui generis status of normative facts, which is consistent with the claim that contingent normative facts obtain in virtue of non-normative facts. According to what I call unsupplemented partial ground approach, non-derivative normative facts have non-normative partial grounds, but are not fully grounded in any collection of facts. This view entails that an explanatory gap separates the normative from the non-normative domain. I argue that this account provides non-naturalists with a metaphysically coherent response to the challenge of accounting for explanatory dependence relations between two domains while positing metaphysical discontinuity (explanatory challenge), and avoids serious objections that alternative non-naturalist accounts face. Moreover, I show that the unsupplemented partial ground approach is an attractive option for the popular Reasons-First approach, which is often, but I argue prematurely, considered a particularly promising account for non-naturalists.

My cat Percy and I both engage in inquiry. For example, we both might wonder where the food is, and look around systematically in an effort to find the food. Indeed, we might even recruit others to help us search for the food, and so engage in collaborative inquiry concerning the location of the food. But such inquiry, even when collaborative, does not amount to research. Why not? What distinguishes research from the kinds of inquiry in which Percy and I can both engage? You might think that research involves the exercise of distinctive skills or capacities, or that it involves focus on a special range of topics. But how can we specify the relevant skills, or the relevant range of topics? This paper articulates and defends an account of research that answers these questions. According to the present account, research is a form of inquiry that is guided by the judgment that the answer to this very inquiry matters to our theoretical understanding.

Discussions of statistical criteria for fairness commonly convey the normative significance of calibration within groups by invoking what risk scores “mean.” On the Same Meaning picture, group-calibrated scores “mean the same thing” (on average) across individuals from different groups and accordingly, guard against disparate treatment of individuals based on group membership. My contention is that calibration guarantees no such thing. Since concrete actual people belong to many groups, calibration cannot ensure the kind of consistent score interpretation that the Same Meaning picture implies matters for fairness, unless calibration is met within every group to which an individual belongs. Alas only perfect predictors may meet this bar. The Same Meaning picture thus commits a reference class fallacy by inferring from calibration within some group to the “meaning” or evidential value of an individual’s score, because they are a member of that group. The reference class answer it presumes does not only lack justification; it is very likely wrong. I then show that the reference class problem besets not just calibration but other group statistical criteria that claim a close connection to fairness. Reflecting on the origins of this oversight opens a wider lens onto the predominant methodology in algorithmic fairness based on stylized cases.