I build on Adam Smith’s account of the impartial spectator in The Theory of Moral Sentiments in order to offer a modest ideal observer theory of moral judgment that is adequate in the following sense: the account specifies the hypothetical conditions that guarantee the authoritativeness of an agent’s (or agents’) responses in constituting the standard in question, and, if an actual agent or an actual community of agents are not under those conditions, their responses are not authoritative in setting this (...) standard. However, in the account that I provide, the hypothetical conditions can themselves be constructed from the psychology and interactions of actual human beings. In other words, facts about the morally appropriate and inappropriate are determined from hypothetical conditions that––while agents in a given society might have yet to attain them––can be constructed from those agents’ shared experiences. Thus, the account offers both an attainable standard of moral judgment and a standard that can transcend the biases of the society which gave rise to it. I also defend the account against three challenges: (a) ideal observer theories do not offer the right kind of motivation to act on the verdicts of the ideal observer; (b) ideal observer theories cannot explain why the idealization in question is well-motivated and not objectionably ad hoc; (c) the standard used in ideal observer theories cannot be defended upon further reflection, because we would need a non-arbitrary, second-order standard to govern our reflection on the first-order standard of moral judgment. (shrink)

: In this paper I attempt to defuse a set of epistemic worries commonly raised against ideal observer theories. The worries arise because of the omniscience often attributed to ideal observers – how can we, as finite humans, ever have access to the moral judgements or reactions of omniscient beings? I argue that many of the same concerns arise with respect to other moral theories (and that these concerns do not in fact reveal genuine flaws in any of these theories), (...) and further, that we can and often do have knowledge of the reactions of ideal observers (according to standard, prominent theories in the domain of epistemology). (shrink)

Antimicrobial resistance (AMR) is a global public health disaster driven largely by antibiotic use in human health care. Doctors considering whether to prescribe antibiotics face an ethical conflict between upholding individual patient health and advancing public health aims. Existing literature mainly examines whether patients awaiting consultations desire or expect to receive antibiotic prescriptions, but does not report views of the wider public regarding conditions under which doctors should prescribe antibiotics. It also does not explore the ethical significance of public views (...) or their sensitivity to awareness of AMR risks or the standpoint (self-interested or impartial) taken by participants. Methods: An online survey was conducted with a sample of the U.S. public (n = 158). Participants were asked to indicate what relative priority should be given to individual patients and society-at-large from various standpoints and in various contexts, including antibiotic prescription. Results: Of the participants, 50.3% thought that doctors should generally prioritize individual patients over society, whereas 32.0% prioritized society over individual patients. When asked in the context of AMR, 39.2% prioritized individuals whereas 45.5% prioritized society. Participants were significantly less willing to prioritize society over individuals when they themselves were the patient, both in general (p = .001) and in relation to AMR specifically (p = .006). Conclusions: Participants’ attitudes were more oriented to society and sensitive to collective responsibility when informed about the social costs of antibiotic use and when considered from a third-person rather than first-person perspective. That is, as participants came closer to taking the perspective of an informed and impartial “ideal observer,” their support for prioritizing society increased. Our findings suggest that, insofar as antibiotic policies and practices should be informed by attitudes that are impartial and well-informed, there is significant support for prioritizing society. (shrink)

The Golden Rule is regularly used in ordinary life, across many different cultures, to acquire new moral knowledge. At the same time, the Golden Rule is widely ignored both in ethics and metaethics because it seems to be an implausible normative theory. Most philosophers who have paid it any attention have thought that, at best, it is an initially tempting thought whose appeal should be explained by the ultimately correct normative theory. My aim in this paper is to attend to (...) an alternative possibility: the Golden Rule teaches us something about metaethics, in the form of moral epistemology, rather than normative ethics. I will argue that sentimentalism, the view that the emotions are an essential source of moral knowledge, provides a compelling explanation of the usefulness of the Golden Rule. Before giving the sentimentalist explanation, I explain why proposed alternative sources of moral knowledge provide less compelling explanations of its usefulness. (shrink)

:Preference-aggregation problems arise in various contexts. One such context, little explored by social choice theorists, is metaethical. ‘Ideal-advisor’ accounts, which have played a major role in metaethics, propose that moral facts are constituted by the idealized preferences of a community of advisors. Such accounts give rise to a preference-aggregation problem: namely, aggregating the advisors’ moral preferences. Do we have reason to believe that the advisors, albeit idealized, can still diverge in their rankings of a given set of alternatives? If so, (...) what are the moral facts when the advisors do diverge? These questions are investigated here using the tools of Arrovian social choice theory. (shrink)

The dominant practice of AI alignment assumes (1) that preferences are an adequate representation of human values, (2) that human rationality can be understood in terms of maximizing the satisfaction of preferences, and (3) that AI systems should be aligned with the preferences of one or more humans to ensure that they behave safely and in accordance with our values. Whether implicitly followed or explicitly endorsed, these commitments constitute what we term apreferentistapproach to AI alignment. In this paper, we characterize (...) and challenge the preferentist approach, describing conceptual and technical alternatives that are ripe for further research. We first survey the limits of rational choice theory as a descriptive model, explaining how preferences fail to capture the thick semantic content of human values, and how utility representations neglect the possible incommensurability of those values. We then critique the normativity of expected utility theory (EUT) for humans and AI, drawing upon arguments showing how rational agents need not comply with EUT, while highlighting how EUT is silent on which preferences are normatively acceptable. Finally, we argue that these limitations motivate a reframing of the targets of AI alignment: Instead of alignment with the preferences of a human user, developer, or humanity-writ-large, AI systems should be aligned with normative standards appropriate to their social roles, such as the role of a general-purpose assistant. Furthermore, these standards should be negotiated and agreed upon by all relevant stakeholders. On this alternative conception of alignment, a multiplicity of AI systems will be able to serve diverse ends, aligned with normative standards that promote mutual benefit and limit harm despite our plural and divergent values. (shrink)

Expressivism, as applied to a certain class of statements, evaluative ones, for instance, is constituted by two doctrines, only the first of which will concern me in this paper. Evaluative statements, according to this doctrine, aren’t propositional (susceptible of truth or falsity). In this paper, I will argue that one of the vexing problems (that I label the “logical problem”) this doctrine engenders for the expressivist is equally pressing for some cognitivists (who think evaluative statements do have a truth-value). I (...) will present the difficulty and argue that some constructivists, who are cognitivists, cannot contend with it at all, and others must resort to more complex ways than the one available to other cognitivists. (shrink)

In this paper I show that one version of motivational internalism follows from the so-called ideal observer theory. Let us call the version of the ideal observer theory used in this essay (IOT). According to (IOT), it is necessarily the case that it ought to be that A if and only if every ideal observer wants it to be the case that A. We shall call the version of motivational internalism that follows from (IOT) (moral) conditional belief motivational internalism (CBMI). (...) According to (CBMI), it is necessarily the case that, for every x: if x were an ideal observer, it would be the case that x believes that it ought to be that A only if x wants it to be the case that A. Given that it is necessarily true that no ideal observer has any false beliefs, (IOT) entails (CBMI). Or, so I shall argue. (shrink)

I show that an overlooked feature of our moral life—moral status—provides a route to vindicating naturalist moral realism in much the same way that the Humean theory of motivation and judgment internalism are used to undermine it. Moral status presents two explanatory burdens for metaethical views. First, a given view must provide an ecumenical explanation of moral status, which does not depend on the truth of its metaethical claims (say, that there are mind-independent facts about moral status). Second, its explanation (...) must be consistent with persistent normative ethical disagreement about what constitutes moral status. I conclude that naturalist moral realism succeeds, while quasi-realism fails because it cannot meet the latter requirement. This argument has three results: we have a new route for metaethical vindication more generally and for naturalist moral realism in particular; quasi-realism's plausibility is undermined by an inability to explain disagreement, but not for the familiar reasons. (shrink)

Peter van Inwagen defines a successful argument in philosophy as one that can be used to convert an audience of ideal agnostics in an ideal debate. Sarah McGrath and Thomas Kelly recently argue that van Inwagen’s definition cannot be correct since the idea of ideal agnostics is incoherent with regard to an absolute paradigm of a successful philosophical argument. This paper defends van Inwagen’s definition against McGrath and Kelly’s objection.

Self-consciousness in non-human animals is a complex phenomenon which raises both conceptual and methodological problems. First, what do we mean by the concept of ‘self-consciousness’? Secondly, what is the best experimental approach to self-consciousness? This paper gives a short overview of the concept of self-consciousness in section 1. We can understand the concept of self-consciousness as capturing the ability of subjects to consciously think about themselves as themselves. If this is accurate, then it is prudent to look at a broad (...) bundle of abilities which are related to this ability. The paper then focuses on advantages and problems of the multipronged approach, a kind of empirical approach which uses several different test criteria to determine whether a subject is self-conscious or not. Section 2 briefly contrasts this kind of approach with other possibilities. The main disparity is to singlepronged approaches, which have been discussed and dismissed in greater detail elsewhere. Section 3 discusses the advantages of a multipronged approach, whereas section 4 is dedicated to the discussion of some of its problems. (shrink)