AAP 2026 CONFERENCE

The idea that scientific realism and anti-realism are, or are closely related to, epistemic stances (clusters of values, attitudes, goals and commitments), has grown in popularity following the work of van Fraassen, Chakravartty and others. Yet the dream of discovering a decisive stance-independent argument for scientific realism as a thesis – an argument that is, or should be, rationally compelling for anyone, whatever their prior philosophical commitments, who is willing to consider the arguments and evidence in an impartial way – dies hard, as recent attempts in this direction by Eronen, Strevens and others have shown. In this paper I argue these attempts all fail. All arguments for scientific realism as a thesis presuppose the realist stance. I also suggest that, paradoxically, arguments for versions of scientific anti-realism may also presuppose the realist stance, inasmuch as they assume that the specific realist thesis in question is on the table as a genuine option, even if it is ultimately rejected. For those who reject the realist stance, such realist theses are not even live options, so arguments against them are beside the point.

This paper argues that the strongest form of causal necessity, which David Hume advanced as unjustified, does in fact exist. It is found, paradoxically, in the relation between causal omissions and their effects. I also propose a reformulation of the ‘c causes e iff…’ analysis into a new analysis that best accounts for causal necessity. In brief, when any causal factor, intrinsic to a causal system, is omitted, a change necessarily occurs.

I propose a contextualized conception of proportionality, which requires bringing the concrete context of answering/raising a particular causal inference question into the picture when assessing proportionality. So, the new formula is this: a cause-variable C is proportional to an effect-variable E relative to a given context T. This conception is bolstered by a brief exposition of recent scientific practice in causal feature learning. Moreover, it gets further support by showing how it readily and elegantly resolves a threat posed by Franklin-Hall (2016).

Richard Jeffrey (1956) famously articulated an internal critique to the argument from inductive risk (AIR) and offered an alternative picture of scientific advice, which we call “the Bayesian picture of scientific advice” (BPSA) with two essential commitments: scientists should only communicate their subjective probabilities (vs. outright beliefs) in hypotheses and doing so upholds a political division of labour (i.e., scientists bring the epistemic input; policy-makers bring the evaluative judgments).

We argue that communicating credences doesn’t deliver the division of labour—the idea that such a prize is secured in the Bayesian picture is an artifact of the idealizations behind the debate around the AIR. Basically: scientists’ role in policy advice goes well beyond reporting credences (or outright beliefs for that matter) for a hypotheses previously specified by policy makers. Scientists are necessarily involved in the framing of policy (decision) problems, i.e., in the curation of the policy actions, the states-of-nature, and outcomes that are worth considering. (They are even needed to come up with utility numbers!) This is easy to see when looking at well-studied cases of scientific advice such as large-scale environmental assessments—we focus on the IPCC reports in the talk—but the point generalizes.

The distribution question about animal consciousness concerns which animal species are conscious. Philosophers and scientists alike have been hung up on the idea that we must first define consciousness if this question is to be answered. Examples in the history of science suggest, however, that this may put the conceptual cart before the empirical horse. Inspired by Hasok Chang's work on inventing temperature, I will argue that answering the distribution question requires the invention of better measures before a good enough theory to support a consensus definition can be developed. To this end, I will extend the ""signature approach"" to comparative animal cognition, developed by Alex Taylor, Amalia Bastos, Rachael Brown and myself. The signature approach applied to comparative studies of animal consciousness shares some affinities with ""marker"" approaches to animal consciousness offered recently by philosophers including Jonathan Birch, Kristin Andrews, and Albert Newen. But whereas marker approaches seek to build dimensional profiles of different species according to the degree to which they display various capacities thought to be related to consciousness, the signature approach is better suited to developing reliable measures that support detailed comparisons of the processes underlying these capacities, and ultimately better theories.

Can empirical psychology—qua scientific discipline—incorporate objective moral truth? That is, are appeals to objective moral truth scientifically acceptable, viable, and legitimate—i.e., are they scientifically adequate? I consider this question through an examination of a particular case, namely, whether objective moral error is a viable scientific kind. I first examine objective moral errors per comprehensive moral theories proffered as the objective moral truth (“Objective CMT”). Importantly, I show that moral philosophical adequacy is not necessarily sufficient for scientific adequacy. I conclude that—currently—such objective moral errors are scientifically inadequate because, lacking a CMT well-established as the objective moral truth, condoning such errors yields the untenable result of different researchers using incompatible standards-of-moral-correctness that generate incommensurable versions of the kind. I reject two potential partners-in-guilt. Finally, I examine objective moral errors that contravene specific moral principles proffered as objective moral truths. I argue that because of the strength of metaethical skepticism within a scientific context, coupled with the availability of viable alternatives to objective moral truths—namely, standards-of-interest—all potential objective moral errors, as well as all appeals to objective moral truth, are scientifically inadequate and incompatible with empirical psychology (qua science).

In his small, often-neglected book Verteidigungsrede fur die Philosophie (1966), philosopher Josef Pieper offers an ingenious criticism of scientism, the thesis that all knowledge is from science. While proponents of scientism are few and far between for well-known problems such as its self-referential incoherence, science's descriptive and explanatory limits, extra-scientific assumptions, and failure to take seriously the qualitative features of reality (plus its materialist underpinnings, though not all are worried on this matter, no pun intended), nonetheless Pieper's criticism displaces both scientism and the move towards so-called "weak scientism" (the thesis that science is broader than natural sciences) and "epistemic opportunism" (the thesis that we should be optimists about scientific success). Following the demise of the Early Vienna Circle and logical positivism on what "empirical" and "observation" mean, Pieper's argument is that scientism intrinsically carries with it an implausible concept of experience, and so should be rejected. Here is the structure of the argument. Because science is an empirical (a posteriori) endeavor, it requires an intelligible notion of experience (or observation). Scientism must say that what exhausts experience is simply the natural world as experienced by our senses. However, this leaves out much of the world as we experience it, not only in its extra-scientific qualities (Schrödinger provides some entertaining examples) but also in the objects of experience themselves ("moral experience", as some call it, is a paradigmatic example). The theoretical advantages of this broad(er) account of experience (which is incompatible with scientism) are its alignment with moral-epistemic virtues like epistemic justice, as well as science's praise for dispassionate objectivity. Broader accounts of experience carry concerns as well, especially Platonic concerns about appearances versus reality, but all this shows - says Pieper - is that the philosophical act is indispensable.

Psychiatry faces profound challenges a quarter of the way into the twenty-first century. Most notably, there are various philosophical disputes pertaining to a) dimensional vs. categorical models of mental disorder, b) the status of psychiatric kinds, c) states vs. traits as the central constructs of psychiatry, and d) the language of “mental disorder” vs. “mental variation.” Furthermore, these ontological disputes are accompanied by methodological disputes regarding which causal factors are most relevant to formulating generalizations about particular mental disorders. Meanwhile, the DSM faces both a validity crisis and a comorbidity crisis. These problems have motivated some in the field to formulate new research traditions– such as RDoC and HiTOP– which offer distinct and novel approaches to the subject matter of psychiatry. The central claim that I advance here is that contemporary psychiatry approximates a pre-paradigmatic Kuhnian science. I take Kuhn’s theory of scientific practice and change as an idealized model– one which abstracts away from the details of particular episodes in the histories of particular sciences, but which nevertheless presents an “ideal type” for how scientific progress often occurs. Two alternative explanations that I will address here are 1) the Human Science Explanation and 2) the Medical Science Explanation.

Some proponents of multiple realization attempt to eat their realized cake while having their reduction, too. The constrained identity account of multiple realization offered by Polger and Shapiro (2016) is such an attempt, and in this paper I argue that it, and attempts like it, will lead to a complex dilemma for proponents of hybrid identity/realization accounts. Either they adhere to Kim’s (1992) principle of the causal individuation of kinds, in which case they must deal with Kim’s own reductionist dilemma for multiple realization, or they follow Polger and Shapiro (2016) in weakening Kim’s principle to the causally-relevant individuation of kinds. However, this leads to a relativisation of kinds to human interests, and thus to a pragmatist approach to science. Since multiple realizationists tend to be realists (after all, they think one set of entities at a basic level of reality makes another set of entities real), this is a problem for them. After diagnosing the problem they face, I offer them only unpalatable solutions: accept one or other fork of the dilemma and agree with Kim’s extreme reductivism or accept a pragmatism which admits only a watery realism.

I have argued that uncertainty assessment of climate model predictions should typically be of the extent to which they are epistemically possible and that, in some cases where they are epistemically possible, the possibilities should further be ranked as to how remote they are. I have also argued that, in the climate science context, an epistemic possibility should be taken to be a possibility that is not recognised to be excluded by what is known and is compatible with knowledge that approximates the basic way things are in the domain the possibility is about. In the present paper, I explain my position on assessing uncertainty in climate science and consider and respond to two challenges to its application, specifically, those of how to operationalise my notion of epistemic possibility and how to classify possibilities that fall short of being epistemically possible. I illustrate my view and responses in the case of the assessment of the possibility of marine ice-cliff instability induced sea-level rise.

In this paper I provide a novel argument against the Value-free Ideal (VFI) and explore some of its implications. I begin by arguing that no existing critique of the VFI targets the relations of inductive support between evidence and hypotheses (relations of confirmation). In fact, many critics of the VFI, like Heather Douglas, explicitly state that relations of confirmation remain value-free (Douglas, 2000, p. 656). However, I argue that confirmation is value-laden. After briefly surveying different inductive logics, I claim that the best prospects for a value-free account of confirmation rely on the probability calculus. For these accounts to be value-free, two conditions must be met: (1) probabilities must themselves be value-free, and (2) the choice of confirmation function (a function of probabilities) must also be value-free. Condition (1) fails because all interpretations of probability face the reference class problem (Hájek, 2007), and choosing a reference class requires non-epistemic value judgements. Condition (2) also fails, since the choice of confirmation function is underdetermined by evidence, theory, and epistemic values, requiring further non-epistemic value judgements. Hence, confirmation is value-laden. I conclude by exploring the implications of this conclusion for contemporary defences of the VFI (e.g. Menon and Stegenga, 2023).

REFERENCES

Douglas, H. (2000). Inductive Risk and Values in Science. Philosophy of Science, 67(4):559–579.
Hájek, A. (2007). The reference class problem is your problem too. Synthese, 156(3):563–585.
Menon, T. and Stegenga, J. (2023). Sisyphean science: why value freedom is worth pursuing. European Journal for Philosophy of Science, 13(4):48.