Whether y obtains under the counterfactual supposition that x is thought to depend on whether y obtains in the most similar world(s) in which x obtains. Graphical causal models have proved useful in developing a principled notion of similarity between worlds, but this notion is limited insofar as it does not apply to counterfactual suppositions about causal structure. Here, we explore the possibility of filling this lacuna by introducing a notion of similarity between causal graphs. Since there are multiple principled senses in which graphs can be similar, we introduce multiple similarity metrics and multiple ways to prioritize these metrics.

Philosophers of biology generally agree that causal specificity tracks biological importance: more specific causes are more important. I argue that this correlation does not hold in much research aimed at biomedical intervention. Applying James Woodward’s analysis of causal specificity to the development and design of HIV treatments, I show that drugs that are less causally specific produce better therapeutic outcomes and are more highly valued. Thus, I conclude that the importance of biological causes does not track their specificity but instead varies depending on the biological practice we consider and the goals of that practice.

This article provides a detailed analysis and explores the prospects of the arguments for higher-level causal autonomy available for the proponents of the mechanistic framework. Three different arguments (context based, organization based, and constraint based) are distinguished. After clarifying previously raised worries with regard to the first two arguments, the article focuses on the newest version of the third argument that has recently been revived by William Bechtel. By using Bechtel’s own case study, it is shown that not even reference to constraints can establish the causal autonomy of higher mechanistic levels.

There is a fairly long-standing distinction between what are called the ideographic as opposed to nomothetic sciences. The nomothetic sciences, such as physics, offer explanations in terms of the laws and regular operations of nature. The ideographic sciences, such as natural history (or, more controversially, evolutionary biology), cast explanations in terms of narratives. This article offers an account of what is involved in offering an explanatory narrative in the historical (ideographic) sciences. I argue that narrative explanations involve two chief components: a possibility space and an explanatory causal mechanism.

I analyze the exploratory function of two main modeling practices: targetless fictional models and hypothetical perspectival models. In both cases, I argue, modelers invite us to imagine or conceive something about the target system, which is known to be either nonexistent (fictional models) or just hypothetical (in perspectival models). I clarify the kind of imagining or conceiving involved in each modeling practice, and I show how each—in its own right—delivers important modal knowledge. I illustrate these two kinds of exploratory models with Maxwell’s ether model and supersymmetric particle models at the Large Hadron Collider.

This essay considers a mixed-effects modeling practice and its implications for the philosophical debate surrounding reductive explanation. Mixed-effects modeling is a species of the multilevel modeling practice, where a single model incorporates simultaneously two (or even more) levels of explanatory variables to explain a phenomenon of interest. I argue that this practice makes the position of explanatory reductionism held by many philosophers untenable because it violates two central tenets of explanatory reductionism: single-level preference and lower-level obsession.

Current debates about the replication crisis in psychology take it for granted that direct replication is valuable, largely focusing on its role in uncovering questionable statistical practices. This article takes a broader look at the notion of replication in psychological experiments. It is argued that all experimentation/replication involves individuation judgments and that research in experimental psychology frequently turns on probing the adequacy of such judgments. In this vein, I highlight the ubiquity of conceptual and material questions in research, arguing that replication has its place but is not as central to psychological research as it is sometimes taken to be.

The Pessimistic Induction over the history of science argues that because most past theories judged empirically successful in their time are not even approximately true, most present ones probably are not approximately true either. But why did past scientists accept those incorrect theories? Kyle Stanford’s Problem of Unconceived Alternatives is one answer to that question: scientists are bad at exhausting the space of plausible hypotheses to explain the available evidence. Here, I offer another answer: the Problem of Misleading Evidence. I argue that this proposal enjoys some advantages over Stanford’s, although both are probably needed to explain the historical record.

Historical inductions, that is, the pessimistic metainduction and the problem of unconceived alternatives, are critically analyzed via John D. Norton’s material theory of induction and subsequently rejected as noncogent arguments. It is suggested that the material theory is amenable to a local version of the pessimistic metainduction, for example, in the context of some medical studies.

Recent philosophy of measurement has emphasized the existence of both diachronic and synchronic “loops,” or feedback processes, in the epistemic achievements of measurement. A widespread response has been to conclude that measurement outcomes do not convey interest-independent facts about the world and that only a coherentist epistemology of measurement is viable. In contrast, I argue that a form of measurement realism is consistent with these results. The insight is that antecedent structure in measuring spaces constrains our empirical procedures such that successful measurement conveys a limited but veridical knowledge of “fixed points,” or modally stable, interest-independent features of the world.

This article considers the following question: What is the relationship between supervenience and reduction? I investigate this formally: first, by introducing a recent argument by Christian List to the effect that one can have supervenience without reduction; then, by considering how the notion of Nagelian reduction can be related to the formal apparatus of definability and translation theory; then, by showing how, in the context of propositional theories, topological constraints on supervenience serve to enforce reducibility; and, finally, by showing how constraints derived from the theory of ultraproducts can enforce reducibility in the context of first-order theories.

In recent years, several authors have called to ground descriptive and normative decision theory on neuropsychological measures of utility. In this article, I combine insights from the best available neuropsychological findings, leading philosophical conceptions of welfare, and contemporary decision theory to rebut these prominent calls. I argue for two claims of general interest to philosophers, choice modelers and policy makers. First, severe conceptual, epistemic, and evidential problems plague ongoing attempts to develop accurate and reliable neuropsychological measures of utility. Second, even if these problems are solved, neuropsychological measures of utility lack the potential to directly inform welfare analyses and policy evaluations.

The problem of theory choice and model selection is hard but still important when useful truths are underdetermined, perhaps not by all kinds of data but by the kinds of data we can have access to ethically or practicably—even if we have an infinity of such data. This article addresses a crucial instance of that problem: the problem of inferring causal structures from nonexperimental, nontemporal data without assuming the so-called causal Faithfulness condition or the like. A new account of epistemic evaluation is developed to solve that problem and justify a standard practice of causal inference in data science.

This article poses a challenge to Schurz’s proposed metainductive justification of induction. It is argued that Schurz’s argument requires a notion of optimality that can deal with an expanding pool of prediction strategies.

This article aims to complement causal model approaches to causal explanation by James Woodward, Joseph Halpern and Judea Pearl, and Michael Strevens. It centers on a strengthened Ramsey Test of conditionals: $\alpha \gg \gamma$ if and only if, after suspending judgment about α and γ, an agent can infer γ from the supposition of α (in the context of further beliefs in the background). Holger Andreas and Mario Günther used this conditional as the starting point of an analysis of ‘because’ in natural language. I refine this analysis so as to yield a fully fledged account of (deterministic) causal explanation.

I draw on speech act theory to understand the speech acts appropriate to the multiple aims of scientific practice and the role of nonepistemic values in evaluating speech acts made relative to those aims. First, I consider work that distinguishes explanatory speech acts from descriptive speech acts within scientific practice. I then show how speech act theory provides a framework to make sense of explaining’s and describing’s distinct felicity conditions. Finally, I argue that if explaining aims to convey understanding to particular audiences rather than describe literally across contexts, then evaluating explanatory speech acts directed to nonscientists involves nonepistemic criteria.

We offer a new account of the role of values in theory choice that captures a temporal dimension to the values themselves. We argue that nonepistemic values sometimes serve as “inquiry tickets,” justifying scientists’ pursuit of certain questions in the short run, while the answers to those questions mitigate transient underdetermination in the long run. Our account of inquiry tickets shows that the role of nonepistemic values need not be restricted to belief or acceptance in order to be relevant to hypothesis choice: the relevance of nonepistemic values to a particular cognitive attitude with respect to h varies over time.

I argue that collaborators do not need to reach broad agreement over the justification of a consensus claim. This is because maintaining a diversity of justifiers within a scientific collaboration has important epistemic value. I develop a view of collective justification that depends on the diversity of epistemic perspectives present in a group. I argue that a group can be collectively justified in asserting that P as long as the disagreement among collaborators over the reasons for P is itself justified. In conclusion, I make a case for multimethod collaborative research and work through an example in the social sciences.

Women are cited less frequently than men in a variety of scientific fields. Drawing theoretical resources from Fricker and Hookway, I argue that these gendered citation practices constitute a form of participatory epistemic injustice insofar as they prevent female scientists from fully engaging in the epistemic practices of science. Furthermore, Longino’s notion of “uptake” gives us a way of understanding gendered citation practices as an epistemic harm accrued not simply by individuals but by scientific communities as a whole. Finally, I discuss the cumulative harms of this kind of participatory epistemic injustice for individuals and for marginalized groups.