My aim here is (1) to argue that the usual argument for thinking that dysfunction has no place in a dispositionalist approach to functions is deeply flawed and (2) to develop a positive account of the explanatory role dysfunction attributions play in dispositionalist-style functional analysis. I also argue that while my account undermines one common motivation for preferring an etiological over a dispositionalist approach, perhaps more interestingly, it also blurs the boundary between the two and opens a path to unifying them.

This note scouts a broad but underexplored class of explanations found in contemporary computer science. These explanations, which I call limitative explanations, explain why certain problems cannot be solved computationally. Limitative explanations are philosophically rich, but have not received the attention they deserve. The primary goals of this note are to isolate limitative explanations and provide a preliminary account of what makes them explanatory. On the account I favor, limitative explanations are a kind of non-causal mathematical explanation which depend on highly idealized models of computation.

Like Austin’s “performatives,” some models are used not merely to represent but also to change their targets in various ways. This article argues that Austin’s analysis can inform model evaluation: If models are evaluated with respect to their adequacy-for-purpose, and if performativity can in some cases be regarded as a model purpose (a proposition that is defended, using mechanism design as an example), it follows that these models can be evaluated with respect to their “felicity,” that is whether their use has achieved this purpose. Finally, I respond to epistemic and ethical concerns that might block this conclusion.

Forber and Smead (2014) analyze how increasing the fitness benefits associated with prosocial behavior can increase the fitness of spiteful individuals relative to their prosocial counterparts, so that selection favors spite over prosociality. This poses a problem for the evolution of prosocial behavior: As the benefits of prosocial behavior increase, it becomes more likely that spite, not prosocial behavior, will evolve in any given population. In this article, I develop two game-theoretic models that, taken together, illustrate how synergistic costs and benefits may provide partial solutions to Forber and Smead’s paradox.

While philosophers have focused on epistemological and ethical challenges of using artificial intelligence (AI) in science, scientists have focused largely on opportunities. I argue that this disconnect between philosophical pessimism and scientific optimism is driven by failures to critically examine the practice of AI-infused science. To appreciate the epistemic justification for AI-powered breakthroughs, philosophers must analyze the role of AI as part of a wider process of discovery. I demonstrate the importance of this with two cases from the scientific literature, and show that epistemic opacity need not diminish AI’s capacity to lead scientists to significant and justifiable breakthroughs.

The philosophy of historical sciences investigates their distinct objects of study, epistemic challenges, and methodological solutions. Rethinking astronomy in this light offers a contribution. First, the methodology of historical sciences adds to a more adequate description of how astronomers study and utilize token events. Second, astronomy faces a typical difficulty in identifying traces of some past events and has developed a delicate solution. This enriches the idea of trace and suggests a methodology that relies on iterations between data-driven approaches and theory-driven approaches, together with the cross-validation between multiple relevant historical events or datasets.

Scientism has recently experienced a resurgence of interest in philosophy. One version of scientism often defended is ontological scientism—the view that any kind or property not mentioned in the theories of science has only a subordinate, secondary kind of reality. It is worth noting that a dominant tradition in the history of philosophy of science—classical American pragmatism—undertook decades of critical engagement with contemporaneous scientistic beliefs, many of which resemble those being debated at the present time. This antiscientistic philosophy has multiple points of relevance for contemporary debates and defenses of ontological scientism.

There remains no consensus among social scientists as to how to measure and understand forms of information deprivation such as misinformation. Machine learning and statistical analyses of information deprivation typically contain problematic operationalizations which are too often biased towards epistemic elites’ conceptions that can undermine their empirical adequacy. A mature science of information deprivation should include considerable citizen involvement that is sensitive to the value-ladenness of information quality, and doing so may improve the predictive and explanatory power of extant models.

I present new counterexamples to the asymmetry of grounding: we have prima facie reason to think that some conditional probabilities partially ground their inverse conditional probabilities, and vice versa. These new counterexamples may require that we reject the asymmetry of grounding or, alternatively, that we reject one or more of the assumptions (regarding, e.g., the correct interpretation of probability) that enable the counterexamples. Either way, by reflecting on these purported counterexamples to grounding asymmetry, we learn something important, either about the formal properties of grounding or about the nature of probability (or both).

Various errors can affect scientific code and detecting them is a central concern within computational science. Could formal verification methods, which are now available tools, be widely adopted to guarantee the general reliability of scientific code? After discussing their benefits and drawbacks, we claim that, absent significant changes as regards features like their user-friendliness and versatility, these methods are unlikely to be adopted throughout computational science, beyond certain specific contexts for which they are well-suited. This issue exemplifies the epistemological heterogeneity of computational science: Profoundly different practices can be appropriate to meet the reliability challenge that rises for scientific code.

This is an essay review of Jill North’s book Physics, Structure, and Reality. It focuses on two of the main topics of the book. The first is North’s idea that we can use coordinates as a window into the structure that a theory posits; the second is North’s argument for the inequivalence of Lagrangian and Newtonian mechanics.

Transparency has been proposed as a way of handling value-ladenness in machine learning (ML). This article highlights limits to this strategy. I distinguish three kinds of transparency: epistemic transparency, retrospective value transparency, and prospective value transparency. This corresponds to different approaches to transparency in ML, including so-called explainable artificial intelligence and governance based on disclosing information about the design process. I discuss three sources of value-ladenness in ML—problem formulation, inductive risk, and specification gaming—and argue that retrospective value transparency is only well-suited for dealing with the first, while the third raises serious challenges even for prospective value transparency.

The framework of Solomonoff prediction assigns prior probability to hypotheses inversely proportional to their Kolmogorov complexity. There are two well-known problems. First, the Solomonoff prior is relative to a choice of universal Turing machine. Second, the Solomonoff prior is not computable. However, there are responses to both problems. Different Solomonoff priors converge with more and more data. Further, there are computable approximations to the Solomonoff prior. I argue that there is a tension between these two responses. This is because computable approximations to Solomonoff prediction do not always converge.

First, I show that the semantic thesis of scientific realism may be relaxed significantly—to allow that some scientific discourse is not truth-valued—without making any concessions concerning the epistemic or methodological theses that lie at realism’s core. Second, I illustrate how relaxing the semantic thesis allows realists to avoid positing abstract entities and to fend off objections to the “no miracles” argument from positions such as cognitive instrumentalism. Third, I argue that the semantic thesis of scientific realism should be relaxed because it is possible for scientific statements to be partially true, and hence approximately true, without being false.

Definitions of animal welfare often invoke consciousness or sentience. Marian Stamp Dawkins has argued that to define animal welfare this way is a mistake. In Dawkins’s alternative view, an animal with good welfare is one that is healthy and “has what it wants.” The dispute highlights a source of strain on the concept of animal welfare: consciousness-involving definitions are better able to capture the normative significance of welfare, whereas consciousness-free definitions facilitate the validation of welfare indicators. I reflect on how the field should respond to this strain, ultimately recommending against splitting the concept and in favor of consciousness-involving definitions.

This paper explores a distinction among causal relationships that has yet to receive attention in the philosophical literature, namely, whether causal relationships are reversible or irreversible. We provide an analysis of this distinction and show how it has important implications for causal inference and modeling. This work also clarifies how various familiar puzzles involving preemption and over-determination play out differently depending on whether the causation involved is reversible.

Under what conditions does machine learning (ML) model opacity inhibit the possibility of explaining and understanding phenomena? In this article, I argue that nonepistemic values give shape to the ML opacity problem even if we keep researcher interests fixed. Treating ML models as an instance of doing model-based science to explain and understand phenomena reveals that there is (i) an external opacity problem, where the presence of inductive risk imposes higher standards on externally validating models, and (ii) an internal opacity problem, where greater inductive risk demands a higher level of transparency regarding the inferences the model makes.

In the quest to discover the neural bases of cognition, rigorous behavioral tools are equally as important as sophisticated tools for neural intervention. This paper evaluates several episodes in the development of a novel behavioral tool for rodent cognitive testing, the rodent touchscreen operant chamber. Using conceptual tools on offer in the philosophical literature on exploratory experimentation and control, I illuminate how optimization of this behavioral tool and an understanding of the causal knowledge it may be used to generate historically has involved dynamic interplay between community-driven exploratory practices operating in parallel with causal hypothesis-driven research.