Randomized controlled trials (RCTs) are widely taken as the gold standard for establishing causal conclusions. Ideally conducted they ensure that the treatment ‘causes’ the outcome—in the experiment. But where else? This is the venerable question of external validity. I point out that the question comes in two importantly different forms: Is the specific causal conclusion warranted by the experiment true in a target situation? What will be the result of implementing the treatment there? This paper explains how the probabilistic theory (...) of causality implies that RCTs can establish causal conclusions and thereby provides an account of what exactly that causal conclusion is. Clarifying the exact form of the conclusion shows just what is necessary for it to hold in a new setting and also how much more is needed to see what the actual outcome would be there were the treatment implemented. (shrink)

Explanations are backed by many different relations: causation, grounding, and arguably others too. But why are these different relations capable of backing explanations? In virtue of what are they explanatory? In this paper, I propose and defend a monistic account of explanation-backing relations. On my account, there is a single relation which backs all cases of explanation, and which explains why those other relations are explanation-backing.

It is argued that causal understanding originates in experiences of acting on objects. Such experiences have consistent features that can be used as clues to causal identification and judgment. These are singular clues, meaning that they can be detected in single instances. A catalog of 14 singular clues is proposed. The clues function as heuristics for generating causal judgments under uncertainty and are a pervasive source of bias in causal judgment. More sophisticated clues such as mechanism clues and repeated interventions (...) are derived from the 14. Research on the use of empirical information and conditional probabilities to identify causes has used scenarios in which several of the clues are present, and the use of empirical association information for causal judgment depends on the presence of singular clues. It is the singular clues and their origin that are basic to causal understanding, not multiple instance clues such as empirical association, contingency, and conditional probabilities. (shrink)

We propose a formal framework for interpretable machine learning. Combining elements from statistical learning, causal interventionism, and decision theory, we design an idealisedexplanation gamein which players collaborate to find the best explanation(s) for a given algorithmic prediction. Through an iterative procedure of questions and answers, the players establish a three-dimensional Pareto frontier that describes the optimal trade-offs between explanatory accuracy, simplicity, and relevance. Multiple rounds are played at different levels of abstraction, allowing the players to explore overlapping causal patterns of (...) variable granularity and scope. We characterise the conditions under which such a game is almost surely guaranteed to converge on a (conditionally) optimal explanation surface in polynomial time, and highlight obstacles that will tend to prevent the players from advancing beyond certain explanatory thresholds. The game serves a descriptive and a normative function, establishing a conceptual space in which to analyse and compare existing proposals, as well as design new and improved solutions. (shrink)

This paper offers a practical argument for metaphysical emergence. The main message is that the growing reliance on so-called irrational scientific methods provides evidence that objects of science are indecomposable and as such, are better described by metaphysical emergence as opposed to the prevalent reductionistic metaphysics. I show that a potential counterargument that science will eventually reduce everything to physics has little weight given where science is heading with its current methodological trend. I substantiate my arguments by detailed examples from (...) biological engineering, but the conclusions are extendable beyond that discipline. (shrink)

ABSTRACTMeasurements of environmental toxicity from long-term exposure to oil contamination have delivered inaccurate and contradictory results regarding the potential harms for humans and ecosyste...

Jonathan Schaffer (2004 ) proposes an ingenious amendment to David Lewis's semantics for counterfactuals. This amendment explicitly invokes the notion of causal independence, thus giving up Lewis's ambitions for a reductive counterfactual account of causation. But in return, it rescues Lewis's semantics from extant counterexamples. I present a new counterexample that defeats even Schaffer's amendment. Further, I argue that a better approach would be to follow the causal modelling literature and evaluate counterfactuals via an explicit postulated causal structure. This alternative (...) approach easily resolves the new counterexample, as well as all the previous ones. Up to now, its perceived drawback relative to Lewis's scheme has been its non-reductiveness. But since the same drawback applies equally to Schaffer's amended scheme, this becomes no longer a point of comparative disadvantage. (shrink)

This paper expounds that besides the well-known spatio-temporal problem there is a causal problem of entanglement: even when one neglects spatio-temporal constraints, the peculiar statistics of EPR/B experiment is inconsistent with usual principles of causal explanation as stated by the theory of causal Bayes nets. The conflict amounts to a dilemma that either there are uncaused correlations or there are caused independences . I argue that the central ideas of causal explanations can be saved if one accepts the latter horn (...) and explains the unfaithful independences by a stable fine-tuning of the causal parameters. (shrink)

This essay examines the relation between causation and causal explanation. It distinguishes two prominent roles that causes play within the sciences. On the one hand, causes may work as metaphysical posits. From this standpoint, mainstream in contemporary philosophy, causation provides the ‘raw material’ for explanation. On the other hand, causes may be conceived as explanatory postulates, theoretical hypotheses lacking any substantial ontological commitment. This unduly neglected distinction provides the conceptual resources to revisit longstanding philosophical issues, such as overdetermination and causal (...) pluralism. It also inspires a provocative reframing of Russell’s famous, if notoriously elusive, remarks on the nature of causation. (shrink)

This paper written as a dialogue between two interlocutors, Julie and a Student, deals with Understanding and its role in the social sciences. The fictional dialogue takes place in Hannover, Germany, and the interlocutors are exchanging arguments about Verstehen and how it should be conceptualized in the philosophy of the social sciences. A range of different approaches is discussed and a naturalistic strategy emerges as a defensible alternative.

Many investigations into the world, including philosophical ones, aim to discover causal knowledge, and many experimental methods have been developed to assist in causal discovery. More recently, algorithms have emerged that can also learn causal structure from purely or mostly observational data, as well as experimental data. These methods have started to be applied in various philosophical contexts, such as debates about our concepts of free will and determinism. This paper provides a “user's guide” to these methods, though not in (...) the sense of specifying exact button presses in a software package. Instead, we explain the larger “pipeline” within which these methods are used and discuss key steps in moving from initial research idea to validated causal structure. (shrink)

Policy makers and economists alike failed to predict the financial crisis of 2008. Their failure is due not only to the difficulties in predicting events in a complex world, but to the self-referential character of modern macroeconomics. Instead of seeking new empirical insights about economic behavior, macroeconomists have become creators of computer games—content to develop models that are internally consistent but have no necessary connection to the real world. Economic modeling aspires to be scientific in its deductive consistency and rigor. (...) However, even the natural sciences require a good dose of induction, which, in turn, entails varying one's method of study based on the phenomenon one is studying. (shrink)

Philosophers of science have long been concerned with the question of what a given scientific theory tells us about the contents of the world, but relatively little attention has been paid to how we set out to build theories and to the relevance of pre-theoretical methodology on a theory’s interpretation. In the traditional view, the form and content of a mature theory can be separated from any tentative ontological assumptions that went into its development. For this reason, the target of (...) interpretation is taken to be the mature theory and nothing more. On this view, positions on ontology emerge only once a theory is to hand, not as part of the process of theory building. Serious attention to theory creation suggests this is too simple. In particular, data collection and experimentation are influenced both by theory and by assumptions about the entities thought to be the target of study. Initial reasoning about possible ontologies has an influence on the choice of theoretical variables as well as on the judgments of the methodology appropriate to investigate them. (shrink)

In order to investigate the relationship between scientific evidence and social commitments this article addresses three questions: does low dose ionizing radiation cause cancer? Is the answer to this question different in a social setting than in a scientific context? What are the consequences of the answers of 1 and 2 for the relationship between epistemology and ethics as played out in radiation protection? Conceptual analysis with basis in the philosophy of science, in particular traditional theories of causality. Whether low (...) dose ionizing radiation causes cancer deeply depends on what we mean by causality. According to traditional scientific conceptions of causality it is not warranted to say that low dose ionizing radiation causes cancer. Standard approaches in radiation protection, however, imply that there is a causal connection, which is due to the strong social commitment in the field. There is a close relationship between social and scientific conceptions of causality, posing a series of challenges: one being that scientists covertly become moral experts, another one that the general public can be misinformed. There is a difference between causality in science and in policy making. Mixing these conceptions, as sometimes is done in radiation protection, can be misleading. Whether low dose ionizing radiation causes cancer is a social and not only a scientific issue. As such those who are warranted to have a say. (shrink)

This paper supports the literature which argues that derivational robustness can have epistemic import in highly idealized economic models. The defense is based on a particular example from mathematical economic theory, the dynamic Walrasian general equilibrium model. It is argued that derivational robustness first increased and later decreased the credibility of the Walrasian model. The example demonstrates that derivational robustness correctly describes the practices of a particular group of influential economic theorists and provides support for the arguments of philosophers who (...) have offered a general epistemic justification of such practices. (shrink)

Robustness is often presented as a guideline for distinguishing the true or real from mere appearances or artifacts. Most of recent discussions of robustness have focused on the kind of derivational robustness analysis introduced by Levins, while the related but distinct idea of robustness as multiple accessibility, defended by Wimsatt, has received less attention. In this paper, I argue that the latter kind of robustness, when properly understood, can provide justification for ontological commitments. The idea is that we are justified (...) in believing that things studied by science are real insofar as we have robust evidence for them. I develop and analyze this idea in detail, and based on concrete examples show that it plays an important role in science. Finally, I demonstrate how robustness can be used to clarify the debate on scientific realism and to formulate new arguments. (shrink)

Our best sciences are frequently held to be one way, perhaps the optimal way, to learn about the world’s higher-level ontology and structure. I first argue that which scientific theory is “best” depends in part on our goals or purposes. As a result, it is theoretically possible to have two scientific theories of the same domain, where each theory is best for some goal, but where the two theories posit incompatible ontologies. That is, it is possible for us to have (...) goal-dependent pluralism in our scientific ontologies. This ontological pluralism arises simply from our inability to directly know the world’s objects, rather than any particular claims about our cognitive limits, values, or social structures. I then present two case studies in which this possibility actually occurs—one based on simulations and theoretical analyses of constructed causal systems, and one from actual scientific investigations into the proper ontology for ocean regions. (shrink)

Causal reasoning is an aspect of learning, reasoning, and decision-making that involves the cognitive ability to discover relationships between causal relata, learn and understand these causal relationships, and make use of this causal knowledge in prediction, explanation, decision-making, and reasoning in terms of counterfactuals. Can we fully automate causal reasoning? One might feel inclined, on the basis of certain groundbreaking advances in causal epistemology, to reply in the affirmative. The aim of this paper is to demonstrate that one still has (...) good skeptical grounds for resisting any conclusions in favour of the automation of causal reasoning. If by causal reasoning is meant the entirety of the process through which we discover causal relationships and make use of this knowledge in prediction, explanation, decision-making, and reasoning in terms of counterfactuals, then one relies besides on tacit knowledge, as might be constituted by or derived from the epistemic faculty virtues and abilities of the causal reasoner, the value systems and character traits of the causal reasoner, the implicit knowledge base available to the causal reasoner, and the habits that sustain our causal reasoning practices. While certain aspects of causal reasoning may be axiomatized and formalized and algorithms may be implemented to approximate causal reasoning, one has to remain skeptical about whether causal reasoning may be fully automated. This demonstration will involve an engagement with Meno’s Paradox. (shrink)

I seek to provide a systematic and comprehensive framework for the description and analysis of scientific practice—a philosophical grammar of scientific practice, ‘grammar’ as meant by the later Wittgenstein. I begin with the recognition that all scientific work, including pure theorizing, consists of actions, of the physical, mental, and ‘paper-and-pencil’ varieties. When we set out to see what it is that one actually does in scientific work, the following set of questions naturally emerge: who is doing what, why, and how? (...) More specifically, we must arrive at some coherent philosophical accounts of the following elements of scientific practice: the agent—free, embodied, and constantly in second-person interactions with other agents; the purposes and proximate aims of the agent; types of activities that the agent engages in; ontological principles necessarily presumed for the performance of particular activities; instruments and other resources that the agent pulls together for the performance of each activity. After sketching the general framework, I also give some illustrative contrasts between the more traditional descriptions of scientific practice and the kind of descriptions enabled by the proposed framework. (shrink)

What kinds of evidence reliably support predictions of effectiveness for health and social care interventions? There is increasing reliance, not only for health care policy and practice but also for more general social and economic policy deliberation, on evidence that comes from studies whose basic logic is that of JS Mill's method of difference. These include randomized controlled trials, case–control studies, cohort studies, and some uses of causal Bayes nets and counterfactual-licensing models like ones commonly developed in econometrics. The topic (...) of this paper is the 'external validity' of causal conclusions from these kinds of studies. We shall argue two claims. Claim, negative: external validity is the wrong idea; claim, positive: 'capacities' are almost always the right idea, if there is a right idea to be had. If we are right about these claims, it makes big problems for policy decisions. Many advice guides for grading policy predictions give top grades to a proposed policy if it has two good Mill's-method-of difference studies that support it. But if capacities are to serve as the conduit for support from a method-of-difference study to an effectiveness prediction, much more evidence, and much different in kind, is required. We will illustrate the complexities involved with the case of multisystemic therapy, an internationally adopted intervention to try to diminish antisocial behaviour in young people. (shrink)

How can philosophy of science be of more practical use? One thing we can do is provide practicable advice about how to determine when one empirical claim is relevant to the truth of another; i.e., about evidential relevance. This matters especially for evidence-based policy, where advice is thin—and misleading—about how to tell what counts as evidence for policy effectiveness. This paper argues that good efficacy results (as in randomized controlled trials), which are all the rage now, are only a very (...) small part of the story. To tell what facts are relevant for judging policy effectiveness, we need to construct causal scenarios about will happen when the policy is implemented. (shrink)

Wolfgang Spohn’s Frege prize lecture, like the work on which it is based, is a tour de force of rich, elegant, coherent argument about how the projected world that we experience is constructed. But we do not live in this projected world nor reason about it. The things Spohn constructs are there from the start—or so my Stanford School pragmatism teaches. This paper explores a deep difference in philosophical approaches—Spohn’s elegant proofs versus the stocky, tangled arguments I advocate—and illustrates how (...) these play out in far more detailed disputes about the nature of causality and causal inference. (shrink)

This paper argues that machine learning and epidemiology are on collision course over causation. The discipline of epidemiology lays great emphasis on causation, while ML research does not. Some epidemiologists have proposed imposing what amounts to a causal constraint on ML in epidemiology, requiring it either to engage in causal inference or restrict itself to mere projection. We whittle down the issues to the question of whether causal knowledge is necessary for underwriting predictions about the outcomes of public health interventions. (...) While there is great plausibility to the idea that it is, conviction that something is impossible does not by itself motivate a constraint to forbid trying. We disambiguate the possible motivations for such a constraint into definitional, metaphysical, epistemological, and pragmatic considerations and argue that “Proceed with caution” is the outcome of each. We then argue that there are positive reasons to proceed, albeit cautiously. Causal inference enforces existing classification schema prior to the testing of associational claims, but associations and classification schema are more plausibly discovered in a back-and-forth process of gaining reflective equilibrium. ML instantiates this kind of process, we argue, and thus offers the welcome prospect of uncovering meaningful new concepts in epidemiology and public health—provided it is not causally constrained. (shrink)