Analyzing causation in terms of Woodward's interventionist theory and describing the structure of the world in terms of causal powers are usually regarded as quite different projects in contemporary philosophy. Interventionists aim to give an account of how causal relations can be empirically discovered and described, without committing themselves to views about what causation really is. Causal powers theorists engage in precisely the latter project, aiming to describe the metaphysical structure of the world. In this paper, I argue that interventionism (...) can benefit from incorporating considerations about causal powers. I describe a previously undiscussed problem for Woodward's definition of an intervention variable that arises when interventionist causal models include disjunctive variables. This problem is solved not by excluding disjunctive (or logically compound) variables per se, but by excluding variables whose values represent disjunctions of too different causal powers. This suggests that interventionism and causal powers theories may be less distant from each other than is often assumed. (shrink)

What features will something have if it counts as an explanation? And will something count as an explanation if it has those features? In the second half of the 20th century, philosophers of science set for themselves the task of answering such questions, just as a priori conceptual analysis was generally falling out of favor. And as it did, most philosophers of science just moved on to more manageable questions about the varieties of explanation and discipline-specific scientific explanation. Often, such (...) shifts are sound strategies for problem-solving. But leaving fallow certain basic conceptual issues can also result in foundational debates. (shrink)

Markus (2021) argues that the causal modelling frameworks of Pearl and Rubin are not ‘strongly equivalent’, in the sense of saying ‘the same thing in different ways’. Here I rebut Markus’ arguments against strong equivalence. The differences between the frameworks are best illuminated not by appeal to their causal semantics, but rather reflect pragmatic modelling choices.

Necessity and sufficiency are the building blocks of all successful explanations. Yet despite their importance, these notions have been conceptually underdeveloped and inconsistently applied in explainable artificial intelligence, a fast-growing research area that is so far lacking in firm theoretical foundations. In this article, an expanded version of a paper originally presented at the 37th Conference on Uncertainty in Artificial Intelligence, we attempt to fill this gap. Building on work in logic, probability, and causality, we establish the central role of (...) necessity and sufficiency in XAI, unifying seemingly disparate methods in a single formal framework. We propose a novel formulation of these concepts, and demonstrate its advantages over leading alternatives. We present a sound and complete algorithm for computing explanatory factors with respect to a given context and set of agentive preferences, allowing users to identify necessary and sufficient conditions for desired outcomes at minimal cost. Experiments on real and simulated data confirm our method’s competitive performance against state of the art XAI tools on a diverse array of tasks. (shrink)

Nancy Cartwright begins her recent book, Hunting Causes and Using Them, by noting that while a few years ago real causal claims were in dispute, nowadays “causality is back, and with a vengeance.” In the case of the social sciences, Keith Morrison writes that “Social science asks ‘why?’. Detecting causality or its corollary—prediction—is the jewel in the crown of social science research.” With respect to the health sciences, Judea Pearl writes that the “research questions that motivate most studies in the (...) health sciences are causal in nature.” However, not all data used by people interested in making causal claims come from experiments that use random assignment to control and treatment groups. Indeed, much research in the social and health science depends on non-experimental, observational data. Thus, one of the most important problems in the social and health sciences concerns making warranted causal claims using non-experimental, observational data; viz., “Can observational data be used to make etiological inferences leading to warranted causal claims?” This paper examines one method of warranting causal claims that is especially widespread in epidemiology and the health sciences generally—the use of causal criteria. It is argued that cases of complex causation generally, and redundant causation—both causal overdetermination and causal preemption—specifically, undermine the use of such criteria to warrant causal claims. (shrink)

The existing philosophical views on what is the meaning of causality adequate to medicine are vastly divided. We approach this question and offer two arguments in favor of pluralism regarding concepts of causality. First, we analyze the three main types of research designs (randomized-controlled trials, observational epidemiology and laboratory research). We argue, using examples, that they allow for making causal conclusions that are best understood differently in each case (in agreement with a version of manipulationist, probabilistic and mechanistic definitions, respectively). (...) Second, we analyze clinical practice and argue that these manipulationist, probabilistic and mechanistic causal claims can be used as evidence for different therapeutic decisions. We differentiate among ‘predicting’ that does not change the relata of causal claims, (mechanistic) ‘interferences’, and ‘interventions’ in the strict sense that act on causes to change effects. The central conclusion is that causal claims agreeing with diverse concepts of causality can deliver evidence for different types of therapeutic decisions. (shrink)

Behavior genetics is a controversial science. For decades, scholars have sought to understand the role of heredity in human behavior and life-course outcomes. Recently, technological advances and the rapid expansion of genomic databases have facilitated the discovery of genes associated with human phenotypes such as educational attainment and substance use disorders. To maximize the potential of this flourishing science, and to minimize potential harms, careful analysis of what it would mean for genes to be causes of human behavior is needed. (...) In this paper, we advance a framework for identifying instances of genetic causes, interpreting those causal relationships, and applying them to advance causal knowledge more generally in the social sciences. Central to thinking about genes as causes is counterfactual reasoning, the cornerstone of causal thinking in statistics, medicine, and philosophy. We argue that within-family genetic effects represent the product of a counterfactual comparison in the same way as average treatment effects (ATEs) from randomized controlled trials (RCTs). Both ATEs from RCTs and within-family genetic effects are shallow causes: They operate within intricate causal systems (non-unitary), produce heterogeneous effects across individuals (non-uniform), and are not mechanistically informative (non-explanatory). Despite these limitations, shallow causal knowledge can be used to improve understanding of the etiology of human behavior and to explore sources of heterogeneity and fade-out in treatment effects. (shrink)

Most defenders of the new mechanistic approach accept ontic constraints for successful scientific explanation (Illari 2013; Craver 2014). The minimal claim is that scientific explanations have objective truthmakers, namely mechanisms that exist in the physical world independently of any observer and that cause or constitute the phenomena-to- be-explained. How can this idea be applied to type-level explanations? Many authors at least implicitly assume that in order for mechanisms to be the truthmakers of type-level explanation they need to be regular (Andersen (...) 2012; Sheredos 2015). One problem of this assumption is that most mechanisms are (highly) stochastic in the sense that they “fail more often than they succeed” (Bogen 2005; Andersen 2012). How can a mechanism type whose instances are more likely not to produce an instance of a particular phenomenon type be the truthmaker of the explanation of that particular phenomenon type? In this paper, I will give an answer to this question. I will analyze the notion of regularity and I will discuss Andersen's suggestion for how to cope with stochastic mechanisms. I will argue that her suggestion cannot account for all kinds of stochastic mechanisms and does not provide an answer as to why regularity grounds type-level explanation. According to my analysis, a mechanistic type- level explanation is true if and only if at least one of the following two conditions is satisfied: the mechanism brings about the phenomenon more often than any other phenomenon (comparative regularity) or the phenomenon is more often brought about by the mechanism than by any other mechanism/causal sequence (comparative reverse regularity). (shrink)

There are simple mechanical systems that elude causal representation. We describe one that cannot be represented in a single directed acyclic graph. Our case suggests limitations on the use of causal graphs for causal inference and makes salient the point that causal relations among variables depend upon details of causal setups, including values of variables.

Current philosophical accounts of causation suggest that the same causal assertion can have different meanings. Yet, in actual social-scientific practice, the possible meanings of some causal generalisations intended to support policy prescriptions are not always spelled out. In line with a standard referentialist approach to semantics, we propose and elaborate on four questions to systematically elucidate the meaning of causal generalisations. The analysis can be useful to a host of agents, including social scientists, policy-makers, and philosophers aiming at being socially (...) relevant. To illustrate our proposal, we analyse the complexities related to the meaning of causal generalisations in the context of a concrete case of economic research which is explicitly intended to guide public policy, namely, the OECD research on the causes of unemployment. (shrink)

There are several different ways in which chance affects evolutionary change. That all of these processes are called “random genetic drift” is in part a due to common elements across these different processes, but is also a product of historical borrowing of models and language across different levels of organization in the biological hierarchy. A history of the concept of drift will reveal the variety of contexts in which drift has played an explanatory role in biology, and will shed light (...) on some of the philosophical controversy surrounding whether drift is a cause of evolutionary change. (shrink)

One part of the true theory of actual causation is a set of conditions responsible for eliminating all of the non-causes of an effect that can be discerned at the level of counterfactual structure. I defend a proposal for this part of the theory.

Questions about the metaphysics of causation may be usefully divided as follows. First, there are questions about the nature of the causal relata, including (1.1) whether they are in spacetime immanence), (1.2) how fine grained they are individuation), and (1.3) how many there are adicity). Second, there are questions about the metaphysics of the causal relation, including (2.1) what is the difference between causally related and causally unrelated sequences connection), (2.2) what is the difference between sequences related as cause to (...) effect, and those related as effect to cause or as joint effects of a common cause direction), and (2.3) what is the difference between sequences involving the cause, and those involving mere conditions selection). (shrink)

This thesis studies Judea Pearl’s logic of counterfactuals derived from the causal modeling framework, in comparison to the influential Stanlnaker-Lewis counterfactual logics. My study focuses on a characteristic principle in Pearl’s logic, named reversibility. The principle, as Pearl pointed out, goes beyond Lewis’s logic. Indeed, it also goes beyond the stronger logic of Stanlnaker, which is more analogous to Pearl’s logic. The first result of this thesis is an extension of Stanlnaker’s logic incorporating reversibility. It will be observed that the (...) translation of reversibility from Pearl’s language to the standard language for conditional logic deserves some attention. In particular, a straightforward translation following Pearl’s suggestion would render reversibility incompatible with Stanlnaker’s logic. A new translation of reversibility will be proposed, and an extension of Stanlnaker’s logic with the inclusion of the translated reversibility will be investigated. More importantly, it will be shown that the extended Stanlnaker’s logic is sound and complete with respect to a modified Stanlnaker’s semantics. The extension of Stanlnaker’s logic has an interesting implication. Zhang, Lam, and de Clercq have shown that special case of reversibility, despite its name, actually states an important kind of irreversibility: counterfactual dependence between distinct events is irreversible. In other words, reversibility entails that there is no cycle of counterfactual dependence altogether generalizations of reversibility. However, as shown in Zhang et al., Pearls’ logic does not rule out cycles of counterfactual dependence altogether. It in fact allows cycles that involve three or more distinct events. This is peculiar because the status of cyclic counterfactual dependence seems no more metaphysically secure than that of mutual counterfactual dependence. This consideration leads to an exploration of logics that rule out all cycles of counterfactual dependence. A surprising result is that the extension of Stanlnaker’s logic is precisely a logic of this sort. (shrink)