The most serious candidates for common causes that fail to screen off and thus violate the causal Markov condition refer to quantum phenomena. In her seminal debate with Hausman and Woodward, Cartwright early on focussed on unfortunate non-quantum examples. Especially, Hausman and Woodward’s redescriptions of quantum cases saving the CMC remain unchallenged. This paper takes up this lose end of the discussion and aims to resolve the debate in favour of Cartwright’s position. It systematically considers redescriptions of ICC structures, including (...) those by Hausman and Woodward, and explains why these are inappropriate, when quantum mechanics is true. It first shows that all cases of purported quantum ICCs are cases of entanglement and then, using the tools of causal modelling, it provides an analysis of the quantum mechanical formalism for the case that the collapse of entangled systems is best described as a causal model with an ICC. (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 chapter contains sections titled: Highlights of Past Literature Current Work Future Work.

How do multiple reasons combine to support a conclusion about what to do or believe? This question raises two challenges: How can we represent the strength of a reason? How do the strengths of multiple reasons combine? Analogous challenges about confirmation have been answered using probabilistic tools. Can reductive and nonreductive theories of reasons use these tools to answer their challenges? Yes, or more exactly: reductive theories can answer both challenges. Nonreductive theories, with the help of a result in confirmation (...) theory, can answer one, and there are grounds for optimism that they can answer the other. (shrink)

There is a long tradition of thinking of thermodynamics, not as a theory of fundamental physics, but as a theory of how manipulations of a physical system may be used to obtain desired effects, such as mechanical work. On this view, the basic concepts of thermodynamics, heat and work, and with them, the concept of entropy, are relative to a class of envisaged manipulations. This article is a sketch and defense of a science of manipulations and their effects on physical (...) systems. I call this science thermo-dynamics, or ΘΔcs\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\Theta \Delta }^{\text{cs}}$$\end{document}, for short, to highlight that it may be different from the science of thermodynamics, as the reader conceives it. An upshot of the discussion is a clarification of the roles of the Gibbs and von Neumann entropies. Light is also shed on the use of coarse-grained entropies. (shrink)

Reichenbach’s ‘principle of the common cause’ is a foundational assumption of some important recent contributions to quantitative social science methodology but no similar principle appears in econometrics. Angrist et al. has argued that the principle is necessary for instrumental variables methods in econometrics, and Angrist Krueger builds a framework using it that he proposes as a means of resolving an important methodological dispute among econometricians. Through analysis of instrumental variables methods and the issue of multicollinearity, we aim to show that (...) the relationship of the principle to econometric methods is more nuanced than implied by previous work but nevertheless may make a valuable contribution to the coherence and validity of existing methods. (shrink)

Social research, from economics to demography and epidemiology, makes extensive use of statistical models in order to establish causal relations. The question arises as to what guarantees the causal interpretation of such models. In this paper we focus on econometrics and advance the view that causal models are ‘augmented’ statistical models that incorporate important causal information which contributes to their causal interpretation. The primary objective of this paper is to argue that causal claims are established on the basis of a (...) plurality of evidence. We discuss the consequences of ‘evidential pluralism’ in the context of econometric modelling. (shrink)

Adjusting for covariates is a well-established method to estimate the total causal effect of an exposure variable on an outcome of interest. Depending on the causal structure of the mechanism under study, there may be different adjustment sets, equally valid from a theoretical perspective, leading to identical causal effects. However, in practice, with finite data, estimators built on different sets may display different precisions. To investigate the extent of this variability, we consider the simplest non-trivial non-linear model of a v-structure (...) on three nodes for binary data. We explicitly compute and compare the variance of the two possible different causal estimators. Further, by going beyond leading-order asymptotics, we show that there are parameter regimes where the set with the asymptotically optimal variance does depend on the edge coefficients, a result that is not captured by the recent leading-order developments for general causal models. As a practical consequence, the adjustment set selection needs to account for the relative magnitude of the relationships between variables with respect to the sample size and cannot rely on purely graphical criteria. (shrink)

It is often argued that higher-level special-science properties cannot be causally efficacious since the lower-level physical properties on which they supervene are doing all the causal work. This claim is usually derived from an exclusion principle stating that if a higher-level property F supervenes on a physical property F* that is causally sufficient for a property G, then F cannot cause G. We employ an account of causation as difference-making to show that the truth or falsity of this principle is (...) a contingent matter and derive necessary and sufficient conditions under which a version of it holds. We argue that one important instance of the principle, far from undermining non-reductive physicalism, actually supports the causal autonomy of certain higher-level properties. (shrink)

Causalism is the thesis that natural selection can cause evolution. A standard argument for causalism involves showing that a hypothetical intervention on some population-level property that is identified with natural selection will result in evolution. In a pair of articles, one of which recently appeared in the pages of this journal, Jun Otsuka has put forward a quite different argument for causalism. Otsuka attempts to show that natural selection can cause evolution by considering a hypothetical intervention on an individual-level property. (...) Specifically, Otsuka identifies natural selection with the causal relationship between a trait and fitness, claims an intervention on the strength of this relationship can cause evolution, then concludes that natural selection can cause evolution. Below I describe why Otsuka’s argument for causalism is unconvincing. Central to my criticism is that Otsuka’s argument works only if one adopts an indefensible account of natural selection, according to which natural selection can occur in the absence of trait or fitness variation. I go on to explain why any attempt to demonstrate the truth of causalism via a hypothetical intervention on an individual-level property would appear to require one to adopt an account of natural selection that is inadequate for the same reason. This in turn means the plausibility of causalism does indeed depend on the plausibility of the claim that population-level properties, which supervene on the properties of the individuals in the population, can be causally efficacious. (shrink)

The contextual approach is a prominent framework for thinking about group selection. Here, I highlight ambiguity about what the contextual approach is. Then, I discuss problematic entailments the contextual approach has for what processes count as group selection—entailments more troublesome than typically noted. However, Sober and Wilson’s version of the Price approach, which is the main alternative to the contextual approach, is problematic too: it leads to an underappreciated paradox called the vanishing selection problem and thereby generates the wrong qualitative (...) account of whether group selection is occurring in a certain family of cases. In response, I develop an account of group selection that can deal with the counterexamples to both the contextual approach and the Price approach. I then discuss the role that contextual analysis can continue to play in the discussion of individual fitness and metapopulation evolution. (shrink)

This article describes a heuristic argument for understanding certain physical systems in terms of properties that resemble the beliefs and goals of folk psychology. The argument rests on very simple assumptions. The core of the argument is that predictions about certain events can legitimately be based on assumptions about later events, resembling Aristotelian ‘final causation’; however, more nuanced causal entities must be introduced into these types of explanation in order for them to remain consistent with a causally local Universe.

Walking through the supermarket, I see the avocados. I know they are avocados. Similarly, if you see a pumpkin on my office desk, you can know it’s a pumpkin from its looks. The phenomenology in such cases is that of “just seeing” that such and such. This phenomenology might suggest that the knowledge gained is immediate. This paper argues, to the contrary, that in these target cases, the knowledge is mediate, depending as it does on one’s knowledge of what the (...) relevant kind of thing looks like. To make the case requires examining the nature of knowing what Fs look like. Is such knowledge to be understood as knowledge of a fact, or rather as a kind of ability? From the conclusion that the knowledge in the target cases is not immediate, the paper concludes that perception does not afford us immediate knowledge concerning objects’ kinds. (shrink)

One of the primary motivations behind moral anti-realism is a deep-rooted scepticism about moral knowledge. Moral realists attempt counter this worry by sketching a plausible moral epistemology. One of the most radical proposals in the recent literature is that we know moral facts by perception – we can literally see that an action is wrong, etc. A serious objection to moral perception is the causal objection. It is widely conceded that perception requires a causal connection between the perceived and the (...) perceiver. But, the objection continues, we are not in appropriate causal contact with moral properties. Therefore, we cannot perceive moral properties. This papers demonstrates that the causal objection is unsound whether moral properties turn out to be secondary, natural properties; non-secondary, natural properties; or non-natural properties. 1. (shrink)

The Principle of the Common Cause is usually understood to provide causal explanations for probabilistic correlations obtaining between causally unrelated events. In this study, an extended interpretation of the principle is proposed, according to which common causes should be invoked to explain positive correlations whose values depart from the ones that one would expect to obtain in accordance to her probabilistic expectations. In addition, a probabilistic model for common causes is tailored which satisfies the generalized version of the principle, at (...) the same time including the standard conjunctive-fork model as a special case. (shrink)

In medicine and the social sciences, researchers must frequently integrate the findings of many observational studies, which measure overlapping collections of variables. For instance, learning how to prevent obesity requires combining studies that investigate obesity and diet with others that investigate obesity and exercise. Recently developed causal discovery algorithms provide techniques for integrating many studies, but little is known about what can be learned from such algorithms. This article argues that there are causal facts that one could learn by conducting (...) a large study but which could not be learned by combining many smaller studies. Moreover, I characterize the frequency with which combining many studies increases underdetermination and exactly how much information is lost. 1 Introduction2 Causal Inference from Observational Data3 Piecemeal Causal Inference4 The Extent and Frequency of the Problem of Piecemeal Induction5 Conclusion. (shrink)

Research on human causal induction has shown that people have general prior assumptions about causal strength and about how causes interact with the background. We propose that these prior assumptions about the parameters of causal systems do not only manifest themselves in estimations of causal strength or the selection of causes but also when deciding between alternative causal structures. In three experiments, we requested subjects to choose which of two observable variables was the cause and which the effect. We found (...) strong evidence that learners have interindividually variable but intraindividually stable priors about causal parameters that express a preference for causal determinism. These priors predict which structure subjects preferentially select. The priors can be manipulated experimentally and appear to be domain-general. Heuristic strategies of structure induction are suggested that can be viewed as simplified implementations of the priors. (shrink)

In medicine and the social sciences, researchers often measure only a handful of variables simultaneously. The underlying assumption behind this methodology is that combining the results of dozens of smaller studies can, in principle, yield as much information as one large study, in which dozens of variables are measured simultaneously. Mayo-Wilson :864–874, 2011, Br J Philos Sci 65:213–249, 2013. https://doi.org/10.1093/bjps/axs030) shows that assumption is false when causal theories are inferred from observational data. This paper extends Mayo-Wilson’s results to cases in (...) which experimental data is available. I prove several new theorems that show that, as the number of variables under investigation grows, experiments do not improve, in the worst-case, one’s ability to identify the true causal model if one can measure only a few variables at a time. However, stronger statistical assumptions significantly aid causal discovery in piecemeal inquiry, even if such assumptions are unhelpful when all variables can be measured simultaneously. (shrink)

Currently, two frameworks of causal reasoning compete: Whereas dependency theories focus on dependencies between causes and effects, dispositional theories model causation as an interaction between agents and patients endowed with intrinsic dispositions. One important finding providing a bridge between these two frameworks is that failures of causes to generate their effects tend to be differentially attributed to agents and patients regardless of their location on either the cause or the effect side. To model different types of error attribution, we augmented (...) a causal Bayes net model with separate error sources for causes and effects. In several experiments, we tested this new model using the size of Markov violations as the empirical indicator of differential assumptions about the sources of error. As predicted by the model, the size of Markov violations was influenced by the location of the agents and was moderated by the causal structure and the type of causal variables. (shrink)

This paper is about mechanisms and models, and how they interact. In part, it is a response to recent discussion in philosophy of biology regarding whether natural selection is a mechanism. We suggest that this debate is indicative of a more general problem that occurs when scientists produce mechanistic models of populations and their behaviour. We can make sense of claims that there are mechanisms that drive population-level phenomena such as macroeconomics, natural selection, ecology, and epidemiology. But talk of mechanisms (...) and mechanistic explanation evokes objects with well-defined and localisable parts which interact in discrete ways, while models of populations include parts and interactions that are neither local nor discrete in any actual populations. This apparent tension can be resolved by carefully distinguishing between the properties of a model and those of the system it represents. To this end, we provide an analysis that recognises the flexible relationship between a mechanistic model and its target system. In turn, this reveals a surprising feature of mechanistic representation and explanation: it can occur even when there is a mismatch between the mechanism of the model and that of its target. Our analysis reframes the debate, providing an alternative way to interpret scientists’ mechanism-talk , which initially motivated the issue. We suggest that the relevant question is not whether any population-level phenomenon such as natural selection is a mechanism, but whether it can be usefully modelled as though it were a particular type of mechanism. (shrink)

In this paper I analyze the process by which modelers in systems biology arrive at an adequate representation of the biological structures thought to underlie data gathered from high-throughput experiments. Contrary to views that causal claims and explanations are rare in systems biology, I argue that in many studies of gene regulatory networks modelers aim at a representation of causal structure. In addressing modeling challenges, they draw on assumptions informed by theory and pragmatic considerations in a manner that is guided (...) by an interventionist conception of causal structure. While doubts have been raised about the applicability of this notion of causality to complex biological systems, it is here seen to be an adequate guide to inquiry. (shrink)

We introduce an expected-value theory of linguistic modality that makes reference to expected utility and a likelihood-based confirmation measure for deontics and epistemics, respectively. The account is a probabilistic semantics for deontics and epistemics, yet it proposes that deontics and epistemics share a common core modal semantics, as in traditional possible-worlds analysis of modality. We argue that this account is not only theoretically advantageous, but also has far-reaching empirical consequences. In particular, we predict modal versions of reasoning fallacies from the (...) heuristics and biases literature. Additionally, we derive the modal semantics in an entirely transparent manner, as it is based on the compositional semantics of Korean modal expressions that are morphosyntactically decomposed into a conditional and an evaluative predicate. (shrink)

Causal inference plays a central role in behavioral science. Historically, behavioral science methodologies have typically sought to infer a single causal relation. Each of the major approaches to causal inference in the behavioral sciences follows this pattern. Nonetheless, such approaches sometimes differ in the causal relation that they infer. Incremental causal inference offers an alternative to this conceptualization of causal inference that divides the inference into a series of incremental steps. Different steps infer different causal relations. Incremental causal inference is (...) consistent with both causal pluralism and anti-pluralism. However, anti-pluralism places greater constraints the possible topology of sequential inferences. Arguments against causal inference include questioning consistency with causation as an explanatory principle, charging undue complexity, and questioning the need for it. Arguments in favor of incremental inference include better explanation of diverse causal inferences in behavioral science, tailored causal inference, and more detailed and explicit description of causal inference. Incremental causal inference offers a viable and potentially fruitful alternative to approaches limited to a single causal relation. (shrink)

Some explanations appeal to facts about the causal structure of a system in order to shed light on a particular phenomenon; these are explanations which do more than cite the causes X and Y of some state-of-affairs Z, but rather appeal to “macro-level” causal features—for example the fact that A causes B as well as C, or perhaps that D is a strong inhibitor of E—in order to explain Z. Appeals to these kinds of “macro-level” causal features appear in a (...) wide variety of social scientific and biological research; statements about features such as “patriarchy,” “healthcare infrastructure,” and “functioning DNA repair mechanism,” for instance, can be understood as claims about what would be different in a system with a different causal structure. I suggest interpreting counterfactual questions involving structural features as questions about alternative parameter settings of causal models, and propose an extension of the usual interventionist framework for causal explanation which enables scientists to explore the consequences of interventions on “macro-level” structure. (shrink)

It has been just over 100 years since the birth of Alan Turing and more than 65 years since he published in Mind his seminal paper, Computing Machinery and Intelligence. In the Mind paper, Turing asked a number of questions, including whether computers could ever be said to have the power of “thinking”. Turing also set up a number of criteria—including his imitation game—under which a human could judge whether a computer could be said to be “intelligent”. Turing’s paper, as (...) well as his important mathematical and computational insights of the 1930s and 1940s led to his popular acclaim as the “Father of Artificial Intelligence”. In the years since his paper was published, however, no computational system has fully satisfied Turing’s challenge. In this paper we focus on a different question, ignored in, but inspired by Turing’s work: How might the Artificial Intelligence practitioner implement “intelligence” on a computational device? Over the past 60 years, although the AI community has not produced a general-purpose computational intelligence, it has constructed a large number of important artifacts, as well as taken several philosophical stances able to shed light on the nature and implementation of intelligence. This paper contends that the construction of any human artifact includes an implicit epistemic stance. In AI this stance is found in commitments to particular knowledge representations and search strategies that lead to a product’s successes as well as its limitations. Finally, we suggest that computational and human intelligence are two different natural kinds, in the philosophical sense, and elaborate on this point in the conclusion. (shrink)

Teleological explanations (TEs) account for the existence or properties of an entity in terms of a function: we have hearts because they pump blood, and telephones for communication. While many teleological explanations seem appropriate, others are clearly not warranted-for example, that rain exists for plants to grow. Five experiments explore the theoretical commitments that underlie teleological explanations. With the analysis of [Wright, L. (1976). Teleological Explanations. Berkeley, CA: University of California Press] from philosophy as a point of departure, we examine (...) in Experiment 1 whether teleological explanations are interpreted causally, and confirm that TEs are only accepted when the function invoked in the explanation played a causal role in bringing about what is being explained. However, we also find that playing a causal role is not sufficient for all participants to accept TEs. Experiment 2 shows that this is not because participants fail to appreciate the causal structure of the scenarios used as stimuli. In Experiments 3-5 we show that the additional requirement for TE acceptance is that the process by which the function played a causal role must be general in the sense of conforming to a predictable pattern. These findings motivate a proposal, Explanation for Export, which suggests that a psychological function of explanation is to highlight information likely to subserve future prediction and intervention. We relate our proposal to normative accounts of explanation from philosophy of science, as well as to claims from psychology and artificial intelligence. (shrink)

Humans often attribute the things that happen to one or another actual cause. In this chapter, we survey some recent philosophical and psychological research on causal attribution. We pay special attention to the relation between graphical causal modeling and theories of causal attribution. We think that the study of causal attribution is one place where formal and experimental techniques nicely complement one another.

In the last decade, several researchers have proposed theories of actual causation that make use of structural equations and directed graphs. Many of these researchers are committed to a widely-endorsed folk attribution desideratum, according to which an important constraint on the acceptability of a theory of actual causation is agreement between the deliverances of the theory with respect to specific cases and the reports of untutored individuals about those same cases. In the present article, we consider a small collection of (...) related theories of actual causation, including a purely structural theory and two theories that supplement the structural equations with considerations of defaults, typicality, and normality. We argue that each of these three theories are meant to satisfy the FAD, and then we present empirical evidence that they fail to do so for several variations on a simple scenario from the literature. Drawing on our previous work on the responsibility view of folk causal attribitons, we conclude by offering a solution that allows the latter two theories to satisfy the FAD for these cases. The solution is to give up on concerns with typicality and focus on injunctive norms in supplementing the graphical modeling machinery. (shrink)

Several prominent, contemporary theories of actual causation maintain that in order for something to count as an actual cause (in the circumstances) of some known effect, the potential cause must be a difference-maker with respect to the effect in some restricted range of circumstances. Although the theories disagree about how to restrict the range of circumstances that must be considered in deciding whether something counts as an actual cause of a known effect, the theories agree that at least some counterfactual (...) circumstances must be considered. I argue that the theories are still too permissive in the range of counterfactual circumstances they admit for consideration, and I present simple counter-examples that make use of this overpermissiveness. (shrink)

Can an agent deliberating about an action A hold a meaningful credence that she will do A? 'No', say some authors, for 'Deliberation Crowds Out Prediction' (DCOP). Others disagree, but we argue here that such disagreements are often terminological. We explain why DCOP holds in a Ramseyian operationalist model of credence, but show that it is trivial to extend this model so that DCOP fails. We then discuss a model due to Joyce, and show that Joyce's rejection of DCOP rests (...) on terminological choices about terms such as 'intention', 'prediction', and 'belief'. Once these choices are in view, they reveal underlying agreement between Joyce and the DCOP-favouring tradition that descends from Ramsey. Joyce's Evidential Autonomy Thesis (EAT) is effectively DCOP, in different terminological clothing. Both principles rest on the so-called 'transparency' of first-person present-tensed reflection on one's own mental states. (shrink)

We offer a new argument for the claim that there can be non-degenerate objective chance (“true randomness”) in a deterministic world. Using a formal model of the relationship between different levels of description of a system, we show how objective chance at a higher level can coexist with its absence at a lower level. Unlike previous arguments for the level-specificity of chance, our argument shows, in a precise sense, that higher-level chance does not collapse into epistemic probability, despite higher-level properties (...) supervening on lower-level ones. We show that the distinction between objective chance and epistemic probability can be drawn, and operationalized, at every level of description. There is, therefore, not a single distinction between objective and epistemic probability, but a family of such distinctions. (shrink)

Scientists often think of the world as a dynamical system, a stochastic process, or a generalization of such a system. Prominent examples of systems are the system of planets orbiting the sun or any other classical mechanical system, a hydrogen atom or any other quantum–mechanical system, and the earth’s atmosphere or any other statistical mechanical system. We introduce a general and unified framework for describing such systems and show how it can be used to examine some familiar philosophical questions, including (...) the following: how can we define nomological possibility, necessity, determinism, and indeterminism; what are symmetries and laws; what regularities must a system display to make scientific inference possible; how might principles of parsimony such as Occam’s Razor help when we make such inferences; what is the role of space and time in a system; and might they be emergent features? Our framework is intended to serve as a toolbox for the formal analysis of systems that is applicable in several areas of philosophy. (shrink)

Newcomb’s problem involves a decision-maker faced with a choice and a predictor forecasting this choice. The agents’ interaction seems to generate a choice dilemma once the decision-maker seeks to apply two basic principles of rational choice theory : maximize expected utility ; adopt the dominant strategy. We review unsuccessful attempts at pacifying the dilemma by excluding Newcomb’s problem as an RCT-application, by restricting MEU and ADS, and by allowing for backward causation. A probability approach shows that Newcomb’s original problem-formulation lacks (...) causal information. This makes it impossible to specify the probability structure of Newcomb’s univocally. Once missing information is added, Newcomb’s problem and RCT re-align, thus explaining Newcomb’s problem as a seeming dilemma. Building on Wolpert and Benford :1637–1646, 2013), we supply additional details and offer a crucial correction to their formal proof. (shrink)

Proponents of mechanistic explanation all acknowledge the importance of organization. But they have also tended to emphasize specificity with respect to parts and operations in mechanisms. We argue that in understanding one important mode of organization—patterns of causal connectivity—a successful explanatory strategy abstracts from the specifics of the mechanism and invokes tools such as those of graph theory to explain how mechanisms with a particular mode of connectivity will behave. We discuss the connection between organization, abstraction, and mechanistic explanation and (...) illustrate our claims by looking at an example from recent research on so-called network motifs. (shrink)

Possible worlds semantics for conditionals leave open the problem of how to construct models for realistic domains. In this paper, we show how to adapt logics of action and change such as John McCarthy’s Situation Calculus to conditional logics. We illustrate the idea by presenting models for conditionals whose antecedents combine a declarative condition with a hypothetical action.

Independence of Conditionals (IC) has recently been proposed as a basic rule for causal structure learning. If a Bayesian network represents the causal structure, its Conditional Probability Distributions (CPDs) should be algorithmically independent. In this paper we compare IC with causal faithfulness (FF), stating that only those conditional independences that are implied by the causal Markov condition hold true. The latter is a basic postulate in common approaches to causal structure learning. The common spirit of FF and IC is to (...) reject causal graphs for which the joint distribution looks ‘non-generic’. The difference lies in the notion of genericity: FF sometimes rejects models just because one of the CPDs is simple, for instance if the CPD describes a deterministic relation. IC does not behave in this undesirable way. It only rejects a model when there is a non-generic relation between different CPDs although each CPD looks generic when considered separately. Moreover, it detects relations between CPDs that cannot be captured by conditional independences. IC therefore helps in distinguishing causal graphs that induce the same conditional independences (i.e., they belong to the same Markov equivalence class). The usual justification for FF implicitly assumes a prior that is a probability density on the parameter space. IC can be justified by Solomonoff’s universal prior, assigning non-zero probability to those points in parameter space that have a finite description. In this way, it favours simple CPDs, and therefore respects Occam’s razor. Since Kolmogorov complexity is uncomputable, IC is not directly applicable in practice. We argue that it is nevertheless helpful, since it has already served as inspiration and justification for novel causal inference algorithms. (shrink)

This is part A of a paper in which we defend a semantics for counterfactuals which is probabilistic in the sense that the truth condition for counterfactuals refers to a probability measure. Because of its probabilistic nature, it allows a counterfactual ‘ifAthenB’ to be true even in the presence of relevant ‘Aand notB’-worlds, as long such exceptions are not too widely spread. The semantics is made precise and studied in different versions which are related to each other by representation theorems. (...) Despite its probabilistic nature, we show that the semantics and the resulting system of logic may be regarded as a naturalistically vindicated variant of David Lewis’ truth-conditional semantics and logic of counterfactuals. At the same time, the semantics overlaps in various ways with the non-truth-conditional suppositional theory for conditionals that derives from Ernest Adams’ work. We argue that counterfactuals have two kinds of pragmatic meanings and come attached with two types of degrees of acceptability or belief, one being suppositional, the other one being truth based as determined by our probabilistic semantics; these degrees could not always coincide due to a new triviality result for counterfactuals, and they should not be identified in the light of their different interpretation and pragmatic purpose. However, for plain assertability the difference between them does not matter. Hence, if the suppositional theory of counterfactuals is formulated with sufficient care, our truth-conditional theory of counterfactuals is consistent with it. The results of our investigation are used to assess a claim considered by Hawthorne and Hájek, that is, the thesis that most ordinary counterfactuals are false. (shrink)

Reasoning about causation in fact is an essential element of attributing legal responsibility. Therefore, the automation of the attribution of legal responsibility requires a modelling effort aimed at the following: a thorough understanding of the relation between the legal concepts of responsibility and of causation in fact; a thorough understanding of the relation between causation in fact and the common sense concept of causation; and, finally, the specification of an ontology of the concepts that are minimally required for (automatic) common (...) sense reasoning about causation. This article offers a worked-out example of the indicated analysis. Such example consists of: a definition of the legal concept of responsibility (in terms of liability and accountability); a definition of the legal concept of causation in fact (in terms of the initiation of physical processes by an agent and of the provision of reasons and/or opportunities to other agents); CausatiOnt, an AI-like ontology of the common sense (causal) concepts that are minimally needed for reasoning about the legal concept of causation in fact (in particular, the concepts of category, dimension, object, agent, process, event and act). (shrink)

Computational machineries dedicated to the attribution of legal responsibility should be based on (or, make use of) a stack of definitions relating the notion of legal responsibility to a number of suitably chosen causal notions. This paper presents a general analysis of legal responsibility and of causation in fact based on Hart and Honoré’s work. Some physical aspects of causation in fact are then treated within the “lite” version of DOLCE foundational ontology written in OWL-DL, a standard description logic for (...) the Semantic Web. (shrink)

The basic Bayesian model of credence states, where each individual’s belief state is represented by a single probability measure, has been criticized as psychologically implausible, unable to represent the intuitive distinction between precise and imprecise probabilities, and normatively unjustifiable due to a need to adopt arbitrary, unmotivated priors. These arguments are often used to motivate a model on which imprecise credal states are represented by sets of probability measures. I connect this debate with recent work in Bayesian cognitive science, where (...) probabilistic models are typically provided with explicit hierarchical structure. Hierarchical Bayesian models are immune to many classic arguments against single-measure models. They represent grades of imprecision in probability assignments automatically, have strong psychological motivation, and can be normatively justified even when certain arbitrary decisions are required. In addition, hierarchical models show much more plausible learning behavior than flat representations in terms of sets of measures, which—on standard assumptions about update—rule out simple cases of learning from a starting point of total ignorance. (shrink)

This paper presents corpus and experimental data that problematize the traditional analysis of must as a strong necessity modal, as recently revived and defended by von Fintel and Gillies :351–383, 2010). I provide naturalistic examples showing that must p can be used alongside an explicit denial of knowledge of p or certainty in p, and that it can be conjoined with an expression indicating that p is not certain or that not-p is possible. I also report the results of an (...) experiment involving lotteries, where most participants endorsed a sentence of the form must not-p despite being instructed that p is a possibility. Crucially, endorsement was much higher for must in this context than for matched sentences with knowledge or certainty expressions. These results indicate that the requirements for felicitous use of must are weaker than for know and certain rather than being at least as strong, as the epistemic necessity theory would predict. However, it is possible to account for these data while retaining the key insights of von Fintel and Gillies’ analysis of the evidential component of must. I discuss several existing accounts that could be construed in this way and explain why none is completely satisfactory. I then propose a new model that embeds an existing scalar theory into a probabilistic model of informational dynamics structured around questions and answers. (shrink)

We derive a probabilistic account of the vagueness and context-sensitivity of scalar adjectives from a Bayesian approach to communication and interpretation. We describe an iterated-reasoning architecture for pragmatic interpretation and illustrate it with a simple scalar implicature example. We then show how to enrich the apparatus to handle pragmatic reasoning about the values of free variables, explore its predictions about the interpretation of scalar adjectives, and show how this model implements Edgington’s Vagueness: a reader, 1997) account of the sorites paradox, (...) with variations. The Bayesian approach has a number of explanatory virtues: in particular, it does not require any special-purpose machinery for handling vagueness, and it is integrated with a promising new approach to pragmatics and other areas of cognitive science. (shrink)

Varied evidence confirms more strongly than less varied evidence, ceteris paribus. This epistemological Variety of Evidence Thesis enjoys widespread intuitive support. We put forward a novel explication of one notion of varied evidence and the Variety of Evidence Thesis within Bayesian models of scientific inference by appealing to measures of entropy. Our explication of the Variety of Evidence Thesis holds in many of our models which also pronounce on disconfirmatory and discordant evidence. We argue that our models pronounce rightly. Against (...) a backdrop of failures of the Variety of Evidence Thesis, the intuitive case for the Variety of Evidence Thesis emerges strengthened. Our models do however not support the general case for the thesis since our explication of it fails to hold in certain cases. The parameter space of this failure is explored and an explanation for the failure is offered. (shrink)

How do people attribute responsibility in situations where the contributions of multiple agents combine to produce a joint outcome? The prevalence of over-determination in such cases makes this a difficult problem for counterfactual theories of causal responsibility. In this article, we explore a general framework for assigning responsibility in multiple agent contexts. We draw on the structural model account of actual causation (e.g., Halpern & Pearl, 2005) and its extension to responsibility judgments (Chockler & Halpern, 2004). We review the main (...) theoretical and empirical issues that arise from this literature and propose a novel model of intuitive judgments of responsibility. This model is a function of both pivotality (whether an agent made a difference to the outcome) and criticality (how important the agent is perceived to be for the outcome, before any actions are taken). The model explains empirical results from previous studies and is supported by a new experiment that manipulates both pivotality and criticality. We also discuss possible extensions of this model to deal with a broader range of causal situations. Overall, our approach emphasizes the close interrelations between causality, counterfactuals, and responsibility attributions. (shrink)

Can the phenomena of associative learning be replaced wholesale by a propositional reasoning system? Mitchell et al. make a strong case against an automatic, unconscious, and encapsulated associative system. However, their propositional account fails to distinguish inferences based on actions from those based on observation. Causal Bayes networks remedy this shortcoming, and also provide an overarching framework for both learning and reasoning. On this account, causal representations are primary, but associative learning processes are not excluded a priori.