Praise for the French-language edition of Causalite · et lois de la nature : ‘‘ . . . it is a pleasure to read Kistler’s book and . . . its argument is very well developed. It is a remarkable example of the standards of clarity and precision that are achieved in today’s analytical philosophy of science.’’ Michael Esfeld, University of Lausanne: review in Dialectica ‘‘ . . . a wonderfully rich book by Max Kistler . . . not only a skilled writer in the history of philosophy; he also makes important and novel contributions both to the theory of causation and to the philosophy of laws of nature. . . . This book is already packed with good arguments. My impression is that Causalite · et lois de la nature is important read- ing for all philosophers with an interest in laws of nature and causa- tion. I will certainly be on the watch both for the translation of it and for forthcoming work by Max Kistler.’’ Johannes Persson, Lund University, review in Mind Causation and Laws of Nature Causation is important. It is, as Hume said, the cement of the universe, and lies at the heart of our conceptual structure. Causation is one of the most fundamental tools we have for organizing our apprehension of the external world and ourselves. But philosophers’ disagreement about the correct interpretation of causation is as limitless as their agreement about its importance. The history of attempts to elucidate the nature of this concept and to situate it with respect to other fundamental concepts is almost as long as the history of philosophy itself. In this first English translation of Causalite · et lois de la nature Max Kistler seeks to reconstruct a unified concept of causation that is general enough to adequately deal with both elementary physical processes and the macroscopic level of phenomena we encounter in everyday life. It will be of great interest to philosophers of science and metaphysics, and also to stu- dents and scholars of philosophy of mind where concepts of causation and law play a prominent role. Max Kistler is Maı ˆtre de conferences, in the Department of Philosophy at the Universite Paris X-Nanterre, France and member of Institut Jean Nicod, Paris. Routledge Studies in Contemporary Philosophy 1 Email and Ethics Style and ethical relations in computer-mediated communication Emma Rooksby 2 Causation and Laws of Nature Max Kistler Causation and Laws of Nature Max Kistler I~ ~?io~!!~n~~;up LONDON AND NEW YORK First published in French: Causalite · et lois de la nature # Librarie Philosophiquie J. Vrin, Paris, 1999 http://www.vrin.com English-language translation first published 2006 by Routledge Routledge is an imprint of the Taylor & Francis Group, an informa business English language translation # Routledge, 2006 Typeset in Times by Taylor & Francis Books British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library Library of Congress Cataloging in Publication Data A catalog record for this book has been requested ISBN 13: 978-0-415–39859–6 (hbk) 2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN 711 Third Avenue, New York, NY 10017, USA Published 2017 by Routledge The Open Access version of this book, available at www.tandfebooks.com, has been made available under a Creative Commons Attribution-Non Commercial-No Derivatives 4.0 license. Contents Acknowledgments viii Introduction 1 1 What is a causal relation? 9 2 Laws of nature and universal generalizations 75 3 Applicability conditions and the concept of ‘‘strict law’’ 99 4 Consequences 122 5 The nomological theory of causation and causal responsibility 139 6 Efficacious properties and the instantiation of laws 165 7 Causal responsibility and its applications 188 Conclusion 213 Notes 224 References 251 Index 261 Acknowledgments This book is a modified version of Causalite · et lois de la nature , published in 1999, which itself grew out of my PhD dissertation from 1995, at EHESS (Ecole des Hautes Etudes en Sciences Sociales). I express my warmest thanks to my supervisor Joe ¤lle Proust, whose generosity never failed, with both encouragement and sharp criticism. Many other people at the Institut Jean Nicod in Paris (which was then part of CREA, Centre de recherches en e ·piste ·mologie applique ·e) have helped me overcome difficulties, doubts and hesitations. I thank the French Ministry of Research and the Ecole Polytechnique for their financial help during the preparation of this work. I am very grateful to Joan Cullen who has read the present English ver- sion and helped me improve it in style and clarity. The errors remain of course mine. Introduction It is not very controversial to say that causation is one of the most funda- mental conceptual tools we have for organizing our apprehension of the external world and ourselves. But philosophers’ disagreement about the correct interpretation of causation is as vast as their accord as to its importance. The history of attempts to elucidate the nature of this concept and to situate it with respect to other fundamental concepts is almost as long as the history of philosophy. Our endeavour can only succeed if we begin by setting ourselves a well- defined aim. The question we shall try to answer is the following: can we reconstruct a concept of causation that claims both objectivity and com- patibility with a world view grounded in contemporary science? My strategy will be to answer this question in the affirmative by effecting a reconstruc- tion of the concept of causation, articulating its presuppositions and testing its fecundity in the analysis of causal explanation, as it exists both in the sciences and in commonsense. In spite of the enormous efforts given over to the analysis of causation, there is a persistent variance as to what may be concluded. Abandoning the hope of finding a satisfying analysis of the concept, many authors whose work is not directly focused on the concept of causation use it both exten- sively and with little rigour. For lack of a consensus on the ‘‘correct’’ inter- pretation of causation, several incompatible conceptions exist alongside of commonsense intuitions. Among these, three are particularly influential: 1 ) Often causation is identified with determination. This identification stems presumably from the assimilation of causation with the traditional Principle of Causality. If the Principle of Causality states that ‘‘every event has a cause’’, that ‘‘every determinate phenomenon has a determinate cause’’, or that ‘‘the same causes always have the same effects’’, it is appar- ently easy to slip from this Principle to the thesis that it is equivalent to say that the event e has a cause and to say that the fact that e has happened has been determined by the circumstances, i.e. by the set of events preceding e (or, for that matter, by the set of events following e ). According to this conception, causation is assimilated to the natural course of events, taken in a deterministic sense. To say that everything has a cause is tantamount to saying that nothing happens without a reason, sufficient to determine its nature. To say that all phenomena are subject to the constraint of causation means, according to this conception, not only that nothing happens by chance, but that there are no irreducibly indeterministic processes. On the contrary, on the supposition that such processes exist, e.g. in radioactive decomposition, this conception implies that the fact that the atomic nucleus n decomposes at instant t is not only not determined by laws of nature but furthermore that it has no cause. 2) The dominant doctrine in contemporary philosophy of science – a position that can safely be qualified as ‘‘classical’’ – differs from the position I have just sketched by acknowledging the existence of indeterministic pro- cesses. According to the deductive-nomological model, to explain (or to predict) a phenomenon is to show that there is a description of the phe- nomenon which is such that it can be logically deduced from a nomological statement, together with a description of the initial conditions specifying the circumstances in which the phenomenon occurred (or is expected to occur). According to the classical position, to be cause or effect is nothing other than to be subject to a law of nature. However, this conception avoids con- fusing causation and determination by acknowledging, within the domain of phenomena covered by laws, and therefore within the domain of causa- tion, the existence of phenomena that are subject only to probabilistic but not deterministic laws. An important variant of the classical doctrine consists in interpreting a phenomenon’s falling under a law as resulting from a subjective act, groun- ded on laws which are themselves created, not discovered, by the human mind. In this framework, the causal relation is reduced to a relation whose existence is due to an act of judgment by a subject. Without scientists trying to put order into the multitude and diversity of phenomena, it would make no sense to speak of laws of nature or of causation. The strategy of reducing causation to falling under a law – in an instrumentalist interpretation of the latter concept – has led certain empiricist philosophers of science to an anti- realist theory of causation. According to Hume, causal judgments are grounded in the psychological mechanism that transforms observations of regularities into habits and expectations. For those contemporary philoso- phers of science who work in the tradition of logical empiricism – whose position we have above qualified as ‘‘classical’’ and of which Carl Hempel is an outstanding representative – causal judgments are justified by their inte- gration within the framework of the set of scientific judgments. This is a significant difference between Hume and Hempel: by being identified with falling under a law used in explanatory arguments, causation has retrieved its place in the scientific conception of the world. 3) The classical conception is opposed to a tradition which goes back more directly to Hume and which has in this century been defended by Carnap and Russell. Until today, many find it attractive for its simplicity. Causation is conceived of as a concept belonging exclusively to common- 2 Introduction sense and one that cannot possibly be replaced by any scientific concept: instead of being reducible to other scientifically definable notions – what the other two conceptions try to show – nothing, within a scientific world view, corresponds to the commonsense notion of causation. Therefore it should be eliminated from the scientific world view. An explanation is scientific only to the extent to which it is not causal. When Bertrand Russell presented this thesis in 1912, he took it for gran- ted that mathematical physics was the paradigmatic science and that the emergence of a unified science, which would follow the explanatory model of mathematical physics, was only a question of time. When all sciences have achieved maturity, scientific explanations will take the form of deduc- tions of measurable quantities from functional laws and quantitative descriptions of phenomena. Today we no longer feel it is legitimate to sup- pose that in the near future biology and psychology will follow the example of mathematical physics. Still, it was on the basis of this conviction that Russell had hoped to ban the concept of causation from the whole of science. This book elaborates a realist conception of causation that is opposed to all three conceptions outlined above. This realist conception vindicates the obstinate presence of the concept of causation in explanation in all the dis- ciplines of applied science. For one can easily observe that not only in psy- chology and biology, but also in solid state physics and geology, many explanations are causal. This is the case each time one switches from the theoretical study of a model to the explanation of the real phenomena the model is supposed to describe. In this way I hope to show that causation can be considered as a scientifically respectable concept. This would allow us to reject Russell’s exaggerated parallel drawn between the concept of causation and the English monarchy, judging that both had been allowed to survive only because they had wrongly been thought to do no harm (cf. Russell 1912, p. 173). My conception of causation is based on a very simple hypothesis. It says that two events are related as cause and effect if and only if a determined amount of a conserved quantity is transferred from one event to another. It goes without saying that I have not developed this conception from scratch. The underlying intuition comes directly from commonsense. Only the ela- boration of details will show whether the hypothesis is fruitful. Attempting to analyse the concept of causation directly may well raise suspicion in some quarters. The long and complex history of philosophical work on this notion is well known. Contemporary philosophers of science who favour the third position sketched above think it is vain to place any hope in the theoretical fecundity of the concept of causation. Let me quote the harsh judgment Ernest Nagel has made on this topic: ‘‘It would be an ungrateful and pointless task to canvass even partially the variety of senses that have been attached to the word ‘cause’ – varying from the ancient legal associations of the word, through the popular conception of Introduction 3 causes as efficient agents, to the more sophisticated modern notions of cause as invariable functional dependence. The fact that the term has this wide spectrum of uses immediately rules out the possibility that there is just one correct and privileged explanation for it’’ (Nagel 1961, p. 73). I think it is possible to resist this judgment, although the challenge is intimidating. As Nagel rightly notes, the concept of causation has, throughout the history of Western philosophy, accumulated such a wealth of meaning that it may seem futile to try to ground modern naturalism on this concept. Nevertheless, it is still possible to construct a theory of causation which justifies the commonsense idea that causes are efficacious while also being compatible with contemporary science. I do agree, however, with Nagel’s implicit critique that the identification of every cause with an agent reveals naive anthropomorphism and I join Nagel in rejecting con- temporary anthropomorphic conceptions of causation according to which human actions are paradigmatic causes, while nature outside mankind contains causal relations only in a derivative sense (see Chapter 1). Simi- larly, the analysis of the concept of law proposed in Chapter 2 should, if it is correct, allow us to avoid the objection to have loaded the word ‘‘cause’’ with ‘‘ancient legal associations’’, by making the concept of causation dependent on the concept of law. We are not obliged to deny that scientific law is a conceptual descendant of human, or divine, law (see Ruby 1986; van Fraassen 1989, Introduction; Lloyd 1995), in order to claim that our reconstruction of the concept of law is free of such associations. An important risk, run in particular by the classical position, is to see causation in all scientific explanations. With an eye on its first appearance in history, I suggest calling this the ‘‘Aristotelian threat’’: to the extent that all explanations are considered as causal, the specificity of the concept of causa- tion diminishes until, in the end, the expression ‘‘causal explanation’’ becomes a pleonasm. A concept is useful only to the extent that it expresses a relevant distinction: it is instructive to call an explanation causal only to the extent that there are also non-causal explanations. It is only to the extent that there are nomic regularities which are nevertheless not causal that the statement that a given relation is causal has any content. Otherwise the concept of causation would be redundant and we would be well advised to abandon it. I do not intend to claim that Aristotle himself is a victim of this ‘‘Aris- totelian threat’’. For Aristotle, as for the pre-Socratic philosophers whom he criticizes in his philosophy of nature, the question whether nature should be explained in causal terms does not even arise. For them, the concept of causation rather provides the framework within which the debate on the best philosophical analysis of natural phenomena and their explanation takes place. In modern times, the concept of causation no longer plays the role of a general framework for all scientific explanations. What Aristotle considered as explanations in terms of formal causes are today often considered as paradigms of non-causal explanations. This is the case of mathematical 4 Introduction explanations in terms of relations between numbers (Aristotle, Physics II, 3, 194b28; trans. 1970, p. 28), and of logical explanations in terms of the logical relation between premises and conclusion (Aristotle, Physics II, 3, 195a16; trans. 1970, p. 29). However, might Aristotle’s causal methodology not still be valid for the explanation of natural phenomena? With respect to nature it might still look plausible today, that explaining is pointing out causes. Should we therefore let ourselves be guided by his principle? We think we understand a thing simpliciter (and not in the sophistic fashion accidentally) whenever we think we are aware both that the explanation [i.e. cause: the Greek word ‘‘aitia’’ can be translated both as ‘‘cause’’ and as ‘‘explanation’’] because of which the object is its explanation, and that it is not possible for this to be otherwise. It is clear, then, that to understand is something of this sort. (Aristotle, Posterior Analytics 71b10, trans. 1984, p. 115) My answer is that we should not accept Aristotle’s principle: it is possible to provide a scientific explanation, in the full sense of the word, of a natural phenomenon without designating its cause or even one of its causes. The concept of causation that will be worked out in this book is much more restrictive than Aristotle’s. On the basis of this new analysis, one can replace the Aristotelian methodological imperative by a ‘‘naturalistic’’ imperative that constitutes the background of all research on natural phenomena: for a phenomenon to be natural is to be capable of a scientific analysis. However, this analysis can more generally be nomological, i.e. grounded on laws of nature; it need not be causal. I shall proceed as follows. I begin by elaborating a concept of causation that is both very general with respect to its domain of application and very restrictive due to its physical character (Chapter 1). This analysis is intended to do justice to causation as it reigns in the physical world. The expression ‘‘physical world’’ is meant to include all of nature, particularly living beings, and among them, those capable of representing their environment. This naturalist project aims at reconciling requirements that might seem incom- patible: giving the word ‘‘cause’’ a sense that is both narrowly physical, yet broad enough to make room for the possibility of explaining causally, nat- ural phenomena that differ radically in type and complexity. It may turn out that the framework constituted by this concept of causality is simply not rich enough to even question whether or not the interactions of perception and behaviour living beings entertain with their environment are of the same type as physical causal interactions. Even this seemingly adverse result would be important to the extent that this tentative would at least indicate which aspects of these phenomena resist a causal analysis. Our analysis of the physical concept of causation quite naturally shows us how to proceed, as several other fundamental concepts used in this analysis must be clarified. We begin with the concept of a law of nature: In Chapter Introduction 5 1, physical causation is analysed as a relation of transference between events, but the concept of transference presupposes the concept of a law of conservation, for only what is conserved can be transferred. Chapter 2 will develop an account of laws according to which they are universal relations between property instances. We shall reject the idea of a dichotomy of two types of laws differing with respect to their logical structure: the laws of fundamental physics and the laws of the so-called special sciences. True, laws of the latter type do not apply to all natural objects but only to a restricted subset of them. But the analysis of the application relation of a law to concrete situations (Chapter 3) will show that the difference with respect to the size of domain of application does not entail any difference with respect to logical structure: this structure is the same for physical laws and for laws of special sciences. This step will turn out to be crucial for the construction of a theory up to the task we have set ourselves: construe the concept of causation so that it is general enough to be applied both to interactions between elementary par- ticles and to an animal’s or a human’s perception of its environment. The complex properties of a perceived scene have a causal impact on perceiving subjects in the same sense in which the properties of elementary particles have a causal impact on those particles’ interactions. Now, the dichotomy between the world of microphysics and the world of the macroscopic phe- nomena of our experience is often justified by the idea of a nomological difference: it is often taken for granted that only simple physical phenomena obey strict laws without exceptions, whereas more complex macroscopic phenomena are irreducibly irregular or hazardous. This is why the laws governing their evolution and interactions are taken to be non-strict or valid only ‘‘ceteris paribus’’. I shall, on the contrary, argue that there is no reason to consider all physical laws as strict, in the sense of allowing no exceptions, nor to attri- bute to the laws of the special sciences the property of holding only ‘‘ceteris paribus’’, i.e. under the proviso that, between one application of the law and the next, ‘‘all things (non explicitly specified by the law) remain the same’’. I shall challenge this dichotomy, on the basis of an analysis of the relation between a law and its applications to particular situations. This will lead us to a new interpretation both of the origin of exceptions and of what are traditionally taken to be ‘‘ceteris paribus laws’’. We shall in particular find that the hedging or ceteris paribus clauses are not part of the laws them- selves, but rather characterize the conditions under which a given law applies to a concrete situation. Most physical laws only apply if certain conditions are satisfied. Such physical laws are not exempt from exceptions any more than biological or psychological laws. This justifies abandoning the very idea of a ‘‘ceteris paribus law’’, to the extent that this notion gets its meaning from its opposition to the notion of a strict law that applies, without exceptions, always and everywhere. This analysis of laws has important implications, some of which will be developed in Chapter 4. 6 Introduction The earlier chapters in the book give us the conceptual instruments required to deal with a second aspect of causation, the notion of causal responsibility , grounded both on the concept of causation in terms of transference between events and on the concept of law. This is a richer and more powerful tool for the analysis of complex causal relations than the simpler causation between events. It allows us both to satisfy the require- ment that every natural phenomenon must in principle yield to a causal explanation, and to take into account the fact that not all laws of nature are microphysical laws. Indeed, objects interact causally not only in virtue of their microphysical properties but in virtue of all their nomological proper- ties. When I heat water in a closed bottle, the fact that the pressure reaches a certain point can be causally responsible for the fact that the bottle explodes. But pressure is an essentially macroscopic property: it makes no sense to attribute pressure to microphysical objects. The relation of causal responsibility brings into play all the nomological properties of a given event. Just like the expression ‘‘law of nature’’, the expression ‘‘causal responsi- bility’’ is borrowed from human law. Our reconstruction of the latter con- cept must be just as independent of the legal associations the expression brings to mind as the analysis of the former. The structure of the relation of causal responsibility is more complex than that of the causal relation between events. Its terms are not events themselves but rather facts . The fact that the cause–event has certain properties is causally responsible for the fact the effect–event possesses certain properties. Chapter 5 analyses the logical structure of the relation of causal responsibility, and the way it is grounded on the causal relation between events and on the instantiation of a law of nature in a particular situation. The account of causation in terms of transference and causal responsibility is compared to the traditional ‘‘nomological theory of causation’’ which reduces causation to the instan- tiation of a law, or to explanation in terms of such a law. We clarify the relation between causation linking events and causal responsibility linking facts by studying the patterns of inference between statements of both types, and address some important objections against the idea that both facts and events can be terms, or ‘‘relata’’, of causation. Chapter 6 offers an analysis of the nature of the properties apt to be efficacious in causation. It develops the idea that the intrinsic natural properties of events are the sources of their efficacy, where a property is ‘‘intrinsic’’ if it is entirely determined at the space–time location occupied by the event possessing it. A property is ‘‘natural’’ if it participates in a law of nature. It is also argued that the distinction between causal and non-causal is better construed as applying not to laws, but rather to the particular situations to which laws apply. We criticize one important proposal to understand causation in terms of a particular species of causal laws: Too- ley’s version of the thesis that causes, by instantiating causal laws, raise the probability of their effects. Introduction 7 Finally, Chapter 7 explores the fecundity of the concept of causal responsibility in clarifying some important problems encountered by the analysis of causal phenomena and statements reporting them. Among other things, we shall enquire whether there is something true in the statement that the cause of Socrates’ death is the fact that he drank hemlock at dawn , although the emphasis lies on the wrong expression, whether it can be true both that an ice cube in a glass cools the water around it and that the water warms the ice cube up, whether someone’s not falling can be a cause of her not dying, whether switching out the light causes the light’s being out, and whether a massage can be a cause of death even if that massage is in fact beneficial and slows down the patient’s dying. It is hoped that our reconstruction of the concept of causation defends it successfully against Russell’s verdict that no rational core can be extracted from this concept, purifying it from the confusion and anthropomorphist connotations it possesses in commonsense. We shall see that the concept of causation can be given a clear sense, which justifies its broad use both in commonsense and scientific explanations. Causal explanation is an impor- tant specific kind of scientific explanation. But there are also non-causal scientific explanations. Therefore, causation and explanation remain distinct though related concepts. 8 Introduction 1 What is a causal relation? The concept of causation plays a central role in many incompatible philoso- phical theories of nature. The aim of this chapter is to reconstruct the con- cept of causation without, however, pretending to do justice to the wealth of aspects the concept has accumulated due to its paramount importance. I propose to tackle the concept of causation by taking up David Hume’s analysis. This choice is promising although it may at first seem paradoxical. Hume’s analysis ends up throwing into doubt the idea that causal relations are objective . My aim is to resist this conclusion which eliminates causation as an objective relation, and we approach this through a dialogue with Hume. Of the three necessary conditions posited by Hume and taken to be together sufficient for something to be perceived as a cause, to wit, con- tiguity, precedence and regular repetition, we shall begin with the last, namely regular repetition. It is indeed this condition that leads Hume to conceive of the causal relation as having a subjective rather than an objec- tive ground. We shall examine two alternative conceptions to Hume’s, which I call respectively singularist and anthropocentric. Both try to avoid Hume’s conclusion although they stick to the main line of his analysis. Having found them unsatisfactory, we shall propose our own analysis of the objec- tive causal relation: causation is basically grounded on the transference of an amount of a conserved quantity. My proposal for the reduction of causation to a physical relation can be expressed by the following equivalence: (S) Two events c and e are related as cause and effect if and only if there is at least one physical quantity P , subject to a conservation law, exemplified in c and e , of which a determinate quantity is transferred between c and e Condition (S) is intended to be necessary and sufficient for the existence of a causal relation between events c and e . It offers a reduction of the concept of causal relation albeit an a posteriori one. The thesis is that the causal relation is reducible in point of fact to the relation of transference of a conserved quantity. Still, this is no purely conceptual identity: it could be refuted by the empirical discovery of causal relations without any transfer- ence of conserved quantities. We shall bring out the most important consequences of this proposition by comparing it to Hume’s analysis with respect to the following questions: Within a theory of causation that claims to be compatible with con- temporary science, is it still correct to require that the cause be contiguous with the effect or should we rather admit the possibility of causal action at a distance? Similarly, is it still correct to require that the effect follow the cause in time or should we rather admit that ‘‘backwards’’ causation is a possibility that reasoning alone cannot and should not exclude? Further- more, does our analysis allow us to retain Hume’s argument against the possibility of simultaneous causation? In order to answer these questions we shall have to analyse more precisely the nature of the terms of the causal relation. To this end, we shall conceive of an event as the content of a space–time zone (in other words, as of what fills a space–time zone). This immediately raises the question whether we can agree with Hume when he considers that causes and effects are objects . Next, we shall answer an objection raising a doubt about the persistence of the transferred quantities through time. Our reply will lead to a clarification of the dependency of the concept of causation on that of a law of nature. At the same time, the fact that the concept of causation introduced here depends only on the laws of conservation , and the fact that our analysis does not reduce the causal rela- tion to the instance of a law, show that our analysis does not belong to the category of nomological theories of causation, which constitute the most important contemporary paradigm of analyses of causation. However, we shall see later that the analysis given by (S) is not sufficient. The concept of physical causation which is the object of the reduction state- ment (S) is fundamental to be sure, but it is not the only aspect of causation. In the context of causal explanation it is necessary to mention certain properties of cause and effect events. To be able to do that we shall intro- duce the concept of causal responsibility . It brings into play the fact that the cause event c possesses a property F that is efficacious in bringing about the effect, and the fact that the effect e possesses a property G by virtue of the influence of the cause. However, the relation of causal responsibility depends on the relation of causation between events c and e , which will be worked out in this chapter and which is reducible according to (S). In this sense, causation between events, which we shall analyse presently, is most fundamental. Causal responsibility depends furthermore on a nomological relation between a property F of the cause c and a property G of the effect e Causal responsibility does not directly relate the events c and e , but rather certain facts involving these events. I shall defend the thesis that causal statements expressing the concept of causal responsibility have the following logical structure: 1 ( O ) C F ( Fc, Ge ). 10 What is a causal relation? (O) says that the fact that c is F (or the fact that c possesses property F , in short ‘‘ Fc ’’) is causally responsible for the fact that e is G (in short, ‘‘ Ge ’’). I shall show that (O) is equivalent to: ð O 0 Þ C E ð c ; e Þ ^ Fc ^ Ge ^ L ð F ; G Þ : ( O’ ) says first that the events c and e are causally linked according to (S) ( in short, ‘‘ C E ( c,e )’’), second, that the events c and e possess respectively the properties F and G (in short, Fc and Ge ) and third, that c ’s possession of F brings about, by virtue of a law of nature linking F to G (in short, L(F,G )), e ’s possession of G In most causal statements playing an explanatory role the concept of causation that is immediately relevant is the concept expressed by (O). However, as ( O’ ) shows, (O) presupposes (S), which is therefore the core of the causal relation. The analysis of the concept of causation by Hume What makes one object cause another? David Hume has raised a powerful challenge to the idea that causation has an objective grounding. Any ana- lysis of this concept that aims at giving an objective value to causal state- ments must answer to Hume’s objections. The first lesson Hume teaches us about causation is that it is not a quality or property of objects 2 but rather a relation (Hume 1955, p. 75): this is shown by the fact that everything that exists can be either cause or effect; 3 to judge whether something exists, there is no other means than to check its capacity of entering into causal relations with other things. No particular property can therefore be characteristic of all causes or of all effects. On the contrary, what makes a cause of something is a certain relation in which it stands to another thing, its effect. The relevant question is: what is the nature of this relation? Hume’s research on the concept of causation is focussed on the idea of causation as it exists in human understanding. One of the fundamental theses of Hume’s radical empiricism is that we can achieve objective knowledge neither of the external world nor of ourselves. Thinking that the objects of perception possess an independent identity allowing them to subsist through time and to exist without us perceiving them is an illusion (Hume 1955, p. 16, p. 209–218). The same is true of the identity of the self (Hume 1955, p. 251–263). From Hume’s point of view it would therefore be vain to try to find an objective grounding of the relation between cause and effect. When he says that the object of his analysis is the idea of causation (Hume 1955, p. 74), the word ‘‘idea’’ has a psychological sense. For Hume, every idea has its origin in a perception; indeed Hume says literally that ideas are a kind of perception, distinguished from impressions only by the fact that they are What is a causal relation? 11 less strong and lively. Ideas are ‘‘faint images’’ (Hume 1955, p. 1) of sense impressions, and these impressions give them all the content they have. I do not share Hume’s radical empiricist starting point. The most important difference between the framework adopted here and Hume’s is that I admit the existence of all those objects and properties which are described by scientific theories, whereas Hume restricts the supposition of existence to perceptions only (Hume 1955, p. 66–68, p. 210). Nevertheless, Hume’s analysis of the idea of causation constitutes a useful starting point for our own research on causation, precisely to the extent to which it con- tains a direct 4 and radical attack against the thesis that causal judgments have objective content. Hume’s starts by pointing out two aspects of the causal relation, namely the contiguity of the cause and the effect and the precedence in time of the cause with respect to the effect (Hume 1955, p. 75–76). He takes the first property of contiguity between cause and effect as obvious and requiring no defence; however, he tries to provide a proof of the second property of pre- cedence of the cause with respect to the effect. His reasoning is this. Let us admit by hypothesis, says Hume, that a cause can be strictly simultaneous with its effect. From this hypothesis we can infer that a cause that produces its effect only after some delay, is not really the (only) cause of that effect, for the fact that there is a delay during which the cause is already present, but not the effect, shows that the cause was not sufficient for the effect. Necessarily there is something else than the supposed cause which comes into play after the delay, and which brings about the effect at that moment. To admit the possibility that the effect comes about at a moment separated by a given delay, from the impact of the cause, without any new cause acting at that moment, would, according to Hume, be tantamount to rejecting the universal validity of the principle of sufficient reason. This principle is however ‘‘an establish’d maxim both in natural and moral phi- losophy’’ (Hume 1955, p. 76): it is unacceptable to suppose that the cause exists for a certain amount of time without producing its effect, but that it produces the effect at some later moment without any further intervention from outside. 5 Contiguity and succession in time are not the only and, above all, not the most important ingredients making up the notion of causation. These properties are necessary but not sufficient because ‘‘an object may be con- tiguous and prior to another, without being consider’d as its cause’’ (Hume 1955, p. 77). And ‘‘contiguity and succession are not sufficient to make us pronounce any two objects to be cause and effect’’ (Hume 1955, p. 87). If, for example, a leaf falls on the back of a cat watching out for a mouse, and if the cat leaps precisely when the leaf touches it, it is nevertheless con- ceivable that the leaf has no causal impact on triggering the cat’s behaviour. We can very well imagine that the cause that makes the cat leap is its per- ceiving a movement at the spot where it expects the mouse to hide. In this situation, the event of the leaf’s touching the cat is contiguous and imme- 12 What is a causal relation?