Hardcover
-
PICK UP IN STORECheck Availability at Nearby Stores
Available within 2 business hours
Related collections and offers
Overview
Ben-Menahem's approach reveals that causation is just as relevant to explaining why certain events fail to occur as it is to explaining events that do occur. She investigates the conceptual differences between, and interrelations of, members of the causal family, thereby clarifying problems at the heart of the philosophy of science. Ben-Menahem argues that the distinction between determinism and stability is pertinent to the philosophy of history and the foundations of statistical mechanics, and that the interplay of determinism and locality is crucial for understanding quantum mechanics. Providing historical perspective, she traces the causal constraints of contemporary science to traditional intuitions about causation, and demonstrates how the teleological appearance of some constraints is explained away in current scientific theories such as quantum mechanics.
Causation in Science represents a bold challenge to both causal eliminativism and causal reductionism—the notions that causation has no place in science and that higher-level causal claims are reducible to the causal claims of fundamental physics.
Product Details
| ISBN-13: | 9780691174938 |
|---|---|
| Publisher: | Princeton University Press |
| Publication date: | 06/12/2018 |
| Pages: | 224 |
| Product dimensions: | 6.12(w) x 9.25(h) x (d) |
About the Author
Read an Excerpt
CHAPTER 1
From Causal Relations to Causal Constraints
THIS BOOK EXAMINES the family of causal notions and causal constraints employed in fundamental science, and analyzes some of the conceptual relations between them. It argues that the concepts of determinism, locality, stability, and symmetry, as well as conservation laws and variation principles, constitute a complex web of constraints that circumscribe the causal structure of our world. It argues, further, that mapping out the various links between these causal constraints is an indispensable, though neglected, aspect of the project of understanding causation. The book thus seeks to shift our attention from causal relations between individual events (or properties of events) to the more general causal constraints found in science, and the relations between them. In so doing, it does not purport to replace causal relations with causal constraints in every context, but rather to suggest a broader perspective on causation and a new research program for the philosophy of causation.
Philosophical analysis of complex concepts usually begins with definitions. The exploration of causality is no exception. Enormous effort has been devoted to formulating the "right" definition of causation and defeating rival definitions. Regularity theories, counterfactual analyses, interventionist/manipulation accounts, probabilistic theories, transmission accounts, and explanation-oriented accounts are regular contenders in this ongoing competition, which comprises much of the literature on causation. Each of these definitions captures important characteristics of the notion of cause, but also raises difficulties that advocates of competing conceptions are quick to seize on. Needless to say, the contending accounts are never conclusively defeated by such difficulties; their advocates find ways to patch them as necessary. Nevertheless, the "attack/patch" cycle has an adverse cumulative impact as the difficulties pile up. More generally, concern over definitions and their weaknesses has led philosophers to devote a great deal of attention to intriguing yet marginal "hard cases." Adding "epicycles" may salvage a threatened definition of causation, but sheds little light on the ways in which causal notions are actually used, and in particular, on how they're used in scientific contexts. Science seeks to identify constraints that distinguish what may happen, or is bound to happen, from what is excluded from happening. Hence the notion of causal constraint, which is broader than the notion of cause, is at the center of my analysis. Even when searching for individual cause-events (effect-events), awareness of the framework of constraints that these individual events must satisfy is vital. And because there is no single causal constraint that is operative in science, but rather several different constraints, a study of the relationships between the various constraints is called for. I do not take the notion of cause to be reducible to any one of the constraints in question, or to a particular combination of them. The evolution of causal constraints — and thus of our understanding of causation — is as open-ended as the evolution of science in general. The difference between the "causal constraints" approach to causation, and the traditional approach, will become sharper as the book proceeds.
I will not review the current philosophical accounts of causation, and the difficulties they pose, in any detail. The Oxford Handbook of Causation (Beebee, Hitchcock, and Menzies 2009) gives an admirably balanced account of the literature. But it will be useful to briefly identify the main contending proposals, and the key issues they bring to the fore, as these key issues underscore my claim that it is time for a different approach to causation.
• Regularity theories, also known as Humean theories, reduce causation to lawful behavior and therefore assimilate causation to determinism. Nonetheless, it is not laws that constitute causes, but the events that fall under them. Roughly, Hume's definition of causation set down three conditions: contiguity in space, succession, and constant conjunction of the same event types. The succession condition can in turn be divided into a condition of contiguity in time, and an asymmetry condition to the effect that the cause must precede the effect. Physics has had to discard the spatial and temporal contiguity requirement due to its emptiness in continuous spacetime, and thus in mathematical theories that involve a continuum (such as theories employing differential and partial differential equations). But the remaining conditions are independent of the contiguity requirement and permit extension of the cause–effect relation to distant events: any event c that is regularly or lawfully followed by an event e can be considered the cause of e.
The connection Hume established between causation and lawful behavior has had a lasting impact on the philosophy of science. Yet regularity theories, though still among the leading accounts of causation, have also garnered much criticism. Objections target the very connection between regularity and causation, denying either the necessity or the sufficiency of regularity for causation. The claim that regularity is unnecessary for causation entails the acceptability of singular — that is, nonrepeatable — cause–effect relations. Although I do not deny the feasibility of such singular causal relations, their existence is peripheral to my primary concern. When focusing on science rather than, say, human actions, it is largely possible to remain within the boundaries of lawful causation. The converse claim — that regularity is insufficient for causation — is backed by several arguments. For one thing, regularities, even when they appear to be lawlike, may reflect accidental rather than causal connections. For another, regularities as such lack the asymmetry typical of causal relations. There are also examples like the tower and its shadow, which, despite the nonaccidental nature of the regularity in question, speak against the identification of lawful regularities with causation. The height of the tower and the length of the shadow are correlated by laws, but we see the shadow's length as caused (and explained) by the tower's height, and not the other way around. The example suggests a distinction between causal and noncausal regularities, where only the former are truly explanatory. The concern that regularity falls short of causation often motivates the requirement that causal connections, unlike mere regularities, must be embodied in concrete mechanisms. While the distinction between lawful and accidental regularities is crucial for science (and in that sense the critique of regularity theories is warranted), a connection between causation and concrete mechanisms is often lacking. When it comes to very general causal constraints, such as the relativistic limit on the speed of interaction, the search for an underlying mechanism is futile. The distinction between laws and mere regularities can be supplemented by a hierarchy that differentiates lower-order constraints, constraints on facts, from higher-order constraints, constraints on laws. Although constraints on laws do not fit the Humean scheme, they should be seen as causal constraints; see chapters 5 and 6. But even if we were able to fend off all the standard objections to the regularity account of causation, it would, from the perspective of this book, remain inadequate. Except for determinism, the constraints that make up the family of causal concepts (henceforth, causal family) cannot be expressed in the language of regular succession of individual events.
• The counterfactual account championed by David Lewis (1973) analyzes the causal relation between event c (the cause) and event e (the effect) in terms of the counterfactual "had c not occurred, e would not have occurred." In addition to the formidable problem of analyzing counterfactuals, and the metaphysical assumptions this analysis mandates, the counterfactual account faces the challenge of overdetermination. Recall the standard example (a typical hard, though marginal, case) of two desert travelers who set out, separately, to murder a third, one pouring poison into the victim's water flask, the other puncturing it, so that the victim will die of either poisoning or dehydration. The counterfactual conditional "Had x not punctured the water flask, z would not have died," fails to identify the cause, for (assuming the cause of death to be dehydration) it would not be true that had the water flask not been punctured, the victim would not have died. I should stress that I do not adduce these problems to critique counterfactual considerations in general, but to critique their adequacy as definitive criteria of causation. I take counterfactuals to be indispensable for reasoning, and will use them extensively in chapter 2. But counterfactuals are also used in contexts that have nothing to do with causation: "If I were you, I would accept the offer"; "This triangle (pointing, for example, to a triangle with sides 3 cm, 4 cm, and 6 cm in length) is not right-angled — if it were, it would satisfy the Pythagorean theorem." Because of their broader applicability, counterfactuals cannot be relied on to pick out causal relations.
• Process accounts of causation focus on prolonged progressions rather than instantaneous events, and tie causation to a particular process, such as energy transfer from one system or state to another (Fair 1979; Salmon 1984; Dowe 1992, 2000). This approach can handle the case of Jane's happiness being due to John's response to her message but has difficulty with the seemingly parallel case of Jane's unhappiness being due to John's failure to respond. In other words, the process approach is unable to account for failures and omissions.
• Probabilistic accounts of causation (Suppes 1970; Kvart 1986) have the great advantage of extending causal discourse to non-deterministic contexts. From the probabilistic perspective, a cause need only raise the probability of the effect-event's occurring; there is no need for it to determine or induce its occurrence. On the other hand, however, probabilistic accounts engender paradoxes of their own — namely, probabilistic correlations that do not seem to reflect causal relations, or events that seem entitled, from the intuitive point of view, to be considered causes of certain "effect" events, yet appear to lower, rather than raise, the probability of their occurrence.
• Interventionist or manipulation accounts of causation (also known as agency accounts) have been in vogue for several decades (Menzies and Price 1993; Woodward 2003; Hitchcock 2007). Here, causes are identified as the factors that, when manipulated, change the result that would have ensued in the absence of that intervention. In identifying causes as necessary conditions of effects, the manipulation account has much in common with the counterfactual account. In focusing on human intervention, however, it has given rise to the objections of anthropocentrism, and — since intervention is actually a causal concept — circularity. One merit of the interventionist criterion is that it distinguishes causal relations from correlations that are merely accidental. Together with insights from the counterfactual and probabilistic accounts, it has stimulated elegant work on causal networks and their graphic representations (Pearl 2000). Causal networks are highly valuable in a variety of practical contexts — legal, medical, economic, policymaking, and so on — where distinguishing effective from ineffective intervention is essential. Nevertheless, the manipulation account, I contend, is far too limited to provide a comprehensive understanding of causal processes in the world. This point requires elaboration.
Consider, first, the contrast between the regularity account and the manipulation account. In a standard case, where natural laws and initial conditions determine a certain result (a certain chemical reaction, say), we can often manipulate the initial conditions, but not the laws. On the manipulation account, therefore, the initial conditions constitute the only relevant factor for a causal account of the process. By contrast, the regularity theorist ascribes a crucial role to the laws — to what we cannot manipulate. The initial conditions can be considered causes only because they are invariably followed by the same trajectories, that is, they are considered causes only because of the existence of laws that are non-manipulable constraints. And although (for the reasons mentioned above) I do not consider the regularity account, as it stands, fully satisfactory, there is something fundamentally correct about the intuition that constraints which we cannot manipulate are an inherent feature of causal descriptions and explanations. Furthermore, invoking laws is by no means the only context in which we ascribe a causal role to the non-manipulable. The electric charge of the electron, which though non-manipulable is causally efficacious, is a case in point. The constraints we will examine in this book are typically not subject to human intervention, but they enable us to grasp and predict the dynamics of unfolding events, and exclude infinitely many alternatives to what actually transpires. In this sense, they also constrain our interventions — manipulations and interventions are always carried out within a general framework of constraints. The manipulation account of causation thus already presupposes the preconditions of possible manipulation, preconditions that a complete causal story must take into account and render explicit.
In view of the difficulties that beset each of these accounts, I have relinquished the search for the definition of causation, instead taking causation to be a cluster concept comprising a broad range of causal notions. My primary focus will be the causal notions employed in science, which include the much-discussed notion of determinism, but also notions such as stability and locality, which philosophers tend to neglect. In turning to pluralism, I am not, of course, alone: a pluralistic attitude to causation has been advocated (sometimes only in passing) by Reichenbach (1956); Anscombe (1971); Cartwright (1983, 2004); and Godfrey-Smith (2009), not to mention Aristotle, who introduces his presentation of the four causal categories by noting that the number of causes matches that of the things "comprehended under the question 'why'" (Physics II, 198a15–16). Skyrms (1984) speaks aptly of an "amiable jumble" of causal notions that can, but need not, work together.
But the declaration of pluralism is only a starting point. To do justice to causation, recognition of the variety of causal notions must be augmented with detailed investigation of their usage, especially in fundamental science. Science, like daily life, presents us with a spectrum of causal notions and constraints. These scientific constraints are often in some way "descended" from the more intuitive constraints of daily life, though differing from them significantly in precision and scope. The invocation of causal notions in scientific contexts is particularly noteworthy in view of arguments that challenge causation's place in fundamental science, or relegate it to "folk science" (Russell 1913; Norton 2007). I will return to the Russell-Norton position further on, but for now, let me point out that the argument I make in this book is that as soon as we shift our attention from the familiar paradigms of breaking a glass or tickling a baby to determinism, locality, stability, and conservation laws, it becomes evident that causal notions permeate fundamental science.
Critique of causal discourse also comes from another direction. These critics concede causation's place in fundamental physics, but deny it elsewhere, arguing that higher-level realms, such as the realms of biological, mental, and social events, are causally inert. Alleged causal relations on these levels, or between higher-level events and events at the fundamental level, are (according to this view) all reducible to the causal relations of physics. This argument is rebutted in chapter 7.
Our first encounter with causal notions does not come from fundamental science. We acquire the notion of cause early in life in relatively simple situations, such as sucking, pulling, pushing, holding, biting, and so on. These actions involve several elements of the cluster concept of cause, so that many of the features explicated in the aforementioned accounts of causation are operative. A child pulls the string of a toy that plays a tune. The interaction is both regular — whenever the string is pulled the tune plays — and local — no action at a distance; it instantiates both the manipulative and the counterfactual conditional accounts of causation — had the string not been pulled, no tune would have been played; and there is no creation ex nihilo — the power was provided by tugging the string, energy was transferred from the child's hand to the string, and from the string to the musical instrument. Infants are unable to articulate these concepts, but may acquire a rudimentary grasp of some of them, and learn to associate the features in question with each other, so as to form a more complex sense of causation. Later on, with exposure to less paradigmatic causal nexuses, and to science, intuitive ideas give way to more explicit notions, occasionally undoing the automatic associations, or establishing new ones. A child who is at one stage prone to "magical" thinking, for instance, believing that merely wishing someone ill suffices to actually bring about harm, will, with age and experience, likely revise this conception.
(Continues…)
Excerpted from "Causation in Science"
by .
Copyright © 2018 Princeton University Press.
Excerpted by permission of PRINCETON UNIVERSITY PRESS.
All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.
Table of Contents
Preface ix
Acknowledgments xiii
1 From Causal Relations to Causal Constraints 1
2 Determinism and Stability 35
3 Determinism and Stability in Physics 61
4 Determinism and Locality 82
5 Symmetries and Conservation Laws 105
6 The Principle of Least Action: From Teleology to Causality 134
7 Causation and Reduction 158
References 183
Index 193
What People are Saying About This
"Causation in Science puts forward a new way of thinking about causality and is a shining example of intelligent and sophisticated philosophical analysis. I highly recommend this book."—Margaret Morrison, author of Reconstructing Reality: Models, Mathematics, and Simulations"An important and novel scholarly contribution. Against those who would relegate causality to ‘folk science,' Ben-Menahem convincingly argues that causal notions and their associated constraints permeate science, moreover that this is true both in fundamental sciences such as physics as well as in the higher-level sciences."—Michael Cuffaro, Rotman Institute of Philosophy, University of Western Ontario