Introduction: The Contours of Contemporary Free Will Debates
Abstract and Keywords
This article describes the contours of contemporary free will debates and provides an overview of the topics in this volume. It considers the following issues: free will and conflicting views about persons; the determinist question and modern science; the compatibility question and arguments for incompatibilism; classical compatibilism; moral responsibility and alternative possibilities; new compatibilist approaches to freedom and responsibility; libertarian or incompatibilist theories of free will; hard determinism, successor views, and other nonstandard theories; neuroscience and free will; and theological determinism and fatalism.
There is a disputation [that will continue] till mankind is raised from the dead, between the Necessitarians and the partisans of Free Will
Jalalu'ddin Rumi, twelfth-century Persian poet
The problem of free will and necessity (or determinism) is “perhaps the most voluminously debated of all philosophical problems,” according to a recent history of philosophy.1 This situation has not changed at the end of the twentieth century and the beginning of a new millennium. Indeed, debates about free will have become more voluminous in the past century, especially in the latter half of it—so much so that it has become difficult to keep up with the latest developments. This handbook was compiled as a remedy in the form of a sourcebook or guide to current work on free will and related subjects for those who wish to keep up with the latest research.
The focus of the volume is on writings of the past thirty to forty years, an era of reborn interest in traditional issues regarding free will in the context of new developments in the sciences, philosophy, and humanistic studies. While references (p. 4) are frequent throughout this volume to major thinkers of the past who have discussed free will, the emphasis is on recent research.2 Many of the writers of the following essays are long-time contributors to contemporary debates about free will; others are younger scholars who are beginning to make significant contributions. By surveying and evaluating recent writings, the hope is that their essays will serve as a guide to the latest work and a resource for future research.
What is often called “the free will issue” or “the problem of free will,” when viewed in historical perspective, is related to a cluster of philosophical issues—all of them to be dealt with to some degree in this volume.3 These include issues about (1) moral agency and responsibility, dignity, desert, accountability, and blameworthiness in ethics; (2) the nature and limits of human freedom, autonomy, coercion, and control in social and political theory; issues about (3) compulsion, addiction, self-control, self-deception, and weakness of will in philosophical psychology; (4) criminal liability, responsibility, and punishment in legal theory; (5) the relation of mind to body, consciousness, the nature of action,4 and personhood in the philosophy of mind and the cognitive and neurosciences; (6) the nature of rationality and rational choice in philosophy and social theory; (7) questions about divine foreknowledge, predestination, evil, and human freedom in theology and philosophy of religion; and (8) general metaphysical issues about necessity and possibility, determinism, time and chance, quantum reality, laws of nature, causation, and explanation in philosophy and the sciences. Obviously, this volume does not discuss every aspect of these complex issues, but it does attempt to show how contemporary debates about free will are related to them.
In the remainder of this introduction, I describe the contours of contemporary free will debates, placing them—and the essays to follow—in historical and dialectical perspective.
1. Free Will and Conflicting Views About Persons
The problem of free will arises when humans reach a certain higher stage of self-consciousness about how profoundly the world may influence their behavior in ways of which they were unaware (Kane 1996: 95–6). Various authors have described this stage of self-consciousness as the recognition of a conflict between two perspectives we may have on ourselves and our place in the universe (p. 5) (P. F. Strawson 1962; Nagel 1986; Bok 1998; Blackburn 1999). From a personal or practical standpoint, we see ourselves as free agents capable of influencing the world in various ways. Open alternatives seem to lie before us. We reason or deliberate among them and choose. We feel it is “up to us” what we choose and how we act; and this means that we could have chosen or acted otherwise—for, as Aristotle succinctly put it, “[W]hen acting is up to us, so is not acting” (1915b: 1113b6). This “up to us-ness” also suggests that the origins or sources of our actions are in us and not in something else over which we have no control—whether that something else is fate or God, the laws of nature, birth or upbringing, or other humans.5
These two features of the personal or practical standpoint are pivotal to what has traditionally been called free will: we believe we have free will when (a) it is “up to us” what we choose from an array of alternative possibilities and (b) the origin or source of our choices and actions is in us and not in anyone or anything else over which we have no control. Because of these features free will is frequently associated with other valued notions such as moral responsibility, autonomy, genuine creativity, self-control, personal worth or dignity, and genuine desert for our deeds or accomplishments (Anglin 1990; Kane 1996: ch. 6). These two features of free will also lie behind various reactive attitudes that we naturally assume toward our behavior and that of others from a personal standpoint (P. F. Strawson 1962). Gratitude, resentment, admiration, indignation, and other such reactive attitudes seem to depend upon the assumption that the acts for which we feel grateful, resentful, or admiring originated in the persons to whom we direct these attitudes. We believe that it was up to them whether they performed those acts or not (cf. Nathan: 1992: 46).
But something happens to this familiar picture of ourselves and other persons when we view ourselves from various impersonal, objective or theoretical perspectives (Nagel 1986: 110). Perhaps we only seem to “move ourselves” in a primordial way when in fact our actions are caused by physical forces over which we have no control (Trusted 1984). Perhaps our choices from among alternative possibilities are determined by unconscious motives and other psychological springs of action of which we are unaware (Hospers 1958). These thoughts take many forms in human history, but in all their forms they threaten our self-image and cause a corresponding crisis in human thinking (Farrer 1967, Kenny 1978). Such is the case when we learn that much of our character and behavior is influenced by heredity or environment (Felt 1994), or that our thoughts and behavior can be covertly influenced by social conditioning (Waller 1990; Double 1991), or by subtle chemical imbalances of the neurotransmitters or hormones of our brains or bodies.
Free will becomes an issue when, by reflections such as these, humans realize how profoundly the world may influence them in ways previously unknown. (p. 6) The advent of doctrines of determinism or necessity in the history of ideas is an indication that this higher stage of awareness has been reached—which accounts for the importance of such doctrines in the long history of debates about free will (Woody 1998).6 Determinist or necessitarian threats to free will have taken many historical forms—fatalist, theological, physical or scientific, psychological, and logical—all of which are discussed in this volume. But a core notion runs through all these forms of determinism, which explains why these doctrines appear to threaten free will. Any event is determined, according to this core notion, just in case there are conditions (such as the decrees of fate, the foreordaining acts of God, antecedent physical causes plus laws of nature) whose joint occurrence is (logically) sufficient for the occurrence of the event: it must be the case that if these determining conditions jointly obtain, the determined event occurs. Determination is thus a kind of conditional necessity that can be described in various ways. In the language of modal logicians, the determined event occurs in every logically possible world in which the determining conditions (e.g., antecedent physical causes plus laws of nature) obtain. In more familiar terms, the occurrence of the determined event is inevitable, given these determining conditions.
Historical doctrines of determinism refer to different kinds of determining conditions, but they all imply that every event (including every human choice or action) is determined in this general sense.7 One can understand as a consequence why such doctrines pose a threat to free will. If one or another form of determinism were true, it seems that it would not be (a) “up to us” what we chose from an array of alternative possibilities, since only one alternative would be possible; and it seems that (b) the origin or source of our choices and actions would not be “in us” but in conditions, such as the decrees of fate, the foreordaining acts of God, or antecedent causes and laws, over which we had no control. But these apparent conflicts can only be the first word on a subject as difficult as this one. Many philosophers, especially in modern times, have argued that, despite intuitions to the contrary, determinism (in all of its guises) poses no threat to free will, or at least to any free will “worth wanting,” as Daniel Dennett (1984) has put it.8
As a consequence, debates about free will in the modern era (since the seventeenth century) have been dominated by two questions, not one—the “Determinist Question”: “Is determinism true?” and the “Compatibility Question”: “Is free will compatible (or incompatible) with determinism?” Answers to these questions give rise to two of the major divisions in contemporary free will debates, that between determinists and indeterminists, on the one hand, and that between compatibilists and incompatibilists, on the other. Let us look at the two questions in turn.
(p. 7) 2. The Determinist Question and Modern Science
One may legitimately wonder why worries about determinism persist at all in the twentieth-first century, when the physical sciences—once the stronghold of determinist thinking—seem to have turned away from determinism. Modern quantum physics, according to its usual interpretations, has introduced indeterminism into the physical world, giving us a more sophisticated version of the Epicurean chance “swerve of the atoms” than the ancient philosophers could ever have conceived. We have come a long way since the beginning of the nineteenth century, when Pierre Simon, marquis de Laplace, could claim that discoveries in mechanics and astronomy unified by Newton's theory of gravitation have made it possible
Twentieth-century physics threatened this Laplacean or Newtonian determinist vision in several related ways. Quantum theory, according to its usual interpretations, denies that elementary particles composing the “system of the world” have exact positions and momenta that could be simultaneously known by any such intelligence (the Heisenberg Uncertainty Principle); and it implies that much of the behavior of elementary particles, from quantum jumps in atoms to radioactive decay, is not precisely predictable and can be explained only by probabilistic, not deterministic, laws. Moreover, the uncertainty and indeterminacy of the quantum world, according to the orthodox view of it, is not merely due to our limitations as knowers but to the nature of the physical world itself.
to comprehend in the same analytical expressions the past and future states of the system of the world.… Given for an instant an intelligence which could comprehend all the forces by which nature is animated and the respective situation of the beings who compose it—an intelligence sufficiently vast to submit these data to analysis—it would embrace in the same formula the movements of the greatest bodies of the universe and those of the lightest atom; for it nothing would be uncertain and the future, as the past, would be present to its eyes. (1951: 3–4)
In the light of these indeterministic developments of twentieth-century physics, one may wonder why physical or natural determinism continues to be regarded as a serious threat to free will, as evident in many essays of the volume.9 Indeed, it is an important fact about the intellectual history of the twentieth century that, while universal determinism has been in retreat in the physical sciences, determinist (and compatibilist) views of human behavior have been thriving (while antideterminist and incompatibilist views of free will continue to be on the defensive).
What accounts for these apparently paradoxical trends? There are four reasons, I believe, why indeterministic developments in modern physics have not (p. 8) disposed of determinist threats to free will, all of them on display in this volume. First, there has been, and continues to be, considerable debate about the conceptual foundations of quantum physics and much disagreement about how it is to be interpreted. Orthodox interpretations of quantum phenomena are indeterministic, but they have not gone unchallenged. These issues about determinism and indeterminism in modern physics and related sciences, and their implications for the free will problem, are the subject of two essays of this volume, by David Hodgson and Robert Bishop.
Second, contemporary determinists about free will often concede that if modern physics is correct, the behavior of elementary particles is not always determined (see Honderich 1988; Weatherford 1991; Pereboom 1995). Yet they insist that this has little bearing on how we should think about human behavior, since quantum indeterminacy is comparatively negligible in macroscopic physical systems as large as the human brain and body. Since physical systems involving many particles and higher energies tend to be regular and predictable in their behavior for the most part, according to quantum physics itself, modern determinists argue that we can continue to regard human behavior as determined at the macroscopic level “for all practical purposes” (or “near-determined,” as one of them has put it10) even if microphysics should turn out to be indeterministic; and this is all that determinists need to affirm in free will debates. (For this line of argument, see the essays in this volume by Ted Honderich and Derk Pereboom; and for discussion of conflicting views about the role indeterminism might play in macroscopic systems, see the essays of Hodgson and Bishop.)
Third, one often hears the argument in contemporary free will debates that if quantum jumps or other undetermined events did sometimes have non-negligible effects on the brain or behavior, this would be of no help to defenders of an incompatibilist free will. Such undetermined effects would be unpredictable and uncontrollable by the agents, like the unanticipated emergence of a thought or the uncontrolled jerking of an arm—just the opposite of the way we envision free and responsible actions (for example, Dennett 1984; G. Strawson 1986; Honderich 1988; Double 1991). This argument has been made in response to suggestions by prominent twentieth century scientists (such as Nobel laureates Louis De Broglie and A. H. Compton  in physics and Sir John Eccles  in neu-rophysiology) that room might be made for free will in nature if undetermined events in the brain were somehow amplified to have large-scale effects on human choice and action. Unfortunately, such a modernized version of the Epicurean chance swerve of the atoms seems to be vulnerable to the same criticisms as its ancient counterpart. It seems that such undetermined events in the brain or body would occur spontaneously and would be more of a nuisance—or perhaps a curse, like epilepsy—than an enhancement of an agent's freedom. (For this line of argument and others about the limitations of indeterminist free will, see the essays (p. 9) in this volume by Galen Strawson, Honderich, and Christopher Taylor and Daniel Dennett.)
The fourth and final reason why indeterministic developments of twentieth-century physics have not undermined determinist thinking about human behavior is perhaps the most important. While determinism has been in retreat in the physical sciences during the twentieth century, developments in sciences other than physics—in biology, neuroscience, psychology, psychiatry, social and behavior sciences—have been moving in the opposite direction. They have convinced many persons that more of our behavior is determined by causes unknown to us and beyond our control than previously believed. These scientific developments are many, but they clearly include a greatly enhanced knowledge of the influence of genetics and heredity upon human behavior (the recent mapping of the human genome is a symbolic as well as real indication of this influence, naturally arousing fears of future control of behavior by genetic manipulation); greater awareness of biochemical influences on the brain; the susceptibility of human moods and behavior to drugs; the advent of psychoanalysis and other theories of unconscious motivation; development of computers and intelligent machines that mimic aspects of human cognition in deterministic ways; comparative studies of animal and human behavior suggesting that much of our motivational and behavioral repertoire is a product of our evolutionary history; influences of psychological, social, and cultural conditioning upon upbringing and subsequent behavior, and so on. (The impact of such trends on contemporary free will debates is discussed in essays by Taylor and Dennett, Paul Russell, Richard Double, Benjamin Libet, and Henrik Walter.)
In sum, there continues to be considerable debate about determinism and indeterminism in the physical world, and about the relationship of both to human behavior, while contemporary sciences other than physics provide continuing support for deterministic thinking about human behavior. Worries about determinism in human affairs therefore persist with good reason in contemporary debates about free will.11
3. The Compatibility Question and Arguments for Incompatibilism
These worries about determinism make the second pivotal question of modern free will debates, the Compatibility Question, all the more important. Is free will (p. 10) compatible (or incompatible) with determinism? The free will problem arose historically because it was assumed that there was some kind of conflict between free will and determinism. If it turns out, to the contrary, that determinism is no threat to free will because the two can be reconciled, then worries about determinism would be misplaced. The traditional problem of free will would not only be solved, but in a manner “dissolved,” for the supposed conflict with determinism that gave rise to it in the first place would have been shown to be illusory.
Such a “dissolutionist” strategy has been the reigning strategy of modern compatibilists about free will since Thomas Hobbes in the seventeenth century. The strategy led in the twentieth century to claims by logical positivists and others that the traditional problem of free will is a “pseudo-problem” or a “dead issue” and ought to be laid to rest. This has not happened, of course; debates about free will are more alive than ever today. But compatibilist views have had a powerful influence throughout the twentieth century. The idea that free will and determinism are compatible continues to be a majority view among philosophers and scientists because it seems to offer a simple resolution of the conflict between ordinary views of human behavior from a practical standpoint and theoretical images of human beings in the natural and social sciences. (Philosophers always believe they have made progress when they discover something we don't have to worry about—or when they discover something we do have to worry about.)
So while the debate over the Compatibility Question has not ended, the burden of proof has shifted back to those who believe in a traditional free will that is incompatible with determinism. One cannot simply assume that if determinism is true, we would lack freedom or free will in an important sense. Arguments must now be provided to show this; and one of the interesting developments of the past thirty years is that new arguments for incompatibilism have indeed been proposed to meet the challenge. These incompatibilist arguments have in turn provoked more sophisticated compatibilist responses, and new theories on both sides of the Compatibility Question, as we shall now see.
Two features of free will were mentioned earlier that seem to imply its incompatibility with determinism—(a) it is “up to us” what we choose from an array of alternative possibilities and (b) the origin or source of our choices and actions is in us and not in anyone or anything else over which we have no control. Most modern arguments for the incompatibility of free will and determinism have proceeded from feature (a)—the requirement that an agent acted freely, or of his or her own free will, only if the agent had alternative possibilities, or could have done otherwise.12 Let us refer to this requirement as the AP condition (for “alternative possibilities”) or simply AP. (It is also sometimes called the “could have done otherwise” condition or the “avoidability” condition.)
The case for incompatibility from this AP (or “could have done otherwise”) condition has two premises: (p. 11)
1. The existence of alternative possibilities (or the agent's power to do otherwise) is a necessary condition for acting freely, or acting “of one's own free will.”
2. Determinism is not compatible with alternative possibilities (it precludes the power to do otherwise).
Since it follows immediately from these premises that determinism is not compatible with acting freely, or acting of one's own free will, the case for incompatibilism (and the case against it) must focus on one or another of these premises. In fact, there have been heated and labyrinthine debates in recent philosophy about both premises. Premise 1 is just the AP condition itself (free will requires alternative possibilities or the power to do otherwise) and it has been subjected to searching criticisms. But I shall begin with premise 2, which is usually regarded as the most crucial (and vulnerable) premise since it asserts the incompatibility of determinism with the power to do otherwise.
The most widely discussed argument in support of premise 2 in recent philosophy has been the so-called “Modal” or “Consequence” Argument for incompatibilism. This argument was first formulated in varying ways by Carl Ginet (1966, 1980), David Wiggins (1973), Peter van Inwagen (1975, 1983), James Lamb (1977), and (in a theological form) by Nelson Pike (1965).13 Alternative formulations have since been proposed and defended by many others.14 Van Inwagen (1983), who offers three versions of the argument, regards these as versions of the same basic argument, which he calls the “Consequence Argument,” and states informally as follows:
If determinism is true, then our acts are the consequences of the laws of nature and events in the remote past. But it is not up to us what went on before we were born; and neither is it up to us what the laws of nature are. Therefore, the consequences of these things (including our present acts) are not up to us. (p.16)
To see the connection between this argument and premise 2, recall that something's being “up to us” implies our having the power to bring it about or not to bring it about. Given this assumption, the claims in the quote that “it is not up to us what went on before we were born” or “what the laws of nature are” imply that there is nothing we could have done to alter the past before we were born or the laws of nature. But if determinism is true, the past before we were born and the laws jointly entail our present actions. So it seems that there is nothing we can do to alter our present actions. We do not have the power to do otherwise and hence lack alternative possibilities.
This Modal or Consequence Argument for incompatibilism is the topic of two essays in this volume, one by Tomis Kapitan, a compatibilist critic of the argument, the other by Peter van Inwagen, one of its best-known incompatibilist (p. 12) defenders. The first half of Kapitan's essay surveys various formulations of the Consequence Argument and criticisms made against it over the past three decades. Like many of the argument's critics, Kapitan believes its soundness depends upon how one interprets modal notions such as “power” or “ability” (to bring something about) and “could have done otherwise”; and the argument also depends on how one interprets related conditional statements about what would or might have happened if various abilities or powers had been exercised. Kapitan explores these topics in the latter part of his essay and, in the light of them, considers possible compatibilist strategies for answering the argument.
In the first half of his essay, van Inwagen restates and reaffirms a formal version of the Consequence Argument first presented in van Inwagen (1983) (which has become the most widely discussed version of the argument in the 1990s) and then defends this version against a recent objection by Thomas McKay and David Johnson (1996). In the second part of his essay, van Inwagen turns to a different topic that we will consider later—how one is to make sense of the incompatibilist or nondeterminist kind of freedom that the Consequence Argument seems to require. Van Inwagen believes that no one to date has been able to give an intelligible account of incompatibilist freedom; and he has doubts about the possibility of doing so. Yet because he also thinks the Consequence Argument is undeniably sound, he argues that we must continue to believe in an undetermined free will even if we do not know how to give an intelligible account of it.
4. Classical Compatibilism: Interpretations of “Can,” “Power,” and “Could Have Done Otherwise”
Most compatibilists believe that the Consequence Argument and all arguments for incompatibilism can be defeated by giving a proper analysis of what it means to say that agents can (or have the power or ability to) do something; and consequently there has been much debate in recent philosophy about the meaning of these notions. Traditionally, compatibilists themselves have defined freedom in terms of “can,” “power,” and “ability.” To be free, most compatibilists have insisted, means in ordinary language (1) to have the power or ability to do what we will (desire or choose) to do, and this entails (2) an absence of constraints or impediments preventing us from doing what we will, desire, or choose. The constraints or impediments they have in mind include physical restraints, lack of (p. 13) opportunity, duress or coercion, physical or mental impairment, and the like. You lack the freedom to meet a friend in a cafe across town if you are tied to a chair, are in a jail cell, lack transportation, someone is holding a gun to your head, or you are paralyzed. Compatibilists have typically insisted that (1) and (2) capture what freedom means in everyday life—that is, an absence of such constraints and hence the power (= ability plus opportunity) to do whatever you will or want to do.
A view that defines freedom in this way has been called “classical compatibilism” by Gary Watson (1975); and this is a useful designation. Classical compatibilists include well-known philosophers of the modern era such as Thomas Hobbes, David Hume, John Stuart Mill, and numerous twentieth-century figures (such as A. J. Ayer 1954, Moritz Schlick 1966 and Donald Davidson 1973).15 Despite differences in detail, we can say that what these classical compatibilists have in common is that they define the freedom to do something in terms of (1) and (2). What do they say about the freedom to do otherwise? It is also defined by classical compatibilists in terms of (1) and (2). You are free to do otherwise than meet your friend when you (1) have the power or ability to avoid meeting him, which entails in turn that (2) there are no constraints or impediments preventing you from avoiding the meeting (e.g., no one is forcing you at gunpoint to meet him). Of course, an absence of constraints and hence the freedom to do something does not mean you will actually do it. But for classical compatibilists it does mean that you would do it, if you wanted or desired to do it. Thus they hold that (1) and (2) entail a third feature of classical compatibilism, namely, that terms such as can, power, ability, and freedom should be given a conditional or hypothetical analysis: (3) that an agent can (has the power, is able, is free, to) do something means that the agent would do it, if the agent wanted (or desired or chose) to do it.
Such conditional or hypothetical analyses of can, power, and freedom were not invented by compatibilists to thwart the Consequence Argument. They were invented long before that argument in an effort to represent ordinary notions of freedom. But, if conditional analyses are correct, they would effectively thwart the Consequence Argument and other arguments for the incompatibility of freedom and determinism that appeal to alternative possibilities, or the power to do otherwise. For if the power to do otherwise means only that you would have done otherwise if you had wanted or desired, it would be consistent with determinism. It might be true that you would have done otherwise if you had wanted, though it is determined that you did not in fact want otherwise. Likewise, if the power to do otherwise has only such a conditional meaning, it would not require changing the past or violating laws of nature. To say “you could have done otherwise” would only amount to the counterfactual claim that you would have done otherwise, if (contrary to fact) the past (or the laws) had been different in some way, for example, if you had wanted or desired or chosen otherwise.16
(p. 14) So if conditional or hypothetical analyses of power and freedom favored by classical compatibilists are correct, the Consequence Argument would fail. But such analyses are themselves controversial and have also been subject to searching criticisms in contemporary philosophy. Current debates about conditional analyses—and generally about the meanings of can, power, ability and “could have done otherwise”—are surveyed and critically evaluated in another essay of this volume, by Bernard Berofsky. Berofsky's survey begins with G. E. Moore (the first important figure of the twentieth century to put forward a conditional analysis of can) but focuses primarily on the forty-year period since the publication of J. L. Austin's influential essay “Ifs and Cans” (1961), which criticized Moore's view. Though Berofsky is a compatibilist himself, he has been among the critics of conditional or hypothetical analyses of power and ability. He thinks compatibilists should look elsewhere if they wish to blunt the force of incompatibilist arguments and he explains his own compatibilist alternative at the end of his essay.
Despite the difficulties with conditional analyses of can, power, and freedom chronicled in Berofsky's essay, conditional analyses are far from dead among contemporary philosophers. Many compatibilists continue to believe that the spirit, if not the letter, of conditional or hypothetical analyses of power and freedom can be salvaged by focusing on more sophisticated interpretations of modal and counterfactual claims about what might or might not occur in possible worlds that are similar to the actual world. (See, for example, Lehrer 1976; Lewis 1981; Falk 1981; Flint 1987; Vihvelin 1991; Audi 1993; M. White 1993; Peacocke 1999; and the essays in this volume by Kapitan, Paul Russell and Taylor and Dennett). Yet conditional analyses of can, power, and freedom have frequently been on the defensive in recent philosophy; and many other compatibilists now try to avoid them altogether while seeking other ways to undermine arguments for incompatibilism. To these new alternative compatibilist views we now turn.
5. Moral Responsibility and Alternative Possibilities: Frankfurt-Style Examples and Semi-Compatibilism
Recall the two premises mentioned earlier, on which arguments for the incompatibility of free will and determinism have been based, (1) The existence of alternative possibilities (or the agent's power to do otherwise) is a necessary condition (p. 15) for acting freely, or acting “of one's own free will.” (2) Determinism is not compatible with alternative possibilities (it precludes the power to do otherwise). We have been focusing on arguments for and against premise 2. Premise 1 would appear to be less vulnerable. It simply states the AP, or alternative possibilities, condition—that free will requires alternative possibilities or could have done otherwise—a claim that persons on both sides of the compatibility debate (until recently) have tended to accept. Even most compatibilists in the past were willing to grant that freedom required the power to do otherwise. They assumed their task was to give an analysis or interpretation of this power (conditional or otherwise) that would show it to be compatible with determinism.
This situation has changed dramatically in the past thirty years. Many contemporary compatibilists on free will and determinism—influenced in part by the difficulties of conditional analyses, but more by new arguments—would now deny that alternative possibilities or the power to do otherwise are needed for free will in the first place, in the sense in which free will is required for moral responsibility. If this denial is correct and premise 1 (the AP condition) is false, then any argument for the incompatibility of free will and determinism based on alternative possibilities—including the Consequence Argument—could not get off the ground.
Recent support for the denial of premise 1, or AP, has come from two distinct kinds of argument, which should be distinguished. The first kind of argument appeals to what David Shatz (1997) has called “character examples”17; the second kind appeals to what have come to be known as “Frankfurt-style examples (or cases)”—named after Harry Frankfurt, who introduced the first example of this kind in an influential article “Alternate Possibilities and Moral Responsibility” (1969).
One of the best known arguments of the first kind—appealing to character examples—comes from Daniel Dennett (1984). Dennett argues that the “could have done otherwise” principle (our AP) is false, citing as a counterexample the case of Martin Luther. According to Dennett, when Luther said, “Here I stand. I can do no other,” upon finally breaking with the church in Rome, he meant “that his conscience made it impossible for him to recant” (p. 133). Suppose, says Dennett, that Luther was literally right about this: he could not then and there have done otherwise because his act was determined by his character and motives. This would not matter to Luther's free will or responsibility, Dennett argues, for “we simply do not exempt someone from blame or praise for an act because we think he could do no other” or because we think his act was determined by his character (ibid.). In saying, “I can do no other,” Luther was not renouncing free will or moral responsibility, but rather taking full responsibility for acting of his own free will. Dennett concludes that neither free will nor moral responsibility require “could have done otherwise”; and hence neither requires the falsity of determinism.
(p. 16) “Character examples” of this kind have their source in David Hume's well-known observation that we cannot be held responsible (or be said to act from “our own” free wills) unless our actions are to a considerable degree determined by our characters or motives in regular ways. The obvious truth of this claim gives character examples, like Dennett's Luther example, an undeniable force. Yet there is reason to think such character examples do not provide conclusive evidence that free will does not require alternative possibilities and hence that free will is compatible with determinism. For the following response to such examples is available to incompatibilists and has been made by some of them in recent debates (see Kane 1985, 1996; van Inwagen, 1989; Shatz 1997; Ekstrom 2000).
It may be true that Luther's “Here I stand” might have been a morally responsible act done “of his own free will,” even if he could not have done otherwise at the time he performed it and even if his act was determined by his then-existing character and motives. But this would be true only to the extent that one could assume other things about the background of Luther's action that made him responsible or accountable for it—namely, that he was responsible by virtue of earlier choices and actions for making himself into the kind of person he now was, with this character and these motives, and that he could have done otherwise with respect to at least some of those earlier acts. If this were not so, one might argue, there would have been nothing he could have ever done to make himself different than he was—a consequence that is difficult to reconcile with the claim that he is morally responsible for being what he is.18 (The implications of this line of reasoning for issues about moral responsibility, desert, and freedom are discussed—from differing points of view—in several essays of this volume, including, those of Paul Russell, Galen Strawson, Laura Ekstrom, and Robert Kane.)
This argument, even if correct, would not show of course that character examples like that of Luther lack significance—far from it. Such examples seem to show that alternative possibilities need not be required for every morally responsible act done of our own free wills; and, if correct, this would be a significant implication. Yet the preceding argument also seems to show that, if we take a broader view of an agent's life history, rather than focusing on individual acts in isolation, it does not necessarily follow that free will and moral responsibility do not require alternative possibilities at all, that is, at any times, in the course of an agent's life.19 A stronger argument would be needed to show this; character examples alone do not suffice.
This leads to examples of the second kind mentioned earlier—Frankfurt-style examples or Frankfurt-style cases—which, according to many philosophers, do provide the stronger argument needed to show that alternative possibilities are not required at all for free will or moral responsibility. Examples of this kind were originally introduced by Frankfurt (1969) with the intent of undermining what he called the “Principle of Alternative Possibilities” (PAP): “[A] person is morally (p. 17) responsible for what he has done only if he could have done otherwise” (p. 829). This is our AP condition applied to “morally responsible” acts rather than to acts done “of our own free will.” (The AP variant would be “a person acts of his own free will only if he could have done otherwise.”) The two principles (PAP and AP) would be equivalent, if the moral responsibility at issue (in PAP) were precisely the kind that free will (in AP) is suppose to confer; and this assumption has been commonly made in free will debates. But we shall see that this assumption (linking moral responsibility and free will) has also come into question in contemporary free will debates, specifically in connection with examples of the Frankfurt type.
Frankfurt-type examples typically involve a controller who can make an agent do whatever the controller wants (perhaps by direct control over the agent's brain). The controller will not intervene, however, if the agent is going to do on his own what the controller wants. Frankfurt argues that if the controller does not intervene because the agent performs the desired action entirely on his own, the agent can then be morally responsible (having acted on his own)—even though the agent literally could not have done otherwise (because the controller would not have let him). If this is so, the Principle of Alternative Possibilities, PAP, would be false: the agent would be morally responsible, though he could not in fact have done otherwise. If PAP and AP turn out to be equivalent, AP would be false as well. Neither moral responsibility nor free will (in the moral responsibility-entailing sense) would require alternative possibilities; and arguments for incompatibilism, such as the Consequence Argument, would be thwarted.
Frankfurt-style examples of this sort involving “pre-emptive” (or “counter-factual”) control have proliferated since they were first introduced. The literature on them is now enormous and has had a significant impact on contemporary free will debates. Note that Frankfurt-style examples provide extra leverage against PAP and AP that character examples do not provide. For one might go on to imagine a “global” Frankfurt controller hovering over agents throughout their entire lifetimes, so that the agents never could have done otherwise; and yet the controller never in fact intervenes because the agents always do on their own what the controller wants. Such a global controller would be a mere observer of events, never actually intervening in the agents' affairs (a mere “counterfactual intervener,” in John Fischer's words). It seems that the agents would act “on their own” throughout their entire lifetimes and would be responsible for many of their actions even though they never could have done otherwise and never had any alternative possibilities.20
Contemporary debates about the implications of Frankfurt-style examples are the subject of three essays of this volume, by John Martin Fischer, Laura Ekstrom, and David Widerker. Fischer, whose prior writings have contributed as much as any contemporary philosopher to our understanding of the implications of these (p. 18) examples, provides a comprehensive survey of arguments about Frankfurt-style examples over the past thirty years. He considers various strategies by which critics of these examples have tried to rescue the Principle of Alternative Possibilities, PAP (or variations of it) from arguments based on such examples; and various responses to these strategies. Fischer himself is a defender of Frankfurt-style examples, who believes that moral responsibility does not require alternative possibilities (that is, he denies PAP). But, surprisingly, he is also an advocate of his own version of the Consequence Argument (Fischer 1994) and believes that freedom does imply alternative possibilities (that is, he affirms AP). This view, which Fischer calls semi-compatibilism, is defended by him and also by Mark Ravizza in a number of recent writings (Fischer 1994; Ravizza 1994; Fischer and Ravizza 1998).21 It amounts to the claim that moral responsibility is compatible with determinism (since it does not require the power to do otherwise), while freedom (which does require the power to do otherwise) is not compatible with determinism.22
Laura Ekstrom and David Widerker look at Frankfurt-style examples from an opposing incompatibilist or libertarian perspective. Ekstrom, along with other incompatibilist critics of these examples (for example, van Inwagen 1978, 1983; Kane 1985, 1996b; Lamb 1993; Widerker 1995a, and b, Ginet 1996; Copp 1997; Wyma 1997), has argued that Frankfurt-style examples do not refute every relevant form of PAP and do not show that moral responsibility is compatible with determinism (cf. Ekstrom 2000: ch. 6). In her essay for this volume, she defends a number of objections to Frankfurt-style examples, arguing that intuitions to the effect that agents are morally responsible in such examples beg the question against those who believe free will is incompatible with determinism. Ekstrom also discusses a new Frankfurt-style example put forward in an influential article by Alfred Mele and David Robb (1998), which was designed to answer objections to earlier Frankfurt-style cases by incompatibilist critics such as Widerker (1995a and b) and Kane (1985, 1996a).
Widerker is the author of several articles (notably, 1995a and b) that have had a significant impact on recent debates about Frankfurt-style exsmples. In his essay for this volume, he defends the main theses of these articles by responding to new Frankfurt-style examples put forward in the past decade to answer his and other incompatibilist objections. Since Widerker has discussed the Mele/Robb example elsewhere (2000a), and it is dealt with by Fischer and Ekstrom in this volume, he focuses on other recent Frankfurt-style exsmples designed to answer incompstibilist objections—exsmples suggested by Eleonore Stump (1996a, 1999b), David Hunt (1996a), and others. Widerker's essay also discusses some theological implications of Frankfurt-style cases. He concludes with a general argument (called the “W-defense”) designed to show that it would be unreasonable to hold an agent morally blameworthy for an action if the agent could not have avoided performing the action. In the light of this defense, Widerker argues for a version of PAP for (p. 19) at least one kind of moral responsibility—moral blameworthiness—against supporters of Frankfurt, such as Fischer (1994).
6. Beyond Classical Compatibilism: New Compatibilist Approaches to Freedom and Responsibility
In addition to semi-compatibilism, a host of other new compatibilist views of both freedom and responsibility have been introduced in the past forty years in the attempt to answer objections to classical compatibilism. These new compatibilist theories are described and critically evaluated in two further essays of this volume, by Ishtiyaque Haji and Paul Russell.
Haji's essay deals with two broad categories of contemporary compatibilist views of freedom and responsibility, which he calls reactive attitude theories and mesh theories. Mesh theories are further divided into “hierarchical theories,” “valuational theories,” “reason theories,” and others. Compatibilist theories of the first category—reactive attitude theories—have their roots in another seminal essay of modern free will debates, P. F. Strawson's “Freedom and Resentment” (1962). Strawson argues that free will issues focus pivotally on the conditions required to hold persons responsible for their actions; and he argues that responsibility is constituted by persons adopting certain reactive attitudes toward themselves and others—attitudes such as resentment, admiration, gratitude, indignation, guilt, and the like. To be responsible, according to Strawson, is to be a fit subject of such attitudes. It is to be enmeshed in a “form of life” (to use Ludwig Wittgenstein's apt expression for this view) in which such reactive attitudes play a constitutive role.
Moreover, this form of life of which the reactive attitudes are constitutive is such that, according to Strawson, we would not give it up even if we found that determinism was true, because we could not give up assessing ourselves and others in terms of the reactive attitudes if we continued to live a human form of life. So Strawson contends that the freedom and responsibility required to live a human life (whatever else they may involve) must be compatible with determinism. Freedom and responsibility do not require some mysterious indeterminist or “contra-causal” free will, as incompatibilists claim. This Strawsonian reactive attitude view has inspired considerable debate since the 1960s, which is documented in Haji's essay. It has also gained new adherents in the 1990s, one of whom is R. Jay Wallace, (p. 20) whose book Responsibility and the Moral Sentiments (1994) is the most thoroughly developed Strawsonian view in the recent literature. Wallace's view is also critically evaluated in Haji's essay.
Mesh theories, which form another influential class of new compatibilist theories, insist that the freedom required for responsibility is a function of the appropriate “mesh” or connection between agents' choices or actions, on the one hand, and their reasons or motives for acting, on the other. The most widely discussed of mesh theories are the hierarchical theories of motivation of Gerald Dworkin (1970), Harry Frankfurt (1971), Wright Neely (1974), and others. In his seminal essay “Freedom of the Will and the Concept of a Person” (1971), Frankfurt argued that persons, unlike similar animals, “have the capacity for reflective self-evaluation that is manifested in the formation of second-order desires” (p. 7)— desires to have or not to have various first-order desires. Free will and responsibility require that we assess our first-order desires or motives and form “second-order volitions” about which of our first-order desires should move us to action. Our “wills”—the first-order desires that move us to action—are free, according to Frankfurt, when they conform with our second-order volitions, so that we have the will (first-order desires) we want (second-order desires) to have and in that sense we “identify” with our will.
Classical compatibilism is deficient, according to hierarchical theorists such as Frankfurt, because it gives us only a theory of freedom of action (being able to do what we will) without a theory of freedom of will in terms of the conformity of first-order motives to higher-order motives (being able, so to speak, to will what we will). Hierarchical theories remain compatibilist, however, since they define free will in terms of a conformity (or “mesh”) between desires at different levels without requiring that desires at any level be undetermined. It does not matter, as Frankfurt puts it, how we came to have the wills we want to have, whether by a deterministic process or not. What matters is that we have the wills we want and the power to realize them in action. That is what makes us free.23
Hierarchical views are an improvement in many ways over classical compatibilism since they provide a compatibilist account of free will as well as of free action and a richer picture of the human person. But hierarchical views are not without problems; and they have also been subjected to searching criticisms, which Haji considers in his essay. Some of the criticisms of hierarchical views have given rise to other “mesh theories” that depart from the hierarchical model in various ways. Among further mesh theories discussed in Haji's essay are “valuational” theories, such as that of Gary Watson (1975), and Susan Wolf's “reason view” (1990). For Watson, the relevant mesh required for free agency is not necessarily between higher-and lower-order desires, but between an agent's “valuational system” (beliefs about what is good or ought to be done), which has its source in (p. 21) the agent's reason, and the “motivational system” (desires and other motives), which has its source in appetite. Watson thus revives the ancient Platonic opposition between reason and desire—arguing that freedom consists in a certain conformity of desire to reason.
Susan Wolf's “reason view” takes this approach in yet another direction that also has ancient roots. She argues that freedom consists in being able to do the right thing for the right reasons, which requires in turn the ability to appreciate “the True and the Good.” Wolf's theory thus has a stronger normative component than many other mesh theories. According to her, you are free only when you are doing the right thing for the right reasons. Another recent theory that shares this particular kind of normative component and has affinities to Wolf's, though it is original, is put forward by Phillip Pettit and Michael Smith 1996. Normative theories of freedom of somewhat different kinds have also been defended by Michael Slote (1980) and Paul Benson (1987), among others.24 Wolf's reason view has some unusual and controversial implications that Haji also evaluates in his essay. For example, her view contains an “asymmetry thesis” according to which we act freely when we do the right thing for the right reasons, but do not act freely when we act wrongly or otherwise fail to do the right thing for the right reasons. Finally, Haji's essay discusses yet another compatibilist mesh theory of recent vintage put forward by Hilary Bok (1998).
Paul Russell's essay considers a number of other contemporary compatibilist views, some of which fit into Haji's categories, but most of which are not easily classified. Russell organizes his essay around themes from Daniel Dennett's influential compatibilist work, Elbow Room (1984), and in the light of these themes considers other compatibilist views along with Dennett's, including those of Paul Benson (1987), Martha Klein (1990), John Fischer and Mark Ravizza (1998),25 Robert Audi (1993), and Kevin Magill (1997). Through these authors and a number of other authors cited in his essay, Russell discusses a variety of topics that have been of concern to contemporary compatibilists, such as control, reflexivity, responsiveness to reasons, “moral luck,” the place of character in moral evaluation, ultimacy, blameworthiness, and normative elements of freedom.
As noted, Russell's discussion of Dennett focuses on the latter's earlier Elbow Room (1984). Dennett's more recent views may be seen in the essay he himself has contributed to this volume in collaboration with Christopher Taylor. Taylor and Dennett argue in defense of compatibilism that objections to compatibilist accounts of free agency are based on a flawed understanding of the relation of such notions as possibility and causation to freedom and agency; and they undertake an analysis of the relevant notions of possibility and causation to show this. Taylor and Dennett also employ analogies to the functioning of sophisticated computers to argue that the flexibility, reflexivity, and creativity that free will requires are consistent with the hypothesis that the behavior of humans, like that (p. 22) of intelligent machines, is determined. As their essay illustrates, appeals to intelligent machines and computer simulations of human cognition and behavior have come to play an increasingly important role in modern debates about free will.
7. Libertarian or Incompatibilist Theories of Free Will: The Intelligibility Question
Let us now turn from compatibilist theories to contemporary incompatibilist views of free will. Those who believe that free will is incompatible with determinism and who also affirm that free will exists (thus denying the truth of determinism) are usually referred to as libertarians in twentieth-century writings on free will. Libertarianism in this sense is not to be confused with the political doctrine of the same name. In free will contexts, libertarians are those who believe in the existence of a traditional antideterminist (or incompatibilist) free will, which does not necessarily commit them to political beliefs about freedom associated with political libertarianism. To avoid confusion, it would be more accurate to speak in free will debates of “free will libertarians” or “libertarians about free will” or “defenders of an incompatibilist free will”—making clear that the designation libertarian and its cognates is an abbreviation for these longer expressions—as it is assumed to be throughout this volume.
Contemporary free will libertarians must not only answer the Determinist and Compatibility questions by denying determinism and denying the compatibility of free will and determinism. They face an even more daunting task of answering a third pivotal question that has been at the heart of modern debates about free will and may be called the Intelligibility Question. Can one make sense of a freedom or free will that is incompatible with determinism? Is such an incompatibilist freedom coherent or intelligible, or is it, as many critics contend, essentially mysterious and terminally obscure?
The threat to free will from the perspective of this Intelligibility Question does not come from determinism, but from its opposite, indeterminism. If free will is not compatible with determinism, it does not seem to be compatible with indeterminism either. (One might say that the Compatibility Question is about the first half of this ancient dilemma, while the Intelligibility Question is about the second half.) An event that is undetermined might occur or not occur, given the entire past. Thus, whether or not it actually occurs, given its past, would seem to (p. 23) be a matter of chance. But chance events are not under the control of anything, hence not under the control of the agent. How then could they be free and responsible actions? Reflections such as these have led to charges that undetermined choices or actions would be “arbitrary,” “capricious,” “random,” “irrational,” “uncontrolled,” “inexplicable,” or merely “matters of luck or chance,” not really free and responsible actions at all. It appears that the indeterminism that libertarians demand for free will would not in fact enhance freedom but would undermine it.
One of the significant features of recent free will debates is that an increasing amount of attention has been given to this Intelligibility Question concerning libertarian free will. It is one thing for libertarians to put forth arguments for incompatibilism or to point out flaws in compatibilist accounts of free agency (as they have often done); it is quite another to give a positive account of the libertarian free agency that will show how such a free will can be reconciled with indeterminism and how it is to be related to modern views of human behavior in the natural and human sciences. Recent efforts to give positive accounts of incompatibilist or libertarian free agency are discussed in four essays of this volume, by Timothy O'Connor, Randolph Clarke, Carl Ginet, and Robert Kane.
It is instructive in reading these essays to sort recent libertarian theories into two broad categories—(1) agent-causal (or AC) theories and (2) teleogical intelligibility (or TI) theories (see Kane 1989). Agent-causal or AC theories, in O'Connor's words (1995a: 7), posit “a sui generis form of causation by an agent that is irreducible (ontologically as well as conceptually) to event-causal processes within the agent.” (I shall follow a common practice in recent writings on free will of hyphenating expressions such as “agent-cause” and “agent-causation” when talking about AC theories to indicate that a special kind of relation is intended.) AC theories have taken many historical forms. Indeed they have been the most common kind of libertarian theory until recent times. But, despite differences, they are usually motivated by a common line of reasoning. Since undetermined free acts might occur or not occur, given all the same prior events or states of affairs involving the agents, if we to avoid saying such acts merely happen by chance, we must posit some additional causal factor over and above (and not reducible to) prior events or state of affairs to account for their occurrence. This additional causal factor would be the agent itself, which cannot be caused in turn by prior events because it is not an event and therefore not of the right type to be the effect of any cause. The agent-cause, to use Roderick Chisholm's notable expression (1982b: 30), must be a kind of “prime mover unmoved.”
Libertarian theories of the teleological intelligibility (or TI) variety, by contrast, attempt to make undetermined free actions intelligible in terms of reasons or motives and intentions or purposes (hence teleologically intelligible), without postulating sui generis kinds of agency or causation that cannot be spelled out in terms of events or states of affairs involving the agent. TI theories in turn fall into two categories, depending on how they interpret the relation of reasons and intentions (p. 24) to actions: (1) simple indeterminist (or noncausalist) TI theories maintain, again from O'Connor (1995: 7), “that free agency doesn't require there to be any sort of causal connection (even of an indeterministic variety) between the agent ['s reasons] and his free actions”; while (2) causal indeterminist (or event-causal) TI theories maintain that agents cause their “free actions via [their] reasons for doing so, but indeterministically” (ibid.).
As a consequence of these distinctions, recent positive libertarian theories have often been sorted into three categories: agent-causal, simple indeterminist, and causal indeterminist theories (see O'Connor 1993, 1995a; Clarke 1995, Ekstrom 1999), the latter two being TI theories. But it is interesting to note that current agent-causal or AC theories can also be divided into two categories, depending on whether they interpret the relation of reasons to actions noncausally or causally. Some AC theorists maintain that reasons for acting play an essential (probabilistic) causal role in agent-causation (for example, Clarke 1993, 1996a), while other agent-cause theorists question this (see, e.g., O'Connor 1995a, 2000). So there are really at least four types of libertarian theory in the contemporary literature: AC theories of noncausalist and causalist kinds and TI theories of noncausalist (simple indeterminist) and causalist (causal indeterminist) kinds. All four types are discussed in the four essays of this volume on current libertarian views. Indeed, the authors of the four essays represent each of the theories: O'Connor and Clarke are AC theorists of the noncausalist and causalist kinds, respectively; while Ginet and Kane are TI theorists of the noncausalist and causalist kinds, respectively.
O'Connor's essay provides an overview of recent agent-causal (or AC) theories, explaining what motivates them through a discussion of mechanism, teleology and agency. He considers different accounts of the agent-causal relation by, among others, libertarians such as C. A. Campbell (1967), Roderick Chisholm (1966, 1976a), Richard Taylor (1967), John Thorp (1980), Michael Zimmerman (1984), Richard Swinburne (1986),26 Godfrey Vesey (in Flew and Vesey 1987), Alan Donagan (1987), William Rowe (1991),27 Randolph Clarke, (1993, 1996a) and O'Connor himself (1995a, 2000). O'Connor also poses the question whether agent-causal theories require a substance dualism of mind and body—as many philosophers have suspected they must, since they posit a suigeneris causal relation between an agent and action that is irreducible to ordinary modes of causation. O'Connor argues that AC theories do not necessarily require substance dualism but may require some sort of strong emergence of mind from matter. He also discusses some contemporary dualist accounts of free agency in the light of this question (for example, those of John Eccles and Karl Popper 1977 and Richard Swinburne 1986, among others28).
Clarke's essay critically examines TI theories of both the simple indeterminist and causal indeterminist kinds. (He prefers to call the former “noncausal” theories (p. 25) and the latter “event-causal” theories—two alternative designations that are also common in contemporary discussions.) With regard to simple indeterminist or noncausal TI theories, Clarke critically examines two representative views, those of Carl Ginet (1990) and Hugh McCann (1998) (while also citing other views of this sort, for example, of Storrs McCall 1994 and Stewart Goetz 1997). Clarke poses questions about how well these theories account for the relation of reasons to action and how well they are able to account for the causal role of agents in the control of free actions. His discussion of the second kind of TI theory, causal indeterminist or event-causal theories, is more complicated in that there are two distinct forms such theories have taken. The general possibility of causal indeterminist libertarian theories (as alternatives to both agent-causation and simple indeterminism) was first suggested, though not worked out, by David Wiggins (1973), Daniel Dennett (1978), Richard Sorabji (1980), and Robert Nozick (1981). Attempts to develop such theories in more detail have taken two forms in the 1980s and 1990s.
In one form, the causal indeterminism is placed earlier in the deliberative process, in the coming to mind of considerations for choice or in the formation of preferences. It is these processes that are said to be undetermined. Dennett (1978) and Kane (1985) first suggested views of this sort (which have been called “Valerian” libertarianisms29), but neither unqualifiedly endorsed them.30 (Indeed, Dennett, a compatibilist, suggested such a view only to criticize it.) Clarke critically examines two more recent versions of this kind of causal indeterminist view, those of Alfred Mele (1995) and Laura Ekstrom (2000).31 For causal indeterminist theories of the second kind, the indeterminism is not only placed earlier in the deliberative process, but also later, in choices themselves and in efforts of will preceding choice and action. Clarke also critically examines the most developed version of this second form of causal indeterminist theory, the view of Robert Kane (1985, 1996a). With regard to both versions of causal indeterminism, Clarke (as he did with simple indeterminist theories) considers questions about how well they are able to account for the rationality and control agents are supposed to exercise over their free actions. He concludes his essay with an assessment of the evidence for the indeterminism in nature which these and other libertarian theories would require.
Ginet's essay focuses on the issue that distinguishes causalist from noncausalist forms of both AC or TI theories—that is, the issue of how explanations of actions in terms of reasons or motives (beliefs, desires, intentions, and other motivating psychological attitudes) are related to causal explanations of behavior. This has been a central issue in the philosophy of action generally for the past forty years; and Ginet has been a major contributor to the debates about it. Causalists hold that reasons explanations are a form of causal explanation and require that there be a causal connection between the agent's reasons or motives and the actions (p. 26) they explain. Noncausalists deny that reasons explanation are a form of causal explanation and deny that such explanations require a causal connection between reasons and the actions they explain.
Ginet is a noncausalist and he defends a noncausalist account of reasons explanation in his essay, as he has in other writings (for example, Ginet 1990). He notes that the issue has an obvious bearing on the free will problem, for if reasons can explain actions noncausally, then actions could be explained without the supposition that they are either caused or determined. But noncausalist accounts of reasons explanations have been controversial and have been criticized by compatibilists and many incompatibilists as well, ever since the publication of Donald Davidson's seminal article on the subject, “Actions, Reasons and Causes,” in 1963. Ginet surveys these debates since Davidson's essay and undertakes a defense of noncausalism against its critics. In the process, he also criticizes Clarke's (and O'Connor's) agent-causal views, which Ginet, as a TI theorist and simple indeterminist, also rejects.
Kane agrees with Ginet that a special kind of agent- or nonevent causation is not needed to account for libertarian free will. As a TI theorist, he too rejects sui generis forms of causation, such as AC theories postulate. But, as a causal indeterminist, Kane disagrees with Ginet on the causalist issue: he does not think libertarians should deny that reasons explanation are a kind of causal explanation; nor need they deny that there are causal relations between reasons and actions in order to give an adequate account of incompatibilist free will. What matters is that the relevant causal relations involving reasons not always be deterministic (they may sometimes be nondeterministic or probabilistic). In short, “undetermined” need not mean “uncaused; and reasons, like other causes, may “incline without necessitating.” If “undetermined” did mean “uncaused,” one could see why libertarians might be tempted to posit (as AC theorists do) some extra kind of causation or agency to account for how free actions can be caused or produced, given that they are undetermined by events. Libertarian freedom, Kane contends, must be indeterminist, but it need not be “contra-causal.”32
But it is one thing to make such claims, and another to give an adequate account of libertarian free agency without appealing either to some special kind of agent-or nonevent causation or to the claim that reasons explanation are not causal. Kane undertakes this task in his essay and attempts to show how a TI account of free will of such a kind might be reconciled with modern conceptions of human beings in the natural and human sciences. His essay also addresses the Compatibility Question in addition to the Intelligibility Question, suggesting a novel route to incompatibilism that avoids direct appeals to alternative possibilities (AP) and the Consequence Argument, relying instead on a notion of “ultimate responsibility” (UR).
(p. 27) 8. Hard Determinism, Successor Views, and Other Nonstandard Theories
Not all of those who believe that free will is incompatible with determinism affirm the existence of free will (as libertarians do). Some incompatibilists also believe that determinism is true and so are committed to denying the existence of free will. Those who take such a stand are commonly referred to as hard determinists. The designation originated with the American philosopher and psychologist William James (1907), who distinguished “soft” from “hard” determinists. Both groups believe that all human behavior is determined. But soft determinists are compatibilists who insist that determinism does not undermine any free will or responsibility worth having, while hard determinists are incompatibilists who take a harder line: since determinism is true, free will does not exist in a sense required for genuine responsibility, accountability, blameworthiness, or desert.
Few thinkers have been willing to embrace such a hard determinist position unqualifiedly, since it would require wholesale changes in the way we think about human relations and attitudes, how we treat criminals and criminal behavior, and so on. This has not prevented hard determinism from being unequivocably endorsed by some (for example, Baron d'Holbach in the eighteenth century and Clarence Darrow and Paul Edwards 1958 in the twentieth), but unequivocal endorsement has been rare. The principle at work seems to have been that of the Victorian lady who exclaimed when she first heard of Darwin's theory. “Descended from the apes,” she said: “let's hope it isn't true. But if it is, let's hope it does not become generally known.”
Nonetheless, a core or kernel of the traditional hard determinist position has persisted into the twentieth century and continues to play a significant role in contemporary free will debates. This kernel persists in what might be called Successor Views to classical hard determinism. These Successor Views cannot strictly speaking be called hard determinist in the classical sense; and those who hold them would disown that title. But these views do contain a kernel of the classical hard determinist position that—detached from its traditional moorings—presents a powerful challenge to both compatibilist and libertarian views of free will. This kernel is defended in one form or another by four contributors to this volume, Galen Strawson, Ted Honderich, Derk Pereboom, and Saul Smilansky. None of them would accept the label of hard determinist with its classical connotations33; and their views differ from each other. But each accepts an important kernel of traditional hard determinism; and each accordingly presents a challenging alternative to both contemporary compatibilist and libertarian views.
The kernel may be identified as follows. Classical hard determinism consists of three theses: (1) free will (in the strong sense required for ultimate responsibility (p. 28) and desert) is not compatible with determinism; (2) there is no free will in this strong sense because (3) all events are determined by natural causes (that is, determinism is true). What I am calling Successor Views to classical hard determinism accept (1) and (2), but remain noncommittal about (3)—whether universal determinism is true. Aware of developments in twentieth-century physics, advocates of Successor Views are less confident than classical hard determinists about the truth of universal determinism; and they prefer to leave that question to the scientists. They remain convinced, however, that (1) free will—in what Galen Strawson (1986) calls the “true-responsibility entailing” sense—is incompatible with determinism, and that (2) there is, no such incompatibilist or libertarian free will.
This is the kernel—theses (1) and (2). Though this kernel is clearly not hard “determinism” in the classical sense, because it does not unqualifiedly affirm universal determinism (3), it still represents a pretty “hard” view since it rejects free will in the “true-responsibility entailing” sense. According to it, persons cannot be responsible or deserving for what they do in the ultimate sense assumed by believers in traditional free will. But what is especially interesting about this kernel is that it puts advocates of Successor Views who hold it at odds with both contemporary libertarian and compatibilist views. For, anyone holding the kernel holds (against compatibilism) that free will in the true-responsibility entailing sense is not compatible with determinism—thesis (1)—and (against libertarianism) that incompatibilist or libertarian free will does not exist—thesis (2).
But why do advocates of Successor Views think libertarian free will does not exist, if they remain noncommittal about the truth of universal determinism? One answer lies in the dilemma of free will mentioned earlier: if free will is not compatible with determinism, it does not seem to be compatible with indeterminism either. In response to this dilemma, compatibilists try to refute the first horn (by arguing that free will can be reconciled with determinism) while libertarians try to refute the second horn (arguing that free will can be reconciled with indeterminism). Many Successor Views (Pereboom's is an exception) reject both reconciliation projects and accept both horns of the dilemma. Thus, for these Successor Views, another thesis tends to play the role played by the thesis of universal determinism (3) in classical hard determinism, namely (3′): free will is not compatible with determinism and it is not compatible with indeterminism either. The first part of (3′) is just (1) and the second half implies (2); so the kernel is reached in a different way without assuming the truth of determinism.
But this is not quite the whole story about Successor Views either. While they remain noncommittal about whether determinism holds universally in nature, most advocates of Successor Views do in fact believe that human behavior is regular and determined for the most part and that if indeterminism did exist in the microphysical world, its macroscopic effects on human behavior would be (p. 29) negligible and of no significance for free will. If the effects of indeterminism on the human brain and behavior are insignificant, as most advocates of Successor Views believe, then the traditional problems posed by determinism for free will remain unchanged, whatever one's view about the microphysical world. More important, most advocates of Successor Views insist that if indeterminism at the micro-level did sometimes have macroscopic effects on human behavior, it would be of “no help” to believers in free will, since such indeterminism would not enhance, but would only diminish, freedom and responsibility.
As noted, theses (1) and (2) put advocates of Successor Views at odds with both libertarians and compatibilists. We should not be surprised, therefore, to find the four contributors to this volume who defend Successor Views—Strawson, Honderich, Pereboom, and Smilansky—arguing against both existing libertarian and compatibilist solutions to the free will problem. In his essay, for example, Galen Strawson defends his influential argument that libertarian free will is an impossibility, whether determinism is true or not (Strawson 1986, 1994a, 2000). Strawson argues that either libertarian free will requires an impossible infinite series of backtracking free choices by which we formed ourselves or it terminates in mere chance events over which we have no control. Strawson has also been a critic of positive accounts of libertarian free agency, like those considered in the previous section. And he has criticized compatibilist views as well (including the “reactive attitude” view of his father, P. F. Strawson) for failing to give us all that we want in the way of true responsibility-entailing freedom (1986: ch. 5).
In the process of revisiting arguments against incompatibilist or libertarian accounts of freedom in this volume, Strawson focuses attention on a different line of argument for incompatibilism from those we have so far considered. Recall the two features of traditional free will that seem to imply its incompatibility with determinism: (a) it must be “up to us” what we choose from an array of alternative possibilities and (b) the origin or source of our choices and actions must be “in us” and not in anyone or anything else over which we have no control. As noted, most arguments for incompatibilism (of which the Consequence Argument is the prime example) have proceeded from feature (a). But another noteworthy feature of recent free will debates, especially of the past two decades, is that more attention has been directed toward the comparatively neglected condition (b), the idea that agents must be the “ultimate sources” of their own wills and actions in a sense that entails that they, and they alone, are ultimately responsible for being the kinds of persons they are.34 In his essay, Strawson subjects this so-called “ultimacy” or “ultimate responsibility” condition to searching examination. Compatibilists are criticized for failing to recognize the importance of ultimacy for free will, while libertarians are criticized for thinking it is a realizable condition. (It is instructive in this regard to read Strawson's essay in conjunction with the preceding essay in this volume by Kane. Both philosophers argue that notions of ultimate (p. 30) responsibility and desert are crucial for free will, but they differ about whether the freedom required for ultimate responsibility and desert is intelligible and can be realized in the actual world.35)
Ted Honderich also argues for theses (1) and (2) of the kernel of classical hard determinism, as he has done in other influential writings (1988, 1993). Honderich thinks that the traditional notion of free will requires a power of ultimate “origination” of choices or actions that is incompatible with determinism (thesis ).36 But he also argues that no such power of ultimate origination is possible or could exist in the real world (thesis ). Of all the Successor Views canvased in this volume, Honderich's comes closest to affirming the whole package of classical hard determinism, but even he does not quite affirm it all. The first clause of the title of his essay, “Determinism as true,” suggests that he does, but it requires interpretation. The determinism Honderich affirms is a qualified “macro-determinism” of human behavior that, he says, is consistent with micro-indeterminism of the kind that standard quantum theories postulate. Honderich's sympathies do clearly lie on the side of universal determinism, since he also questions whether quantum physics should be interpreted indeterministically and whether it will turn out to be the last word about the physical world. But he does not take a final stand on these issues or on the truth of universal determinism, as classical hard determinists do. Honderich's main concern in the latter part of his essay (which he calls “the real problem” of free will) is how humans should react to the realization that their behavior is mostly determined and libertarian free will impossible. Some important “life-hopes” must be abandoned, he believes, if determinism is true, but many other life-hopes that matter to us can be retained.37
Derk Pereboom also defends both theses of the kernel of classical hard determinism—that (1) genuine free will is incompatible with determinism and (2) libertarian free will does not exist, offering his own arguments for each thesis. Pereboom calls the resulting position—which amounts to a rejection of both compatibilism and libertarianism—”hard incompatibilism” (see Pereboom 2001). He candidly admits that accepting such a position involves “relinquishing our ordinary view of ourselves as blameworthy for immoral actions and praiseworthy for actions that are morally exemplary.”38 The question that chiefly concerns Pereboom in his essay is whether affirming such a view would have the dire consequences many people fear for a host of everyday concerns that matter to us—for example, for moral reform and education, for crime prevention, interpersonal relations, our reactive attitudes of indignation, guilt, gratitude, love and repentance, the ways we treat others, including children, and generally, for our form of life. Pereboom discusses each of these topics, arguing that the consequences of hard incompatibilism would not be as destructive as many people believe and would be compensated by benefits in the form of more humane treatment of others.
(p. 31) Saul Smilansky is another philosopher who holds the kernel of hard determinism without being a hard determinist in the traditional sense. But Smilansky's view is unusual among contemporary views of free will, including the other Successor Views we have been discussing. His view is defined by two radical theses, both of which he defends in his essay for this volume and in a recent work (Smilansky 2000). The first thesis, Fundamental Dualism, states that we can and should be both incompatibilists and compatibilists about freedom and responsibility. There is no reason, Smilansky argues, why it should not be the case that certain forms of moral responsibility, desert, and blame require libertarian free will, while other forms can be sustained without it. Thus, if libertarian free will is impossible (as he believes), there is no reason why we have to choose between hard determinism or compatibilism. We can hold a mixed view that embraces what is true in both hard determinism and compatibilism, while denying that either has the whole truth.
Smilansky's second thesis, Illusionism, is even more radical. In contrast to both Honderich and Pereboom, Smilansky thinks the consequences for humanity of widespread belief that we lack libertarian free will would be dire and destructive. Illusion on free will is therefore morally necessary, he argues (this is the thesis of Illusionism). It is not that Smilansky thinks we need to induce illusory beliefs in people—in Brave New World fashion—but rather that such beliefs are already “in place” (for example, most people either don't question whether they have libertarian free will or, if they are compatibilists, assume they have all the freedom and responsibility they need); and these illusory beliefs play a largely positive social and moral role, he thinks. Recognizing that this thesis of Illusionism is likely to meet with considerable resistance (to put it mildly), Smilansky offers a series of arguments in the latter part of his essay to show the necessity of illusion by explaining the difficulties that would prevail without it. In this connection, he considers issues concerning guilt and innocence, value and worth, remorse and integrity, and related issues.
Richard Double defends yet another nonstandard view on free will which falls entirely outside the category of Successor Views we have been considering. Double calls his view free will subjectivism. The goal of his essay is to show how such a view and the free will problem in general are related to metaethical questions about the objectivity and subjectivity of value. In the process, he also considers how free will debates are influenced by differences in metaphilosophy—by differing views about the nature of philosophy (see Double, 1991, 1996a). Double thinks that people generally designate by “free will” the amount and kind of freedom that is required for moral responsibility. But ascriptions of moral responsibility, he also believes, do not attribute objective properties to persons; rather they express our subjective moral and evaluative attitudes toward people and their behavior. Consequently, ascribing free will to persons is also not a matter of ascribing (p. 32) some objective property they may or may not possess, but of expressing subjective attitudes toward them and signaling how they will be treated—for example, through reactive attitudes, verbal recrimination, praise and blame, retributive punishment or reward, and so on.
Double relates this free will subjectivism to metaethical views about value in the first part of his essay and argues that such a view best explains the persistent and seemingly irresolvable disagreements that have characterized debates about free will. In the second part of his essay, he compares and contrasts his view with those of other prominent contributors to contemporary free will debates, including several contributors to this volume: B. F. Skinner (1948, 1971), Daniel Dennett (1984), Bruce Waller (1990), Galen Strawson (1986), P. F. Strawson (1962), Thomas Nagel (1986), Ted Honderich (1993), and Peter Unger (1984).
Finally, Alfred Mele defends another nonstandard position on free will that does not fit into any of these categories. In his essay for this volume, Mele arrives at this position by a discussion of three topics that have been intertwined with contemporary debates about free will: autonomy, self-control, and weakness of will. Mele follows Aristotle in taking self-control (enkrateia) and weakness of will (ak-rasia) to be contraries—weakness of will being a deficiency of self-control. His essay begins with a survey of recent debates about the nature of self-control and weakness of will—debates to which Mele himself has been a significant contributor (Mele 1987, 1995).39 This survey leads him to an account of what he takes to be an “ideally self-controlled agent.”
Mele then poses the question whether an ideally self-controlled agent so conceived would necessarily also be autonomous (that is, self-governing or self-legislating) in a manner that many contemporary philosophers would associate with free will. He discusses this question in the context of the growing corpus of recent philosophical writing on the topic of autonomy (including works by Feinberg 1986; Dworkin 1988; Benn 1988; Lehrer 1997; Haworth 1986; Lindley 1986; Christman 1991; Berofsky 1995; among others). Mele argues that autonomy—and hence also free will—requires more than self-control, including ideal self-control, and he considers the additional conditions required, showing how contemporary discussions of autonomy are interwined with debates about free will (compare Mele 1995). Do these further conditions for genuine autonomy require that we choose between compatibilist and incompatibilist accounts of autonomy (and hence free will)? Mele thinks not, because one can give a “robust, satisfiable” set of adequate conditions for both compatibilist autonomy and incompatibilist autonomy. One can thus remain agnostic on the Compatibility Question regarding autonomy and free will without giving up the belief that there are autonomous human beings. He calls this view “agnostic autonomism.”
(p. 33) 9. Neuroscience and Free Will
Contemporary debates about determinism in human behavior have by no means been confined to the implications of modern physics. As noted in section 2, while determinism was in retreat in the physical sciences during the twentieth century, developments in sciences other than physics—in biology, neuroscience, psychology, psychiatry, social and behavior sciences—have convinced many persons that more of their behavior is determined by causes unknown to them and beyond their control than previously believed. Of particular significance among these scientific developments is the growing knowledge of genetics and physiology, of biochemical influences on the brain, including the susceptibility of human moods and behavior to drugs and biochemical sources of psychiatric disorders. All this, coupled with advances in understanding human cognition and neural networks, has led to a growing interest during the past two decades in the implications of the cognitive sciences and neurosciences for traditional issues about free will. This interest is reflected in two essays of this volume, by Benjamin Libet and Henrik Walter.
Benjamin Libet, professor of psychology at the University of California, San Francisco, has been a pioneer in the neurophysiological study of volition and willed action. In 1958 he began a series of groundbreaking experimental studies in human subjects (in collaboration with neurosurgeon Bertram Feinstein) relating brain activities to the appearance or production of conscious experience and willed action that have been much discussed by philosophers as well as scientists. In his essay for this volume, Libet discusses the implications of these experiments for traditional debates about free will, with special reference to the role that consciousness plays in free voluntary action.
Libet and his collaborators found that voluntary acts are preceded by a specific electrical charge in the brain (the “readiness potential”) which begins several hundred milliseconds before the human subjects become consciously aware of their intention to act. This suggests that the volitional process is initiated unconsciously. While some philosophers and scientists have been tempted to conclude from Libet's findings that willed actions are determined by unconscious forces and hence that our awareness of conscious control is illusory, Libet himself has a more nuanced view of the results, which he presents in this essay. He believes there is still a role for consciousness in controlling the outcome of willed actions, since consciousness can veto the act once underway. Thus, free will is not necessarily excluded, though novel neuroscientific findings place constraints on how free will could operate and how we are to make sense of it in terms of current research on the brain.
Henrik Walter is a neuropsychiatrist and philosopher who has also written perceptively about the implications for neuroscientific research for issues about (p. 34) free will. His contribution to this volume consists of excerpts from his Neurophilosophie des Willensfreiheit, published in Germany in 1996 and in English as Neurophilosophy of Free Will in 2001. In the excerpts, Walter discusses the role of the frontal cortex (and particularly the prefrontal association cortex) of the brain in the planning of actions, the selection from various options, and the organizing of behavior over time. He discusses neurological evidence for and against the claim that this region of the brain is the seat of the “will” in human beings, which some scientists have suggested (for example, Crick 1994). Walter's view is that the functions of willing (deliberation, planning, and the like) are distributed throughout the brain, though the prefrontal areas play a pivotal role since they provide a link between the cortical regions involved in higher cognitive functioning and other parts of the brain that are the sources of emotions, feelings, and motor reactions.
Walter also discusses fascinating neurological evidence suggesting that interruption of circuits involving the frontal cortex and related parts of the brain due to lesions or other deficiencies gives rise to disturbances in the feeling of agency. These include, among other examples, “alien hand syndrome” (where the patient's hand seems to have a “will of its own”), obsessive-compulsive disorders, and the “self-disorder” of schizophrenics, where patients feel that certain experiences and mental actions no longer belong to themselves or are produced outside of themselves. In the book from which these excerpts are taken, Walter defends a compatibilist view of free will. He sides with those who believe that libertarian free will cannot be made intelligible. But he thinks that justice can be done to many libertarian intuitions, if we take a neurophilosophical approach to the notions of autonomy and free will.
For further references to research in the neurosciences relevant to current debates about free will, readers should consult the bibliographical references in the essays of Libet and Walter as well as two noteworthy recent anthologies, The Volitional Brain: Towards a Neuroscience of Free Will (1999), edited by Libet, Anthony Freeman, and Keith Sutherland, and Neurobiology of Decision Making (1956), edited by A. Damasio, H. Damasio, and Y. Cristen.
10. Theological Determinism and Fatalism
While determinist threats to free will from the natural and human sciences have taken center stage in modern free will debates, the scientific challenges have not been the only ones of importance to current debates. Other historically important (p. 35) determinist threats to free will of continuing interest are dealt with in two further essays of this volume—theological determinism (in Linda Zagzebski's “Recent Work on Divine Foreknowledge and Free Will”) and fatalism, or logical determinism (in Mark Bernstein's “Fatalism”).
The theological implications of the free will problem have been a central preoccupation of many religious traditions, Jewish, Christian, and Muslim, among others, as well as a central preoccupation of Western intellectual history in general, especially since St. Augustine's seminal work, On the Free Choice of the Will. In his classic poem Paradise Lost, John Milton describes the angels debating about predestination and free will—wondering how they could have freely chosen to serve or reject God, given that God had made them what they were and had complete foreknowledge of what they would do. Milton tells us that the angels debating this issue were in “Endless Mazes lost,” not a comforting thought for us mortals.40
Many theologians through the centuries have believed that God's power, omniscience, and providence would be unacceptably compromised if one did not affirm that all events in the universe, including human choices and actions, were foreordained and foreknown by God. But many other theologians argued, with equal force, that if God did in fact foreordain or foreknow all human choices and actions, then no one could have chosen or acted differently, making it hard to see how humans could have ultimate control over their actions in a manner that would justify divine rewards or punishments. In such a case, the ultimate responsibility for good or bad deeds, and hence responsibility for evil, would devolve to God—an unacceptable consequence for traditional theists.
In the past thirty years, there has been renewed interest among philosophers and theologians in these issues of theological determinism or theological fatalism; and contemporary debates about them have surpassed even medieval discussions in labyrinthine complexity. Most of the recent literature on this topic has focused on the relation of divine foreknowledge and divine providence to human freedom and the implications of these topics to such things as divine omniscience and power, prophecy, petitionary prayer, the relation of time and eternity, and numerous other religious concerns. Linda Zagzebski's essay is a comprehensive and illuminating guide to the contemporary literature on theological determinism by a philosopher who has herself contributed significantly to current discussions of the religious implications of free will issues.
Mark Bernstein's essay deals with yet another historically important source of determinist thinking. The earliest of determinist or necessitarian doctrines that posed a threat to free will involved fate, conceived either as an impersonal cosmic force, or in the words of the ancient philosopher Empedocles, “an oracle of Necessity, an ancient decree of the gods, eternal, sealed fast with oaths.”41 Ancient concerns about fatalism were taken one step farther early in the development of Western philosophy when ancient Greek thinkers of the Megarian and (later) the (p. 36) Stoic schools conceived the idea that laws of logic alone might imply that human wills are fated and not free. If every proposition must be true or false (as the logical law of bivalence requires), and if this is the case for propositions about the future as well, then it seems that every future event would be fated either to occur or not to occur. If the proposition “a sea fight will occur tomorrow” were true today (to use an example made famous by Aristotle), then a sea fight could not but occur tomorrow. If the proposition were false, then a sea fight could not occur. Either way, the outcome would be necessitated by the past, together with the requirement that every proposition be true or false. This esoteric doctrine of logical fatalism or logical determinism has exercised thinkers for centuries and continues to be discussed in contemporary philosophy. Recent discussions of it are the subject of Mark Bernstein's essay.
11. Plan of The Volume
This is a sourcebook. Each essay can be read on its own and the references within each essay direct the reader to further writings in that topic area. While the essays can be read in any order, some naturally go together and are grouped into sections, guided by the three central questions discussed in this introduction.
The Determinist Question. The first four essays consider various determinist threats to free will and the contemporary debates they have generated, from theological and fatalist doctrines that posed the earliest threats (part I: essays of Zagzebski and Bernstein) to considerations of determinism and indeterminism in the modern physical sciences (part II: essays by Hodgson and Bishop).
The Compatibility Question. The Modal or Consequence Argument for incompatibilism, the most widely discussed recent argument for the incompatibility of free will and determinism, is discussed in part III (essays by Kapitan and van Inwagen). Part IV considers compatibilist responses to the Consequence Argument and surveys a variety of contemporary compatibilist perspectives on freedom and responsibility (essays by Berofsky, Haji, Russell and Taylor and Dennett). Part V considers issues about moral responsibility and alternative possibilities posed by so called “Frankfurt-style examples” and discusses doctrines of “semi-compatibilism” (essays by Fischer, Ekstrom, and Widerker).
The Intelligibility Question. The essays of Part VI consider various incompatibilist or libertarian perspectives on free will and agency and address the question of whether traditional doctrines of free will that require indeterminism can be made intelligible. They also consider issues about the nature and explanation of (p. 37) action, the relation of reasons to causes, control, rationality, and metaphysical issues about mind and body, agency and personhood (essays by O'Connor, Clarke, Ginet and Kane).
Part VII considers recent nonstandard views on free will, including Successor Views to hard determinism. The essays of this part also discuss a variety of further topics related to free will, including metaethical issues about the objectivity and subjectivity of value, morality and ultimate desert, criminal punishment, autonomy, self-control and weakness of will, illusion, and metaphilosophy (essays by Strawson, Honderich, Pereboom, Smilansky, Double, and Mele).
Finally, Part VIII takes a look at all three questions—determinism, compatibility, and intelligibility—from the perspective of the neurosciences, which have begun to influence debates about free will in the past decade and are likely to have further influence in the immediate future (essays by Libet and Walter).
(2.) I contemplated including essays on the history of free will debates in non-Western as well as in Western cultures but found this task excessive for one volume. Another volume with a historical or comparative focus would be a valuable project in its own right, but it also goes beyond the scope of this work. Some recent works on historical figures in the Western tradition do have relevance to contemporary debates about free will, and many are cited in various essays of this volume. Some prominent examples (by no means a complete list) include, on Aristotle, Hardie (1968), Kenny (1969b), Sorabji (1980), Irwin (1980), (1988), Fine (1981), Broadie (1991), Meyer (1993); on other ancient thinkers, Mi. White (1985); on Augustine, Babcock (1988), MacDonald (1999), Hunt (1996a, 1999); on Thomas Aquinas, Stump (1990), Kenny (1993), Loughran (1994), Gallagher (1994), Pink (1997), MacDonald (1998): on medieval thinkers in the period from Aquinas to Scotus, Kent, (1995); on Ockham, M. Adams (1987); on Molina and Molinism, Freddoso (1983, 1988): on Locke, Yaffe (forthcoming); on Leibniz, R. Adams (1982, 1994), Blumenfeld (1988a), Sleigh (1990, 1994), Paull (1992), Murray (1995); on Thomas Reid, Rowe (1991), O'Connor (1994); on Hume, Russell (1995); and on Kant, Wood (1984); Allison (1990).
(3.) Introductions to the problem of free will which appeared in the past twenty years include C. Williams (1980), Trusted (1984), Flew and Vesey (1987), Thornton (1990), Honderich (1993), Felt (1994), Dilman (1999), Ekstrom (2000). Anthologies of readings on free will in the same twenty-year period include Watson (1982), Fischer (1986), Fischer and Ravizza (1993), O'Connor (1995), Mongkin and Kellner (1998), Ekstrom (2001a), Kane (2001), and for theological issues about free will and divine foreknowledge, Fischer (1989).
(4.) There is an enormous literature on the “philosophy of action” (or “theory of action”) in the period covered by this volume, much of it relevant to issues about free will and frequently cited in this volume. influential book-length treatments and anthologies include Goldman (1970), von Wright (1971), Castaneda (1975), Brand and Walto (1976), Tuomela (1977), Aune (1977), Thomson (1977), Thalberg (1977), Davis (1979), Peacocke (1979), Davidson (1980), Hornsby (1980), O'shaughnessy (1980), Brand (1984), Bratman (1987), Donegan (1987), Frankfurt (1988), Dretske (1988), J. Bishop (1989), Velleman (1989), Wilson (1989), Ginet (1990), Lennon (1990), Schick (1991), Mele (1992), (1997b), Audi (1993), Bennett (1995), McCann (1998).
(5.) To be sure, this personal or practical standpoint involves other presuppositions as well, for example, that we as persons are enduring objects with a continuing identity in time. If one reduces persons to successions of causally related physical and mental events—denying they are enduring substances—as does Derek Parfit (1984) or much of the Buddhist tradition, then it may be argued that free agency and free will go by the board as well, In the normal way we understand them from a practical standpoint. (See Timothy O'Connor's essay in this volume for a discussion of some of these issues.) In an insightful article, Mark Siderits (1987) argues that the Buddhist reductionist conception of persons yields a novel compatibilist position on free will.
(7.) For further discussion of definitions of determinism, see the essay by Robert C. Bishop in this volume. The sense of determiism 1 have in mind here that is relevant to free will is what Jordan Howard Sobel (1998) calls determinism, by “ancient causes.” Sobel identifies ninety varieties of determinism, but he indicates that the ones that pose problems for free will are those that imply that all events have ancient causes (i.e., they are events that, for any past time, have sufficient antecedent causes earlier than that time). Other influential works on determinism of the past thirty years include Berofsky (1971), Montague (1974), Earman (1986), and Honderich (1988).
(8.) Those persons, like Dennett, who believe that freedom in every sense worth having is compatible with determinism, usually note that notions of freedom in ordinary language, such as freedom from coercion, from addiction, physical restraint, or political oppression are all consistent with determinism. An interesting variation of this compatibilist strategy is exhaustively pursued by Christine Swanton (1992), who lists scores of ordinary statements in which we talk about agents' being free and tries to devise a “coherence theory” of freedom, as she calls it, satisfying all of them.
(9.) E.g., those of Hodgson, Bishop, Kapitan, Russell, Taylor and Dennett, Honderich, Pereboom, Strawson, Smilansky, Libet, and Walter.
(11.) I have been speaking here of various kinds of scientific or causal forms of determinism (by antecedent causes and laws of nature), which have been the focus of most of the attention in modern debates about free will. There continue to be worries about other sources of determinism of theological and logical, rather than causal, varieties in contemporary philosophy. These are discussed in a later section of this introduction and in essays in this volume by Linda Zagzebski (on theological determinism) and Mark Bernstein (on logical fatalism), chs. 2 and 3, respectively.
(12.) While the vast majority of contemporary arguments for incompatibilism have taken this route, there are exceptions. A few contemporary philosophers have chosen to argue for incompatibilism by way of condition (b)—the requirement of ultimate origins or sources—rather than from condition (a) for alternative possibilities. For discusion of this alternative approach to incompatibilism, see section 8 of this introduction and essays by Strawson, Kane, and Russell, chs. 19, 18, and 10, respectively. In addition, there has been a tradition of arging that a notion of indeterminacy is loically built into the notion of a free choice (see Mackay 1967; Popper 1972), or that Godel's incompleteness theorem has implications concerning the compatibility of human reasoning and determinism (Lucas 1970; Penrose 1989). Finally, an unusual self-referential argument for incompatibilism is presented in Boyle, Grisez, and Tollefson (1976).
(13.) Pike's theological form of the argument is discussed in Linda Zagzebski's essay (ch. 2) in this volume which discusses the theological aspects of the free will problem. I return to Zagzebski's essay and the theological issues later in this introduction.
(14.) E.g., Widerker (1987), Talbott (1988), Hasker (1989), Ginet (1990), O'Connor (1993, 2000), Fischer (1994), Warfield (1996) and Finch and Warfield (1998). Differences of various formulations and their implications are discussed in Kapitan's essay, ch. 6.
(15.) For further discussion of classical compatibilism, see the essays of Russell and Berofsky.
(16.) The role and interpretation of such counterfactual conditionals have therefore been important toics of discussion in free will debates. See the essays in this volume by Kapitan, Berofsky and Taylor and Dennett, chs. 6, 8, and 11, respectively.
(17.) Shatz (1997) is an insightful discussion of character examples and one of the most thorough discussions of this topic in the recent literature. See also van Inwagen (1989) and Kane (1996: ch. 3).
(19.) Robert Audi is a compatibilist who, unlike most other recent compatibilists, has emphasized the importance of responsibility for character as well as for individual actions (see, e.g., Audi 1991b for a perceptive account of responsibility for character). Audi's view allows him to concede the points made in this paragraph about a need for the power to do otherwise, but he is also one of those compatibilists who offers a compatibilist account of the power to do otherwise (1974, 1993) rather than denying its importance for responsibility.
(20.) One can see why Frankfurt-style examples have also had an impact on debates about theological determinism. For discussion of their theological implications, see the essays in this volume of Zagzebski and Widerker, chs. 2 and 14, respectively.
(21.) Semi-compatibilists must of course show what moral responsibility does require if it does not require the power to do otherwise. Fischer (1994) and Fischer and Ravizza (1998) attempt to do this in terms of notions of (what they call) guidance control and reasons-responsiveness. There is a discussion of their use of these notions in the essay by Russell in this volume.
(22.) The possibility of separating moral responsibility and free will in this regard is a new wrinkle in free will debates. Once one conceives the possibility of disentangling them, one might consider going the other way—regarding moral responsibility as incompatible with determinism, while freedom is compatible with determinism. Such a view has in fact a modern defender in Bruce Waller (1990), whose view is discussed in Richard Double's essay for this volume ch. 23. The separation might also be considered in theological contexts. Some incompatibilists, such as Eleonore Stump (1990, 1996a) and Linda Zagzebski (1991, 2000), influenced by Frankfurt-style examples, have also entertained the view that, while free will is incompatible with causal determinism, moral responsibility is not. See Zagzebski's essay in this volume, ch. 2.
(24.) A number of other distinguished philosophers, such as Bernard Williams (1986) and Thomas Scanlon (1988), defend normative approaches to the free will problem of compatibilist kinds that differ in certain respects from those mentioned in these paragraphs. Williams is a compatibilist, but he does not think determinism can be reconciled with our “ethical conceptual scheme … l as it stands” (1986: 12), which is strongly influenced by Judeao-Christian and Kantian ideas of moral duty and conscience. Incompatibilism does seem to fit this scheme, Williams thinks. But he also thinks that modern “morality” conceived in this Kantian or deontological manner is deficient and thinks we have to “recast our ethical conceptions” (p. 13) by returning to ancient Aristotelian models of the ethical life which he argues are compatible with determinism. One's views about free will therefore depend on how one conceives the ethical life. (See Richard Double's essay in this volume on this point, ch. 23). Likewise, Scanlon's compatibilist view (1988) is related to his influential contractualist ethical theory. What gives free choice and action their special value and moral significance, Scanlon holds, is the desire to regulate one's behavior by standards that other persons could not reasonably reject in an informed and unforced contractual agreement and the desire to justify one's behavior to others in accord with such standards. What the satisfaction of these desires requires by way of freedom and responsibility is the capacity for critically reflective, rational selfgovernance (which Scanlon spells out in a manner similar to hierachical theorists like Frankfurt). Yet another compatibilist who believes that the view one takes on the freedom/determinism issue depends upon one's conception of ethics and the good life is Richard Warner (1987).
(25.) Fischer and Ravizza are of course only (semi-)compatibilists about moral responsibility, not freedom. But their account of responsibility in terms of “reasons-responsiveness” employs strategies, examined by Russell, that throw light on compatibilist views in general.
(26.) O'Connor says that while Swinburne appears to be an AC theorist, some of his statements suggest otherwise.
(27.) Rowe (1987, 1991) develops his view as an interpretation and defense of the theory of Thomas Reid, the seventeenth-century philosopher who is regarded by many AC theorists as the modern originator and inspiration for their view.
(29.) The expression “Valerian” refers to the French poet Paul Valery, who argued that freedom and creativity involved the “intelligent selection” from a number of undetermined alternatives. Dennett (1978) quotes Valery to this effect; and subsequently Mark Bernstein dubbed views of this sort “Valerian libertarianisms.” See also Double (1988a) and Kane (1988).
(32.) Even where “causal” has the ordinary sense of “event-causal.”
(33.) Honderich comes closest, but even the determinism he affirms is not the universal determinism of classical hard determinism, as I explain later.
(34.) Examples of philosophers who have focused more attention on this so-called “ultimacy” condition include Paul Gomberg (1975), Richard Sorabji (1980), Robert Nozick (1981), Robert Kane (1985, 1996a), Thomas Nagel (1986), Ted Honderich (1988), W. S. Anglin (1990), Martha Klein (1990), Derk Pereboom 2001, and Strawson himself (1986). Not all of these figures think this condition can be realized (indeed many of them argue against the possibility of its realization); but they all think it is a significant feature of free will in the traditional sense.
(36.) The assertion in the title of his essay that “Compatibilism and incompatibilism are False” might suggest that Honderich is denying thesis (1) of the kernel, but this is not so. What he is claiming is that compatibilists go wrong when they insist that compatibilist freedom is the only kind worth having and incompatibilists go wrong when they claim incompatibilist freedom is the only kind worth having. Both kinds support significant “life-hopes.”
(37.) Another philosopher who has construed the free will issue in terms of conflicts between certain of our beliefs and our deeply held desires is Nicholas Nathan. In his work Will and World (1992), Nathan develops a general view about the nature of philosophical issues as involving conflicts between beliefs and desires, applying such a metaphilosophical theory not only to the question of free will, but to other philosophical problems as well.
(38.) This volume, p. 479.
(39.) There is a rich recent literature on the problem of weakness of will or akrasia, much of which Mele refers to in his chapter. Significant recent works include Audi (1979), Bigelow, Dodds, and Pargetter (1990), Charlton (1988), Dunn (1987), Davidson (1970), Gosling (1990), Hurley (1993), Jackson (1984), Kenny (1975), Mortimore (1971), Pears (1985), Pugmire (1994), K. Robinson (1991), A. Rorty (1980a), Walker (1989), and Watson (1977).
(40.) These theological questions about free will also played a pivotal role in the Arab retrieval of ancient philosophical texts, so important in the development of late medieval Western culture. When Muslim scholars (about a century after the death of Mohammed) asked the caliphs if they could look at the ancient scrolls of the Greek philosophers hidden away in the libraries of the Middle East, one of their primary motives was to see if the pagan philosophers would provide insight into the vexing question of predestination and free will, which the Koran did not resolve.