John Martin Fischer
John Martin Fischer is best known for the idea of "semicompatibilism" - the idea that moral responsibility is compatible with determinism, whether free will is or is not compatible.
The concept is similar but not identical to Randolph Clarke's idea of a "narrow incompatibilist." A narrow incompatibilist is an incompatibilist on free will and a compatibilist on moral responsibility. Confusingly, this can include those who believe in free will and those who deny free will. Semicompatibilists assert only their belief in moral responsibility. They are agnostic on free will and argue that moral responsibility exists whether determinism or indeterminism is "true." A broad incompatibilist sees determinism as incompatible with both free will and moral responsibility. Broad incompatibilists thus include (very confusingly) both those who accept and those who deny free will and moral responsibility. Those who deny one or both are variously called "hard incompatibilists," "illusionists," or "impossibilists." Here is a taxonomy of determinist and compatibilist positions showing where semicompatibilism fits. Many of these philosophers reduce free will to the "control condition" for moral responsibility. This is to make freedom dependent on moral responsibility, which we call an ethical fallacy. As Fischer says:
Some philosophers do not distinguish between freedom and moral responsibility. Put a bit more carefully, they tend to begin with the notion of moral responsibility, and "work back" to a notion of freedom; this notion of freedom is not given independent content (separate from the analysis of moral responsibility). For such philosophers, "freedom" refers to whatever conditions are involved in choosing or acting in such a way as to be morally responsible.5Free will is of course a prerequisite for responsibility. Questions about free will are scientific questions about the physical nature of minds. The question of moral responsibility is a moral and ethical question, not a question for physical science. We must separate the problem of free will from the issue of moral responsibility. Fischer has written three books on moral responsibility and compiled what is the largest anthology of articles on free will, determinism, and moral responsibility - his four-volume, 46-contributor, 72-entry, 1300+ pages, Free Will, a reference work in the Routledge Critical Concepts in Philosophy series.4 Although it is titled "Free Will," the material is mostly about moral responsibility. Fischer describes the development of the idea of semicompatibilism:1
In ancient times — some fifteen years ago — I suggested that Frankfurt-type examples call into question the principle of transfer of nonresponsibility (which I then called, a bit too narrowly, the 'principle of transfer of blamelessness', following John Taurek's usage in his fascinating Ph.D. dissertation at UCLA in 1972). In the introductory essay to my anthology, Moral Responsibility, I presented a somewhat informal version of van Inwagen's modal principle (which he called principle 'B'), and (following van Inwagen) explained how it could be employed as part of a 'direct' argument for the incompatibility of causal determinism and moral responsibility (i.e., an argument for the incompatibility claim that does not employ the claim that causal determinism rules out alternative possibilities):Also, I told the following story:Of course, this case contains the distinctive characteristics of a Frankfurt-type case: a fail-safe arrangement that plays no actual role but the presence of which nevertheless ensures the actual result. I then suggested that the Frankfurt-type examples are plausible counterexamples to the principle of transfer of nonresponsibility, even though they would not be counterexamples to the parallel modal principle employed in the argument for the incompatibility of causal determinism and alternative possibilities (the principle of the transfer of powerlessness):Thus, semicompatibilism was born. Here I wish to defend the basic intuition, which I still believe is correct, that the principle of transfer of blamelessness (or, more broadly, nonresponsibility) is called into question by the Frankfurt-type cases, and that it cannot be employed in an uncontroversial, decisive argument against the compatibility of causal determinism and moral responsibility.
Fischer's technical interests have been in the area of theological determinism (compatibilism of God's foreknowledge and free will), logical determinism (the idea that truths in the "fixed past" constrain present actions), and the recent work of Harry Frankfurt to deny the existence of alternative possibilities yet affirm moral responsibility.
The set of Free Will books reflects these interests strongly, with five entries from Fischer, Frankfurt, and Peter van Inwagen (a Frankfurt critic). There are four from libertarian Robert Kane, but only one about his free-will model, the rest are on responsibility and Frankfurt examples. Daniel Dennett appears only once, criticizing Kane's indeterministic decision-making model. Two excerpts from Laura Waddell Ekstrom's Free Will - Varieties of Libertarianism provide other incompatibilist models.
Volume III, Part 1 of Free Will is devoted to libertarian (incompatibilist) accounts of free will. Here is how Fischer describes it:
Given the appeal of the fundamental argument for incompatibilism about causal determinism and the sort of freedom that involves genuine access to alternative pathways into the future, one might be attracted to an indeterministic notion of freedom. The libertarian believes that causal determinism would rule out free will (this is incompatibilism), but that causal determinism is false. Further, the libertarian holds that we do in fact have free will. Just as with the compatibilist, it is not enough for the libertarian simply to assert our freedom — it is desirable to provide some sort of account of such freedom. There are two basic strategies for coming up with an indeterministic analysis of freedom. The first is the "event-causal approach." On this approach, prior events do in fact cause our decisions and behavior, but via indeterministic sequences. The structure and nature of some of these sequences confers control and moral responsibility on agents. Kane offers a detailed and sophisticated event-causal account of indeterministic freedom ("Free will: New directions for an ancient problem"). It is important to Kane to render his libertarianism consistent with materialism about the mind, and, indeed, with contemporary neuroscience and physics. Some libertarian approaches, particularly in the past, have been committed to extravagant or unscientific excesses; Kane is at pains to avoid such implausible commitments. Dennett offers a critique of Kane's analysis of freedom. Laura Ekstrom offers an alternative, event-causal model of freedom ("Excerpt from 'Varieties of libertarianism'"). O'Connor presents a critique of event-causal indeterministic models. A fundamental worry, according to O'Connor, is that the event causal approach does not confer control on the agent. Clarke ("Event-causal accounts and the problem of explanation"), and Kane ("Responsibility, luck, and chance: reflections on free will and indeterminism") discuss and respond to some such worries. The following selections by Clarke ("Agent causation and event causation in the production of free action") and O'Connor ("Excerpt from 'The metaphysics of free will'") develop cutting-edge versions of agent-causal libertarianism. On this view, agents cause certain events (choices, decisions, actions), but they are not caused by prior events to cause these events. Further, when an agent causes such events, the causation by the agent is basic: the agent's causal role cannot be reduced to that of an event or set of events. Whereas this approach may seem attractive to those who are concerned that event-causal theories do not offer enough in the way of control, there are various problems with agent-causation, Some of these problems are explored in Ekstrom ("Excerpt from 'Varieties of libertarianism'") and Pereboom ("'Empirical objections to agent causal libertarianism'").6
Fischer's view of moral responsibility takes its basic starting point from Peter Strawson's 1962 essay Freedom and Resentment.
In his landmark essay, "Freedom and resentment," Peter Strawson argues that the concept of moral responsibility is analyzed in terms of a set of attitudes he termed the "reactive attitudes": indignation, resentment, hatred, respect, gratitude, and love. A morally responsible agent is the target (or perhaps appropriate target) of such attitudes. Practices such as punishment presuppose and express the reactive attitudes. Strawson suggests that human life without moral responsibility, so understood, would be extremely unattractive, if coherently conceivable at all. Further, he suggests that his analysis of moral responsibility can help to show that moral responsibility is compatible with causal determinism.7 That we apply the reactive attitudes toward ourselves and other persons signals something extremely important: we take a certain distinctive perspective toward persons (as opposed to nonpersons). We are engaged with persons. In contrast to the perspective from which we view other persons, our perspective toward nonpersons tends to be "objective." We treat nonpersons as objects to be used, exploited, manipulated, or perhaps just enjoyed. But we do not have attitudes such as resentment or love toward them; rather, we view them from a more detached and uninvolved - a more objective - perspective.8
How does Fischer make the case for "semicompatibilsm" appealing to reflective thinkers? In his 1998 book Responsibility and Control, written with Mark Ravizza, Fischer hopes to avoid the implications of strict causal determinism. Determinism is such a dark idea, one that "hard determinist" Ted Honderich calls a "black thing."9 And in any case, Fischer says, we don't really know whether determinism is true or false
Ordinarily, we simply assume that we and other human beings are persons and are at least sometimes morally responsible agents. Thus, we assume that we (most of us) at least sometimes have the kind of control that grounds moral responsibility and personhood. Typically, this assumption is deemed so obvious as not to command any attention or elicit even the slightest bit of controversy. But imagine now that a certain doctrine turns out to be true: the doctrine of causal determinism. Roughly, causal determinism is the view that all events can in principle be fully explained by reference to past states of the world and the laws of nature. Slightly more carefully, "Causal determinism is the thesis that, for any given time, a complete statement of the facts about that time, together with a complete statement of the laws of nature, entails every truth as to what happens after that time." We certainly do not know that this doctrine is true; indeed, many contemporary physicists would claim it is false. But then again we do not know that it is false. For all we know, contemporary physicists could announce that their previous theories were defective, and that the indeterminacies those theories posited were the result of inadequacies of information and analysis. They could in the future develop an entirely deterministic theory of the universe. 10Fischer separates an agent's control into two kinds he calls "legislative control" - the kind needed to choose between alternative possibilities, and "guidance control" - the kind of control needed to initiate or originate an action, by being "reasons responsive" and taking ownership of the action, meaning the agent can say the action was " Our contention is that even if causal determinism were true, there is a strong impetus to think that human beings should still be properly considered persons, morally responsible, and at least sometimes in control of their behavior. That is, even if we discovered that causal determinism were true, there is a strong tendency to think that this sort of discovery should not make us abandon our view of ourselves as persons and morally responsible agents.up to me." Fischer is convinced enough by Harry Frankfurt's arguments against the Principle of Alternate Possibilities to give up the need for legislative control. All Fischer says he needs for his "semicompatibilism" is guidance control in the "actual sequence" of events. Hypothetical alternate sequences are unimportant. Incompatibilist critics attack the idea of guidance control with what they call "sourcehood" or "source incompatibilism," which requires that the source of an action be "internal" to the agent. Fischer notes that if causal determinism is true, then our behavior is the result of external causally deterministic sequences that began well before we were even born. Compare Peter van Inwagen's Consequence Argument. Fischer says source incompatibilists call for some kind of "gaps" in the actual causal sequence to make room for free will. Fischer does not go into any detail on these gaps, bur they may be something like a causa sui. Fischer calls the kind of "total control" demanded "a total fantasy - metaphysical megalomania." 12
So total control is a chimaera. It is manifestly ludicrous to aspire to it or to regret its absence. The locus of control is not wholly within us. We do not exist in a protective bubble of control. Rather, we are thoroughly and pervasively subject to luck: actual causal factors entirely out of our control are such that, if they were not to occur, things at least might be very different. Quite apart from any special assumption about causal determinism, we can see that from a broader perspective, it is entirely a matter of luck or arbitrary that I behave as I do (or even that I developed into an agent at all — or have maintained that status). Although it is perfectly reasonable to wish to be the source of one's choices and behavior, it is not reasonable to interpret the relevant notion of sourcehood in terms of total control and internality.13 Nietzsche famously said, "the causa sui is the best self-contradiction that has been conceived so far; it is a sort of rape and perversion of logic." The quotation is from Twilight of the Idols, or: How to Philosophize with a Hammer, section 8, "The Four Great Errors." Now I am not sure that the causa sui would make my Top Ten List of Good (or perhaps Egregious) Self-Contradictions, but to be the cause of oneself (in a stringent way) is surely an unreasonable aspiration. Whereas some philosophers would claim (with Nietzsche) that being a causa sui is both ludicrous and part of commonsense, I would urge that we note that being the "initiator" or "source" of our choices and behavior is indeed part of commonsense, but that it is inchoate and undeveloped in commonsense. We should not be quick to attribute a ludicrous and obviously self-contradictory notion to commonsense. Rather, we should seek to capture the kernel of truth embedded in our ordinary conceptual scheme and articulate it in a more plausible, attractive way. (p.70) Since taking responsibility is one of the components of guidance control, we need to argue that the conditions we have specified for taking responsibility are compatible with causal determinism. Clearly, causal determinism does not rule out an individual's believing that he is an agent (in our sense) and that the given social practices render him a fair target for the reactive attitudes in certain circumstances.14 Things get a bit more complicated when we turn to individuals who are philosophically sophisticated and have reflected on the relationship between causal determinism and the fairness of the application of the reactive attitudes. So, imagine that an individual has immersed himself in the debates about causal determinism, free will, and moral responsibility. Presumably, most reflective individuals will not be confident about what to think here. That is, most reflective individuals will find considerable force in arguments on both sides. We believe that many (although not all) of these open-minded individuals can be brought to adopt a certain stance. More specifically, they can be brought to think that it is at least plausible that causal determinism does not rule out the aptness of the reactive attitudes. Further, they can be convinced – if they need to be convinced – that, for all practical purposes, they should "put aside" their doubts about the consistency of causal determinism with the aptness of the reactive attitudes. 15 Why should such an individual deem himself a prima facie plausible candidate for the reactive attitudes, and be willing to put aside metaphysical worries? We believe that the considerations developed thus far in this book can move a reflective individual in precisely this direction. First, we have sought to defend the idea that the sort of control that involves alternative possibilities is not required for moral responsibility. Thus, we have attempted to remove what is probably the most significant objection to the compatibility of causal determinism and the appropriateness of the reactive attitudes. This should move reflective, open-minded individuals toward adopting the stance we have specified. Further, we have contended that moral responsibility is grounded in a kind of control – guidance control – with two components. The first component is moderate reasons-responsiveness of the mechanism leading to the behavior in question. And we have argued that this sort of responsiveness is entirely compatible with causal determinism. Of course, the second component remains – the ownership condition. But we would suggest that the Frankfurt-type examples are also illuminating here. 16 We concede that some individuals will not be convinced. For example, some individuals are "natural incompatibilists"; when they adopt the assumption of causal determinism, they might be dubbed "natural hard determinists." Such individuals will not deem themselves apt targets for the reactive attitudes, and thus they will not take responsibility for the kinds of mechanisms that lead to their behavior. Thus, on our account, they will not be morally responsible for their behavior. But we do not take this to be a defect of our theory. Indeed, it follows straightforwardly from the fact (noted earlier) that we agree with Galen Strawson in embracing a "subjectivist" approach to moral responsibility. Recall that this sort of approach requires that an agent have a certain kind of view of himself, in order to be morally responsible for his behavior. And this is precisely the case, on our account of taking responsibility (and moral responsibility). In order to be morally responsible, a person must see himself as an agent who is an appropriate candidate for the reactive attitudes.17
In their 2007 book Four Views on Free Will, Fischer, Robert Kane, Saul Smilansky, and Manuel Vargas, present their positions and comment on one another's views. Fischer makes it clear that he is trying to develop a compatibilist position based on a priori metaphysical truths that will remain defensible even if causal determinism is found to be true. He says that a compatibilist need not flip-flop metaphysically and give up his assumption,
"even if he were to wake up to the headline, "Causal Determinism is True!" (and he were convinced of its truth). Nor need the compatibilist give up any of his basic metaphysical views — apparently apriori metaphysical truths that support his views about free will — simply because the theoretical physicists have established that the relevant probabilities are 100 percent rather than 99 percent. Wouldn't it be bizarre to give up a principle such as that the past is fixed and out of our control or that logical truths are fixed and out of our control, simply because one has been convinced that the probabilities in question are 100 percent rather than 99 percent. A compatibilist need not "flipflop" in this weird and unappealing way." 18
Since the time of Peter van Inwagen's 1983 classic Essay on Free Will which introduced the Consequence Argument, Fischer has been arguing the case for a compatibilism that focuses on moral responsibility and agent control rather than compatibilist free will per se. Fischer also has been influenced by Harry Frankfurt's attack on what Frankfurt called the Principle of Alternate Possibilities (PAP). Before Frankfurt, compatibilists and incompatibilists alike had argued that alternative possibilities seemed to be a condition not only for free will but for moral responsibility. Frankfurt's clever examples changed the debate from compatibilism vs. incompatibilism to the very existence of alternative possibilities. Although attacks and counterattacks continue, Frankfurt-style examples have become far too arcane and unlikely to win support outside a small number of compatibilists and incompatibilists. Nevertheless, Fischer has tried to carve out a position called semicompatibilism, which de-emphasizes alternative possibilities and emphasizes agent control. Fischer hopes that semicompatibilism will be resistant to any discovery by science that strict causal determinism is true. He does this by dividing the needed agent control into two parts, "regulative control" and "guidance control." Regulative control involves alternative possibilities, which lead to what Fischer calls "alternative sequences" of action. Fischer thinks he can simply deny that agents have regulative control, and bypass the question of alternative possibilities, based on Frankfurt-style examples. Although Fischer generally supports Frankfurt-style examples, he is the author of one of the cleverest counterattacks, the idea that the mere possibility that the agent might try an alternative gives rise to a "flicker of freedom." 19 Fischer wants to focus our attention on the more critical guidance control, which describes the "reasons-responsiveness" and "sourcehood" involved in the "actual sequence" of events leading up to the agent's action. For Fischer, no alternative sequences, however many and however they flicker with freedom, are as relevant as the actual sequence. Being the source of our actions allows us to say that our actions are "up to us," that we can take ownership of our actions. This is what Fischer regards as the "freedom-relevant condition." It is what Robert Kane calls our "ultimate responsibility (UR)." And it is what Manuel Vargas calls the "self-governance condition" in his Revisionism. Kane, Vargas, Derk Pereboom and Fischer co-authored the recent book Four Views on Free Will. Pereboom also focuses on moral responsibility like Fischer, but he disagrees with Fischer that moral desert justifies praise and blame, reward and punishment. At the most, says Pereboom, responsibility can justify that we can be "legitimately called to moral improvement." Desert implies retributivism. Pereboom says the most we can justify is moral rehabilitation, for its beneficial consequences to society. Parallel to Fischer's guidance control (sourcehood in the actual sequence) and regulative control (alternative possibilities), Pereboom distinguishes source incompatibilism from leeway incompatibilism, the latter corresponding to Fischer's regulative control, which depends on the existence of alternative possibilities. Although Fischer is officially agnostic on the ancient problem of free will versus determinism, he shows a strong commitment to causality and determinism over his years of defending compatibility with determinism. Nevertheless, Fischer's dividing of agent control issues into regulative control (involving alternative possibilities) and guidance control (what happens in the actual sequence) is an excellent approach that allows us to situate the indeterminism that many thinkers feel is critical to any libertarian model. Fischer notes that indeterminism in the alternative possibilities might generate "flickers of freedom." And he says clearly (Four Views, p.74) that guidance control is not enhanced by positing indeterminism. In his 1998 book Responsibility and Control, written with Mark Ravizza, Fischer describes what he calls the Direct and Indirect Arguments for incompatibilism. The Indirect Argument says that determinism rules out alternative possibilities. From his semicompatibilist view, that does not threaten moral responsibility. Only in the Direct Argument for incompatibilism does determinism rule out moral responsibility. So might Fischer agree with a view that 1) allows the "freedom-relevant condition" (reasons responsiveness and ownership) in the actual sequence to be governed by what he calls "almost causal determinism" (Responsibility and Control, p.15n) and 2) allows indeterminism in the generation of the alternative possibilities (flickers of freedom)? That is the view we offer in the I-Phi Cogito model. Although they do not endorse it themselves, Daniel Dennett and Alfred Mele have also offered this view as something libertarians should like. Indeterminism is important only in microscopic structures, but that is enough to introduce noise and randomness into our thoughts, especially when we are rapidly generating alternatives for action by random combinations of past experiences. But our brain and our neurons can suppress microscopic noise when they need to, insuring what we call adequate determinism, what Fischer calls almost causal determinism, and what Ted Honderich calls near determinism - in our willed actions. In Robert Kane's contribution to Four Views on Free Will, he correctly identifies noise in messages as generated indeterministically, but mistakenly thinks these are merely a "hindrance or obstacle" that raises our level of effort when making his rare but morally significant "self-forming actions." The role of indeterminism in free will is better seen as simply generating Fischer's AP "flickers of freedom." These alternative possibilities are then the "free" part of "free will" (Fischer's regulative control). The "will" part (Fischer's guidance control) is "almost causally" determined to be reasons responsive and to take ownership for the determination to act in a fashion consistent with the agent's character and values. Event-causal libertarians like Kane, Mark Balaguer, and Laura Waddell Ekstrom think this kind of freedom is not enough. And agent-causal libertarians like Randolph Clarke and Timothy O'Connor want even more "metaphysical" freedom. They say that if the will is determined to act in a rational way consistent with its character and values, then the agent will make exactly the same decision in exactly the same circumstances. Such consistency of action does not bother the common sense thinker or the compatibilist (even a hard incompatibilist?) philosopher. Kane, Balaguer, Ekstrom, and others continue to invoke some indeterminism in the decision process itself. As Daniel Dennett recommended as early as 1978 (in Brainstorms) and Alfred Mele has been promoting as a "modest libertarianism" in his recent books (Autonomous Agency and Free Will and Luck), indeterminism is best kept in the early stage of a two-stage process. We first need free (alternative possibilities) and then will (adequately determined actions) in a temporal sequence. First chance, then choice. I think that John Martin Fischer's guidance control, perfectly compatible with his "almost causal determinism," validates not only his semicompatibilist view of moral responsibility, but also supports the common sense or popular view of free will that is found in the opinion surveys of experimental philosophers Joshua Knobe and Shaun Nichols. While limited compared to "metaphysical" freedom, this view is consistent with a broadly scientific world view, a requirement for any systematic revisionism that Manuel Vargas calls "naturalistic plausibility." 20 Ironically perhaps, this view would be the very opposite of a revisionism, in the sense that the diagnostic (descriptive) analysis of common sense would agree remarkably well with what Vargas calls the prescriptive view for philosophers. Or perhaps it is the philosophers' views that need revision? As an illustration of just how naturalistically plausible this new view of free will is, consider the case of biological evolution. The evidence is overwhelming that variations in the gene pool are driven by random mutations in the DNA. Many of these mutations are caused by indeterministic quantum mechanical events, cosmic ray collisions for the most part. Think of the mutations as alternative possibilities for new species. An adequately determined process of natural selection then weeds out those random variations that can reproduce themselves and compete with their ancestors. First chance, then selection. Indeed, the story of life is maintaining some information stability (parts of our DNA have been the same for 2.8 billion years) in a chaotic environment - and not the pseudo-random deterministic chaos of the computer theorists, but real irreducible chaos. Only a believer in metaphysical determinism would deny the evolutionary evidence for indeterminism and two stages, the first microscopic and random (chance) the second macroscopic and adequately determined (choice). Sadly, such a metaphysical belief is the intelligent design position of the creationists. Of course we are discussing only science, not logical certainty. So we can also ameliorate John Martin Fischer's nightmare of waking up one morning to a New York Times headline "Causal Determinism Is True." 21 Nothing in science is logically true, in the sense of true in all possible worlds, true by the principle of non-contradiction or the weaker law of the excluded middle. It is the excluded middle argument that leads us to the muddled standard argument against free will. Our two-stage argument is quite old. We can trace it back to William James (1884 in The Dilemma of Determinism), Henri Poincaré (1906), Arthur Holly Compton (1935), and Karl Popper (1961). What does Information Philosophy have to do with the two-stage model? Information is the principal reason that biology is not reducible to chemistry and physics. Information is what makes an organism an individual, each with a different history. No atom or molecule has a history. Information is what makes us ourselves. Increasing information is involved in all "emergent" phenomena. In information philosophy, the future is unpredictable for two basic reasons. First, quantum mechanics shows that some events are not predictable. The world is causal but not determined. Second, the early universe does not contain the information of later times, just as early primates do not contain the information structures for intelligence and verbal communication, and infants do not contain the knowledge and remembered experience they will have as adults. The universe began in a state of minimal information nearly fourteen billion years ago. Information about the future is always missing, not present until it has been created, after which it is frozen. John Martin Fischer calls this the "Principle of the Fixity of the Past." 22 It suggests that even divine foreknowledge is not present in our open expanding universe, lending support to the religious view called Open Theism.
Fischer's Two-Stage Model of Libertarian Free WillIn 1995, in a debate with David Widerker, Fischer proposed a variation on Daniel Dennett's "Valerian" model. Like Dennett , he had his doubts about whether the model could solve the control problem. Fischer uses the Dennett idea - that the indeterminism comes at an early stage of the overall deliberation-decision process - to locate a Frankfurt-style "prior sign" needed by the hypothetical intervener at a place deterministically linked to the decision and subsequent action. Fischer's main criticism of alternative possibilities for action is that it is implausible to suppose that one's moral responsibility is grounded on the possibility of forming a certain sort of judgment about what is best: a judgment on behalf of doing something there are no good reasons to do. The responsibility for doing good is not grounded in the possibility of doing bad. Note that freedom of action is completely independent of, and merely a prerequisite to, moral responsibility. Otherwise it would be the ethical fallacy. Fischer hopes to develop "another sort of libertarianism." He says he does not have the space to lay out his "second family of libertarian accounts," and gives us very little on how it differs from Dennett. He says "Dennett argues that it is the only sort of libertarianism that is plausible, and I believe that it is at least minimally plausible. I also believe that it is libertarianism." It may simply be a libertarianism with a built-in place for the Frankfurt intervener, in order to support the absence of alternative possibilities and Fischer's semicompatibilism. Here is Fischer's sketch of his main idea.
I wish to develop (in an extremely sketchy way) another sort of libertarianism; on this kind of approach, the relationship between the relevant "sign" or "signal" and the subsequent choice is causally deterministic, but there is nevertheless a lack of causal determination along the sequence that issues in the decision (and action). And I shall point out that this approach also seems to lead to the view that an agent can be morally responsible for making a choice even though he could not have (at any relevant time) made a different choice. I do not have the space here to lay out this second family of libertarian accounts fully or carefully. But I shall simply sketch the main ideas and hope that enough of the content of the approach will emerge to convince the reader that this family of views constitutes a minimally plausible, serious libertarian approach-worth further elaboration and evaluation in the context of the issues under discussion here. In his article, "On Giving Libertarians What They Say They Want," Daniel Dennett has presented this family of approaches; he does not necessarily endorse the view, but presents it as the most plausible and appealing version of libertarianism.7 What is crucial to Dennett's view is that indeterminacy be installed at the appropriate place, and Dennett argues that this is not between the judgment that a particular act is the best among one's alternatives and the subsequent choice. He says, "Clearly, what the libertarian has in mind is indeterminism at some earlier point, prior to the ultimate decision or formation of intention...."8 Rather, Dennett argues that there can be lack of causal determinism (of a certain sort) within the process of deliberation that leads to the agent's judgment as to what is the best option (under the circumstances). He attributes the following thought to the poet, Paul Valery, and claims that it nicely captures the basic idea of the approach he is suggesting on behalf of the libertarian:It takes two to invent anything. The one makes up combinations; the other one chooses, recognizes what he wishes and what is important to him in the mass of the things which the former has imparted to him. What we call genius is much less the work of the first one than the readiness of the second one to grasp the value of what has been laid before him and to choose it.9Dennett goes on to say:When someone is faced with an important decision, something in him generates a variety of more or less relevant considerations bearing on the decision. Some of these considerations, we may suppose, are determined to be generated, but others may be non-deterministically generated....Those considerations that are selected by the agent as having a more than negligible bearing on the decision then figure in a reasoning process, and if the agent is in the main reasonable, those considerations ultimately serve as predictors and explicators of the agent's final decision.10So Dennett's picture suggested on behalf of the libertarian involves some lack of causal determination in the process of deliberation, but no such lack in the link between the judgment as to what is best and the formation of an intention (or the making of a decision). Let me emphasize that I am not in a position here fully to lay out this view (or set of views) or to defend it. Dennett argues that it is the only sort of libertarianism that is plausible, and I believe that it is at least minimally plausible. I also believe that it is libertarianism. Note that Widerker only considers those forms of libertarianism according to which no state of the world (including the judgment as to what is best) prior to the decision causally determines the decision. But this unduly restricts the options open to the libertarian, and it was not the understanding of libertarianism with which I operated in "Responsibility and Control"; there I spoke more broadly of a lack of determination in the actual sequence issuing in the decision and action:11Now if roughly the sort of libertarianism suggested by Dennett is correct, then we can take the prior sign to be the agent's judgment about what is best to do. By hypothesis this sign is deterministically related to the subsequent decision. Given the approach suggested by Dennett, the example of Jones and Black can be developed as follows. Prior to T, Jones engages in deliberation; some aspects of this deliberation - perhaps the precise considerations that emerge or the precise order of Jones' reflections - are not causally deterministic. At T Jones comes to judge that voting for Reagan is best. On the basis of this judgment, at T+i Jones decides to vote for Reagan. Given the libertarian view of the fixity of the past, Jones cannot at T+i refrain from deciding to vote for Reagan. And yet the actual sequence that issued in his decision was not causally deterministic. Further, given the presence of Black and his ability to intervene should Jones form the judgment at T that voting for Carter would be best, it is true in the example that Jones at no relevant time has the ability to decide to vote for Carter (or anyone else). And yet he may be deemed by the libertarian morally responsible for voting for Reagan. But perhaps Widerker would here object that I have simply pushed the debate back to the issue of whether the relevant agent can make a different judgment as to what is best (and how this ability relates to moral responsibility). And I agree that in a full discussion of the relevance of alternative possibilities to moral responsibility one would need carefully to consider these matters.12 Let me say a few brief words here. There obviously are cases (perhaps different from the Jones/Black case) in which it is absolutely clear what one should do - cases in which there are extremely strong reasons to do something and no good reasons not to. For example, a baby has fallen into a swimming pool in front of you and is in immediate danger of drowning. All you have to do is bend over and pick the baby up; this would be extremely easy for you, and we may suppose that there are no other morally relevant reasons. On the picture suggested by Dennett and given the presence of a counterfactual intervener such as Black, if you decide to save the baby, you may well be morally responsible for this decision even though you could not have made a different decision. And this is compatible with lack of causal determination in the sequence leading to the decision: the precise ordering of considerations in the (admittedly brief) deliberations that preceded your decision may have been indeterministic. (Thus far, the analysis implies that things are the same as in the Jones/Black case.) Now is it plausible here to say that it is in virtue of the fact that you could have formed a different judgment as to what is best that you are morally responsible for your decision? That is, does the existence of this alternative possibility ground your moral responsibility for your decision? I do not deny that the alternative possibility exists, but I do very much doubt that it is what grounds your moral responsibility. For what would such an alternative possibility be like? It would be the possibility to judge best something for which there are no good reasons - failing to bend over and save the baby. And it does not seem to me plausible to say that this kind of possibility is what grounds your moral responsibility for your decision.13 Thus, I believe that this case is plausibly construed as a case in which the actual sequence exhibits the lack of causal determination, the agent does not have the ability to make a different decision, and the agent is morally responsible for making his decision; further, the ascription of moral responsibility is not based on the existence of any alternative possibility.
Fischer is a founder and general advisor to the Garden of Forking Paths group blog on free will and moral responsibility.
Fischer's Objection to Indeterminism and Alternative PossibilitiesIn 1999, Fischer wrote a short review of Alfred Mele's 1995 book Autonomous Agents, which gives us a good idea of Fischer's attitude toward two-stage models like our Cogito. Fischer cannot make sense of how indeterministic alternative possibilities could add control compared to an agent who is completely determined. For the determined agent, he says, what "comes to mind," one's prior states — desires, beliefs, values, general dispositions — determine the precise content and ordering of the subsequent doxastic states (that constitute deliberation). Having this all totally determined seems to Fischer to be more control than allowing anything new to enter for the agent to choose from.
On Mele's picture of libertarian agency and control, the precise sequence of doxastic states leading ultimately to a judgment as to what is best is undetermined, although the agent's standing desires, values, and general dispositions constrain the content and presumably the order (to some degree) of the states. Everything else in the sequence — the formation of a best judgment based on one's deliberations, the transitions to an intention, a proximal intention, and an action are all deterministic.Whereas Daniel Dennett has suggested a similar picture, a distinctive twist is added by Mele: he argues that indeterminism of the sort he posits — internal, doxastic indeterminacy — is no worse, in respect to control, than determination in this portion of the sequence leading ultimately to action. This is because the agent typically does not directly control the sequence of doxastic states he undergoes — this is generally a passive rather than an active process. Mele says:...notice that we are not always in (proximal) control of which of our beliefs come to mind anyway, even if determinism is true. Assuming determinism, everything that happens on this front is causally determined, but the causal story often does not place the agent in the driver's seat. (p. 215)So Mele's suggestion is that, even though installing indeterminacy may seem to erode genuine control, installing it at this particular place (at some points in the sequence of doxastic states which constitute deliberation) does not diminish control that otherwise would be present (under the assumption of determinism); a proponent of doxastic indeterminacy is thus not worse off than a proponent of doxastic determination, with respect to control. Additionally, Mele contends that by installing indeterminacy in the sequence leading to action, one is able to preserve the crucial libertarian belief in alternative possibilities or freedom to choose and do otherwise. Of course, one has to be careful about the specification of the temporal index associated with the alternative possibility. It will be true during the period of doxastic indeterminacy in the agent's deliberation that he has the power to form different best judgments and thus to pursue more than one path in the future; but after the period of doxastic indeterminacy (and just prior to the formation of the agent's best judgment), he will no longer have alternative possibilities, insofar as the process is now causally deterministic. My puzzle could be put as follows. How can adding arbitrariness of the sort envisaged — the lack of determination of the beliefs that come to mind during deliberation — to a causally deterministic process yield genuine control? A libertarian of course will contend that an entirely deterministic process does not contain genuine control by the relevant agent. How, then, can installing the sort of indeterminacy envisaged — indeterminacy as to which belief states will come to the agent's mind — transform the sequence from one of lack of control to one containing control? This smacks of alchemy. 13 Perhaps my point could be made in terms of a crucial distinction (of which Mele is certainly aware) between an agent's having control over what happens, and its being the case merely that something different might have occurred. If an agent has genuine control in the sense of possessing alternative possibilities, he can make it the case that one path is followed, or another path is followed, in accordance with what he judges best and chooses. He can deliberately pursue one course of action, or deliberately pursue another; what path the world takes (at least in certain respects) is "up to him." In contrast, when it is merely possible that something different have occurred, the path the world takes need not depend in the relevant way on the agent. In a genuinely random event, presumably there are various metaphysically open possibilities; but by definition no agent has control over what happens. Now it seems to me that, whereas it may well be possible that Mele's libertarian agent do something different from what he actually does, it is not clear that he has genuine control over what he does. Mele admits that the precise sequence of doxastic states can have an effect on what the agent judges best (and then does); given that the sequence is not (entirely) determined by prior states of the agent (although it is constrained by such states), it is not clear that what the agent judges best and then does is genuinely up to him. Mele's point that even on a deterministic model, the agent is not "in the driver's seat" with respect to which considerations come to mind (and thus that the libertarian is here not losing anything with respect to control) is an intriguing and suggestive idea. But presumably the compatibilist will point out that, even though the agent does not directly control what belief-states come to mind (in the sense of choosing them or willing them), they are envisaged as strongly connected to the agent's prior states to the extent that they are a deterministic product of those past states. Under determinism, one's prior states — desires, beliefs, values, general dispositions — determine the precise content and ordering of the subsequent doxastic states (that constitute deliberation), even if the agent does not directly control what doxastic states he will be in (and thus is not in the "driver's seat," in this sense). It may then be possible to argue that one does give up some measure of control, when one shifts from thinking of the doxastic sequence as deterministic to thinking of it as indeterministic: one gives up the notion that the states constituting one's deliberations are an "outflowing" of the agent's prior states in a strong sense. How one assesses Mele's libertarianism seems to me to hinge on whether one believes that the doxastic states that constitute one's deliberations can be a genuine "outflowing" of the agent's prior states, even though they are not causally determined by those prior states. I am not convinced, however, that from the mere fact that the agent is deemed passive in regard to his doxastic states even under determinism, it follows that one does not attenuate the agent's control in positing doxastic indeterminism. Thus, whereas Mele's suggestion on behalf of the libertarian is suggestive and argued with considerable resourcefulness, I am left with nagging doubts about the strategy.
Notes1. Fischer, 2004, p.189 2. Fischer 1986, p.60 3. Fischer 1986, p.41 4. Fischer 1986, p.60-1 5. Fischer, 2005, v.I, p.xxiii 6. Fischer, 2005 v.III, p.1 7. Fischer, 2005 v.I, p.1 8. Fischer and Ravizza, 1999, p.6 9. Honderich, 1998, p.12 10. Inwagen, 1983 pp. 197-198. 11. Fischer and Ravizza, p.14-15 12. Fischer et al., 2007, p.67 13. Fischer et al., 2007, p.68 14. Fischer and Ravizza, p.225 15. Fischer and Ravizza, p.226 16. Fischer and Ravizza, 1999, p.227 17. Fischer and Ravizza, 1999, p.228 18. Fischer et al., 2007, p.47 19. Fischer 1994, p.131-159 20. Fischer et al., 2007, p.153 21. Fischer et al., 2007, p.44 22. Fischer and Ravizza, 1999, p.22 (Responsibility and Control, p.22).
BibliographyFischer, J. M., 1982, "Responsibility and Control," Journal of Philosophy, vol. 89 , pp. 24-40. ---- 1986. "Introduction:Responsibility and Freedom," In J. Fischer, ed., Moral Responsibility. Ithaca, NY: Cornell. ---- 1994. The Metaphysics of Free Will: An Essay on Control, Cambridge, Mass: Blackwell. ---- & Ravizza, M., 1999. Responsibility and Control: A Theory of Moral Responsibility New Ed., Cambridge University Press. ---- 2004, "The Transfer of Nonresponsibility," in Freedom and Determinism, ed. Joseph Keim Campell et al., MIT, Cambridge. ---- 2005, Free Will: Critical Concepts in Philosophy, Routledge, London. ----, Robert Kane, Derk Pereboom, & Manuel Vargas, 2007, Four Views on Free Will, Blackwell. ---- 2008, "Freedom, Foreknowledge, and Frankfurt: A Reply to Vihvelin," Contemporary Debates in Metaphysics, ed. Theodore Sizer, John Hawthorne, Dean W. Zimmerman, Blackwell, pp.302-18. Honderich, T., 1998. The Consequences of Determinism, Clarendon, Oxford Inwagen, P. v., 1983. An Essay on Free Will, Oxford University Press
Review of Autonomous Agents, by Alfred R. Mele
Introduction to Fischer, Free WillThe network of concepts associated with the traditional idea of Free Will has received a huge amount of attention by philosophers (as well as social scientists, legal theorists, and authors of fiction). This is certainly one of the most spectacularly active and vital areas of contemporary philosophical research. Although the topics are enticing, the sheer volume of the work can be daunting. A primary purpose of this collection is to present and organize much of the best contemporary work in philosophy on free will. Regrettably for the editor (but perhaps not all readers!), even four volumes are not nearly enough to include all of the very high quality work on this set of topics, and I have had to make some somewhat arbitrary decisions about topics and selections to exclude. I hope however to provide a kind of "roadmap" to the vast territory of recent work on this fascinating set of subjects. "Free will" covers a large range of phenomena. The term is used differently by different philosophers, and I think that it is most helpful to think of it as an "umbrella-term" used to describe some sort of freedom that connects in important ways with moral responsibility, and, ultimately, person-hood. More specifically, the domain of free will includes various sorts of freedom (freedom of choice, of action, choosing and acting freely, and so forth), and the practices constitutive of moral responsibility (moral praise and blame, punishment and moral reward, and a set of distinctively moral attitudes, such as indignation, resentment, gratitude, respect, and so forth). The term, "free will" is often employed when it is not necessary to be more specific about which element in the domain of phenomena is being discussed; but in particular dialectical contexts it is important at least to keep in mind the various distinct notions. Some philosophers do not distinguish between freedom and moral responsibility. Put a bit more carefully, they tend to begin with the notion of moral responsibility, and "work back" to a notion of freedom; this notion of freedom is not given independent content (separate from the analysis of moral responsibility). For such philosophers, "freedom" refers to whatever conditions are involved in choosing or acting in such a way as to be morally responsible. Most philosophers, however, would be inclined to seek to give separate accounts of freedom and moral responsibility. On this sort of approach, freedom (of a certain specific sort) may well be a necessary condition of moral responsibility, but freedom (of the relevant kind) is given a content apart from merely being whatever non-epistemic ingredient is necessary for moral responsibility. I believe that the latter approach has the virtue that it helps us to avoid conflation of different sorts of freedom, and thus helps us to see that there are importantly different threats to our possession of free will. Putting aside Quinean skepticism about the distinction between analytic and synthetic truths, it is helpful to distinguish the concepts or notions involved in free will from their conditions of application. John Rawls makes a similar distinction in his work on distributive justice (Rawls 1971). Rawls separates what he calls the concept of justice from the conditions of application of this concept. For Rawls, the concept of justice includes the abstract notion of treating like cases alike. But then there are more specific accounts of what the relevant likenesses or dimensions of similarity are, which give rise to particular accounts of the conditions of application of the concept, "distributive justice." These theories include utilitarianism of various sorts, Marxism, and Rawls's own theory of justice, including the "Priority of Liberty" and the "Difference Principle." It is not surprising that there would be a similar analytical structure with respect to notions related to retributive justice (the free will family of phenomena). In Volume I, the collection begins with a discussion of the major elements of the concepts of freedom and moral responsibility. Typically, human beings think of ourselves as both free and morally responsible (apart from specific contexts in which our freedom and moral responsibility are impaired or eliminated). One might distinguish a "forward-looking" and "backward-looking" perspective or viewpoint. From the forward-looking perspective, we engage in planning, deliberation, and practical reasoning concerning the future. Here we tend to presuppose that we are free to choose from among a range of options — that the future is, in Borges phrase, "a garden of forking paths". We are inclined to think of ourselves as having the capacity to select one path into the future, where there are (often at least) various paths extending into the future that are "really" or "genuinely" available to us. The presupposition that our freedom involves the capacity to select from among genuinely open possibilities implies that, although we make a particular choice on a particular occasion, we could have chosen otherwise (we had, at the relevant time, "alternative possibilities"). The assumption that we possess the sort of freedom that includes alternative possibilities is also a feature of the "backward-looking" perspective. This is the viewpoint we take when we evaluate our own and others' behavior. Typically, we assume that an agent is blameworthy or morally responsible for a bit of behavior (an action, an omission, an upshot, and so forth) only if he or she "could have done otherwise". Thus, the concept of a certain sort of freedom, a freedom that involves selection from among really available pathways or possibilities, is deeply ingrained in our conception of ourselves as agents. Additionally, we see ourselves as deeply and importantly different from mere animals and inanimate beings (including sophisticated computers). We human beings (except for those who have specific impairments that exempt them) are morally responsible agents, and we can legitimately be held accountable (or morally responsible) for our behavior (and perhaps elements of our emotions and characters). Human beings are (normally) "persons" —we have a serious right to life and we are (or potentially are) morally responsible agents. The concept of personhood does not exclude the possibility that members of other species or different sorts of creatures (or entities) be considered persons, but it at least entails that (in the normal case) a fully developed human being is a person. Although freedom and moral responsibility are encoded in our concepts of ourselves as agents and specifically as persons, there are different ways of understanding the relevant notions of freedom and moral responsibility. Part 1 of Volume I of the collection begins by charting this territory. In Part 2 we go on to consider a set of worries about whether our ordinary picture of ourselves as free and morally responsible agents is justified. Although we typically think of ourselves as free and morally responsible, we can be led to question the basis of these ordinary views. Of course, we do not have evidence that decisively establishes that we have the sort of freedom that involves alternative possibilities; whatever evidence we have is compatible with our sense of freedom being an illusion. And there are elements of our reflective views of ourselves and the world that can be organized into troubling arguments for skepticism about our freedom and moral responsibility. The skeptical arguments about free will are in some respects parallel to skeptical arguments in epistemology. Although we typically take ourselves to know lots of things about the world, we can be made to worry that our phenomenological evidence issues in illusory beliefs. Perhaps we are dreaming, hallucinating, or in the hands of an evil neurophysiologist who is stimulating our brains electronically. Maybe I'm a "brain-in-a vat" or in a "Matrix-like" simulation, and my experiences do not point accurately to an external world of which they are faithful images. This sort of skepticism would not be troubling, if it did not latch onto important facts about our experiences and the relationship between those experiences and the world. Similarly, skepticism about free will engages important facts (or putative facts) about truth, the past, the laws of nature, and the relationships between these notions and our agency. We begin by considering skeptical arguments that have been called versions of the doctrine of "fatalism." Ancient philosophers worried about fatalism. We shall move to a discussion of the Medieval concern that God's existence would rule out our free will. In a significant respect the skeptical arguments for fatalism and theological incompatibilism are similar: they both rest on the idea that the past is "fixed" and out of our control because it is "over-and-done-with". In Volume II, we continue by investigating the modern worry generated by science, that is, we consider the argument that full scientific predictability of our behavior would rule out freedom (in the sense that involves alternative possibilities). More specifically, in Part 1 we consider whether the doctrine of "causal determinism" is incompatible with human freedom. Again, the basic argument for the incompatibility of causal determinism and human freedom employs the commonsense idea of the fixity of the past (among other ingredients). Although the argument for incompatiblism about causal determinism and such doctrines as the existence of God or causal determinism is powerful, it is not decisive. The collection presents various ways of seeking to block the argument (and thus refute the skeptic about our free will). Consideration of the worries generated by the skeptical arguments from God's existence, and the possibility that the doctrine of causal determinism obtains, helps us to separate two elements of our concept of freedom: the idea that we select from a range of possibilities, and the idea that it is we who select — that we are the source of our behavior. The first is the alternative-possibilities element, and the second is the "source" element. Given that the skeptical arguments are not decisive, it is illuminating to consider various compatibilist and incompatibilist accounts of the conditions of application of the concepts of freedom and moral responsibility. We begin Part 2 by looking at compatibilist accounts of alternative-possibilities freedom. Of course, if some version of the skeptical argument about our free will is correct, then no compatibilist account of the conditions of application of the concepts involved in free will can be acceptable. But we are here putting aside the skeptical argument (given that it is not uncontroversially decisive). Different compatibilist accounts of freedom are considered, as well as critiques of such accounts. These critiques focus on particular compatibilist theories of freedom apart from considerations arising from the general skeptical argument for incompatibilism; so even if one does not accept the skeptical argument (in any version), one can see that there are specific problems with compatibilist accounts of freedom (in the sense that involves alternative possibilities). Given these problems (as well as the general problems presented by the basic argument for incompatibilism), one might try to develop an incompatibilistic account of our freedom (see Volume III). It is here helpful to separate "event—causal" and "agent—causal" approaches to giving an incompatibilistic account of the relevant sort of freedom. In Part I both sorts of approaches are presented, together with specific criticisms. To take stock. We typically think of ourselves as having free will — as being free and morally responsible agents. Ordinarily we take ourselves to have the sort of freedom that involves selection from a range of genuinely available paths. But there are powerful reasons to worry that our ordinary conception of ourselves may be mistaken. These reasons can be combined into various argument-forms that have a similar basic structure. This structure involves such ideas premises as the fixity of the past and the fixity of the natural laws. Although the basic argument for incompatibilism is not decisive, it does have considerable force. It casts into doubt our view of ourselves as having the sort of freedom that involves real access to alternative possibilities. Additionally, both compatibilist and incompatibilist attempts to give accounts of this sort of freedom are problematic in certain ways (despite having thoughtful proponents). It may then be promising to look more closely at the sort of freedom that is connected with moral responsibility (see Part 2). Historically, philosophers have distinguished between the "liberty of indifference" and the "liberty of spontaneity". Whereas the liberty of indifference is a sort of freedom that involves access to alternative possibilities, the liberty of spontaneity is not. The latter sort of freedom is exhibited when an individual freely manifests the distinctive human capacity for choice and action in an unimpaired fashion (roughly speaking). So, the liberty of spontaneity can be displayed even if the relevant agent lacks liberty of indifference, insofar as the barriers to access to alternative possibilities do not impair the agent's activity. John Locke presented an example that begins to prize apart the two sorts of liberty or freedom. Locke discusses a man who is put into a room while he is asleep. Unbeknownst to him, the door to the room is locked, and he cannot leave the room. When the man awakens, he decides to stay in the room, and does so, for his own reasons; the locked door plays absolutely no role in the man's deliberations or behavior. According to Locke, the man stays in the room freely, although he could not have done otherwise. We might say that he freely stayed in the room, although he lacked the freedom to leave the room. In a sense, then, the man exhibited a kind of freedom (similar to liberty of spontaneity), even though he lacked another kind (similar to liberty of indifference). Harry Frankfurt has recently developed more sophisticated versions of Locke's example that seek to address some obvious objections to the conclusions some draw from Locke's example (about the separability of the two sorts of freedom). Someone might point out that, although the man in Locke's example could not have left the room, he could have chosen to leave the room, tried to leave the room, tried to open the door, and so forth. Thus there appear to be some alternative possibilities in Locke's example, even if the man could not have left the room. Frankfurt's examples, involving a signature sort of fail-safe mechanism that can reach into the agent's very brain, seek to address these problems with Locke's version of the example. Frankfurt contends that there can be agents who choose and act freely, even though they lack freedom to choose and do otherwise. Further, Frankfurt suggests that the sort of freedom that involves alternative possibilities is not necessary for our status as morally responsible agents; choosing and acting freely is sufficient for moral responsibility. The Frankfurt examples have produced a huge literature in contemporary philosophy. Some of the major papers in this literature are collected here. The authors discuss whether Frankfurt's cases really show that there can be agents who are morally responsible, even though they lack the sort of freedom that involves selection from a range of really available paths (alternative-possibilities freedom). Whereas some authors seek to defend Frankfurt, others remain unconvinced. Philosophers who agree with Frankfurt disagree about the significance of this conclusion. Some argue that Frankfurt has helped to establish compatibilism about causal determinism (or God's existence) and moral responsibility. On their view, even if causal determinism rules out the sort of freedom that involves alternative possibilities, it does not thereby rule out moral responsibility. Further, there is no other reason to conclude that causal determinism rules out moral responsibility. Recall, however, that above we noted that the ordinary concept of freedom includes selection from a range of paths into the future and also being the source of one's choices and behavior. Even if Frankfurt's examples show (to some degree of plausibility) that alternative possibilities are not required for moral responsibility, the examples do not thereby show that causal determinism is compatible with moral responsibility. Some philosophers are willing to accept Frankfurt's conclusion about the relationship between alternative possibilities and moral responsibility, while insisting that causal determinism would rule out an agent's being the source of his behavior and thus moral responsibility. If indeed acting (choosing) freely is a separate kind of freedom from alternative-possibilities freedom, it would be helpful to have an account of it. After all, Frankfurt-examples merely show, if successful, that there is some sort of freedom that can be exhibited even in the absence of alternative-possibilities freedom. We still need a better understanding of this putative species of freedom. In Volume IV, Part 1, the collection includes various important analyses of acting freely. We have seen that there are powerful skeptical worries about our ordinary, commonsense view of ourselves as free and morally responsible agents. After all, I cannot be certain that causal determinism is not true, and if it were true, there are good reasons to think that, despite my intuitive conception of myself, I lack access to alternative possibilities. Similar considerations apply to God's existence. Additionally, there are specific challenges to particular compatibilistic accounts of freedom and moral responsibility. Seeking refuge in indeterminism may not seem any more attractive, given the problems with specific indeterministic accounts of freedom. These problems can be seen to be quite general — they flow from the concern that indeterminism erodes control, and that control is required for the relevant kind of freedom and for moral responsibility. This dilemma was presented by William James in a particularly striking way. James's "dilemma of determinism" is as follows. Either causal determinism is true, or it is not. If it is true, then we would lack freedom (in the alternative-possibilities and source senses). If it is false, then we would lack freedom in that we would not select the path into the future — we would not be the source of our behavior. Indeterminism appears to entail that it is not the agent who is the locus of control. Such concerns have prompted some philosophers to conclude that free will is a mystery or perhaps that our ordinary notion of free will is fundamentally incoherent (see Part 2). The "Mysterian" view about free will is parallel to such a view about consciousness and the nature of the mind (and its relationship to the brain). Philosophers who hold such a view about free will, or who think that our ordinary views about free will are fundamentally incoherent, disagree about "where to go from here". Some think that we ought to abandon the entire conceptual framework of free will and all of the practices associated with moral responsibility, and to replace them with what are allegedly more accurate, honest conceptualizations and more humane practices. These may simply involve positive and negative reinforcement of behavior with an eye to producing more desirable and socially acceptable behavior. Others believe that we can preserve much of what we care about in a sort of revised conceptualization and associated set of practices. Some such philosophers would prune out practices of what they take to be an unjustified "retributive" component. According to these philosophers, we can have much of what we care about in free will without supposing that we are free and morally responsible in a robust sense. As I said at the beginning, there is much excellent work on the topics mentioned above that I have simply not had the space to include. Also, I have had to limit the topics. Some topics that could not be included due to constraints of space are: neuroscience and free will, "empirically-informed" philosophy and free will, and the connections between free will and distributive justice. Some excellent collections that include important contributions to these and other topics are O'Connor (1995), Pereboom (1997), Kane (2002a,b), Buss and Overton (2002), Widerker and McKenna (2003), and Watson (2003).