Determinism al Dente
Derk Pereboom, University of Vermont
Noûs 29 (1995), pp. 21-45
                                                           
Al dente means "firm to the bite," and that is how Italians eat pasta.  Soft pasta is no more fit to eat than a limp and soggy slice of bread.  As soon as pasta begins to lose its stiffness and becomes just tender enough so that you can bite through without snapping it, it is done.  Once you have learned to cook and eat pasta al dente, you'll accept it no other way.  (Marcella Hazan, The Classic Italian Cookbook, pp. 90-1)

    The demographic profile of the free will debate reveals a majority of soft determinists, who claim that we possess the freedom required for moral responsibility, that determinism is true, and that these views are compatible.  Libertarians, incompatibilist champions of the freedom required for moral responsibility, constitute a minority.  Not only is this the distribution in the contemporary philosophical population, but in Western philosophy it has always been the pattern.  Seldom has hard determinism -- the incompatibilist endorsement of determinism and rejection of the freedom required for moral responsibility -- been defended.  One would expect hard determinism to have few proponents, given its apparent renunciation of morality.  I believe, however, that the argument for hard determinism is powerful, and furthermore, that the reasons against it are not as compelling as they might at first seem.

    The categorization of the determinist position by 'hard' and 'soft' masks some important distinctions, and thus one might devise a more fine-grained scheme.  Actually, within the conceptual space of both hard and soft determinism there is a range of alternative views.  The softest version of soft determinism maintains that we possess the freedom required for moral responsibility, that having this sort of freedom is compatible with determinism, that this freedom includes the ability to do otherwise than what one actually will do, and that even though determinism is true, one is yet deserving of blame upon having performed a wrongful act.  The hardest version of hard determinism claims that since determinism is true, we lack the freedom required for moral responsibility, and hence, not only do we never deserve blame, but, moreover, no moral principles or values apply to us.  But both hard and soft determinism encompass a number of less extreme positions.  The view I wish to defend is somewhat softer than the hardest of the hard determinisms, and in this respect it is similar to some aspects of the position recently developed by Ted Honderich.  In the view we will explore, since determinism is true, we lack the freedom required for moral responsibility.  But although we therefore never deserve blame for having performed a wrongful act, most moral principles and values are not thereby undermined.

    I.  Let us, for the sake of counterargument, devise a soft-determinist position that incorporates the essential features of three widespread compatibilist notions of freedom.  First, perhaps the most prominent compatibilist conception is found in the Humean tradition -- a notion of freedom of action.  In this view, an action is free in the sense required for moral responsibility when it is one the agent really wanted to perform.  More precisely, an action is free in the right sense just in case desires that genuinely belong to the agent make up the immediate causal history of the action.  An action is unfree, by contrast, when, for example, it is performed as a result of brainwashing or some types of mental illness.  In such cases, desires that genuinely belong to the agent do not play the causal role necessary for the action to be genuinely free.

    Second, in Harry Frankfurt's view, to be morally responsible, one's effective desires to perform actions must conform to one's second-order desires.  Frankfurt has us suppose "that a person has done what he wanted to do, that he did it because he wanted to do it, and that the will by which he was moved when he did it was his will because it was the will he wanted."  Such a person, in his view, acted freely in the sense required for moral responsibility.  

    Third, Bernard Gert and Timothy Duggan have argued that the type of freedom required for moral responsibility is the ability to will, or, in John Fischer's development of this view, responsiveness to reasons.  For an action to be free in the right sense, it must result from the agent's rational consideration of reasons relevant to the situation, such that, in at least some alternative circumstances in which there are sufficient reasons for her to do otherwise than she actually does, she would be receptive to these reasons and would have done otherwise by the efficacy of the same deliberative mechanism that actually results in the action.  Hence, I am free in the right sense when I decide to harvest the wheat next week rather than this week, if, in circumstances in which I knew it would rain next week, I would, by the deliberative mechanism that actually results in my deciding to harvest next week, appreciate the different reasons and harvest this week instead.  If my practical reasoning would not differ in varying circumstances, I am neither free nor morally responsible.

    Let us consider a situation involving an action that is free in all of the three senses we have just discussed.  Mr. Green kills Ms. Peacock for the sake of some personal advantage.  His act of murder is caused by desires that are genuinely his, and his desire to kill Ms. Peacock conforms to his second-order desires.  Mr. Green's desires are modified, and some of them arise, by his rational consideration of the relevant reasons, and his process of deliberation is reasons-responsive.  For instance, if he knew that the bad consequences for him resulting from his crime would be much more severe than they are actually likely to be, he would not have murdered Ms. Peacock.  Given that determinism is true, is it plausible that Mr. Green is responsible for his action?  

    In the deterministic view, the first and second-order desires and the reasons-responsive process that result in Mr. Green's crime are inevitable given their causes, and those causes are inevitable given their causes.  In assessing moral responsibility for his act of murder, we wind our way back along the deterministic chain of causes that results in his reasoning and desires, and we eventually reach causal factors that are beyond his control -- causal factors that he could not have produced, altered, or prevented.  The incompatibilist intuition is that if an action results from a deterministic causal process that traces back to factors beyond the control of the agent, he is not morally responsible for the action.

    A compatibilist rejoinder to this intuition is that moral responsibility does not leave Mr. Green behind as the deterministic causal process traces backwards in time.  Even though the chain of sufficient causes for his crime reaches far beyond him, to a time before he ever existed, he retains moral responsibility.  Mr. Green is morally responsible for the act of murder because his first-order desires caused the action, and these first-order desires conform to his second-order desires, and all of these desires are generated in a context of his rational evaluation of reasons.  Since the causal history of his action has the right pattern, he is free and morally responsible.

    Let us consider a series of different ways in which the above type of situation might come about, in order to undermine soft determinism and to support the contrary claim that moral responsibility precludes being determined in virtue of a causal process that traces back to factors beyond the agent's control.

Case 1: Mr. Green is like an ordinary human being, except that he was created by neuroscientists, who can manipulate him directly through the use of radio-like technology.  Suppose these neuroscientists directly manipulate Mr. Green to undertake the process of reasoning by which his desires are modified and produced, and his effective first-order desire to kill Ms. Peacock conforms to his second-order desires.  The neuroscientists manipulate him by, among other things, pushing a series of buttons just before he begins to reason about his situation, thereby causing his reasoning process to be rationally egoistic.  His reasoning process is reasons-responsive, because it would have resulted in different choices in some situations in which the egoistic reasons were otherwise.  Mr. Green does not think and act contrary to character, since the neuroscientists typically manipulate him to be rationally egoistic.

Mr. Green's action would seem to meet the criteria of the various compatibilist theories of freedom we have examined.  But intuitively, he is not morally responsible because he is determined by the neuroscientists' actions, which are beyond his control.  

    The intuitions generated by this case challenge the suppositions of many soft determinists.  Fischer argues that in "case of direct manipulation of the brain, it is likely that the process issuing in the action is not reasons-responsive, whereas the fact that a process is causally deterministic does not in itself bear on whether it is reasons-responsive."  He claims that although Frankfurt's sort of freedom can be induced neurophysiologically, a process that is reasons-responsive cannot.  But Fischer's claim is mistaken.  As long as a process requires only abilities that are physically realized, it can be induced by sufficiently equipped scientists.

One might argue that although in Case 1 the process resulting in the action is reasons-responsive, it is induced by direct manipulation near the time of the action, and this makes the case very much like one of brainwashing.  Or one might contend that Mr. Green's reasons-responsiveness is too superficial, because the neuroscientists could make him lack reasons-responsiveness just by controlling him differently.  It is not clear how deeply these objections cut, but in reply, let us consider a further case:    
 
Case 2: Mr. Green is like an ordinary human being, except that he was created by neuroscientists, who, although they cannot control him directly, have programmed him to be a rational egoist, so that, in any circumstances like those in which he now finds himself, he is causally determined to undertake the reasons-responsive process and to possess the set of first and second-order desires that results in his killing Ms. Peacock.

Case 2 is more similar than Case 1 to the ordinary human situation, since the agent is not directly manipulated near the time of the action.  But again, although the agent is free in each of our compatibilist senses, intuitively he is not morally responsible because he is determined in virtue of the neuroscientists' actions, which are beyond his control.  Furthermore, it would seem unprincipled to claim that whether Mr. Green is moral responsible depends on the length of the temporal interval between the programming and the action.  Whether the programming takes place two seconds or thirty years before the action is irrelevant.  

Case 3: Mr. Green is an ordinary human being, except that he was determined by the rigorous training practices of his home and community to be a rational egoist.  His training took place at too early an age for him to have had the ability prevent or alter the practices that determined his character.  Mr. Green is thereby caused to undertake the reasons-responsive process and to possess the organization of first and second-order desires that result in his killing Ms. Peacock.
If the compatibilist wishes to argue that Mr. Green is morally responsible under these circumstances, he must point out a morally relevant feature present in Case 3 but not in the first two cases, and such a difference is difficult to detect.  In each of these cases Mr. Green is free in all of the compatibilist senses.  Causal determination by agents whose determining activity is beyond Mr. Green's control most plausibly explains his lack of moral responsibility in the first two cases, and accordingly, we would seem forced to concede that he is not morally responsible in the third case as well.

Case 4: Physicalist determinism is true, Mr. Green is a rationally egoistic but (otherwise) ordinary human being, raised in normal circumstances.  Mr. Green's killing of Ms. Peacock comes about as a result of his undertaking the reasons-responsive process of deliberation, and he has the specified organization of first and second-order desires.

Just as in Cases 1-3, Mr. Green's action in Case 4 results from a deterministic causal process that traces back to factors beyond his control.  Given that we are constrained to deny moral responsibility to Mr. Green in the first three cases, what principled reason do we have for holding him morally responsible in this more ordinary case?  One distinguishing feature of Case 4 is that the causal determination of Mr. Green's crime is not, in the last analysis, brought about by other agents.  But if we were to revise the first three cases so that the determination is brought about by a spontaneously generated, mindless machine, the intuition that Mr. Green is not morally responsible would persist.  Hence, the best explanation for this intuition in these first three cases is just that Mr. Green's action results from a deterministic causal process that traces back to factors beyond his control.  Consequently, because Mr. Green is also causally determined in this way in Case 4, we must, despite our initial predilections, conclude that here too Mr. Green is not morally responsible.  And more generally, if every action results from a deterministic causal process that traces back to factors beyond the agent's control, then no agents are ever morally responsible for their actions.  

    The soft determinist might point out that according to ordinary intuitions, in Case 4 Mr. Green is morally responsible, and that these intuitions should be given more weight than we have given them.  But in the incompatibilist view, one consequence of determinism is that ordinary intuitions about moral responsibility in specific cases are based on a mistake.  In making moral judgments in everyday life, we do not assume that agents' choices and actions result from deterministic causal processes that trace back to factors beyond their control.  Our ordinary intuitions do not presuppose that determinism is true, and they may even presuppose that it is false.  Indeed, in Case 4 it is specified that determinism is true, but ordinary intuitions are likely to persist regardless of this stipulation, especially if the implications of determinism are not thoroughly internalized.  If we did assume determinism and internalize its implications, our intuitions might well be different.  Consequently, a reply to incompatibilism requires something more powerful than an analysis of freedom and moral responsibility designed to capture ordinary intuitions about moral responsibility in specific cases.  What is needed is an argument against the fundamental incompatibilist claim, that if one's action results from a deterministic causal process that traces back to factors beyond one's control, to factors that one could not have produced, altered, or prevented, then one is not free in the sense required for moral responsibility.

    II.  It has often been assumed that there is an alternative and equivalent statement of the fundamental incompatibilist claim.  According to this variant formulation, moral responsibility requires that, given all of the factors that precede one's choice, one could have done otherwise than what one actually did.  Furthermore, some have argued that because this variant formulation can be defeated, the incompatibilist view is mistaken.  But the variant formulation is not equivalent to the original, and since the original is more forceful, it would be best to reject the view that a successful challenge to the "responsibility only if she could have done otherwise" intuition also undermines the "responsibility only if her action does not result from a deterministic causal process that traces back to factors beyond her control" intuition.

    As Peter Van Inwagen points out, if physicalist determinism is true, there is a clear sense in which no agent could have done otherwise than what he in fact did.  By Van Inwagen's characterization, physicalist determinism is true just in case a proposition that expresses the entire state of the universe at some instant in time, in conjunction with the physical laws, entails any proposition that expresses the state of the universe at any other instant.  So if physicalist determinism is true, given the entire state of the universe at some instant in time, every subsequent state of the universe is thereby rendered inevitable.  Suppose Ms. White murdered Mr. Green last Tuesday.  Given physicalist determinism, Ms. White's crime is inevitable given the state of the universe 100 years before she was born and the natural laws.  So if Ms. White was able to do otherwise last Tuesday, then she must at that time have been able to alter the state of the universe 100 years before she was born, or to change the natural laws.  Since she was able to do neither, last Tuesday she could not have done otherwise than to murder Mr. Green.

    But soft determinists have argued that one can be morally responsible for one's actions even if one could not have done otherwise.  Frankfurt has devised a case similar to this one:

Ms. Scarlet is seriously considering whether to kill Colonel Mustard.  Meanwhile Professor Plum, a neuroscientist, very much wants the Colonel dead, and is worried that Ms. Scarlet will not choose to kill him.  So Professor Plum has implanted a device in Ms. Scarlet's brain, which, just in case Ms. Scarlet were to be swayed by a reason not to kill Colonel Mustard, would cause her to choose to kill him.  But Ms. Scarlet chooses to kill, and carries out the deed, without even beginning to be swayed by a reason for making the alternative choice.

Our intuition is that Ms. Scarlet is responsible for killing Colonel Mustard, although she could not have done otherwise, and thus, the conclusion of Frankfurt's argument is that the variant intuition is mistaken.  This argument is powerful and resilient.  For example, it succeeds not only against the intuition that moral responsibility requires the ability to do otherwise, but also against the intuition that it requires the ability to choose otherwise.  For Ms. Scarlet could not even have chosen otherwise, because the device would have arrested the deliberative process before it resulted in any alternative choice.  

    Frankfurt's argument strongly suggests that the incompatibilist (and everyone else) must relinquish the "responsibility only if she could have done otherwise" intuition.  As Fischer has shown, however, this type of argument does not establish that the incompatibilist must also abandon the claim that moral responsibility requires that one's action not be causally determined, or, in my formulation, that moral responsibility requires that one's action not result from a deterministic causal process that traces back to factors beyond one's control.  (One might note that Frankfurt does not state that his argument has this result).  In the Frankfurt-style example it is not specified that Ms. Scarlet is causally determined to choose or act as she did.  Our intuition that she is responsible might well depend on the assumption that although the device prevents her from being able to choose to do otherwise, her choice does not result from a deterministic causal process that traces back to factors beyond her control.  And indeed, if it were specified that her choice is caused in this way, incompatibilists, among others, would no longer agree that Ms. Scarlet is morally responsible.

    That one's choice and action result from a deterministic causal process that traces back to factors beyond one's control entails that one cannot choose to do otherwise (in at least one sense), but not vice versa.  For as Fischer points out, it is possible that one's choice not come about as a result of a deterministic process at all, and yet there be mechanisms that prevent one's choosing to do otherwise.  Ms. Scarlet might have been the undetermined agent-cause of the murder of Colonel Mustard even if Professor Plum's device renders her incapable of choosing to do otherwise.  The incompatibilist's most fundamental claim is that moral responsibility requires that one's choice and action not result from a deterministic causal process that traces back to factors beyond one's control.  An argument of the sort that Frankfurt advances cannot dislodge this claim.  This incompatibilist premise does not entail the proposition that moral responsibility requires that one be able to choose to do otherwise, and this proposition, for the reasons Frankfurt has advanced, is best rejected.

    III.  Let us now consider the libertarians, who claim that we have a capacity for indeterministically free action, and that we are thereby morally responsible.  According to one libertarian view, what makes actions free is just their being constituted (partially) of indeterministic natural events.  Lucretius, for example, maintains that actions are free just in virtue of being made up partially of random swerves in the downward paths of atoms.  These swerves, and the actions they underlie, are random (at least) in the sense that they are not determined by any prior state of the universe.  If quantum theory is true, the position and momentum of microparticles exhibit randomness in this same sense, and natural indeterminacy of this sort might also be conceived as the metaphysical foundation of indeterministically free action.  But natural indeterminacies of these types cannot, by themselves, account for freedom of the sort required for moral responsibility.  As has often been pointed out, such random physical events are no more within our control than are causally determined physical events, and thus, we can no more be morally responsible for them than, in the indeterminist opinion, we can be for events that are causally determined.

    Alternatively, many libertarians advocate the theory of agent causation, the view that freedom of action is accounted for not (simply) by randomly occurring events of the sort we have described, but by agents capable of causing their actions deliberately.  In this view, an agent's causation of her action is not itself produced by processes beyond her control.  Positing such agent-causes, in my view, involves no internal incoherence.  There is no internal incoherence in the idea of an agent having a non-Humean causal power to cause her actions deliberately in such a way that her causation of her actions is not itself produced by processes beyond her control.  It is unclear, however, whether we have any reason to believe that such entities exist.

    Furthermore, we have not encountered any divergences from the predictions of our physical theories.  The libertarian could, of course, advocate a theory that embraces such divergences, but this, by itself, would provide a powerful reason to reject such a view.  So let us focus on those theories that attempt to reconcile agent-causation with the predictions of our physical theories.    

    Suppose first that the physical world is a deterministic system.  If this is so, then the physical component of any action -- constituent events describable, for example, by neurophysiology, physiology, chemistry, physics -- will be causally determined.  As Kant argues, it is possible that undetermined, non-physical agents always make free choices for just those potential actions whose physical components are causally determined.  In Kant's view, this possibility is all we need for rational faith in indeterministic freedom.  But is it credible that this possibility is actually realized?  There would certainly be nothing incredible about an undetermined agent-cause making a free choice on some particular occasion for a possible action whose physical component was causally determined.  However, it would be incredible if for any substantial period of human history all free choices made by agent-causes should be for just those possible actions whose physical components are causally determined to occur, and none of these choices should be for the alternatives.  Independent of an idealistic theory according to which agents construct the physical world, the coincidences this view implies are too wild to believe.

    To try to solve this problem of wild coincidences, the libertarian might invoke indeterminacy in nature.  Nevertheless, in ordinary cases, quantum indeterminacy only allows for an extremely small probability of counterfactual events at the scale of human actions.  Suppose, by analogy, that the soda can on the table remains where it is for the next minute.  Given quantum indeterminacy, there is some probability that instead it would spontaneously move one inch to the left sometime during this minute.  But for this event to occur, each of many quantum indeterminacies would have to be resolved in a specific alternative way, the probability of which is extremely small.  The prospects for counterfactual human actions are similarly bad.  Even if quantum indeterminacy results in the indeterminacy of certain neural events, like the firing of individual neurons, so that at certain times both the probability that the neural event will occur and that it will not are significant, the likelihood of physical components of counterfactual actions occurring is insignificant.  The reason is that the making of a decision is an event of a much larger scale than is an event like the firing of a neuron.  When a decision is made, a very large number of individual quantum and neural events are involved, and quantum indeterminacy would not undergird a significant probability for counterfactual events of this magnitude.  

    Let us assume that what determines an indeterministically free agent's choices is how she finally weighs the reasons.  The weighing of each reason will be (partially) realized in a very large complex of neural and quantum events.  But this complex will be too large for quantum indeterminacy to substantiate a significant probability of counterfactual actions.  Suppose an agent actually makes a decision to perform action A rather than action B, and that the physical realization of her weighing of reasons is large-scale neural pattern of type X.  Given quantum indeterminacy, there is some antecedent probability -- the probability, let us say, just as the agent begins to weigh the reasons for action -- that her brain should realize a very different neural pattern upon weighing the reasons, one of type Y, which is correlated with performing action B.  But for a pattern of type B to come about, each of many indeterminacies would have to be resolved in a specific alternative way, the antecedent probability of which is extremely small.  More generally, the antecedent probability of the occurrence of the physical component of any counterfactual action is extremely small.  And it would be too wildly coincidental to believe that for any substantial interval of human history all or even almost all indeterministically free choices made by agent-causes should be for just those possible actions the occurrence of whose physical components has the extremely high antecedent physical probability, and not for any of the alternatives.  Thus the fact that quantum theory allows counterfactual actions to have non-zero antecedent probability fails to remedy the problem of wild coincidences posed by the attempt to reconcile libertarianism with strict determinism.  

    Now it might be objected that the problem of wild coincidences arises only if it turns out that, at the neurophysiological level, counterfactual events do not have significant antecedent probability.  Yet there are examples, such as the moving of the needle on a Geiger counter, of microphysical indeterminacies that are magnified to significantly indeterminate events at the macrolevel.  Perhaps similar magnifications occur in the brain.  Randolph Clarke suggests that a libertarian might take advantage of macrolevel natural indeterminacy of this sort by positing agent causes who have the power to make the difference as to which of a series of naturally possible actions is performed.  Might this picture not offer the libertarian a way out of the wild coincidences problem?  No.  Suppose that physical components of counterfactual actions do have a significant antecedent probability of occurring.  Consider a class of possible actions each of which has a physical component whose antecedent probability of occurring is approximately 0.32.  If indeterminist free action is to be compatible with what our physical theories predict to be overwhelmingly likely, then over a long enough period of time these possible actions would have to be freely chosen almost exactly 32% of the time.  Yet their actually being freely chosen almost exactly 32% of the time would constitute a coincidence no less wild than the coincidence of possible actions whose physical components have an antecedent probability of about 0.99 being freely chosen about 99% of the time.  The problem of wild coincidences, therefore, is independent of the physical components of actions having any particular degree of antecedent probability.

    This point reveals the fundamental difficulty for libertarian agent causation.  Whether determinism is true or whether there is quantum indeterminacy, the antecedent probabilities of the physical components of human actions are fixed.  If determinism is true, the antecedent probability of any such component is either 1 or 0.  According to quantum theory, such probabilities will be different.  But regardless of which view is true, it would be wildly coincidental, and hence too bizarre a scenario to believe, if for any substantial span of human history frequencies of indeterministically free choices should happen to dovetail with determinate physical probabilities.

    The libertarian might reply that physics is likely to be so different in the future that one should not be daunted by this problem.  Who can tell, after all, what physics will look like in the year 2525?  However, any physical theory according to which the antecedent probabilities of physical events are determinate will give rise to the problem of wild coincidences for the libertarian picture.  To avoid this problem, physics would have to allow for physical events with no determinate antecedent probabilities at all.  

    Thus, barring certain revolutionary discoveries in neurophysiology, psychology, or physics it seems unlikely that libertarianism is true.  Accordingly, let us focus our attention on the hard determinist version of incompatibilism.  But first, our discussion of libertarianism reveals the need to revise our characterization of the wider issues: assuming the truth of our best scientific theories, determinism turns out to be false.  However, the kinds of indeterminacies these theories posit provide us with no more control over our actions than we would have if determinism were true.  Our actions may not result from deterministic causal processes that trace back to factors over which we have no control, but yet there are processes, either deterministic or indeterministic, over which we have no control, that produce our actions, and this is enough to rule out freedom of the sort required for moral responsibility.  Hence the fundamental incompatibilist intuition turns out to be "responsibility only if her action is not produced by processes, either deterministic or indeterministic, beyond her control."  For the sake of simplicity and meshing with the traditional discussion, however, I shall continue to describe the position I am defending as "hard determinism," with the understanding that this term is strictly speaking inaccurate, but not in a way that makes a difference to the issues we shall now explore.

    IV.  The alternative to soft determinism and libertarianism is hard determinism, the view that because determinism is true, we lack the freedom required for moral responsibility.  Let us examine this option to ascertain whether it must be as unacceptable as it may initially seem.

    One instinctive reaction to hard determinism is that if it were true, we would have no reason to attempt to accomplish anything -- to try to improve our lives or the prospects of society -- because our deliberations and choices could make no difference.  This challenge has also been directed towards soft determinists, and they have responded persuasively.  Ayer and Dennett, among others, have pointed out that the determination of our deliberations, choices, actions, and their consequences does not undermine their causal efficacy.  The hard determinist can legitimately appropriate this position.  It is true that according to hard determinism we are not free in the sense required for moral responsibility, and therefore, what happens cannot be affected by choices that are free in this sense.  But what happens may nevertheless be caused by the deliberations we engage in and the choices we make.  

    It is undeniable that we feel we have the ability to choose or do otherwise; for example, that you feel that it is now possible for you either to continue or to stop reading this article.  In the hard determinist's judgment, this feeling of freedom is an illusion (and soft determinists of some types agree).  This judgment would be challenged by those who believe that our introspective sense provides us with infallible beliefs about our own abilities.  But it is a familiar fact that such an assessment of introspection is implausible.  Kant, however, provide us with a different reason not to discount the feeling of freedom.  He suggests that engaging in a process of deliberation requires that one suppose that more than one choice for action is causally possible.  This view seems compelling: could one deliberate about which roads to take if one believed that one was causally capable of choosing only one of them?  But according to hard determinism, one cannot choose otherwise than the way one actually does.  Thus, as Van Inwagen argues, whenever one engages in a process of deliberation, one would be making a false supposition, and hence if one were a self-professed hard determinist, one would often have inconsistent beliefs; "anyone who denies the existence of free will must, inevitably, contradict himself with monotonous regularity."

    There are two replies available to the hard determinist.  The first grants that when we deliberate, at the moment of choice we must indeed make the false and unjustified assumption that more than one course of action is open to us.  But it is legitimate to assume this cognitive posture, because the practical gains of engaging in deliberation are significant enough to outweigh the losses of having false and unjustified beliefs.  In this view, deliberation requires us to choose between theoretical and practical irrationality.  One is irrational in the theoretical sense when, for example, one has a belief that has no justification, or a belief one knows to be false, and one is irrational in the practical sense if, for instance, one does something one knows will frustrate what one wants, all things considered.  Hard determinism would seem to leave us with the following choice: either deliberate and have a belief that you know to be false whenever you do, or cease to deliberate.  Practical rationality would appear to have the upper hand.

    It is nevertheless disturbing to maintain that one must be theoretically irrational whenever one deliberates.  There is, however, a more attractive alternative which does not require that one override the canons of theoretical rationality.  The hard determinist might deny that at the moment of choice, one must assume that more than one option is causally possible.  One might instead believe that one's actions are determined by way of one's choice, that one's choices are determined by means of one's deliberation, and that one does not know in advance of deliberation which action one will choose.  As long as one's actions are determined by deliberation and choice, and one does not know beforehand what the result of one's deliberation will be, there will be no interference with the deliberative process.  Indeed, the deliberative process might be jeopardized if one had previous knowledge of the choice that would result.  Perhaps it is even incoherent to suppose that one might know in advance of deliberation which of two roads one will choose, for in such a situation genuine deliberation would be undermined.  But given that one cannot know the results of one's deliberation in advance, the process can go on unimpeded.

    V.  A very prominent feature of our ordinary conception of morality that would be undermined if hard determinism were true is our belief that persons deserve credit and praise when they deliberately perform morally exemplary actions, and that they deserve blame when they deliberately perform wrongful actions.  To deserve blame is to be morally liable to blame by deliberately choosing to do the wrong thing.  Hard determinism rules out one's ever deserving blame for deliberately choosing to act wrongly, for such choices are always produced by processes that are beyond one's control.  

    Someone might argue that even if no one ever deserves blame, it would nevertheless be best for us to think and act as if people sometimes do, because thinking and acting this way is a superb method for promoting moral reform and education.  More generally, even if no one is really morally responsible, it would still be best to hold people morally responsible.  Such a view might be justified on practical grounds, were we confident, for example, that thinking and acting as if people sometimes deserve blame is often necessary for effectively promoting moral reform and education.  But this option would have the hard determinist thinking that someone deserves blame when she also believes him not to, which is an instance of theoretical irrationality, and would have her blaming someone when he does not deserve to be blamed, which would seem to be morally wrong.

    There is, however, an alternative practice for promoting moral reform and education which would suffer neither from irrationality nor apparent immorality.  Instead of blaming people, the determinist might appeal to the practices of moral admonishment and encouragement.  One might, for example, explain to an offender that what he did was wrong, and then encourage him to refrain from performing similar actions in the future.  One need not, in addition, blame him for what he has done.  The hard determinist can maintain that by admonishing and encouraging a wrongdoer one might communicate a sense of what is right, and a respect for persons, and that these attitudes can lead to salutary change.  Hence, one need not hold the wrongdoer morally responsible for what he has done, but rather consider him responsive to moral admonishment and encouragement.  Likewise, although one could not justifiably think of one's own wrongful actions as deserving of blame, one could legitimately regard them as wrongful, and thereby admonish oneself, and resolve to refrain from similar actions in the future.  But like blame of others, blame of self, and more generally, holding oneself morally responsible, would be best avoided.    
   
    But what of the character who regularly and deliberately does wrong, and refuses to make a commitment to doing what is right?  Doesn't the hard determinist have little to say to such a person?  While the hard determinist can only admonish, the advocate of moral responsibility can also blame.  But having recourse to blame in such circumstances is not clearly a significant practical advantage.  One might argue that hard determinism is a threat to moral practice because the character we have described might offer determinism as an excuse for his behavior.  Certainly, the hard determinist would have to accept his excuse, whereas the proponent of moral responsibility would not.  But the practical advantages from this point on do not favor either side.  Both face the task of moral education and imparting a respect for persons, and it is not obvious that the hard determinist has fewer resources for this project than those available to her opposition.

    The hard determinist position implies that the appalling actions of persons are much more similar to earthquakes and epidemics than they are according to views that hold persons morally responsible.  The justification we assume for regarding especially wrongful actions of persons as deeply different from natural disasters is that persons are typically responsible for their actions.  But according to hard determinism, because a person's actions are the result of processes over which he has no control, we cannot consider him responsible for them, just as we cannot hold earthquakes or epidemics responsible for their effects.  One still might legitimately have a feeling of moral concern about what persons do, or about what persons who are reasons-responsive do, which would differ from one's attitudes to earthquakes and epidemics.  This feeling would be legitimate supposing it has no cognitive component that conflicts with hard determinism.  But as I shall soon argue in further detail, the various attitudes that presuppose the cognitive component that persons are morally responsible would be unjustified.  

    Honderich contends rightly, I believe, that in the face of determinism we must eschew retribution, but he also argues that

we can persist in certain responses to the desires and intentions of others, and hence to them.  There is no obstacle to my abhorrence of the desires and intentions of the treacherous husband foreseeing his divorce, or, more important, to my abhorrence of him, a man whose personality and character are consistent with these desires and intentions, and support them.  

But the determinist must be more abstemious here.  Abhorrence of a person because of the actions he has performed at least typically involves blaming him for those actions, which, in turn, presupposes that his actions and character did not result from processes beyond his control.  If one were to discover that an especially wrongful "action" was caused by some non-psychological, physiological reaction in the person, one's abhorrence would tend to vanish, and this would suggest that one's abhorrence was founded in blame.  It is legitimate to feel moral concern in response to a wrongful action, and to be deeply saddened that there are persons with immoral characters, but at least most often one's response of abhorrence, because it involves blaming someone, is unjustified.

    Perhaps one can learn to abhor people because of the wrongful actions they perform without blaming them, just as one might abhor soggy Corn Flakes because of their sogginess without blaming them.  But it is doubtful that developing such an attitude people could be justified on moral grounds if determinism is true.  One might be able to abhor people for their wrongful actions without being theoretically irrational, but it seems unlikely that one would advance the good by fostering this attitude, by contrast, for example, with attitudes such as moral concern or sadness.

    Susan Wolf has argued that whereas deserved blame cannot be justified if determinism is true, deserved praise does not collapse along with it.  As she puts it, she is "committed to the curious claim that being psychologically determined to perform good actions is compatible with deserving praise for them, but that being psychologically determined to perform bad actions is not compatible with deserving blame."  Wolf, in effect, endorses the hard determinist's view about deserved blame, but not about deserved praise.  She cites the following example in support of her view:

Two persons, of equal swimming ability, stand on equally uncrowded beaches.  Each sees an unknown child struggling in the water in the distance.  Each thinks "The child needs my help" and directly swims out to save him.  In each case, we assume that the agent reasons correctly -- the child does need her help -- and that, in swimming out to save him, the agent does the right thing.  We further assume that in one of these cases, the agent has the ability to do otherwise, and in the other case not.

Wolf says that whereas according to the libertarian, only the first of these agents is responsible, "there seems to be nothing of value that the first agent has but the second agent lacks."  Perhaps the second agent does not have the ability to do otherwise because "her understanding of the situation is so good and her moral commitment so strong."  Wolf concludes that the fact that the second agent is determined to do the right thing for the right reasons does not make her any less deserving of praise than the first agent.

    First of all, Wolf's argument is susceptible to an objection inspired by the point Fischer raises in connection with Frankfurt's case.  Given the way Wolf presents her lifesaver case, the reader might yet presuppose that the swimmer who cannot do otherwise is not causally determined to deliberate and act as she does.  If it were specified that her action results from a deterministic causal process that traces back to factors she could not have produced, altered, or prevented -- perhaps by adding that she is controlled by neuroscientists -- the intuition that she deserves praise might well vanish.  Wolf's case may indicate that an agent might deserve praise even if she could not have done otherwise, but it fails to show that an agent deserves praise even if her action results from a deterministic causal process that traces back to factors beyond her control.

    But suppose that the intuition that the second swimmer deserves praise persists even if it is specified that she is causally determined.  The hard determinist can now argue that while according to ordinary intuitions both swimmers deserve praise, the second swimmer really does not.  Ordinarily, we consider persons praiseworthy for their great intelligence, good looks, or native athletic ability, even though these qualities are not due to any agency of theirs, and hence, even though they in no sense really deserve praise for these qualities.  Thus it comes as no surprise that we would ordinarily consider the second swimmer, who is determined to do the right thing for the right reasons, praiseworthy.  She may be considered praiseworthy because she is a good person, and has acted in pursuit of the good, but as in the case of the person of great intelligence, we need not conclude that she is genuinely deserving of praise.

    Sometimes it may well be a good thing to praise someone despite her not deserving it, perhaps because praise can at times simply be an expression of approbation or delight about the actions or accomplishments of another.  By contrast, blaming someone who does not deserve it would seem always to be (at least prima facie) wrong.  The explanation for this disanalogy might be that because blaming typically causes pain, it must be wrong unless it is deserved, whereas since praise is far from painful, it can be appropriate beyond cases in which it is deserved.  Whatever may be the case here, the intuition that the determined swimmer is praiseworthy fails to undermine the hard determinist view, that not only deserved blame but also deserved praise is incompatible with determinism.  

    VI.  Another feature of our ordinary conception of morality that would be threatened if we accepted hard determinism is the belief that statements of the following form are sometimes true: 'Although you did not choose x, you ought, morally, to have chosen x.'  There are different senses of the moral 'ought,' but the central senses might well be undermined in a hard determinist picture.  It would seem that in any case in which one could never have performed an action, it is never true that one ought or ought to have performed the action.  Consequently, if because one is causally determined one can never choose otherwise than the way one actually does, then it is false that one ever morally ought to choose otherwise.  And further, if it is never true that one ought to have chosen otherwise than the way one does, then what would be the point of a system of moral 'ought's?  Hard determinism imperils this system, because it would seem that if 'A ought to choose x' is true at all, it must be true not only when A comes to choose x, but also when A does not come to choose x.

    But even if moral 'ought' statements are never true, moral judgments, such as 'it is morally right for A to do x,' or 'it is a morally good thing for A to do x,' still can be.  Thus, even if one is causally determined to refrain from giving to charity, and even if it is therefore false that one ought to give to charity, it still might still be the right thing or a good thing to do.  Cheating on one's taxes might be a wrong or a bad thing to do, even if one's act is causally determined, and hence, even if it is false that one ought not to do so.  These alternative moral judgments would indeed lack the deontic implications they are typically assumed to have, but nevertheless, they can be retained when moral 'ought' statements are undermined.  In addition, the various benefits of the system of moral 'ought's can be recouped.  For instance, when one is encouraging moral action, one can replace occurrences of 'you ought to do x' with 'it would be right for you to do x,' or with 'it would be a good thing for you to do x.'  Discouragement of wrongful action could be revised analogously.  

    One might argue that if moral 'ought' statements were never true, we could have no reason to do what is right.  But this view is mistaken.  Although it is false that one ought to eat boiled rather than poached eggs, one might still have reason to choose one over the other, perhaps in virtue of one's preference for boiled eggs, or even because one thinks that one type is objectively better than the other, and one has resolved to aspire to excellence.  Similarly, one might treat others with respect because one prefers to do so, or because one has resolved to do what is right, even if it not the case that one ought to do so.  If one has resolved to do what is right, by whatever motivation, one thereby has reason to act in accordance with this resolution.

    It may seem that relinquishing the moral 'ought' together with deserved praise and blame restricts hard determinism to a consequentialist position in ethics.  One might be tempted by the claim that although rejection of the moral 'ought' is consistent with the goodness of certain consequences and, derivatively, with the goodness of actions that bring about such consequences, abandoning the moral 'ought' does rule out principles of right that are based on non-consequentialist considerations.  But this claim seems mistaken for the reason that insofar as they have been developed, the metaphysical bases for non-consequentialist positions do not clearly involve an essential appeal to a notion of freedom unavailable to the hard determinist.  One might argue that the hard determinist is restricted to consequentialism because her rejection of deserved praise and blame confines her to forward-looking ethical views, and such forward-looking views are consequentialist.  But although the hard determinist may not look to the past to assess praise and blame, she can legitimately make judgments about the rightness and wrongness of past actions.  Furthermore, not all forward-looking ethical views are consequentialist.  The Kantian principle, "Act only on that maxim which you can also will to be a universal law" is no less forward-looking than the utilitarian principle, "Act so as to maximize happiness."  The hard determinist seems free to accept non-consequentialist ethical views.

    VII.  If hard determinism is true, how would it be best to regard our reactive attitudes, for example, our resentment and anger upon being betrayed, or our gratitude upon receiving help in trouble?  In the face of a deterministic universe, the Stoics urge self-discipline aimed at eradicating at least the negative reactive attitudes.  David Hume and P. F. Strawson, on the other hand, advance the psychological thesis that our reactive attitudes cannot be affected by a general belief in determinism, or by any such abstract metaphysical view, and that therefore the project of altering or eliminating our reactive attitudes by a determinist conviction would be ineffectual.

    Let us address two issues: first, whether the reactive attitudes really are immune from alteration by a belief in determinism, and second, whether it would be good for them to be altered by such a belief (if they could be).  On the first issue, Gary Watson provides a compelling example, the case of Robert Harris, who brutally murdered two teenage boys in California in 1978.  When we read an account of these murders "we respond to his heartlessness and viciousness with loathing."  But an account of the atrocious abuse he suffered as a child "gives pause to the reactive attitudes."  Upon absorbing such information, not everyone relinquishes his attitude of blame completely, but his attitude is at least typically tempered.  It is not only that we are persuaded to feel pity for the criminal.  In addition, our attitude of blame is mitigated by our coming to believe that the criminal was at least partially determined to behave as he did.  One might claim that although belief in determinism about a particular situation can affect reactive attitudes, the general belief in determinism never can.  But I can think of no reason to accept this view.  Because particular cases of determinism can be vividly described, they can much more readily affect one's attitudes, but there is no reason to believe that the general conviction cannot have a similar effect.

    It would be implausible to maintain that in every case the presence or the intensity of one's reactive attitudes can be affected by a belief in determinism.  Sometimes a wrong committed might be too horrible for such a belief to have any effect on one's subsequent reaction.  The Stoics maintained that we can always prevent or eradicate attitudes like grief and anger, regardless of their intensity, with the aid of a determinist conviction.  But they might well have overestimated the extent of the control we have over our emotional lives.  If someone were brutally to murder your family, it might well be psychologically impossible for you ever to eradicate feelings of intense anger toward the killer.  This fails to show, however, that a determinist conviction cannot affect reactive attitudes, even in typical cases.

    Let us suppose, therefore, that a determinist conviction can affect our reactive attitudes.  Would it be a good thing if they were affected by this means?  According to Strawson, human beings would stand to lose much if reactive attitudes were dislodged by a belief in determinism, for we would then be left with a certain "objectivity of attitude."  A stance of this sort, Strawson believes, conflicts with the types of attitudes required for good interpersonal relationships:

To adopt the objective attitude to another human being is to see him, perhaps, as an object of social policy; as a subject for what, in a wide range of sense, might be called treatment; as something certainly to be taken account, perhaps precautionary account, of; to be managed or handled or cured or trained; perhaps simply to be avoided...  The objective attitude may be emotionally toned in many ways: it may include repulsion or fear, it may include pity or love, though not all kinds of love.  But it cannot include the range of reactive feelings and attitudes which belong to involvement or participation with others in interpersonal human relationships; it cannot include resentment, gratitude, forgiveness, anger, or the sort of love which two adults can sometimes be said to feel reciprocally, for each other.
Strawson is right to believe that objectivity of attitude would destroy interpersonal relationships.  But he is mistaken to think that objectivity of attitude would result or be appropriate if determinism were to undermine the reactive attitudes.  As Honderich argues, a reasonable determinist attitude towards the moral life "recommends no such bloodlessly managerial an attitude toward others."

    In his analysis, Honderich points out that one's reactive attitudes presuppose certain beliefs about the persons to whom they are directed, and that these beliefs can sometimes be undermined by determinist convictions.  I agree, and I would develop the claim in this way.  One's reactive attitudes presuppose beliefs of this sort, and when these presuppositions lack adequate justification, or when one believes them to be false, or when they have little or no justification and conflict with justified beliefs one holds, then maintaining attitudes that have such presuppositions is irrational in the theoretical sense.  Suppose, for example, that you are angry with the guests because they are very late for dinner.  Your anger presupposes the belief that they reasonably could have been on time.  But you come to know that they are late because an airplane crashed on the freeway, and the resulting traffic jam trapped them for an hour.  Given that your presupposition no longer has justification, and since it conflicts with a justified belief you hold, it is theoretically irrational for you to maintain your anger, and you would therefore have to give up your anger to escape irrationality.  

    Now suppose that you have a justified belief that hard determinism is true, and that you are angry with a friend because he has betrayed a confidence.  Your anger presupposes the belief that he deserves blame and that his betrayal was not produced by processes beyond his control.  You have no justification for this presupposition, let us suppose, and it conflicts with your justified belief that his action was produced by processes beyond his control.  Consequently, your anger is irrational in the theoretical sense, and in order to escape this irrationality, you must give up your anger.

    Someone might point out, however, that such anger may not be practically irrational, and since practical and theoretical rationality may conflict, an issue may arise about which sort it would then be best to secure.  If one's anger is practically rational in virtue of playing a part in a system of attitudes required for interpersonal relationships, but it is nevertheless theoretically irrational because of its presuppositions, how would it be best to act?  For Hume and Strawson, the issue would happily be resolved by facts about human psychology, since we would be psychologically incapable of theoretical rationality in such situations.  But since their psychological claim is implausible, the issue again becomes live.

    If the hard determinist were to acknowledge that a determinist conviction could affect the reactive attitudes, but that adopting an objectivity of attitude would be practically irrational in virtue of being destructive to human relationships, she might well override theoretical rationality by retaining her normal reactive attitudes.  If she acted in this way, however, she would be reduced to the uncomfortable position of maintaining attitudes that are theoretically irrational.  But the hard determinist is not clearly forced into such a difficult situation.  For first, although many ordinary reactive attitudes might be irrational, these reactive attitudes are not obviously required for good interpersonal relationships.  Some reactive attitudes, like certain kinds of anger and resentment, may well not be good for relationships at all.  And secondly, the reactive attitudes one would want to retain have analogues that do not have false presuppositions.  Such analogues by no means amount to Strawson's objectivity of attitude, and they are sufficient to sustain good interpersonal relationships.

    In Strawson's view, some of the attitudes most important for interpersonal relationships are resentment, anger, forgiveness, gratitude, and mature love.  As I have suggested, a certain measure of resentment and anger is likely to be beyond our power to affect, and thus even supposing that one is committed to doing what is right and rational, one would still not be able to eradicate all of one's resentment and anger.  As hard determinists, we might expect these attitudes to occur in certain situations, and we might regard them as inevitable and exempt from blame when they do.  But we sometimes have the ability to prevent, alter, or eliminate resentment and anger, and given a belief in hard determinism, we might well do so for the sake of morality and rationality.  Modification of anger and resentment, aided by a determinist conviction, could well be a good thing for relationships (supposing that no unhealthy repression is induced).  At very least, the claim that it would be harmful requires further argument.

    The attitude of forgiveness seems to presuppose that the person being forgiven deserves blame, and therefore, forgiveness is indeed imperiled by hard determinism.  But there are certain features of forgiveness that are not threatened by hard determinism, and these features can adequately take the place this attitude usually has in relationships.  Suppose your companion has wronged you in similar fashion a number of times, and you find yourself unhappy, angry, and resolved to loosen the ties of your relationship.  Subsequently, however, he apologizes to you, which, consistent with hard determinism, signifies his recognition of the wrongness of his behavior, his wish that he had not wronged you, and his genuine commitment to improvement.  As a result, you change your mind and decide to continue the relationship.  In this case, the feature of forgiveness that is consistent with hard determinism is the willingness to cease to regard past wrongful behavior as a reason to weaken or dissolve one's relationship.  In another type of case, you might, independently of the offender's repentance, simply choose to disregard the wrong as a reason to alter the character of your relationship.  This attitude is in no sense undermined by hard determinism.  The sole aspect of forgiveness that is jeopardized by a hard determinist conviction is the willingness to overlook deserved blame or punishment.  But if one has given up belief in deserved blame and punishment, then the willingness to overlook them is no longer needed for relationships.  

    Gratitude would seem to require the supposition that the person to whom one is grateful is morally responsible for the beneficent act, and therefore hard determinism might well undermine gratitude.  But certain aspects of this attitude would be left untouched, aspects that can play the role gratitude commonly has in interpersonal relationships.  No feature of the hard determinist position conflicts with one's being joyful and expressing joy when people are especially considerate, generous, or courageous in one's behalf.  Such expression of joy can produce the sense of mutual well-being and respect frequently brought about by gratitude.  Moreover, just as in the case of gratitude, when one expresses joy for what another person has done, one can do so with the intention of developing a human relationship.

    Finally, the thesis that love between mature persons would be subverted if hard determinism were true requires much more thorough argument than has been provided.  One might note, first of all, that parents love their children rarely, if ever, because these children possess the freedom required for moral responsibility, or because they freely (in this sense) choose the good, or because they deserve to be loved.  But moreover, when adults love each other, it is also seldom, if at all, for these kinds of reasons.  Explanations for love are complex.  Besides moral character and action, factors such as appearance, manner, intelligence, and affinities with persons or events in one's history all have a part.  But suppose we agree that moral character and action are of paramount importance in producing and maintaining love.  Even then, it is unlikely that one's love would be undermined if one were to believe that moral character and action do not come about through free and morally responsible choice.  Love of another involves, most fundamentally, wishing well for the other, taking on many of the aims and desires of the other as one's own, and a desire to be together with the other.  Hard determinism threatens none of this.

    While certain reactive attitudes might well be irrational because of the presuppositions these attitudes have, turning to analogues of the sort we have described is in no sense irrational, and it is far from assuming the objectivity of attitude so destructive to interpersonal relationships.  Furthermore, nothing about hard determinism recommends assuming an objectivity of attitude.  The specter of the objective attitude arises out of the sense that the hard determinist is constrained to view other persons as mere mechanical devices, to be used and not respected.  The hard determinist, however, is not forced to view persons in this way.  She is not compelled to deny that human beings are rational and responsive to reasons, and no feature of her view threatens the appropriateness of respecting persons for their rational capacities.

    Accordingly, someone's thinking and acting in harmony with her hard determinist conviction would not endanger her interpersonal relationships.  She would resist anger, blame, and resentment, but she would not be exempt from pain and unhappiness upon being wronged.  She might, if wronged, admonish, disregard the wrongdoing, or terminate the relationship.  Although she would avoid gratitude, she could enjoy and express joy about other persons' efforts in her behalf.  No obstacle would be posed to her loving others.  Only if, in addition, she had an unappealing tendency to control another, would she see him "as an object of social policy; as a subject for what, in a wide range of sense, might be called treatment; as something certainly to be taken account, perhaps precautionary account, of; to be managed or handled or cured or trained; perhaps simply to be avoided..."  But taking on such an objectivity of attitude would not be justified by her hard determinist conviction.

    VIII.  Given that free will of some sort is required for moral responsibility, then libertarianism, soft determinism, and hard determinism, as typically conceived, are mutually exhaustive positions (if we allow the "deterministic" positions the view that events may result from indeterministic processes of the sort described by quantum mechanics).  Yet each has a consequence that is difficult to accept.  If libertarianism were true, then we would expect violations of the physical laws (as we currently understand them) whenever a free action is performed.  If soft determinism were true, then agents would deserve blame for their wrongdoing even though their actions were produced by processes beyond their control.  If hard determinism were true, agents would not be morally responsible -- agents would never deserve blame for even the most cold-blooded and calmly executed evil actions.  I have argued that hard determinism could be the easiest view to accept.  Hard determinism need not be of the hardest sort.  It need not subvert the commitment to doing what is right, and although it does undermine some of our reactive attitudes, secure analogues of these attitudes are all one requires for good interpersonal relationships.  Consequently, of the three positions, hard determinism might well be the most attractive, and it is surely worthy of more serious consideration than it has been accorded