15. Free Will and Ethics

CHAPTER 15

Free Will and Ethics^*

A PERSISTENT OLD ISSUE

The very topic of ethics requires dipping into metaphysics. Although I cannot settle an old issue, I must recognize it. Are individuals’ actions and even their decisions, desires, and characters fully determined by circumstances ultimately outside their own control? If that were true—if people lack free will and true choice—then personal responsibility would lack meaning. Praise and blame, reward and punishment, would have no application; and ethics as a field of study would lack any genuine subject matter.

This position, right or wrong, seems to have been the position of Immanuel Kant. Throughout his Groundwork (1785/1964), Kant acknowledges an antinomy between freedom of the will and the prevalence of causal laws of nature. He maintains, however, that freedom of the will is a necessary presupposition of morality. He suggests that the antinomy might somehow be resolved through his distinction between the intelligible and sensible worlds (noumenal and phenomenal worlds, in his technical terminology). Experience, filtered through the Kantian “categories” of perception and understanding, imposes the idea of tight causality; but unknowable characteristics of the noumenal world of things in themselves might make freedom of the human will genuine. Confessedly, all this is quite mysterious to me.

The terms “free will” (or “free choice”) and “determinism” have no agreed precise meanings, so I cannot begin by defining them. Exploring what these terms and concepts might mean and how they interrelate is a main task of this appendix.

We must, however, avoid “essentialism.” As criticized by Karl Popper (e.g., 1985, pp. 88—94), essentialism means focusing on one or more pieces of terminology, supposing that each one labels a definite aspect of reality, brooding over these aspects or concepts to grasp the “essence” of each, and perhaps brooding further over whether the realities corresponding to two or more concepts could exist together.¹ Trying to gain knowledge in this way proceeds backward. Perceiving uniformities and diversities in the real world belongs ahead of brooding over words to label them. Conceivably, traditional formulations of the whole supposed issue of free will versus determinism will prove misconceived.

SCHLICK ON THE ISSUE

My admiration for two ethicists in particular, Moritz Schlick and Henry Hazlitt, predisposed me toward their solutions; yet in the end I find them incomplete or otherwise unsatisfying. Schlick regarded the supposed issue as a mere pseudo-problem: determinism and free will reconcile (1930/1961, chap. VII). Causality can operate while leaving individuals some freedom not only over what acts they perform but also over what choices they make. The opposite of freedom is compulsion, and determination does not mean compulsion. In the words of R.E. Hobart, whose views are discussed below, compulsion implies causation but causation does not necessarily imply compulsion.²

Schlick identifies a confusion between descriptive and prescriptive laws. Scientific laws describe how the world works; they do not prescribe events; they do not resemble totally enforced legislation making events unfold as they do. Kepler’s Laws describe how the planets revolve around the sun; they are not prescriptions compelling them to revolve as they do. The law of demand describes how buyers respond to alternative levels of an item’s price (apart from other overriding influences); it does not compel buyers to behave in the way it describes.

These distinctions help explain, then, how a person could conceivably be choosing and acting free from compulsion even when his choices and actions are causally determined and in principle predictable. The opposite, presumably, would be choices made and actions taken by sheer chance or baseless caprice. Choices and actions need not be stochastic or capricious to be properly called free.

If their choices and actions were totally unfree, people could not properly be held answerable for them. We do not blame a person for firing a shot if someone stronger forced the gun into his hand and pulled his finger against the trigger. We do not hold someone guilty of a crime if he was genuinely insane and lacked any control over his decision and action. It would be pointless to hold the man whose hand was forcibly manipulated or the insane person accountable for an action not truly his own. Neither is responsible because neither enjoyed freedom of choice and will over his act. (Clear-cut examples like these should not, however, invite multiplying excuses to relieve persons of responsibility for their actions.)

Responsibility presupposes a point for applying a motive, such as desire to avoid blame or punishment or to win praise or reward. Frequently it makes eminent sense to apply motives to people and hold them responsible for their choices and actions. This could not be true if no grounds existed for attributing freedom to people. Hence there are grounds for belief in freedom in some sense associated with responsibility.

More exactly, perhaps, the whole free-will/determinism controversy is a chimerical basis for questioning ordinary ethical concepts. Schlick’s argument comes across to me as I have summarized it.

C.A. Campbell (1951/1966) finds Schlick’s distinction between descriptive and prescriptive laws irrelevant. The usual reason for thinking that moral freedom presupposes some breach in causal continuity is not a belief that causal laws compel in the way legislation compels but instead is the belief that an unbroken causal chain leaves no one able to choose and act other than as he does.

Moral responsibility is not the same, says Campbell, as scope for sensibly applying motives. Dogs can be trained with punishments and rewards; yet we do not hold dogs morally responsible for what they do. We can judge dead men morally responsible for particular actions without being able to affect those past actions. Perhaps we might reinterpret Schlick as meaning that a person is morally responsible when his motive could in principle be affected by reward or punishment, whether or not the judges or observers are in a position to apply it. But this modification would change Schlick’s theory, which links the whole meaning and importance of moral responsibility “to our potential control of future conduct in the interests of society” (Campbell 1951/1966, p. 115).

Schlick identifies “Who is morally blameworthy?” with “Who is to be punished?”—paradoxically, given his view of punishment as a purely educative measure, without retributive content. We often think it proper to “punish” a person, in Schlick’s educative sense, even without holding him morally blameworthy (Campbell 1951/1966, p. 116). We punish the dog. We punish demonstrators who may be obstructing traffic from motives that even we, the judges, may think noble.

I’ll try to rephrase or interpret Campbell’s objection. Schlick sees instrumental, educative, value in applying rewards and punishments, which he identifies with holding people morally responsible. We could hardly do so unless we attributed some freedom to people. But is this a valid linking of ideas? Perhaps rewards and punishments and their generally good consequences are just particular events in the unbroken causal chain. Metaphysical freedom cannot be established by pointing to the apparent or genuine good consequences of reward and punishment.

Campbell suspects that Schlick and many other philosophers cannot recognize contracausai freedom as prerequisite to moral responsibility because, while denying that freedom, they do accept the commonsense belief in moral responsibility (Campbell 1951/1966, p. 117).

His own purpose, Campbell concludes (p. 135), has been not actually to defend free will but rather to show “that the problem as traditionally posed is a real, and not a pseudo, problem.”

HAZLITT’S RECONCILIATION

Henry Hazlitt (1964/1972, chap. 27) tries to reconcile free will and responsibility with determinism, interpreted as omnipresent cause and effect. He agrees “that everything that happens is a necessary outcome of a preceding state of things” (p. 269). Like Schlick, however, he stresses that causation is not compulsion. Absence, not presence, of causation is what would exempt people from moral responsibility. “It is precisely because we do not decide or act without cause that ethical judgments serve a purpose... . The knowledge that we will be held ‘responsible’ for our acts by others, or even that we will be responsible in our own eyes for the consequences of our acts, must influence those acts, and must tend to influence them in the direction of moral opinion” (p. 275).

Hazlitt warns against confusing determinism with materialism, interpreted as the dogma that all causation, even in human affairs, operates ultimately through physical and chemical processes alone. He especially warns against confusing determinism with fatalism, which he interprets as the dogma that events will unfold as they are bound to do, regardless of how people try to promote or prevent them. Fatalism in this peculiar sense is obviously false. Human decisions, choices, wishes, reflection, and will clearly do influence the course of events. If, contrary to fact, they did not do so, or if they operated only stochastically, outside of causal chains, then notions of responsibility and ethics would have no application.

Hazlitt accepts universal causation, then, but distinguishes sharply between its supposed operation solely in material ways and its operation in ways leaving scope for human decision and will. But can this distinction carry all the weight Hazlitt places on it?

TWO FURTHER ATTEMPTS AT RECONCILIATION

Michael Slote (1990) explains how making and implementing ethical judgments could be sensible even if determinism prevailed, a question he does not tackle. Here I interpret Slote’s argument together with a commentary by Peter van Inwagen (1990) and forgo trying to paraphrase each separately.

We may label a person or a dog and certain actions as “vicious” and guard against and “punish” them. Yet we may recognize that the person’s or dog’s disposition and actions trace to unfortunate genes or mental illness or previous maltreatment, which attenuates or dispels moral culpability. We are not necessarily inconsistent in both recognizing the dispositions and actions as determined yet judging and punishing them as vicious. Our judging and punishing can themselves be links in the chain of deterministic causation and may make the dispositions and actions less vicious than they would otherwise be.

Similarly, we tend to judge actual murder “more wrong” than a failed attempt; we revile and punish an actual murderer more severely than an attempted murderer. Both culprits may have had the same intentions, and only sheer luck may have frustrated one attempt. Still it may make sense to condemn and punish the successful murderer more severely. How a person is judged and punished may thus reasonably depend on more than what he freely willed. Several considerations may warrant distinguishing between actual and attempted murder. Evil intentions may be harder to prove in a failed attempt. The gradation in punishments may help emphasize the public’s solemn condemnation of murder, and in subtle psychological ways it may cause more murder attempts to fail than otherwise would.

An analogy of sorts holds between these considerations and F.A. Hayek’s argument (1960, chap. 6) about merit versus value in determining a person’s income. Even though the market value of a person’s efforts probably does not correspond closely to his moral merit, powerful reasons argue for allowing market supply and demand to establish his income anyway. Such remuneration may usefully guide individuals on how to use their special talents and knowledge, and it may motivate appropriate kinds and degrees of risk-bearing. Above all, perhaps, alternative institutions intended to attune remunerations to moral merit appear very unattractive upon close analysis. Again, a person’s free will and intentions should not be the only factors governing how other people treat him.

Admittedly, full-fledged determinism still poses embarrassment for consequentialist considerations like these. If we recognize that our making and implementing ethical judgments and adopting this or that set of institutions are themselves fully caused and are mere links in a tight causal chain, we run into awkward paradoxes. These pertain to the whole free-will/determinism issue itself, however, rather than to ethical issues in particular.

DETERMINISM AS FATALISM

Determinism in its most extreme version (commonly attributed to Pierre-Simon de Laplace) is fatalism even more comprehensive than the variety rejected by Hazlitt. It recognizes human will and decision as elements in one grand chain of universal causation. Everything that is happening or has happened or will happen has been fated from the beginning of time to happen exactly as it does or did or will. Causation operates tightly in every detail. Even all of a person’s thoughts as he deliberates whether to accept a new job or break off a love affair, and even all other persons’ reactions to his decision, were fated to be exactly as they turn out. Even all philosophical controversies over the free-will issue itself take an exactly predetermined course. Far from denying that ideas and choices have consequences, extreme determinism maintains that even these are links in the great causal chain.

Laplace regards the present state of the universe as the effect of its anterior state and the cause of its next state. Ineluctable necessity rules. Nothing would be uncertain for a sufficiently vast intelligence; the future and the past alike would be present to its eyes. A true act of free will is impossible. Without a determinative motive, not even the nearest thing to a free will could originate even actions considered indifferent. The contrary opinion is an illusion of the mind (Meyerson 1921/1991, pp. 563—564, citing Laplace’s Théorie analytique desprobabilités). (Boyle et al. 1976, pp. 57, 86, give apt quotations from Laplace; further discussion of Laplacean determinism occurs in Popper 1982, pp. xx-xxi, 123—124, and passim; and Georgescu-Roegen 1971, p. 170 and passim.)

Clarence Darrow used to defend his clients with such an argument. The accused criminal is a mere link in the chain. Even his character and his ability or inability to reshape it trace ultimately to causes outside himself, and he is therefore not responsible and not properly punishable for his crimes (Hospers 1961/1966, p. 41).

Can anyone really believe in such tight universal causation? If only Queen Victoria had been a man, the Salic Law would not have separated the hitherto linked crowns of Great Britain and Hanover upon her—his—accession in 1837; and the subsequent history of Germany, Europe, and the world would probably have unfolded much differently from how it actually did. (Reflection on the events of 1866,1870—1871,1914, and 1917 helps explain why.) Much depended, then, on which particular sperm happened to fertilize her mother’s ovum at Victoria’s conception in 1818.³ Yet this micro event and all its momentous consequences were bound to occur exactly as they did. Thus must strict determinism maintain.

No one, to my knowledge, espouses this position consistently. It is just too preposterous—though I may be mistaken in saying so.

INCREASING COMPLEXITY

One reason for calling full determinism preposterous is that the world seems to be getting more complicated over time. It is hard to imagine how the less complicated past might contain all the information necessary to specify the more complicated present and future in complete detail. Such complete specification would constitute at least an equal degree of complexity already prevailing.

The world is getting “more complicated” by any ordinary standard of judgment. More people are living than in the past, with all their individual characteristics and thoughts and actions. The number and intricacy of the works of man are increasing, including the texts of all the books and articles ever written. The state of the world at any instant includes all the information and all the misprints in all these documents, and even the slightest details of all the flyspecks and coffee stains in individual copies.

Affairs on earth interact with affairs throughout the universe. Men or man-made instruments have disturbed the surfaces of the moon and Mars, and rockets have escaped our solar system. Eclipses, comets, planetary movements, and supernovas have affected human activities directly and through popular, religious, and scientific beliefs.

If a later state is fully determined by an earlier state, then that earlier state must contain aspects or properties or patterns or whatever—whose totality I am calling “information”—specifying that later state in complete detail. And if the world is generally getting more complicated over time, then more information is required to specify a later state than an earlier state. It is hard to imagine how all the detailed information necessary to specify the more complicated later state already existed in the simpler earlier state. It is hard to believe that even the tiniest fraction of a second after the Big Bang, the universe already contained detailed coded information about everything that would ever happen thereafter, including the exact configuration of every wisp of cloud I observed during my last airplane trip and including the exact times at which and pressures with which I would strike each key during my current session at my computer keyboard. Full determinism seems still more incredible because it involves each state’s specifying not only one subsequent state but also all the infinitely many intervening states (“infinitely many” if time is continuous).

These points tell against complete causal determination.⁴ Its being hard to conceive of does not, however, constitute disproof. Perhaps increasing overall complexity is a mere illusion. Perhaps greater complexity in some dimensions—in the products of the human mind and in the details of flyspecks on published pages—is offset somehow by reduced complexity in other dimensions. If so, what might they be? Perhaps greater complexity on our earth, which, like our whole solar system, is an open system, is offset somehow by reduced complexity elsewhere in the universe. Even so, wouldn’t the point still hold that increasing complexity on our earth implies incomplete predictability of human affairs? By what mechanism, if any, could any offsetting reduced complexity elsewhere save the complete predictability of human affairs, if only in principle? But perhaps I am wrong about my notion of information required for complete causal specification.

Speaking of total causal determination, we may well pause to ask just what “cause” and “causality” mean. Trying to frame objection-free definitions is a sobering challenge. This very difficulty throws some slight extra embarrassment onto doctrines of a great unbroken causal chain.

In arguing for indeterminacy or openness in the universe, Karl Popper distinguishes among “three worlds.”⁵ World 1 contains physical objects—rocks, trees, structures, living creatures, and physical fields of force. World 2 is the psychological world of fears and hopes, of dispositions to act, and of subjective experiences of all kinds. World 3 contains products of the human mind—artworks, ethical values, social institutions, the intellectual contents of books (books as physical objects belong to World 1), scientific problems, theories, including mistaken theories, and solved and unsolved puzzles. Especially characteristic of World 3 is human knowledge put into words.

Autonomous objects exist even in World 3. Human beings originally conceived of the prime numbers and conjectured about their properties, but the primes and their properties have taken on an objective existence. It is a bare fact, but a logical truth rather than a contingent empirical fact, that 143 prime numbers, no more and no fewer, exist in the range of 100 to 1000. Euclid already proved the infinitude of prime numbers: no largest one exists. But is there a largest pair of twin primes (like 17 and 19, 521 and 523, 1451 and 1453)? No one, the last I heard, has actually proved either a “yes” or a “no” answer. The problem objectively exists as a challenge to human intelligence.

The autonomous objects of World 3 interact with World 1 through the human perceptions, feelings, dispositions, and decisions of World 2. The challenges of pure mathematics lead to results that find applications in computer hardware and software, which in turn function in changing the physical world. Some challenges of her field lead a mathematician to results that enhance her reputation and win her an appointment at a prestigious university, where she has a house—a physical object—built in accordance with her tastes and increased income.

World 3 is intrinsically open or emergent, says Popper (1972/1982, p. 5); any theory holding scientific and artistic creation ultimately explainable by physics and chemistry seems absurd to him. Moreover, interrelations among the three Worlds render the whole universe partly open and emergent.

I am not sure that Popper would agree, but his concept of World 3 in particular, the world of things like scientific theories, does help underline how preposterous it is to suppose that each later state of affairs is totally specified by earlier states. Scientific progress does occur. New knowledge, by its very meaning, was not available in advance; the notion of something being known before it is known is self-contradictory Is it plausible, then, to maintain that all the mathematical and physical knowledge not yet achieved but that will be achieved in the next hundred years somehow already exists in latent form, already somehow coded into the current state of the universe, along with the date and other details of the discovery of each bit of that future knowledge? (Many of the associated challenges already exist as problems belonging to Popper’s World 3, but this is not the same as the preexistence of solutions both to unsolved problems and to problems not yet even formulated.)

CHANCE

Gerd Gigerenzer and coauthors (1989/1993, esp. pp. 59—68,276—285) review apparent and supposed implications of probability theory and statistics for free will versus determinism. These disciplines have been successfully applied on the assumption that some sheer random processes do operate in the world, which is some evidence, if weak, that determinism in nature is incomplete. Scientists and philosophers such as James Clerk Maxwell and Charles Peirce have believed that airtight causality does not operate in every little detail of the universe, that some element of randomness remains, and that free will might occupy this gap somehow. And such gaps might not be confined to the subatomic level that quantum theory deals with. (Here I insert the obligatory allusion to Heisenberg’s indeterminacy principle concerning subatomic randomness.) Maxwell and Peirce (1877, reprinted 1958, p. 95; articles of 1891, reprinted 1955, pp. 9, 319) pointed to the statistical or probabilistic aspect of the kinetic theory of gases, which envisages the constituent molecules moving at different randomly determined velocities and changing velocities as they collide with one another and with the sides of the container. Peirce also noted the random nature of the biological mutations on which natural selection operates.

Karl Popper also argued for the genuineness of chance events even above the subatomic level or even the molecular level. What explains the statistical stability of the heads and tails produced by a penny-tossing machine? Or consider Alfred Landé’s conception of dropping ivory balls onto the center of a suitably positioned steel blade, very nearly half of the balls falling on each side. For a determinist, barred by his doctrine from appealing to randomness and reduced to imagining the mutual cancellation of many small causes, the lawlike statistical process must remain ultimately irreducible and inexplicable (Popper 1982, pp. 96—104).

Quantum-level and other small-scale indeterminacies gain relevance from the fact that micro differences can have macro consequences. Erwin Schrödinger gave the hypothetical example of a cat whose survival or death in an experiment depends on an apparatus detecting particles emitted randomly and infrequently in the decay of a radioactive element. The far-reaching consequences of Queen Victoria’s sex, already mentioned, provide another example. This micro-to-macro principle is further illuminated by the mathematics of chaos, even though the (hypothetical) systems used in expounding chaos theory are fully deterministic.

An element of sheer chance in the universe appears to operate, then, along with the causality that is also evident. Admittedly, the pervasive appearance of chance or randomness does not rigorously rule out complete Laplacean causality. (Laplace himself made contributions to probability theory.) My statistics professor at Columbia University around 1948, Frederick C. Mills, avoided speaking of “chance,” period; he always used some such expression as “the complex of unknown causes called chance.” Perhaps he had good reason for speaking so carefully. (In his 1938 book, p. 436, he lists three assumptions underlying the derivation of the normal curve of error: “(1) The causal forces affecting individual events are numerous, and of approximately equal weight. (2) The causal forces affecting individual events are independent of one another. (3) The operation of the causal forces is such that deviations above the mean of the combined results are balanced as to magnitude and number by deviations below the mean.”)

Another reservation about sheer chance or randomness requires mention. Chance poses no less difficulty for commonsense notions of human freedom and responsibility than tight causality would. To the extent that a person’s actions, decisions, deliberations, inclinations, feelings, experiences, capabilities, and character traits occur by sheer chance, they are no more meaningfully his own, and he is no more truly responsible for them, than would be true if they all traced fully to external causes. Actions and thoughts governed by sheer chance are no more compatible with human dignity and responsibility, as ordinarily conceived, than their being dominated by external causes. Dignity and responsibility, if genuine, presuppose something beyond chance linking events; they presuppose a causal link in which the individual plays some independent part.

While elements of sheer chance in the world do not imply freedom of the will, “the presence of random phenomena at the quantum level does take the sting out of the argument that man cannot will freely because the material world is governed by determinism. Clearly, a completely deterministic world and a man with an absolutely free will are incompatible conditions” (Georgescu-Roegen 1971, p. 177, in part citing H. Margenau, Hermann Weyl, and A.S. Eddington). (I’ll add that not merely an “absolutely” but even a partially free will is incompatible with complete determinism.) The point so far is not that indications of sheer chance in the world establish the case for free will but only that they defuse one particular kind of argument against it.

Causality as opposed to chance is required for any predictability in human behavior. Yet predictability does not rob human beings of the dignity usually associated with free will and responsibility. If anything, the contrary is true. Suppose that a friend of yours had an opportunity to steal $10,000 while escaping suspicion. In fact the money remains unstolen. Which would your friend rather hear from you: “I was sure that you would not steal the money” or “I didn’t have a clue whether or not you would steal it”? Would your friend be insulted by your thinking that his behavior is predictable? (Compare Hobart 1934/1984, p. 81).

Far from being vitiated by elements of stable relations—of causality—among events and circumstances, including character traits and actions, the very concepts of free choice and responsibility presuppose such elements. Often these elements make confident predictions possible, all without undercutting notions of free choice and responsibility.

Sheer chance, in short, not only does not establish the case for human freedom and responsibility but even poses difficulties of its own. Its role in my argument is different and slighter: it undermines one particular argument against freedom of the will.

Once chance has shaken the notion of total causal determination, the path remains open for considering whether something besides chance might also contribute to the evident openness or indeterminacy of the universe. Everyday evidence, considered next, testifies to some sort of free will. That evidence can be questioned, but the questions rely precisely on the determinist doctrine that is itself open to question.

THE EXPERIENCE OF FREE WILL

Everyone’s experience suggests that people’s decisions, talk, writings, and thoughts do influence the course of events. The thoroughgoing determinist or fatalist would not deny this personal experience, but he would question its significance. Our decisions and thoughts, influential though they are, are mere links in unbroken causal chains. Each decision, utterance, and thought is caused by other events and circumstances, including physical conditions, the previous thoughts and utterances of oneself and other people, and one’s own and other people’s character traits, genetic makeups, and current and past environments—according to the determinist. Each of these causal links traces to contemporaneous and earlier links—and so on, presumably, back to the Big Bang.

A hardened habitual criminal could have avoided committing each of his crimes if he had willed not to commit it. But could he have so willed? Well, yes, if his character had been different. Furthermore, it would have been different if his earlier actions and decisions and circumstances had been different. But could they have been different? These earlier character-influencing events and circumstances, perhaps especially including his childhood environment and his genetic makeup, were themselves links in an unbroken causal chain.

The complete determinist is unimpressed, then, by the observation that the criminal could contingently have avoided committing his crimes. Far from proving his responsibility for them, that observation is an uninformative truism. It merely says that if the links in a causal chain had been different from what they actually were and had been fated to be, then the outcome of the chain would have been different from what it in fact was.

Despite these assertions of the (imaginary) fatalist, we all have personal experience with making decisions ourselves. We decide, true enough, largely in the light of external circumstances. Often these include the expected reactions of other people. But it is we ourselves who weigh the considerations pulling one way and another. We know from our own experience with decisions, furthermore, and from what observation suggests about the decisions of other people, that people do respond to prospects of reward and punishment, approval and disapproval. (Surely economists understand about incentives.) Holding people responsible does affect their behavior.

A fatalist could accept this conclusion without abandoning his doctrine. He could agree that if juries, judges, and legislatures generally accepted the Clarence Darrow defense, crime would be more rampant than it is in fact, and the world a more miserable place. Society is fortunate, he could agree, that juries, judges, and legislatures, usually ignoring Clarence Darrow, as they are fated to ignore him, do hold criminals responsible and do punish them. We are fortunate, in other words, that his determinist theory is not generally accepted and implemented. Yet the fatalist could maintain that his theory is correct, that he is fated to propound it exactly as he does, and that—probably fortunately—you and I and most of the rest of us are nevertheless fated to reject it.

How would the theory of strict determinism interpret academic disputes over that theory itself? Taken literally, it would regard each move in the dispute—each conversation, lecture, journal article, criticism of an article, reply to the criticism, and every slight detail in each of these—as simply a particular link in the great causal chain. The determinist philosopher would agree that his latest paper on the topic was fated in every slightest detail to say what it does say, fated not only by what he had heard and read on the topic but by his genes and childhood experiences and innumerable other circumstances. All reactions to his paper are similarly fated. Yet this consideration does not necessarily lead him to abandon the whole issue and turn to some other branch of philosophy or some other line of work. He could stick to the issue, recognizing that he is fated to do so and that his work on it, and others’ reactions to it, are fated to turn out exactly as they do.

This determinist position, then, may not be downright inconsistent with itself, not downright self-refuting. Rather, it is practically incredible.

THE SELF-REFERENTIAL PROBLEM OF DETERMINISM

Joseph M. Boyle, Jr., Germain Grisez, and Olaf Tollefsen (1976) expose the self-contradictory position of one who “rationally affirms” full determinism or, as they say, denies that anyone has any “free choice.” (To “rationally affirm” a proposition goes beyond merely mentioning it or considering it possible; it means holding that it is true or at least more reasonably acceptable than its contradictory.) A philosopher who argues that persons interested in the issue ought rationally to accept the no-free-choice position must believe that although they are not compelled to accept it, they can rationally accept it (“ought implies can”); they have some freedom of choice in the matter. Yet the determinist proposition being urged denies that the persons addressed have any freedom. Either the determinist is thus contradicting himself or else is pointlessly urging people to do what by his own doctrine they cannot do—make the free choice of rationally accepting that doctrine.

Boyle and his coauthors do in effect recognize the possibility, mentioned in the preceding section, that determinism is true and that participants in controversy over it versus free choice are behaving like fully programmed robots whose every slightest verbal move in the game is a fully determined rather than rationally chosen action. “To affirm [the determinist] position in this way, however, is to withdraw from the philosophical controversy” (p. 169).⁶

THE IRREFUTABILITY OF DETERMINISM

Besides being practically incredible and besides putting its proponents in the position of either contradicting themselves or avowing themselves to be mere robots rather than rational controversialists, determinism is irrefutable or unfalsifiable—in the bad sense. It has a built-in immunity to any adverse evidence; its claim to say anything definite about how the world actually works is a sham. No perceptions of persons that they are more than mere cogs in tightly working machinery, that they have some scope for making decisions not totally predetermined by their genetic makeups and past experiences, count for anything; for these very perceptions have themselves been predetermined. Similarly, no number of episodes in which unexpected, astonishing, or unpredictable decisions of particular persons brought major consequences count for anything. Apparent examples of formidable exertions of will count for nothing. The theory itself rules such episodes out as evidence on the grounds that the cited decisions and exertions, as well as their being unexpected, astonishing, or apparently unpredictable, are themselves mere links in the universal causal chain. Examples in which a person seems to have changed his very character by effort of will would not faze the determinist. He would maintain that the person’s exertion of will, and with what degree of success, were themselves predetermined. Determinism does not deny that praise and blame, reward and punishment, can be efficacious in influencing behavior; it simply maintains that these in turn are predetermined.

What adverse evidence of any sort is even conceivable, then, from which the theory does not protect itself in advance? A theory that can accommodate absolutely any evidence does not specify any genuine restrictions on how the real world actually works; its ostensible empirical character is a sham.

Furthermore, the theory does not carry any actual implications for how to live one’s own life or for public policy. Should individuals cultivate a sense of control over their own decisions and actions or, at the other extreme, cultivate a fatalistic outlook? Some psychologists may offer the one line of advice and others the opposite line; but in any case, each is merely offering the advice he is fated to offer. A determinist philosopher is not necessarily bound to advise the fatalistic outlook; for he may recognize the benefits of feelings of autonomy and responsibility and himself feel, furthermore, that he is fated not to undercut such feelings and the benefits flowing from them. Each ordinary individual, similarly, is receiving the advice he is fated to receive and will respond to it, along with other influences, as he is fated to respond.

Should criminals be held more responsible for their actions and more liable to punishment than they currently are or, on other hand, should the Clarence Darrow defense be given greater heed? The first policy shift may reduce crime and make for a healthier society on that account, although it would be unfair to criminals who are, on the determinist theory, mere unfortunate links in a causal chain. (Incidentally, doesn’t determinism undermine even the concept of unfairness?) In either case, policymakers will hear the arguments they are fated to hear and respond as they are fated to do.

In short, the determinist theory not only has built-in immunity to adverse evidence but also lacks implications about how to apply it in practice. It is empty. Individuals, by and large, cannot bring themselves to regard it as meaningful and to conduct their own lives and public policy in accordance with it.

I am saying not that full determinism is wrong but that it is an empty, meaningless doctrine. This conclusion is not, I believe, one of the airy dismissals of philosophical issues that used to characterize a crude logical positivism.

EXTREME POSITIONS AND PARTIAL DETERMINISM

I have tried to show that total fatalistic determinism is empty, perhaps even absurd. We might now try to focus on the opposite extreme position, except that complete free will and absolute indeterminism are downright inconceivable. No conceivable self is free of a biological nature and of the influences imposed by an external world. Still, let us see how far we can get in imagining a self whose will is essentially free.

A self whose character had been determined not by heredity and environment but only internally would be the product of a core self, a miniature self within the self, as R.E. Hobart says (1934/1984). But how could that core self be free from external influences? Only by its character having been determined by a further internal miniature self, and so on in preposterous infinite regress. “To cause his original self a man must have existed before his original self. Is there something humiliating to him in the fact that he is not a contradiction in terms?” (Hobart 1934/1984, p. 505).

In some respects, of course, a persons earlier self does partially shape his later self: his earlier decisions and actions affect his capacity for and inclinations toward later experiences, decisions, and actions. But if a person does improve his qualities, what could merit praise but the ingredient of aspiration and resolution in him that made his effort possible (Hobart 1934/1984, p. 505)? What could merit praise except features of an already existing character that could not have been fully its own creation? One praiseworthy character trait is the capacity to respond suitably to praise, blame, and the concept of responsibility.

Any consistently conceivable self must to at least some extent, then—and experience suggests a large extent—be the product of external forces. What implications follow concerning the freedom, autonomy, dignity, and responsibility of the individual? Hobart faces the question: How can anyone be praised or blamed if heredity and circumstance have ultimately given him his qualities? Does the fact that a person did not create himself bar recognizing his character for what it is? If—inconceivably—someone had somehow made his own “original character,” and a fine one, and if we praised him for it, we would be ascribing a still earlier character to him. Praise or blame for decisions or actions refers to what kind of person took them; there is nothing else for praise or blame to refer to (Hobart 1934/1984, p. 505).

A person’s character at a particular time is what it is. It inclines him to the kinds of intentions and decisions and actions that it does incline him toward; so it meaningfully exposes him to admiration or reprehension, praise or blame. This is true regardless of just how his character came to be what it is. A reprehensible character remains reprehensible even though it can be explained, or explained away, as the product of adverse heredity and environment. The notion of character being admirable or reprehensible only to the extent that it is internally determined, free of external influences, is a self-contradictory notion.

An analogy of sorts holds with a person’s wants and tastes. J.K. Galbraith (1958, esp. chap. XI) made much of what he called the “dependence effect”: many of an individual’s wants in modern society are not wants that he would experience spontaneously if left to himself. Instead, his wants are created by the process of satisfying them. The consumption patterns of other members of society, and notoriously advertising, create wants. Wants that are artificial in this sense cannot be urgent or important, so the implication runs (and, in Galbraith’s view, incomes that might nevertheless be spent on meeting them may properly be taxed heavily to finance really important services of the kinds supplied by government).

F.A. Hayek (1961/1967) calls this argument a non sequitur. Suppose that people would indeed feel no need for something if it were not produced. If that fact did prove the thing of small value, then the highest products of human endeavor, including the arts, literature, and the marvels of high technology, would be of small value. Standards of hygiene and the demand for products with which to meet them, instead of arising spontaneously within each separate individual, are likewise social products.

More generally, the individual himself is the product of social forces, operating largely through language, which conditions his thoughts, values, and activities. Recognizing the individual as a social product in no way denies that happiness and misery, success and frustration, are experienced by individuals; there is no such thing as collective happiness distinct from and transcending the happiness of individuals. Recognizing how society shapes its members in no way imposes collectivist or communitarian rather than individualist thinking and policies.

The analogy, in brief, amounts to this: A person’s tastes are what they are and their gratification or frustration causes him pleasure or unhappiness, even though his tastes are themselves largely the product of external influences. Similarly, a persons character is what it is and does expose him to admiration and praise or reprobation and blame, even though his character, like his tastes, is itself largely the product of external influences.

Praise and reward, blame and punishment, are appropriate to the extent that they are capable in principle of influencing actions, decisions, and character traits, inappropriate otherwise—so Moritz Schlick persuasively argues. Having grown up in ghetto poverty is no valid excuse for robbery, mayhem, or murder; on the other hand, it is pointless to blame a person for actions imposed by congenital deformity or actual insanity. Reward and punishment, praise and blame, all implicitly acknowledge a partial determinism operating in human affairs. (Sometimes, however, a distinction holds between punishment and blame, as in the case of the unruly dog. Individual or collective self-defense against criminally insane persons, as against mad dogs, is not the same as assigning moral culpability. “Punishment” in such a case, like quarantine of a disease-carrier, is not punishment in the fullest sense.)

Praise or blame, reward or punishment, is appropriate for an act committed freely, even and especially for one committed in accordance with the agent’s moral character. Its appropriateness does not hinge on the agent’s character being totally uncaused, whatever that might mean. Praise or blame would be inappropriate if it would have no effect on acts of the type in question and no effect on propensities to commit them.

Partial determinism, which responsibility presupposes, is fundamentally different both from full determinism and from complete (perhaps stochastic) indeterminism. It recognizes that causality does operate in human affairs, as in the rest of the universe. It recognizes that how an individual will decide when facing a particular choice may be heavily or decisively influenced by his genetic makeup and by his past experiences. These influences include the arguments he has heard and the thoughts that have been aroused in his mind, including the concept of responsibility and prospects of praise and blame. Partial determinism does not maintain, however, that absolutely everything is fully predetermined in the minutest detail. It allows some scope for chance and possibly, also, some scope for the autonomy (or whatever it might be called) that doctrines of free will allude to.

Unlike full determinism, the doctrine of only partial determinism, recognizing scope for some sort of free will, does not enjoy built-in immunity to adverse evidence and is not devoid of practical implications. If people never experienced feelings of autonomy in making decisions—if they never experienced situations in which they felt that they personally were weighing conflicting considerations and themselves making decisions, free of total outside compulsion and constraint, and if, on the contrary, they always perceived themselves under identifiable tight compulsions and constraints—then the doctrine of free will would falter. Or if people sometimes did experience feelings of autonomy but could be shown in each case that the feelings were illusory and shown in detail just how their supposed free choices were in fact externally predetermined in full, again the free-will doctrine would be undermined. Most obviously, the doctrine would be discredited if people were always keenly aware of being mere links in a causal chain and if they recognized in detail just what causes were operating on them, including recognizing just how various facts and arguments came to their attention and what weight each of these commanded.

Discrediting evidence of this sort is conceivable, and the free-will doctrine itself does not rule out its significance. That very fact shows that the doctrine is not empty. The absence of such discrediting evidence suggests, furthermore, that the doctrine may be correct. But it does not, of course, prove that it is right; no doctrine about empirical reality can ever be proved absolutely.

CONCLUSION

Discussing free will versus determinism was necessary because many philosophers consider the issue genuine and important, intertwining with the question of moral responsibility and so with ethics in general. The fatalist doctrine of an unbroken chain of tight causal determination operating from the beginning of time, of apparently ever more complex states of the world having been fully specified in advance by the apparently less complex earlier states—that doctrine is practically incredible. The idea of some kind of sheer chance almost imposes itself. Chance enters ethical discussion not because it itself provides scope for responsible human choice but because it undermines the claims of full, fatalistic determinism. Once determinism is shaken, the idea of some sort of free will, operating along side of both causation and chance, gains a possible foothold. Everyday personal experience supports some such idea.

One question, however, remains dangling: Can a person’s will be shaped in any manner other than by chance and by external influences such as heredity, environment, and experiences (including exposure to ideas concerning responsibility, praise, and blame)? Is reflection in one’s own mind such an “other” manner? No, or not unequivocally; for although ample experience testifies to its reality, that reflection is itself conditioned by external influences, including the actions and ideas of other people. Yet some such “other” manner of determination seems to be what the cheerleaders for free will are postulating.

One approach to a solution—to reconciling free will with the sort of determinism that science deals in—appeals to the notion of emergent properties. “Specific combinations, arrangements or interactions of components can give rise to totally new attributes. The whole is more than the sum of its parts.” Diamond and charcoal possess properties quite different from those of their component carbon atoms. A drum made from flat planks can roll. An essay has meaning not contained in the individual ink dots on the printed page. Laws of grammar are quite different from but not incompatible with laws of physics. Similarly, somehow, the human mind is able “to make choices not determined solely by external or genetically fixed factors; the mind is self-programming—it modifies its own processes” (Voss 1995/1996).

I admittedly cannot form a satisfactorily definite conception of what suggestions like that may be getting at. I claim, then, not to have settled the free-will issue but to have kept alive the possibility that if it is not merely a pseudo-problem after all, it anyway is not a problem subversive of ethics. The determinist thesis appears meaningless in the sense of carrying built-in immunity to any conceivable adverse evidence. Since no observations about the world could conceivably clash with it, the thesis does not really say anything about the world and about whether any free will operates in it.

While the free-will/determinism issue thus dangles unsettled, we all find ourselves seized with ineradicable impressions (or illusions) that we enjoy some freedom of action and choice and even of will. When we write as scholars, we simply cannot believe that our every word is precisely predetermined and will draw precisely predetermined reactions from our fellow scholars.

Sometimes we find two or more strands of theory applicable to certain phenomena without our being able—yet, anyway—to reconcile those strands, which may even appear inconsistent. An example concerns the apparent dual nature (wave and particle natures) of light and of electrons. The principle of complementarity, introduced to physics by Niels Bohr, condones applying each strand of theory where it does good service while still hoping to reconcile the different strands, perhaps by modifying one or all (Teller 1980, pp. 93, 105—106, 138—140). Economists formerly did not know, and some would say still do not know, how fully to reconcile three strands of balance-of-payments analysis, the elasticities, absorption, and monetary approaches. It makes sense anyway to apply each approach where it does good service while still seeking a fuller reconciliation among them.

Similarly, in analyzing the worlds of nature and human affairs, we find it reasonable to believe in tight causality or in causality loosened by an element of sheer chance. We also find reason—or at least the pressure of compelling personal experience—to believe in a loosening by some element of free will. The corresponding strands of theory are complementary. Since we cannot really believe that ethics is a field deprived of subject matter, let us continue investigating it.⁷

Let us condemn, though, the tactic of offering mere cheerleading for free will in the guise of argument. (Free will is good, those who doubt it are scoundrels, and we are on the side of the good.) As David Hume said, “this question should be decided by fair arguments before philosophers, [rather] than by declamations before the people” (1739—1740/1961, Bk. 11, Pt. in, last paragraph of Sec. 11).

REFERENCES

Boyle, Joseph M., Jr., Germain Grisez, and Olaf Tollefsen. Free Choice: A Self-Referential Argument. Notre Dame, Ind.: University of Notre Dame Press, 1976.

Campbell, C.A. “Is ‘Freewill’ a Pseudo-Problem?” 1951. In Free Will and Determinism, edited by Bernard Berofsky 112—135. New York: Harper & Row, 1966.

Dworkin, Gerald. “Acting Freely.” 1970. In Reason at Work. Introductory Readings in Philosophy, edited by Steven M. Cahn, Patricia Kitcher, and George Sher, 509—522. New York: Harcourt Brace Jovanovich, 1984.

Eucken, Walter. The Foundations of Economics. Translated by T.W. Hutchison. London: Hodge, 1950.

Galbraith, John Kenneth. The Affluent Society. Boston: Houghton Mifflin, 1958.

Georgescu-Roegen, Nicholas. The Entropy Law and the Economic Process. Cambridge, Mass.: Harvard University Press, 1971.

Gigerenzer, Gerd, Zeno Swijtink, Theodore Porter, Lorraine Daston, John Beatty and Lorenz Krüger. The Empire of Chance. 1989. Cambridge and New York: Cambridge University Press, 1993.

Hayek, FA. “The Non Sequitur of the ‘Dependence Effect’.” 1961. In Studies in Philosophy, Politics and Economics. Chicago: University of Chicago Press, 1967.

———. The Constitution of Liberty. Chicago: University of Chicago Press, 1960.

Hazlitt, Henry. The Foundations of Morality. 1964. Los Angeles: Nash, 1972.

Hobart, R.E. “Free Will as Involving Determination and Inconceivable Without It.” 1934. In Reason at Work: Introductory Readings in Philosophy, edited by Steven M. Cahn, Patricia Kitcher, and George Sher, 494—508. New York: Harcourt Brace Jovanovich, 1984.

Hospers, John. “What Means this Freedom?” 1961. In Free Will and Determinism, edited by Bernard Berofski, 26—45. New York: Harper & Row, 1966.

Hume, David. A Treatise of Human Nature. 1739—1740. Garden City: Doubleday Dolphin, 1961.

Kant, Immanuel. Groundwork of the Metaphysic of Morals. 1785. Translated by H.J. Paton. New York: Harper Torchbooks, 1964.

Lorenz, Edward N. The Essence of Chaos. Seattle: University of Washington Press, 1993.

Meyerson, Èmile. Explanation in the Sciences. 1921. Translated by Mary-Alice Sipfle and David A. Sipfle. Dordrecht and Boston: Kluwer, 1991.

Mills, Frederick C. Statistical Methods Applied to Economics and Business. Rev. ed. New York: Holt, 1938.

Peirce, Charles S. Values in a Universe of Chance. 1877. Edited by Philip P. Wiener. Garden City: Doubleday Anchor Books, 1958.

———. “The Scientific Attitude and Fallibilism.” 1891. In Philosophical Writings of Peirce, edited by Justus Buchler, 42—59. New York: Dover, 1955.

Popper, Karl. “Indeterminism Is Not Enough.” Lecture at the Mont Pélerin Society meeting in Munich, September 1972. Printed in The Open Universe, 30—36. Totowa, N.J.: Rowan and Littlefield, 1982.

———. Unended Quest. Rev. ed. La Salle, Ill.: Open Court, 1982.

———. Popper Selections. Edited by David Miller. Princeton, N.J.: Princeton University Press, 1985.

Schlick, Moritz. Problems of Ethics. 1930. Translated by David Rynin, 1939. New York: Dover, 1961.

Slote, Michael. “Ethics without Free Will.” Social Theory and Practice 16 (Fall 1990): 369—383.

Squire, John Collings, ed. If or History Rewritten. New York: Viking, 1931.

Teller, Edward. The Pursuit of Simplicity. Malibu: Pepperdine University Press, 1980.

van Inwagen, Peter. “Response to Slote.” Social Theory and Practice 16 (Fall 1990): 385—395.

Voss, Peter. “Freewill and Determinism.” 1995. Revised 1996. http://mol.redbarn.org/objectivism/writing/PeterVoss/FreeWillAndDeterminism.html.

---

^* Appendix to Chapter 2, pp. 40-58 and endnotes, in my Ethics as Social Science (Cheltenham, U.K., and Northampton, Mass.: Elgar, 2001).

¹ Walter Eucken (1950, pp. 50-51, 329-330) effectively blasts such essentialism or conceptual realism in economics.

² Just what is compulsion? How may we distinguish acts done under compulsion from free acts for which a person is responsible? In a sense, as Gerald Dworkin (1970/1984) notes, a person does his every act because he prefers it to any alternative open to him under the circumstances—even submitting to a highwayman. Dworkin distinguishes, then, between two sorts of desires or reasons for action. A free act is one motivated by a reason that the agent finds acceptable. A person acts under compulsion when responding to a reason that he does not want to have.

Although Dworkin may be on the right track, his distinction is inexact. One may undergo an operation, free from compulsion, while wishing that the reason for the operation did not exist.

³ This particular example is my own, to the best of my recollection; yet it is in the spirit of essays collected in Squire 1931. There, for example, Winston Churchill speculates on what would have happened if Lee had not [sic] won the battle of Gettysburg, Hilaire Belloc on what would have happened if the cart that in fact blocked Louis XVI’s escape at Varennes in June 1791 had gotten stuck before reaching the crucial place, and Emil Ludwig on what would have happened if German Emperor Frederick III had lived to reign until 1914 and not just for his actual 99 days in 1888.

Such examples mesh nicely with currently popular theories of “chaos” or “complexity.”

⁴ Considerations resembling these appear in Peirce 1958, selection 9, an article I had read and then forgotten many years before first drafting this appendix.

My appeal to increasing complexity and information content may admittedly appear to run afoul of the second law of thermodynamics, the entropy law, and I may be quite wrong. On the other hand, that law in its central context pertains to energy and its degradation; and its rationale is perhaps most clearly set forth with reference to the statistical properties of crowds of nonliving molecules. The law may not fully carry over to the present context. One difference from the context of inanimate processes is that in the evolution of information, as in biological evolution, selection may accomplish a kind of inner directedness. It may be that on our earth and perhaps even in the universe as a whole, neither kind of evolution violates the second law.

⁵ See, for example, Popper 1985, selections 4 and 21; 1982, section 38; and his 1972 lecture, reprinted 1982. Earlier (1908, reprinted 1958, pp. 358-379; and 1908, reprinted 1958, pp. 404-405), Charles S. Peirce had distinguished among “three Universes of Experience.” Popper’s World 3 and Peirce’s first universe correspond fairly well, as do Popper’s World 1 and Peirce’s second universe, but the remaining world and universe correspond loosely at best.

⁶ The argument of Boyle et al. is extremely complex, detailed, and repetitious, contains many cross-references and other obstacles to comprehension, and does, after all, occupy an entire book; so I cannot guarantee that my summary is entirely faithful to their argument.

⁷ Edward N. Lorenz (1993, pp. 159-160) reminds us that we should believe even in an uncomfortable truth rather than in an appealing falsehood. That premise recommends believing in free will. If it is a reality, our choice is correct. If it is not, we still shall not have made an incorrect choice, since, lacking free will, we shall not have made any choice at all.