Hard Incompatibilism and Free-Will Skepticism

Free-will skepticism is the view that no one has free will in the sense required for basic desert moral responsibility — the kind of responsibility that would make a person truly deserving of blame and punishment, or praise and reward, just because of what they have done. Hard incompatibilism (the term is Derk Pereboom's) is its leading contemporary form. It earns the name by being "hard" on both sides of the free-will dilemma: it argues that free will is incompatible with determinism (siding with the incompatibilist against the compatibilist) and that free will is also undermined by indeterminism (siding with the critics of the libertarian). Since determinism and indeterminism are exhaustive, the conclusion follows on every hypothesis about the causal structure of the world: whichever way physics turns out, the desert-grounding freedom is absent.

Crucially, this is not a counsel of despair. Hard incompatibilists generally argue that morality, meaning, agency, and reform survive the loss of basic desert, and that abandoning it would improve much of moral and especially legal life. This page sets out the position, its history, its core arguments, the objections it faces, the misconceptions surrounding it, and its striking practical program.

The position precisely

The view turns on a specific, demanding notion of responsibility, so the definitions must be kept straight.

• Basic-desert moral responsibility. To be responsible in the basic-desert sense is to deserve blame or praise non-instrumentally — not because blaming produces good consequences (deterrence, reform), and not merely because one is an apt target of the reactive attitudes, but simply in virtue of having knowingly done wrong (or right), as a matter of what one fundamentally deserves. This is the responsibility presupposed by retributive punishment ("he deserves to suffer for what he did") and by certain strong forms of guilt and indignation.
• The skeptic's claim is narrow and precise: this kind of responsibility is something no human being ever has. The skeptic typically does not deny weaker, forward-looking senses (answerability, attributability, the aptness of holding-to-account for protective or formative ends).

Hard incompatibilism is thus the conjunction:

$$\underset{\text{anti-compatibilist}}{\underbrace{\text{free will}\nLeftarrow \text{determinism}}}\wedge \underset{\text{anti-libertarian (luck)}}{\underbrace{\text{free will}\nLeftarrow \text{indeterminism}}}⟹\neg \text{free will (basic desert)}.$$

It is one species of the broader genus free-will skepticism, which also includes:

• Hard determinism — the classical skeptic, which reaches the no-free-will conclusion via the premise that determinism is true. Hard incompatibilism is strictly weaker in its assumptions: it does not need determinism, since it blocks the indeterministic escape route too.
• Illusionism (Saul Smilansky) — we lack free will, but the belief in it is a beneficial illusion we should largely preserve.
• Hard-luck / no-self views (Bruce Waller, Galen Strawson) — responsibility is undone by the pervasiveness of luck in who we are, independently of the determinism question.

The two-sided argument

The distinctive structure of hard incompatibilism is that it must close both exits from the dilemma. It does so with two argumentative wings.

Wing 1 — against compatibilism (determinism does not preserve free will)

Here the hard incompatibilist deploys the standard source-incompatibilist arsenal: primarily the Manipulation Argument. Pereboom's four-case argument (Professor Plum) is the centerpiece — a manipulated agent who satisfies every compatibilist condition (reasons-responsiveness, mesh, real-self expression, a suitable history) is intuitively not responsible, and there is no principled difference between him and an ordinary agent in a normal deterministic world. So compatibilist conditions, however refined, are insufficient for basic desert. (The Consequence Argument and Galen Strawson's Basic Argument reinforce the same conclusion from the leeway and ultimacy directions.)

Wing 2 — against libertarianism (indeterminism does not help)

This is what makes the view "harder" than mere incompatibilism. Even granting the libertarian her indeterminism, the hard incompatibilist argues it cannot deliver responsibility:

• Against event-causal libertarianism — the luck / disappearing-agent objection. If the prior state of the agent (her reasons, character, efforts) leaves the choice undetermined, then nothing about the agent settles which way it goes; the rollback replays produce different outcomes with everything about the agent held fixed, so the particular outcome is a matter of luck — and luck cannot ground desert. The agent, as a settling factor, "disappears."
• Against agent-causal libertarianism — the empirical/coherence objection. Agent-causation could, if it existed, supply the missing settler (so Pereboom grants it is the libertarian's best shot and is not incoherent). But it sits in tension with our best physics: an agent-cause that produces free (undetermined) choices would have to interface with the brain's physical processes in a way that should show up as statistical anomalies in neural events — deviations from the probabilities physics predicts — for which there is no evidence and which the scientific picture gives us no reason to expect. So agent-causal libertarianism is, at best, a live but unsupported empirical bet we have no good reason to make.

With both wings in place — compatibilism insufficient, libertarianism unhelpful or unsupported — basic-desert free will is unavailable on any view of the causal order.

A short history

• Ancient–early-modern skeptical strands. Deterministic denials of freedom run from the Stoic fate debates through Spinoza (freedom as illusion born of ignorance of causes — "men think themselves free because they are conscious of their volitions, but ignorant of the causes by which they are determined") to d'Holbach's uncompromising 18th-century hard determinism.
• Clarence Darrow (early 20th c.) brought hard-determinist reasoning into the courtroom, arguing in the Leopold–Loeb case that offenders are products of heredity and environment and so not fit objects of retribution — an early public airing of the practical program.
• Paul Edwards, Galen Strawson (the Basic Argument, 1986/1994), and Bruce Waller (Freedom Without Responsibility, 1990; Against Moral Responsibility, 2011) developed the modern luck- and ultimacy-based skepticism that does not lean on determinism.
• Derk Pereboom (Living Without Free Will, 2001; Free Will, Agency, and Meaning in Life, 2014) named and systematized hard incompatibilism, supplying the four-case manipulation argument and the constructive "life without free will" program.
• Saul Smilansky (Free Will and Illusion, 2000) defended illusionism — skepticism plus the claim that the illusion is socially indispensable.
• Gregg Caruso (Rejecting Retributivism, 2021; the public-health quarantine model) and Caruso–Pereboom (Moral Responsibility Reconsidered) developed the skeptical theory of punishment and its institutional implications; Caruso–Dennett's published debate (Just Deserts, 2021) is the canonical exchange with a compatibilist.
• Neil Levy (Hard Luck, 2011) gave a sustained luck-based argument that neither compatibilist nor libertarian agency escapes the corroding effect of luck.

Arguments for hard incompatibilism

1. The exhaustion (dilemma) argument. Determinism and indeterminism are jointly exhaustive. Compatibilism fails (Wing 1); libertarianism fails (Wing 2). Therefore desert-grounding free will is impossible — not merely absent in fact, but unavailable on any metaphysics. This is the master argument and the source of the view's distinctive strength: it does not depend on settling the physics.
2. The pervasiveness-of-luck argument (Levy, Waller). Who we are is fixed by constitutive luck (the genes, temperament, and formative environment we did not choose) and present/circumstantial luck (the situations we happen to face). Since neither determinism nor indeterminism lets us transcend this luck, and desert cannot rest on luck, basic desert is undermined regardless of the determinism question.
3. The manipulation/origination argument — as in incompatibilism: an agent whose every psychological state traces to factors outside her control is relevantly like a manipulated agent, and so not a fit object of basic desert.
4. The argument from the failure of retributive justification. Independently, the skeptic argues that retributive punishment — suffering imposed because deserved — requires basic desert; since basic desert is unsupported, retributivism loses its justification, and we should prefer forward-looking institutions that do not need it.

The constructive program: living without basic desert

The most important — and most misunderstood — part of the view is that it is revisionary, not nihilistic. Pereboom, Caruso, Waller, and others argue that most of what matters survives:

• Morality and moral judgment survive. We can still judge acts right and wrong, hold axiological assessments, and have moral reasons; what changes is only the inference to basic desert. "He acted wrongly" remains true and action-guiding.
• A reformed responsibility survives. Pereboom's forward-looking account keeps a notion of responsibility grounded in three legitimate aims: protection (of potential victims), moral formation/reconciliation (engaging the wrongdoer to reform and to repair relationships), and the demand for the agent to answer for their conduct — all justifiable without desert.
• The reactive attitudes are modulated, not abolished. Pereboom argues the desert-presupposing attitudes — especially resentment and indignation — should be tempered, while stance-analogues that do not presuppose desert (moral concern, sorrow, disappointment, resolve, a non-retributive "moral protest") can do their work. Love, gratitude, and joy are argued to not essentially depend on desert and so survive intact — answering the Strawsonian worry that giving up desert would dehumanize relationships.
• Punishment is re-grounded. Retributivism is rejected; in its place Caruso defends a public-health–quarantine model: just as we may justly quarantine a person who carries a dangerous contagion — not because they deserve it, but on grounds of others' right to protection, constrained by least-infringement and rehabilitation — society may incapacitate dangerous offenders on a purely protective, non-punitive rationale, while investing in the social determinants of criminal behavior (a public-health approach to crime). This is offered as more humane than desert-based punishment, not more permissive.
• Meaning, achievement, and self-worth survive. Pereboom argues that life's meaning, the value of achievements, and self-respect do not require basic desert; they rest on engagement, relationships, and the goods themselves, which remain fully available.

A frequently cited bonus argument: relinquishing basic desert may bring psychological and social goods — less corrosive anger and vindictiveness, more compassion and willingness to repair, a more humane penal system — echoing Spinoza's thought that understanding the causes of conduct frees us from the bondage of the reactive passions.

Theories of punishment: where deterrence fits

Because re-grounding punishment is the most consequential part of the program, it helps to lay out the theories of punishment explicitly and see which ones the loss of basic desert removes — and which survive. They divide on the direction in which they look for justification.

• Backward-looking — retributivism. Punishment is justified because the offender deserves to suffer for what they did. This is the only theory that requires basic-desert free will, and so it is exactly the theory hard incompatibilism rejects: with basic desert gone, "he deserves it" loses its foundation, and retributive punishment becomes unjustified.
• Forward-looking — consequentialist. Punishment is justified by the future goods it produces, never by desert. Three such goods are standardly distinguished:
  • Deterrence — discouraging crime by attaching a credible cost to it, both for this offender (specific deterrence) and for others who might offend (general deterrence). The classic theory of Hobbes and the utilitarian Bentham.
  • Incapacitation — physically preventing an offender from harming others (e.g. confinement).
  • Rehabilitation — reforming the offender so they no longer wish to offend.

The crucial observation for this debate: none of the forward-looking justifications requires free will or basic desert. They presuppose only that human behavior is causally responsive — that incentives, confinement, and treatment reliably change what people do. That assumption is not merely compatible with determinism; it sits most naturally there (a determined agent is precisely the kind of system that responds predictably to incentives). So deterrence is one of the desert-free tools that survives free-will skepticism — which is why the forward-looking responsibility reconstruction can keep it while discarding retribution.

But hard incompatibilists are typically wary of making deterrence the foundation, for a reason the free-will debate makes vivid. Because deterrence justifies punishment by its effects on others, pure deterrence theories threaten to use the offender as a mere means — in principle licensing the punishment of an innocent, or disproportionately harsh sentences, whenever doing so would deter enough people. This is the Kantian "never merely as a means" objection, and it is sharpest exactly where desert has been removed and can no longer cap what may be done to a person. This is why Caruso's quarantine model grounds itself in incapacitation justified by the right to self-protection (constrained by least-infringement and rehabilitation) rather than in deterrence: it seeks the protective benefit without treating offenders as instruments for influencing third parties, with general deterrence at most a welcome side effect rather than the justifying ground.

In short: of the four classic aims, retribution is the one casualty of free-will skepticism; deterrence, incapacitation, and rehabilitation all remain available — but skeptics tend to build on protection and reform and treat deterrence cautiously, precisely because a desert-free framework no longer has the notion of what the offender deserves to limit how much they may be used.

Objections (and replies)

• The Frankfurt / leeway objection. If Frankfurt cases show responsibility does not require alternative possibilities, one route to skepticism is blocked. Reply: the hard incompatibilist is a source skeptic, not a leeway skeptic — the manipulation and luck arguments target sourcehood and are untouched by Frankfurt cases.
• The compatibilist's hard-line reply to manipulation. McKenna: the manipulated agent is responsible if the conditions are truly met, so Wing 1 fails. Reply: Pereboom argues the no-relevant-difference sorites is robust and the intuition of non-responsibility in Case 1 is too strong to dismiss; the burden is on the compatibilist to locate a principled difference, and historical conditions only push the determination further back without changing its character.
• The "self-defeating / can't-live-it" objection. Belief in desert and the reactive attitudes are argued to be psychologically ineliminable, so the view is unlivable; or: denying responsibility undermines the very practices (promising, trusting) that make moral life possible. Reply: (i) the forward-looking reconstruction shows the practices can be re-grounded; (ii) partial, gradual modulation of the harshest attitudes is both possible and arguably beneficial; (iii) what is ineliminable (love, basic moral concern) is shown not to require desert.
• The Strawsonian objectivity objection. Abandoning the reactive attitudes means adopting the cold "objective attitude," treating persons as objects to be managed rather than fellow members of the moral community. Reply: Pereboom denies the dichotomy — one can drop desert-based resentment while retaining moral protest, concern, and address, which preserve the second-personal, community-respecting stance.
• The "moral motivation collapses" objection. Without desert, why be good, and why would anyone reform? Reply: moral reasons, the desire to do right, concern for others, and forward-looking accountability remain fully operative; deterrence and reform are better served, the skeptic argues, by addressing causes than by retribution.
• The epistemic / burden-of-proof objection. Perhaps we should presume responsibility absent proof of its impossibility. Reply: given the high stakes of basic desert — it licenses deliberately imposing suffering — the skeptic argues the burden runs the other way: we should not inflict desert-based harm unless its presupposition is established, which it is not (a precautionary argument prominent in Caruso).

Common misconceptions

• "Free-will skeptics think nothing matters / anything goes." The opposite: they retain right and wrong, moral reasons, and accountability, and many argue the view yields a more compassionate ethics and a more humane penal system. It rejects basic desert, not morality.
• "It means criminals should go free." No — it re-grounds incapacitation of dangerous offenders on protective (quarantine-analogue) grounds, with rehabilitation and least-infringement constraints. It rejects retributive punishment, not public safety.
• "Hard incompatibilism just is hard determinism." No — hard determinism assumes determinism is true; hard incompatibilism deliberately does not, because it also blocks the indeterministic route (Wing 2). It is the more cautious, metaphysics-independent skepticism.
• "It's refuted because we obviously feel free and can't help blaming people." The feeling and the reactive impulses are exactly what the view predicts and explains; their psychological grip is not evidence that their desert-presupposition is satisfied.
• "Giving up free will means giving up love, gratitude, and meaning." Pereboom argues precisely that these do not depend on basic desert and so survive; what is tempered is desert-based resentment and retribution.
• "It's the same as fatalism — effort is pointless." No — like every position here it distinguishes determinism from fatalism; reform, deliberation, and effort are causally efficacious and central to the forward-looking program. Indeed the view urges us to change the causes of behavior.
• "It denies we are agents at all." It affirms agency, rationality, and self-control as real psychological capacities; it denies only that these suffice for the desert-grounding kind of responsibility.

Where hard incompatibilism sits

Against the master dilemma, hard incompatibilism drops premise (1) — free will exists — but does so on grounds that are independent of premise (2): it argues we lack desert-grounding freedom whether or not determinism is true, because compatibilism is insufficient and the libertarian's indeterminism only adds luck. It therefore agrees with the incompatibilist against the compatibilist, and with the luck objection against the libertarian, and takes both to their joint conclusion. Its strength is dialectical completeness — it answers the dilemma whatever physics says — and its constructive program, which aims to show the conclusion is not only bearable but in important respects better. Its burden is to make good on that program: to show that a moral and legal life shorn of basic desert really can retain accountability, meaning, and the goods of human relationship, against the compatibilist's charge that it gives up something we cannot, and need not, do without.

With hard incompatibilism, the four principal positions on the master dilemma — compatibilism, libertarianism, hard determinism, and hard incompatibilism — are all on the table. The remaining pages take up the further positions that complicate this four-way map (revisionism, illusionism, luck-based skepticism, two-stage models, and the theological debates), and, separately, the bearing of physics and quantum mechanics on whether the indeterminism the libertarian needs is real and usable.