Free Will and Determinism

What It Is

Causal determinism is the thesis that the complete state of the universe at any given time, together with the laws of nature, entails the complete state of the universe at every later time. The state of the world a moment ago, together with the physical laws, fixes the state of the world now; the state now fixes the state in a moment.

The free-will debate asks whether causal determinism, if true, is compatible with two further claims: that agents sometimes act freely and that agents are sometimes morally responsible for what they do. The literature divides into three large positions.

• Hard determinism. Determinism is (at least approximately) true. Free will and moral responsibility require a kind of openness in the future that determinism rules out. So free will and moral responsibility are illusions: nobody is ever genuinely free or morally responsible. Holbach and Spinoza articulate early forms; Derk Pereboom's Living Without Free Will (2001) is the canonical contemporary statement.
• Libertarianism (about free will, not the political doctrine). Free will exists, but it requires alternative possibilities that determinism rules out. So determinism is false, at least where agents act freely. The agent is an originator: some indeterministic event in deliberation makes the action genuinely up to the agent. Robert Kane's event-causal libertarianism and Timothy O'Connor's agent-causal libertarianism are the modern variants.
• Compatibilism. Free will is compatible with determinism. The compatibilist denies that free action requires being able to do otherwise in a deeply metaphysical sense; what it requires is that the action issue from the agent's own deliberation, beliefs, and desires in the right way, whether or not those mental states were themselves the inevitable outcome of prior causes. Hume, Hobbes, and Mill articulate early versions; Harry Frankfurt, P.F. Strawson, and John Martin Fischer are the major twentieth-century figures.

This page assumes What Is Philosophy? and walks the contemporary debate at depth. The intersection with personal identity is in Personal Identity Over Time; the intersection with ethics is in Ethics: Consequentialism, Deontology, Virtue.

The Compatibility Question, Sharpened

The core question. If determinism is true, can an agent ever act freely? Three claims are at stake.

1. Causal determinism (CD). For each event $e$ at time $t$, the conjunction of the state of the universe at $t_0 < t$ and the laws of nature entails $e$.
2. Free action principle (FA). $S$ acts freely in $\phi$-ing iff $S$ could have done otherwise (or some refinement of this).
3. Moral responsibility principle (MR). $S$ is morally responsible for $\phi$-ing only if $S$ acted freely in $\phi$-ing.

Incompatibilists hold that CD and FA cannot both be true. Compatibilists hold they can. Hard determinists hold CD is true and FA is therefore false, with the consequence (via MR) that nobody is ever morally responsible. Libertarians hold FA is true and CD is therefore false (at least at the moment of free action).

The action is in clause (2). Three readings of "could have done otherwise":

• Categorical reading. $S$ could have done otherwise iff, holding fixed the entire prior history of the universe and the laws, there is a metaphysically possible world in which $S$ does otherwise. Incompatibilist reading. Determinism implies this is never true.
• Conditional reading. $S$ could have done otherwise iff, had $S$ chosen otherwise (or had $S$'s desires been different), $S$ would have done otherwise. Classical compatibilist reading (Moore, Hobbes, Hume). Determinism does not rule this out.
• Reasons-responsive reading. $S$ could have done otherwise iff $S$'s action-producing mechanism is sufficiently reasons-responsive that, in a range of nearby counterfactual situations, $S$ would respond to good reasons by acting differently. Fischer's contemporary compatibilism. Determinism does not rule this out.

The dispute between compatibilists and incompatibilists is at bottom a dispute over which reading captures the sense of "could" relevant to moral responsibility.

The Consequence Argument

Peter van Inwagen's consequence argument¹ is the canonical defense of incompatibilism. The argument's intuitive form:

If determinism is true, then our acts are the consequences of the laws of nature and events in the remote past. But it is not up to us what went on before we were born, and neither is it up to us what the laws of nature are. Therefore, the consequences of these things (including our present acts) are not up to us.

The formal version uses a "no-choice" operator $N$ where $N p$ reads "$p$ is true and nobody has, or ever had, any choice about whether $p$." Two rules:

• Rule alpha. $\Box p \vdash N p$. If $p$ is logically necessary, nobody has any choice about $p$.
• Rule beta. From $N p$ and $N(p \to q)$ infer $N q$. If nobody has a choice about $p$, and nobody has a choice about the conditional, then nobody has a choice about $q$.

The argument runs: let $L$ be the conjunction of the laws of nature and $P_0$ the state of the universe at a remote past time; let $P$ be any present truth. Determinism entails $L \land P_0 \to P$. Rule alpha gives $N L$ and $N P_0$. A few applications of rule beta give $N P$ for arbitrary $P$. Therefore, if determinism is true, nobody has a choice about any present truth.

Compatibilist responses. Rule beta has been contested; counterexamples by Michael Slote and Anthony Brueckner attack it on logical grounds. Compatibilists who accept rule beta deny rule alpha applied to laws of nature, or argue that the relevant sense of "could have done otherwise" is not the sense $N$ formalizes.

Frankfurt Cases

Harry Frankfurt's 1969 paper² argued against the principle of alternative possibilities (PAP): the principle that an agent is morally responsible for an action only if she could have done otherwise.

The cases construct scenarios where an agent acts on her own without any external interference, but where a counterfactual intervener stood ready to intervene if she had tried to do otherwise. The standard case: Jones decides to kill Smith and does so on his own. Unbeknownst to Jones, a neurosurgeon Black has implanted a device that would have caused Jones to kill Smith had Jones shown any sign of not deciding to kill Smith. The device is never activated; Jones acts on his own.

Frankfurt's verdict: Jones is morally responsible for the killing. But Jones could not have done otherwise: had he tried, Black would have intervened. So PAP is false.

The case has generated a massive literature on whether the counterfactual intervener really does deprive Jones of alternative possibilities. Several responses argue that some alternative possibility remains (e.g., Jones could have shown a different sign of decision, even if the action itself was fixed). The flicker of freedom strategy: even minimal alternative possibilities suffice for responsibility. Other responses (David Widerker, Robert Kane) argue that Frankfurt cases beg the question against the libertarian: they require Jones's decision to be deterministic to ensure the trigger works, so they assume the very point at issue.

Compatibilists adopt Frankfurt cases as evidence that the freedom required for responsibility is not the freedom to do otherwise; it is the freedom for the action to issue from one's own deliberation. Libertarians deny the case.

The Manipulation Argument

Derk Pereboom's manipulation argument³ is the most serious recent challenge to compatibilism. It presents a sequence of four cases, each adding causal history but preserving the agent's internal deliberative process.

• Case 1. Plum is manipulated moment by moment by neuroscientists who directly produce his deliberation, leading to a decision to murder. The deliberation satisfies every compatibilist condition (reasons-responsiveness, identification with the action, the right kind of mesh between desires and second-order volitions). Plum is intuitively not responsible.
• Case 2. Plum was programmed at the start of his life by the neuroscientists, with the program guaranteeing the decision to murder when the relevant circumstances arise. The deliberation again satisfies all compatibilist conditions.
• Case 3. Plum's character was instilled by his ordinary upbringing in a strict religious commune that guaranteed the decision to murder when the relevant circumstances arise. Again the deliberation satisfies all compatibilist conditions.
• Case 4. Plum is an ordinary determined agent in the actual world. His character resulted from his ordinary upbringing in an ordinary environment, which (together with the laws) guaranteed the decision to murder.

Pereboom's argument. The cases are progressively less obviously cases of manipulation, but at no point does anything morally relevant change. If Plum is not responsible in Case 1 (the manipulation intuition), and nothing morally relevant changes across the sequence, then Plum is not responsible in Case 4 either. So no ordinary determined agent is morally responsible.

Compatibilist responses come in two flavors. Hard-line: accept that Plum is responsible in Case 1 (biting the manipulation bullet); the compatibilist conditions are met. Soft-line: argue that something morally relevant does change across the sequence, locate the right disanalogy, and explain why Cases 1 or 2 differ from Case 4. The soft-line response is harder than it first appears; finding a principled disanalogy that does not amount to demanding indeterminism has proven elusive.

Libertarian Positions

Libertarians hold that free action requires a non-deterministic element in deliberation. Two large sub-families.

Event-causal libertarianism (Kane 1996⁴, Balaguer 2010). Free action requires that, at the moment of decision, the agent's action be caused by mental events (deliberation, desires, beliefs) in a way that is probabilistic rather than deterministic. The indeterminism is located in self-forming actions: moments where an agent's character is in the balance and either of two character-shaping decisions is genuinely open.

The major objection: if the action is genuinely undetermined by the agent's prior mental states, then the action is random with respect to those states. Random action does not enhance the agent's control; it diminishes it. This is the luck objection. Kane's response invokes a complex story about how undetermined choices can still be agent-controlled through "plural rational control."

Agent-causal libertarianism (Chisholm 1964, O'Connor 2000⁵, Lowe 2008). The agent herself, as a substance and not as a sum of events, is the cause of the action. The agent stands outside the event-causal flow as a sui generis kind of causer.

The major objection: agent-causation is mysterious. We have a clear account of event-causation (one event causes another); we do not have an equally clear account of substance-causation. Critics argue the position is a name for the problem rather than a solution.

Recent work in cognitive science of free will has weakened the empirical case against agent-causation in one respect: the Libet experiments (which appeared to show that conscious decision lags behind neural commitment) have been heavily reinterpreted. But the in-principle metaphysical objection remains.

Hard Determinism and Hard Incompatibilism

Hard determinists (Spinoza, Holbach, Strawson "Galen" 1986) accept determinism and conclude moral responsibility is an illusion. Pereboom's hard incompatibilism is a weakened version: he argues that even if the world is indeterministic in the quantum sense, the kind of free will required for moral responsibility is still impossible, because indeterminism gives randomness rather than agent-control.

The position is more pessimistic about responsibility than it first appears, and Pereboom devotes a chapter to what to do when responsibility is gone. Reactive attitudes like resentment and gratitude lose their backward-looking justification; replacement-level practices (forward-looking moral education, social engineering) are recommended. Some hard determinists (Saul Smilansky) argue we should suppress belief in the illusion of free will because the illusion is socially useful; others (Pereboom) argue we should be honest and adapt.

Common Misconceptions

• "Determinism is the same as fatalism." No. Fatalism is the thesis that what happens does so regardless of our deliberations and choices. Determinism is the thesis that what happens is determined by prior states together with laws; the laws may run through the agent's deliberations. Determinism is compatible with deliberation mattering causally; fatalism is not.
• "Quantum indeterminism saves free will." No. Indeterminism at the micro level does not by itself give the agent any extra control. If anything, it adds noise. Libertarian theories require not just indeterminism but indeterminism located in the right place in the deliberative process, and most libertarian theories require an additional theory of how that indeterminism is agent-controlled.
• "Compatibilism just redefines free will to make the problem vanish." This is a charge the incompatibilist sometimes presses; the compatibilist responds that the categorical reading of "could have done otherwise" was never the everyday meaning. The dispute is partly conceptual, partly substantive.
• "If determinism is true, there is no point in deliberating." No. Deliberation is part of the causal chain that determines the action. The deliberation is what determinism runs through, not what determinism bypasses. The compatibilist makes this point against the fatalism conflation above.

Comparisons to Related Views

Go Further

• Stanford Encyclopedia of Philosophy, "Free Will" by Timothy O'Connor and Christopher Franklin. Canonical survey.
• Stanford Encyclopedia of Philosophy, "Compatibilism" by Michael McKenna and Justin Coates.
• Stanford Encyclopedia of Philosophy, "Incompatibilist (Nondeterministic) Theories of Free Will" by Randolph Clarke and Justin Capes.
• Fischer, John Martin, Robert Kane, Derk Pereboom, and Manuel Vargas. Four Views on Free Will. Blackwell, 2007. A four-cornered debate; the best single book for grasping the live positions.
• Kane, Robert, ed. The Oxford Handbook of Free Will. 2nd ed. Oxford University Press, 2011. Reference-level surveys of every major position.
• Pereboom, Derk. Free Will, Agency, and Meaning in Life. Oxford University Press, 2014. The current canonical statement of hard incompatibilism.

For the metaphysics of persons whose actions are being judged, see Personal Identity Over Time. For the responsibility-relevant features of ethical theories, see Ethics: Consequentialism, Deontology, Virtue.