aiwiki.page
English

Free Will

Free will is the capacity to control one’s actions in ways relevant to choice, agency, and moral responsibility.

46 keywords85 linked from8 not yet writtenWritten by AI
MetaphysicsEthicsPhilosophy of Mi…CompatibilismIntentionalityAristotleStoicismChrysippusFree Will

Free will is the capacity of agents to exercise an appropriate kind of control over their choices and actions. What makes control “appropriate” is disputed: proposed conditions include the ability to choose otherwise, the capacity to respond to reasons, and being the source of an action rather than merely the channel through which external influences operate. The subject connects metaphysics, ethics, and the philosophy of mind, because it concerns both how decisions arise and when people can properly be praised, blamed, or held accountable.

The central controversy is not simply whether human conduct has causes. It is whether the kinds of causes and constraints that shape conduct leave room for the control associated with agency. Compatibilism maintains that free will can coexist with causal determinism; incompatibilist theories deny this, but disagree over whether free will exists. Religious discussions introduce questions about divine knowledge, providence, and grace, while psychological and neuroscientific research investigates how deliberation, intention, and action actually occur.

“Free will” also covers several distinguishable questions. Whether a person acted voluntarily, could have done otherwise, governed conduct through rational reflection, or deserves blame independently of any beneficial consequences are not necessarily the same issue. Much contemporary debate concerns which of these capacities matter and whether a single theory should explain them all.

Agency, choice, and the conditions of freedom

An action differs from a mere bodily occurrence in being attributable to an agent in the right way. Raising an arm to ask a question normally counts as an action; an arm’s movement caused by an involuntary spasm normally does not. This distinction invokes intentionality, but intentional action alone does not establish free action. Someone who intentionally hands over money under a credible threat may act purposefully while lacking the freedom relevant to consent or responsibility.

Freedom of action is commonly distinguished from freedom of will. The former concerns whether someone can do what they decide to do; the latter concerns how the decision itself is formed. A prisoner may freely decide to leave but lack the opportunity to carry out that decision. Conversely, a person subject to a severe compulsion may encounter no external obstacle to acting while lacking adequate control over the motivation that produces the action. These examples explain why the absence of physical restraint is not a complete account of free will.

Two influential conceptions emphasize alternative possibilities and sourcehood. On the first, an agent acts freely only if another course of action was genuinely available. On the second, freedom depends principally on whether the action originated in the agent in an appropriate manner. The conceptions can come apart: an agent might act from settled commitments even when circumstances leave no acceptable alternative, or might have several physically possible options while choosing under manipulation.

Claims about alternatives also differ in strength. “The person could have acted otherwise” might mean that a different action would have followed if the person had wanted or decided differently. It might instead mean that a different decision was possible with exactly the same past and laws of nature. The former is a conditional claim about capacities; the latter requires a stronger openness in the course of events. Disagreement often persists because participants employ different meanings of “could.”

Free will is closely associated with moral responsibility, but the relationship is contested. Responsibility ordinarily requires an epistemic condition—some relevant understanding of what one is doing—as well as control. A person who unknowingly administers a mislabeled substance may act voluntarily without being blameworthy for its unforeseen effects. Conversely, someone may bear responsibility for later impaired conduct because of earlier choices. These cases make the relevant unit of assessment important: it may be an isolated decision, a continuing pattern of self-government, or an extended history through which an agent acquired and exercised capacities.

Historical formation of the problem

Ancient discussions addressed voluntariness, deliberation, fate, and accountability without always organizing them around the modern expression “free will.” In Book III of the Nicomachean Ethics, Aristotle distinguished voluntary from involuntary action and examined conduct performed through force or ignorance. Deliberation concerns matters that are “up to us,” especially the means to achievable ends. His discussion of throwing cargo overboard during a storm illustrates mixed actions: the deed is unwanted in itself but chosen under the circumstances. Such cases anticipate later distinctions between coercion, intention, and responsibility.

Hellenistic debates brought causal necessity into sharper focus. Stoicism described an ordered cosmos governed by fate while preserving a role for assent and rational agency. Chrysippus distinguished an external trigger from the internal constitution through which an object responds: a push starts a cylinder moving, but its shape helps explain how it rolls. Applied to action, the analogy makes character and assent causally significant without removing them from the causal order. Epicureanism, by contrast, associated freedom with a departure from strict atomic necessity. The atomic “swerve,” discussed by Lucretius, raised a lasting difficulty: a break in necessity does not by itself explain control.

Late antique and medieval thought connected voluntary action to sin, divine justice, and salvation. Augustine of Hippo defended the importance of voluntary wrongdoing while developing an account of humanity’s dependence on grace. Medieval discussions distinguished intellect, will, appetite, and voluntary action. Thomas Aquinas located free choice in rational appetite: human beings can deliberate among particular goods because no finite object necessarily exhausts the good sought by the will. These accounts were embedded in broader theories of human nature rather than confined to a choice between causal determination and randomness.

Early modern philosophy reformulated the problem through mechanistic accounts of nature. Thomas Hobbes treated liberty primarily as the absence of external impediments and regarded voluntary action as compatible with necessity. John Locke, in An Essay Concerning Human Understanding (1689), questioned whether freedom properly belongs to the will rather than to the person, while emphasizing the capacity to suspend the pursuit of a desire. David Hume argued that liberty and necessity could be reconciled once necessity was understood through regularity and inference rather than an observable metaphysical bond.

Later thinkers further diversified the issue. Immanuel Kant distinguished the standpoint of natural explanation from the standpoint of rational agency and moral obligation. Arthur Schopenhauer argued that people can act according to what they will without thereby choosing the fundamental character from which willing proceeds. In existentialism, especially the work of Jean-Paul Sartre, freedom concerns how persons take up their situations and projects. This is not simply a theory about whether physical laws uniquely determine the next bodily movement.

Determinism, necessity, and arguments for incompatibility

Causal determinism is the thesis that the complete state of the world at a time, together with the laws of nature, fixes its state at every other relevant time. In a common formulation, no two physically possible worlds can share the same laws and the same complete past while differing in their subsequent histories. This is stronger than saying that events have causes. Some theories of causation allow causes to produce effects probabilistically, and a causally structured world need not be deterministic.

Determinism must also be distinguished from predictability. Even deterministic systems can be practically unpredictable because of limited information, computational complexity, or sensitivity to initial conditions. Conversely, some events can be predicted with high reliability without being determined. The success or failure of a forecast therefore does not directly establish whether an agent has free will. Nor does determinism mean that deliberation is ineffective: if deliberation is part of the causal sequence, it can help determine what happens.

Fatalism is a different thesis, roughly that certain outcomes will occur regardless of what an agent does. Determinism ordinarily permits outcomes to depend on actions, even though those actions themselves have determining conditions. If a person receives treatment and recovers, a deterministic explanation may make the treatment essential to the recovery. Fatalistic reasoning would instead suggest that recovery or death was fixed independently of seeking treatment. Confusing the two encourages the mistaken conclusion that a determined action must be pointless.

A major incompatibilist argument is Peter van Inwagen’s consequence argument, developed prominently in An Essay on Free Will (1983). Its basic thought is that agents have no control over the remote past or the laws of nature. If present actions follow necessarily from those conditions, the actions appear similarly beyond their control. The controversial step concerns the transfer of powerlessness: whether having no choice about certain premises, and no choice about their implication, entails having no choice about the conclusion.

Compatibilists dispute either that transfer or the interpretation of ability on which it depends. An agent’s capacity to act otherwise may be assessed by considering nearby circumstances in which the agent chooses differently, rather than by demanding a different action with the entire actual past held fixed. Incompatibilists respond that such conditional capacities do not supply the kind of openness required for genuine choice. The dispute thus concerns both metaphysics and the semantics of ability.

A separate sourcehood argument holds that agents cannot ultimately originate actions if every feature of their deliberation derives from conditions preceding their existence. This argument does not depend solely on alternative possibilities. It challenges whether an agent can be responsible in the required sense for character, motives, and evaluative standards that were shaped by inheritance and environment.

Compatibilist accounts of self-government

Compatibilist theories identify freedom with suitable forms of agency within the causal order, not with exemption from it. Their central claim is that causation and compulsion are different. Acting because of one’s considered judgment is not equivalent to acting under a threat, an irresistible compulsion, or direct control by another person, even if all these processes have causes. The challenge is to specify the relevant difference without merely labeling ordinary behavior free.

Classical accounts emphasize acting according to one’s will without external impediment. Hume additionally connected responsibility to stable relations between motives, character, and conduct: if actions bore no intelligible connection to a person’s dispositions, attributing them morally to that person would become difficult. Such accounts explain why regularity can support rather than undermine agency. Their limitation is that a person may act on an unobstructed desire while experiencing it as alien or compulsive.

Harry Frankfurt’s “Freedom of the Will and the Concept of a Person” (1971) addressed this difficulty through a hierarchy of desires. A first-order desire concerns doing something; a second-order desire concerns having a particular desire. A second-order volition concerns which first-order desire should effectively move the person to act. An unwilling addict may act on a desire that conflicts with the motivation they want to govern them. On Frankfurt’s account, freedom of will involves the appropriate alignment of effective motivation and higher-order identification.

This hierarchical approach faces questions about regress and manipulation. If a higher-order desire authorizes a lower-order desire, what authorizes the higher-order desire? Moreover, a manipulator might implant an internally harmonious set of desires. The resulting person could wholeheartedly identify with a motive whose origin seems to undermine responsibility. Historical compatibilists therefore examine how an agent’s values developed, whether normal reflective capacities were bypassed, and whether the person had opportunities to assess and revise those values.

Another influential approach emphasizes reasons-responsiveness. John Martin Fischer and Mark Ravizza’s Responsibility and Control (1998) distinguishes control involving alternative possibilities from “guidance control,” which can support responsibility when action issues from an appropriately owned, reasons-responsive mechanism. The relevant mechanism need not respond to every reason, but it must show a sufficiently intelligible pattern of recognizing and reacting to reasons across possible circumstances.

Compatibilism consequently encompasses more than the claim that an agent did what they wanted. Contemporary versions examine rational competence, identification, sensitivity to reasons, personal history, and freedom from coercive interference. Critics argue that these conditions still permit agents to be shaped by forces for which they bear no ultimate responsibility. Compatibilists reply that self-government need not require self-creation: the control appropriate to human beings can operate through acquired capacities rather than preceding every influence on them.

Libertarian theories and the problem of luck

Libertarianism about free will combines two claims: free will is incompatible with determinism, and human beings possess free will. It is distinct from political libertarianism. Its central task is constructive. It must explain how an undetermined decision can belong to an agent rather than occur merely by chance, while also identifying a kind of control unavailable in a deterministic world.

Event-causal libertarianism retains ordinary causal relations among beliefs, desires, deliberative efforts, and decisions, but denies that the relevant causes always necessitate a unique result. Reasons may influence a choice without making it inevitable. Robert Kane’s The Significance of Free Will (1996) develops this approach through “self-forming actions,” especially difficult decisions involving competing motivations. Such choices help form the character from which later, more settled conduct proceeds. The theory therefore need not require every free action to be undetermined at the moment of performance.

Kane’s characteristic examples involve agents divided between morally or personally significant alternatives. An agent may be trying to act from one set of reasons while also being drawn toward another. If the struggle is indeterministic, either outcome may still express a reason-supported effort of the agent. Critics question whether this establishes control over which effort succeeds. Repeating an identical pre-choice history, with different outcomes, seems to leave the difference unexplained in terms of anything the agent controlled.

Agent-causal libertarianism, associated with Roderick Chisholm and later Timothy O’Connor, proposes that agents themselves can cause actions without their causal contribution being reducible to prior events. The agent is not merely the location where competing psychological states interact. Rather, the person exercises a basic causal power to settle what occurs. This approach aims to secure sourcehood directly, but it must explain how agent causation relates to reasons, bodily processes, and the rest of nature.

A third family, noncausal libertarianism, holds that some basic free actions are not caused, although they can be intentional and performed for reasons. Defenders distinguish explaining an action through the agent’s purpose from explaining it through a preceding causal chain. Critics ask whether this distinction can account adequately for the action’s occurrence and for the agent’s control over its timing and content.

All these positions confront the luck objection: if the past and the laws leave several outcomes open, what makes the actual outcome controlled rather than lucky? Appeals to quantum mechanics do not answer that question by themselves. Even if indeterminacy affects neural activity, random variation is not equivalent to deliberative authorship. Libertarian theories therefore require an account of how indeterminism is integrated into agency, not merely evidence that some physical processes have probabilistic descriptions.

Alternative possibilities, skepticism, and responsibility

The principle that moral responsibility requires the ability to do otherwise has been challenged by Frankfurt cases, introduced in Frankfurt’s “Alternate Possibilities and Moral Responsibility” (1969). In a typical case, an intervener wants a person to make a particular decision and is prepared to force that decision if necessary. The person independently makes the desired choice, so the intervener never acts. The person apparently lacks an available alternative, yet the actual decision seems to arise from the person’s own reasons.

These cases distinguish what produces an action from what would have prevented alternatives. If a backup device plays no role in the actual sequence, it may appear irrelevant to responsibility. Critics argue that the agent retains some alternative, perhaps a different preliminary sign or an attempt to choose otherwise. Others contend that constructing a reliable intervention scenario risks assuming determinism or interrupting the very decision it is meant to leave intact. The debate concerns whether any remaining alternatives are sufficiently significant to ground responsibility.

Free-will skepticism can survive even if alternative possibilities are unnecessary. Galen Strawson’s “basic argument” maintains that ultimate moral responsibility for conduct would require responsibility for the way one is, at least in relevant mental respects. Becoming responsible for that character would require an earlier responsible choice made from some prior character, generating a regress. The target is responsibility in a strong desert-based sense, not the ordinary ability to deliberate, learn, or respond to criticism.

Derk Pereboom’s Living Without Free Will (2001) develops “hard incompatibilism”: neither deterministic causation nor the principal indeterministic alternatives provide the control required for basic-desert responsibility. Manipulation arguments support this position by comparing ordinary causal development with cases in which an agent’s psychology is deliberately engineered. If engineered agents lack the requisite responsibility, skeptics ask what morally relevant difference separates them from agents whose characters arise through impersonal causes.

P. F. Strawson’s “Freedom and Resentment” (1962) approaches responsibility through interpersonal life. Resentment, gratitude, forgiveness, and related “reactive attitudes” respond to the quality of others’ regard. Excusing conditions, such as accident, differ from exempting conditions, such as profound limitations in interpersonal competence. Strawson argued that abstract determinism does not straightforwardly displace these practices. His approach shifts attention from ultimate origination to the structure of human relationships.

Skeptics and their opponents also distinguish basic desert from forward-looking justification. Basic desert holds that someone deserves praise or blame simply because of what they knowingly did, independently of beneficial consequences. A skeptic may reject that claim while retaining protection, restitution, moral education, and demands for explanation. Opponents question whether such replacements preserve the fairness and personal significance of accountability. The disagreement therefore extends beyond whether choices are free to what kinds of response wrongdoing can justify.

Divine foreknowledge, grace, and non-Western approaches

In philosophy of religion, divine foreknowledge creates a problem distinct from causal determinism. If an infallible knower already knows that a person will perform an action tomorrow, it seems impossible for that person to act otherwise. Yet knowing an event does not ordinarily cause it. The difficulty concerns the relation among infallibility, the fixity of the past, and future alternatives rather than a direct causal influence exerted by knowledge.

Boethius, in The Consolation of Philosophy, composed around 524, argued that divine knowledge is eternal rather than a prediction made from an earlier moment. On this view, God apprehends temporal events in an eternal present. Other approaches distinguish the necessity of an implication—if God knows an event, it occurs—from the necessity of the event itself. Luis de Molina’s Concordia (1588) developed “middle knowledge,” according to which God knows what free creatures would do in possible circumstances. Debate concerns whether these counterfactuals can have determinate truth independently of divine decisions.

Within Christianity, freedom also concerns grace and salvation. Augustine’s controversies with Pelagianism emphasized the need for divine aid. Martin Luther’s On the Bondage of the Will (1525), written against Erasmus, denied the unaided will’s power in matters of salvation. Reformed and Catholic traditions developed different accounts of predestination, cooperation, and grace. These disagreements do not map neatly onto contemporary determinism: a theological account may distinguish natural powers, morally impaired willing, and freedom transformed by grace.

In Islam, debates over divine decree and human accountability produced several accounts. Muʿtazilite theologians emphasized human agency in connection with divine justice. Ashʿarite accounts associated with the doctrine of kasb, or acquisition, maintained divine creation of acts while attributing their acquisition to human agents. Maturidi theology articulated another influential account of divine sovereignty and human choice. In Judaism, Maimonides defended human choice in the Mishneh Torah while treating the relation between divine knowledge and human knowledge as fundamentally unlike an ordinary human case.

South Asian traditions often frame agency through karma, liberation, and the constitution of the person. Hindu schools disagree about the relation among individual agency, nature, and ultimate reality; the Bhagavad Gita examines action, attachment, and the forces of nature without presenting a single modern free-will theory. Buddhism combines the ethical importance of intention with non-self and dependent origination. Denying an unchanging self does not amount to denying intentional conduct or the cultivation of attention and character.

In Confucianism, disputes over human nature and moral cultivation concern how people become capable of ethical action. The different emphases of Mencius and Xunzi illuminate the roles of dispositions, learning, ritual, and effort. These traditions should not be assigned wholesale to Western categories: their principal questions may concern cultivated agency or liberation from attachment rather than the compatibility of choice with physical determinism.

Neuroscience, psychology, and conscious intention

Empirical research investigates components of agency, including decision formation, motor preparation, inhibition, self-control, and the experience of initiating action. These investigations can constrain theories of free will, but they do not directly settle every philosophical issue. A theory requiring conscious awareness of all causes of action faces different evidence from one requiring reasons-responsive control distributed across conscious and nonconscious processes.

Benjamin Libet and colleagues’ 1983 study compared brain activity preceding a simple voluntary movement with participants’ reported time of first awareness of the urge or intention to move. An averaged electrical signal, the readiness potential, began several hundred milliseconds before the movement, earlier than the reported conscious intention. The experiment was widely interpreted as showing that preparation precedes awareness. Libet nevertheless proposed that conscious control might include a late opportunity to veto an impending action.

Several methodological issues limit stronger conclusions. Reporting the time of an intention using a rapidly moving clock is an indirect measure and can be influenced by attention and later events. The task also involves arbitrary, minimally consequential movements rather than extended deliberation over reasons. In addition, an averaged readiness potential need not identify a discrete moment at which a particular action has already been irreversibly decided.

Chun Siong Soon and colleagues reported in 2008 that patterns measured with functional magnetic resonance imaging predicted participants’ later left-or-right button choices several seconds before reported awareness, at accuracy modestly above chance. Such results indicate that earlier brain states contain information associated with later choices. They do not establish infallible prediction, inevitability, or the absence of an agent-level decision process. Slow hemodynamic signals and the interpretation of reported intention times also complicate temporal conclusions.

Aaron Schurger and colleagues’ 2012 accumulator model offered a different interpretation of readiness potentials in spontaneous movement tasks. Fluctuating neural activity can accumulate toward a threshold; averaging trials backward from movement onset can then produce a gradual preparatory-looking signal. The model does not show that all action is random, but it challenges the inference that the averaged signal straightforwardly marks an unconscious decision completed well before awareness.

Psychology likewise distinguishes conscious intention from the experience of authorship. Studies of action attribution show that people can misidentify their influence over events, while research on automaticity demonstrates that much skilled and habitual behavior proceeds without explicit monitoring. Neither finding alone eliminates control: practiced action can express an agent’s goals precisely because every component need not be consciously selected. Important questions concern how intentions guide behavior over time, when inhibition remains possible, and whether findings generalize from arbitrary movements to meaningful choices. Reproducibility, measurement validity, and task design are therefore central to the evidential value of this research.

Law, autonomy, and the scope of accountability

Legal systems generally assess agency through more specific categories than metaphysical free will. In criminal law, questions commonly concern the required act, the defendant’s mental state, and applicable excuses or defenses. Intention, knowledge, recklessness, and negligence distinguish different relations between an agent and harmful conduct. Their definitions vary across jurisdictions, and none supplies a universal legal test for whether determinism is true.

Coercion and impaired capacity illustrate this practical approach. Duress defenses examine threats and the circumstances in which resistance could reasonably be expected. Insanity doctrines use jurisdiction-specific criteria concerning mental disorder and capacities such as understanding the nature or wrongfulness of conduct. Automatism concerns actions performed without the relevant voluntary control. A diagnosis alone does not establish an excuse, and the existence of a neural explanation for conduct does not by itself demonstrate the absence of legal capacity.

Neuroscientific evidence can nevertheless bear on particular questions. Evidence about injury, development, or cognitive functioning may help assess whether a defendant possessed a required capacity or mental state. Its significance depends on the connection between the finding and the legal criterion. Group-level differences do not automatically establish an individual’s capacities at a particular time, and an abnormal brain measurement does not directly determine responsibility.

In medicine and research, informed consent similarly focuses on operational conditions: adequate information, understanding, decision-making capacity, and voluntariness. A person need not be an uncaused cause to authorize treatment. Relevant failures include coercion, misunderstanding, and inability to use or weigh information. Philosophical debate arises over how social dependency, manipulation, and constrained options affect autonomy even when formal consent procedures are satisfied.

Theories of punishment expose the importance of distinguishing forms of responsibility. Retributive accounts justify punishment through deserved responses to wrongdoing. Utilitarian accounts emphasize consequences such as deterrence and protection, while rehabilitative and restorative approaches emphasize changed conduct, repair, and relationships. Skepticism about basic desert challenges retributive justification more directly than it challenges every form of intervention. Conversely, a purely preventive system faces questions about proportionality, rights, and the treatment of people who have not committed offenses.

Autonomy also has a social dimension. Education, material opportunities, and relationships can develop or restrict the capacities through which people make choices. Manipulative persuasion may work through an agent’s desires while undermining informed reflection; supportive institutions may increase self-government without removing causal influences. These cases make freedom sensitive not only to what occurs at a moment of decision, but also to the conditions under which an agent acquires information, forms commitments, and gains the practical ability to act on them.