Sourcehood and Alternative Possibilities

The Problem of Two Conditions of Control

An agent can be responsible for something only in virtue of a relation between the agent and the event for which responsibility is proposed. The relation cannot be reduced to the physical proximity of a body and an outcome. A person's movement caused by a seizure, a decision made under ordinary deliberation, a coerced confession, and an action produced by deliberately engineered preferences may be behaviorally similar while differing in what explains them. Questions of sourcehood ask what makes the action the agent's own. Questions of alternative possibilities ask whether other courses of action were available in the circumstances in which the action occurred. A complete analysis must explain why either property matters and whether one could obtain without the other.

The apparent unity of freedom dissolves into several distinct claims. An agent may be able to perform a different bodily movement if circumstances change; may be able to decide differently while the past and laws remain fixed; may have a counterfactual mechanism that responds differently to sufficient reasons; or may have direct power to settle which of several futures becomes actual. These are different modal commitments. Similarly, to call someone a source might mean that the person's body initiates a movement, that an intentional explanation cites the person's beliefs and desires, that an appropriately functioning deliberative mechanism produces the action, that the mechanism was acquired without bypassing the agent's agency, or that the person is the ultimate originator of the relevant causal chain. These are different explanatory and metaphysical commitments (O'Connor & Franklin, 2022; McKenna & Coates, 2024).

The research question is therefore relational: which sense of alternative possibility, if any, is required for which sense of sourcehood, and which configuration grounds which form of moral responsibility? Answering it requires keeping free action, free willing, accountability, and basic desert conceptually distinguishable. Their connections constitute arguments to be established.

Definitions, Modal Vocabulary, and Philosophical Scope

Sourcehood, in a minimal explanatory sense, is the condition under which an action is appropriately attributable to an agent as issuing from that agent's practical capacities. The definition intentionally leaves open whether these capacities themselves are determined. Strong sourcehood adds an ultimate-origination requirement: the agent must not owe the relevant exercise of control entirely to antecedent conditions beyond the agent's control. The disagreement between modest and ultimate sourcehood is not verbal once each side specifies which historical or causal dependence would undermine responsibility.

An alternative possibility is a genuinely available course other than the actual action, relative to a specified background. The background matters. Holding only bodily capacity fixed leaves many alternatives open; holding an entire past and the laws of a deterministic world fixed can close them. An ability conditional on different desires is not the categorical ability to choose a different desire in identical circumstances. A counterfactual sensitivity to reasons need not imply an open future at the precise instant of choice (van Inwagen, 1983; Lewis, 1981; Fischer & Ravizza, 1998).

The Principle of Alternative Possibilities, or PAP, says that a person is morally responsible for an action only if that person could have done otherwise (Frankfurt, 1969). PAP is a necessity condition for responsibility, not a sufficient condition for it. Showing that someone could act differently does not show that the agent acts with knowledge, appropriate reasons, or ownership. Frankfurt's challenge targets necessity. Source incompatibilism targets the claim that ordinary agent-source relations are enough in a determined world. These distinct targets must not be conflated.

The focal object of this dossier is the philosophical relation between sourcehood and modal openness. Neighboring inquiries about free will as a whole, causal determinism, reasons, coercion, and criminal blame are considered insofar as they illuminate this relation; their comprehensive treatments belong to independent research objects.

Aristotle: The Origin of Voluntary Action

Aristotle gives an early systematic account of voluntary action in the Nicomachean Ethics III.1–5. He distinguishes movement whose initiating principle lies outside the agent, with the agent contributing nothing, from action whose initiating principle is within the agent and whose circumstances the agent understands. His vocabulary includes archē, the initiating principle or origin, hekousion, the voluntary, and prohairesis, choice or deliberate decision. These terms form a conceptual ancestor of sourcehood without amounting to a modern compatibilist theory. The philosophical question is how an action belongs to the agent as an intelligible exercise of a capacity (Aristotle, Nicomachean Ethics III.1, 1110a1–1111a24).

The contrast between compulsion and voluntary action is refined by the example of a threat to a person's family. Someone acting under a tyrant's threat may choose an act that would otherwise be rejected. Aristotle treats such conduct as mixed while generally considering it voluntary at the time of performance because the moving principle remains in the person. This analysis is still relevant: an agent can contribute the immediate initiative under pressure while having severely diminished practical alternatives. The moral classification of the action depends on context as well as physical origin.

Aristotle further distinguishes voluntary action from deliberative choice. Children and animals can engage in voluntary behavior, whereas prohairesis requires a more articulated practical capacity. Choice concerns what is within our power and follows deliberation about means. His discussion of established character complicates the idea of alternative possibility: one may initially have been able to avoid becoming unjust yet later find it difficult or impossible to undo one's character at will, much as a thrown stone cannot be recalled (III.5, 1114a3–1114b25). His account thus recognizes both present causal ownership and temporally distributed responsibility.

Aristotle does not supply the contemporary PAP formulation. He supplies a durable question: how should a theory of praise and blame distinguish an action's internal origin, deliberative selection, known circumstances, and the history by which a person's character was formed?

Necessity, Liberty, and the Early Modern Debate

Early modern disputes brought liberty into direct contact with a mechanistic picture of causation. A crucial transition was the attempt to distinguish necessity, understood as lawful causal connection, from constraint or external impediment. Hobbes maintains in Leviathan XXI that liberty concerns absence of external impediments to motion and action. On this conception, a person can act freely while the will itself has causal antecedents. The conclusion follows because freedom is predicated of action under the description of unimpeded execution, not of an undetermined origination of desire. It is a powerful compatibilist reconstruction, but it leaves open whether an agent whose motives are manufactured by another is free in the responsibility-relevant sense (Hobbes, 1651, ch. 21).

Hume's Treatise of Human Nature II.3.1–2 develops a criticism of the alleged opposition between necessity and ordinary practical agency. He distinguishes the liberty of spontaneity, contrasted with violence, from the liberty of indifference, which would eliminate causal necessity. He argues that regular relations between character, motive, and conduct make human action intelligible. His treatment of moral evaluation presupposes that we can connect behavior to enduring qualities of agents. The explanatory source is the person understood through the person's character and motives, rather than an uncaused interruption of natural order (Hume, 1739/1740, II.3.2).

Hume's argument is often classified as classical compatibilism, but its sourcehood dimension deserves attention. If causation from character is a condition of attribution, randomness threatens responsibility as much as coercion does. Yet if someone else deliberately shapes character, mere character-causation is insufficient to settle who owns the resulting action. The modern manipulation argument sharpens precisely this problem.

These early accounts show that the source question can be posed before any modern Frankfurt example. Their standard of sourcehood is usually local and practical: whether action issues from an agent's motives or choices without relevant interference. The contemporary debate asks whether local ownership is enough.

The Conditional Analysis of Ability and Its Limits

G. E. Moore's Ethics, chapter VI, isolates the ambiguity of 'could have done otherwise.' On a conditional analysis, saying that someone could have acted otherwise may mean that the person would have acted otherwise if the person had chosen otherwise (Moore, 1912, ch. VI). In ordinary practical reasoning, this captures real competencies: someone able to turn left can do so under an appropriate choice, even if the choice to turn right was in fact determined. The analysis allows an important form of ability under deterministic causation.

Its insufficiency appears when choice itself is the disputed object. If the question is whether an agent could have chosen otherwise while the same past and laws held, the conditional shifts rather than answers the question. A person who is unable to choose differently can still satisfy the claim that an impossible alternative choice would have produced a different action. The conditional may also misclassify psychological incapacity when a hypothetical different desire is inconsistent with the agent's condition.

Contemporary dispositional analyses develop more sophisticated accounts of ability, including contextual and modal conditions. They explain how an agent may possess abilities that are unexercised while avoiding the claim that every hypothetical antecedent represents an accessible actual-world path. The central lesson is methodological: specify a background and a kind of power before inferring responsibility from an agent's inability or ability to do otherwise (McKenna & Coates, 2024).

Alternative possibilities cannot be evaluated from a naked counterfactual alone. A theory must distinguish physical opportunity, conditional ability, deliberative accessibility, and the all-in power to select a different future.

A Conditional Ability Is Not an Ability to Choose the Condition

The dispute over 'could have done otherwise' becomes sharper when the choice itself is incorporated into what is to be explained. Suppose an agent is willing and physically able to raise a hand if the agent decides to raise it. The conditional is true even if the agent's actual motivational structure ensures that such a decision never occurs. The consequent action is within a familiar motor capacity, while the antecedent decision may remain inaccessible under the stipulated total past and laws. Conditional freedom can establish that the body was unrestrained, but an incompatibilist can still demand an explanation of freedom in deciding to exercise the capacity.

The reverse inference is equally problematic. An agent who did not perform a morally required act might have been able to decide to perform it in some nearby circumstances while being unaware of the duty in the actual circumstances. Physical and conditional capacities do not by themselves entail epistemic opportunity or justified blame. The philosophical task requires a structured ability predicate: the bearer of the capacity, the act description, the relevant time, the informational situation, and the background conditions must all be specified. This specification prevents the debate from turning on ordinary ambiguity.

The Scope of a Free-Will Inference

The distinction between necessity and sufficiency is indispensable. If access to alternatives is sufficient for responsible agency, the mere availability of an alternative would establish responsibility, a conclusion virtually no major theory accepts. If it is necessary, a case of genuine responsibility without alternatives refutes it. A theory may instead regard alternatives as evidential: normally they indicate uncoerced agency, but their absence can occur for causes irrelevant to the actual decision. These three logical roles of alternatives yield very different predictions about Frankfurt cases and about ordinary excuses.

A further possibility is that leeway is required for certain types of responsibility but not for others. A present obligation to rescue someone may presuppose the ability to rescue, while gratitude for an expression of settled benevolence need not presuppose an alternative of cruelty. The duty at issue, the action's description, and the practice of holding someone responsible help determine which ability is relevant. Responsibility judgments cannot be transferred unchanged from wrongdoing to omission, from blame to praise, or from moral assessment to the ability to satisfy a requirement.

The Consequence Argument and the Fixity of the Past

The strongest modern challenge to leeway under determinism is the Consequence Argument. Let H denote a complete state of the remote past, L the laws of nature, and A a future action. Determinism entails that H and L jointly suffice for A. No present agent has power over H, and ordinary agents lack power to change L. The proposed inference is that the agent lacks power over A. Peter van Inwagen develops versions of this reasoning in An Essay on Free Will (1983), following earlier contributions by Carl Ginet and others (van Inwagen, 1983; Vihvelin, 2022).

The argument derives its force from a transfer principle: if something is outside our control and what it entails is outside our control in the relevant sense, the consequence is also outside our control. Van Inwagen's formal presentation uses an operator for a true proposition about which no one has a choice. A critical issue is whether that operator obeys the proposed inference rules without equivocation. A defender of alternatives can contest the transfer principle or the interpretation of fixing the laws, rather than denying the deterministic entailment.

David Lewis (1981) develops a famous response by distinguishing a person's ability to do something such that a law would have been broken from an ability to perform a law-breaking miracle. The first need not imply the second. In a counterfactual situation where the agent does otherwise, something in the history or laws may differ. The point is not that an ordinary agent violates physics. It is that ability claims can be true without holding the full actual causal history fixed in the way assumed by the Consequence Argument.

A successful Consequence Argument would undermine a categorical sense of leeway. It would not automatically show that ordinary action lacks all forms of reasons-guided control. A failed Consequence Argument would not automatically establish ultimate origination. Its metaphysical target and its ethical significance must be established separately.

Frankfurt's Challenge to the Principle of Alternative Possibilities

Harry Frankfurt's 1969 article, Alternate Possibilities and Moral Responsibility, challenges the necessity of leeway for responsibility. The basic structure of a Frankfurt case involves a monitoring intervener prepared to ensure that an agent performs a certain action if the agent otherwise begins to diverge. In the actual sequence, the agent decides and acts entirely on the agent's own reasons. The intervener never acts. Yet because intervention would have blocked deviation, the agent appears unable to avoid the outcome. Frankfurt concludes that the inability to do otherwise is not by itself sufficient to excuse an agent (Frankfurt, 1969, pp. 829–839).

The point of the example depends on separating the actual explanation of behavior from an idle counterfactual backup. An available intervention can eliminate an alternative without contributing to what actually happened. If the agent's own motives explain the conduct, the source relation seems intact. A threat example in Frankfurt's paper prepares this distinction: whether a threat supplies the motive for action is different from whether it would have compelled the same act had another motive failed.

The inferential structure is precise. PAP asserts that all responsible actions require alternatives. A coherent case with responsibility and no alternatives is a counterexample. The fact that a backup controller existed explains the absence of alternatives; the responsible agent's actual reasons explain the action. For the counterexample to succeed, however, the backup must genuinely eliminate all relevant alternatives while remaining absent from the actual causal process.

Frankfurt's thought experiment is a challenge to a necessary condition. It is not an independent proof that determinism is true or that compatibilism follows. Frankfurt himself subsequently distinguished free willing from responsibility in ways that resist treating the entire article as a final theory of free will (Frankfurt, 1971).

Actual Sequence, Counterfactual Sequence, and Causal Relevance

Frankfurt cases draw philosophical attention to the actual sequence: the motives, choices, deliberations, and mechanisms that really produce an act. The counterfactual sequence describes what would have happened had the agent attempted to diverge. In a standard example the intervener affects the counterfactual sequence while leaving the actual sequence undisturbed. The initial intuition is that moral evaluation tracks actual action and its originating psychology more closely than unrealized avenues of escape.

This observation has explanatory force but also a limit. Merely stating that only the actual sequence matters risks concealing the importance of dispositions. The competence of a driver cannot always be determined by a single successful drive. We ask how the driver responds to obstacles and reasons under relevant variations. A counterfactual pattern can reveal a mechanism's actual type and quality even when the particular variation never occurred. Thus an actual-sequence account need not prohibit counterfactual evidence about the agent; it must distinguish unrealized events that cause nothing from modal facts that help classify the operative mechanism.

Carolina Sartorio (2016) develops a causal actual-sequence view in which the causes of behavior ground freedom. Her account highlights the dependence of freedom assessments on the structure of causation, including absences and sensitivity to reasons. The distinction between what causes the action and what merely would have prevented another action can be more philosophically sophisticated than the distinction between one actual event and every hypothetical event.

The enduring question is how much modal information can inform sourcehood without making freedom depend on leeway of the categorical kind that Frankfurt disputes. This is an internal problem for any source-centered theory, rather than a decisive objection to its central insight.

What Must Remain Fixed for a Counterexample?

A Frankfurt example can be tested through four distinct claims. First, the agent decides from the same actual reasons that would support an ordinary attribution of responsibility. Second, a backup intervener would have ensured the specified outcome had a contrary decision threatened. Third, no possible divergence relevant to responsibility remains. Fourth, the presence of the backup does not itself corrupt the agent's practical capacities. The first two claims create the initial counterexample; the third and fourth are precisely where the philosophical controversy concentrates. A successful case must defend all four.

One difficult variant uses a prior sign of an impending contrary choice. If the sign determines the choice, the case may import causal assumptions that the libertarian disputes. If the sign is fallible, a contrary choice might occur without the sign, leaving the intervener unable to guarantee the outcome. A proposed correction replaces prediction with intervention at an earlier point in deliberation, but then critics can argue that the intervener's presence changes the actual agent's powers. The design of the example is therefore conceptually loaded. Its reliability depends on the structure of deliberation attributed to the agent, not merely on imaginative ingenuity.

The fact that a case is described as impossible under an incompatibilist theory does not, by itself, refute every conceptual role the case could play. Thought experiments can expose tensions within a principle by making relevant relations salient. But a principle about freedom under indeterminism cannot be defeated simply by assuming away the theory's positive conditions. Dialectical success requires shared premises or an independent argument for the disputed assumptions.

The Dilemma Defense and Flickers of Freedom

The most prominent incompatibilist reply to Frankfurt-style cases is the dilemma defense associated with David Widerker (1995) and related work by Carl Ginet and Robert Kane. The intervener must know when to act. If there is a prior sign that infallibly determines the agent's eventual choice, then the example may presuppose a deterministic relation relevant to the very question under dispute. If the sign is not infallible, the agent can perhaps choose otherwise without triggering intervention, leaving a relevant alternative open.

The criticism is often presented as a dilemma between question-begging determinism and incomplete control of alternatives. Its exact force depends on how intervention operates and what the example is meant to show. A Frankfurt case aimed only at disproving PAP may not have to establish libertarian sourcehood. But one intended as dialectically neutral evidence against an incompatibilist understanding of alternatives must avoid assuming the impossibility of undetermined divergence.

Defenders have constructed increasingly intricate examples to remove the difficulty. Critics have replied by identifying 'flickers of freedom': small remaining alternatives, such as the capacity to begin choosing otherwise, to reveal a contrary sign, or to omit the exact act under description. The moral significance of these alternatives is disputed. A residual twitch or sign may be a genuine possibility without providing an ethically meaningful route to different conduct.

Two questions must therefore be separated. Is there any alternative event whatsoever? And is there an alternative sufficient to ground responsibility? PAP is ambiguous unless its relevant alternative is specified. The debate does not end with the detection of a flicker: defenders of Frankfurt cases can concede an insignificant flicker while arguing that it cannot explain why the agent deserves blame. Critics can respond that the flicker marks an undetermined exercise of agency within the actual sequence. Each side owes an account of explanatory relevance.

Robust Alternatives and the Explanatory Demand

A robust alternative is one whose availability contributes intelligibly to why an agent is morally responsible. On some leeway theories, an agent must have been able to avoid wrongdoing with an adequate understanding of the moral significance of that avoidance. On other views it is enough that the agent retained a meaningful possibility of not initiating the culpable act. The robustness requirement responds to the prospect that Frankfurt cases leave tiny but irrelevant options.

The proposal has a cost. If the alternative must already be an option the agent recognizes as morally significant and could reasonably take, responsibility may depend on capacities that are best explained by actual sourcehood. The leeway requirement can become derivative of reasons-responsiveness. Conversely, if any possibility counts, it becomes difficult to explain why morally insignificant alternatives secure responsibility.

An explanatory test is useful. Hold fixed the quality of the agent's actual motives, understanding, and decision-making. Then ask whether removing the residual alternative changes any respect in which the action expresses the agent's practical standpoint. If not, the defender of PAP must explain why responsibility disappears. If the removal changes the kind of self-governance exercised, the defender of Frankfurt can no longer claim the case is causally and normatively unchanged.

The result is a continuing philosophical dispute. Frankfurt examples are strong evidence against an unrestricted slogan requiring the ability to do otherwise; they do not settle every carefully specified leeway principle (O'Connor & Franklin, 2022).

Guidance Control and Regulative Control

John Martin Fischer distinguishes guidance control from regulative control. Regulative control involves access to alternative sequences of action. Guidance control concerns the actual sequence by which action issues from an appropriately functioning mechanism. Fischer's proposal allows moral responsibility when regulative control is unavailable, provided that the agent guides the action through a suitable mechanism (Fischer, 1994; Fischer & Ravizza, 1998).

Fischer and Ravizza's account specifies two central conditions. The actual mechanism must be the agent's own, and it must display moderate reasons-responsiveness. Reasons-receptivity concerns the mechanism's capacity to recognize a comprehensible pattern of reasons; reasons-reactivity concerns its capacity to respond to at least some recognized reasons by producing different behavior in appropriate circumstances. The mechanism need not react differently in all possible cases. A moderate rather than maximal standard is intended to accommodate human agency without treating every failure of sensitivity as the absence of control.

This account has a counterfactual component. It asks how the same mechanism would operate in possible scenarios containing different reasons. That feature does not reintroduce a categorical ability to act otherwise while all antecedent circumstances remain fixed. It introduces modal evaluation of a capacity. The distinction is philosophically consequential: a deterministic agent can possess a reliable capacity to respond to reasons across different conditions even if one complete condition uniquely determines an actual response.

The theory explains why a person acting with ordinary deliberation may differ from a person whose action is produced through an irresistible compulsion. Yet the account must explain how a mechanism becomes the agent's own and why a carefully manipulated mechanism might still fail to count as owned. Historical sourcehood enters exactly here.

Ownership, Taking Responsibility, and the History of Agency

A purely synchronic account of an agent's psychology is vulnerable to manipulation examples. Two agents may display identical current preferences, beliefs, reasoning abilities, and responsiveness, even though one acquired the relevant configuration through sustained critical engagement and the other through covert intervention. Fischer and Ravizza address this by requiring that the agent have taken responsibility for the mechanism through an appropriate history of regarding the self as an agent and as an apt target of responsibility, on evidentially suitable grounds (Fischer & Ravizza, 1998, ch. 8).

This condition reflects an insight already suggested by Aristotle's attention to habituation: the temporal production of character can matter to its ownership. A decision is part of a life, and the life may contain earlier formative actions whose significance reaches into the present. Taking responsibility is itself a historical process, not simply an accompanying feeling of control.

The condition is also exposed to a recursion problem. If someone manipulates the person's process of taking responsibility, the formal requirement might be met while the suspected defect persists. If every stage must be independently owned, an infinite regress threatens. A plausible historical theory must therefore specify which developmental influences undermine agency and which are ordinary enabling conditions. Education, socialization, inherited temperament, and causal dependence cannot all be counted as disqualifying without making finite human agency impossible.

The distinction between formation and bypass is promising. An influence can shape an agent while engaging the agent's evaluative capacities; another influence can install motivational states through routes that bypass or systematically disable those capacities. Yet the distinction becomes difficult when manipulations operate gradually and sympathetically through reasons. The history of agency remains a substantive philosophical task.

Hierarchical Identification and the Authority of the Will

Harry Frankfurt's Freedom of the Will and the Concept of a Person (1971) provides a different route to sourcehood. The central contrast is between first-order desires and second-order volitions concerning which desires should be effective in action. A person is capable of reflecting on motivational impulses and taking a standpoint toward them. When the will that effectively moves the person aligns with a higher-order volition, the agent exhibits a form of identification relevant to freedom.

The model captures differences between acting from an impulse one repudiates and acting from a motive one reflectively endorses. It also explains why simply having several desires does not yet constitute self-government. An agent can experience numerous alternatives internally while lacking the ability to direct the formation of an effective will. Freedom concerns the organization and authority of motivation.

A difficulty arises when reflective endorsement itself has a history. If an external operator installs both an impulse and the second-order approval of that impulse, hierarchical agreement may exist without trustworthy ownership. Moreover, a theory that locates final authority at a higher level can face the question why that level, rather than another, speaks for the agent. Infinite ascent through orders of desire offers no obvious solution.

Gary Watson's account of free agency (1975) develops a related distinction between evaluative and motivational systems. The agent's considered values can conflict with the strongest desire. These accounts bring normative self-assessment into the sourcehood debate, yet neither identity with a desire nor a formal level in a hierarchy automatically guarantees moral responsiveness. A source theory must explain what makes the identification practically authoritative, epistemically informed, and historically appropriate.

Two Forms of Incompatibilist Objection

An alternative-possibilities incompatibilist argues from the unavailability of divergent futures to the absence of freedom. A source incompatibilist argues from the complete determination of an agent's actual deliberative process by factors outside the agent's control. These objections need not stand or fall together. The agent in a Frankfurt case may lack a way to avoid wrongdoing while still deciding on reasons of the agent's own; whether that suffices for responsibility depends on the source requirement, rather than on the number of alternatives. Conversely, an indeterministic decision might preserve options but arise through a random process disconnected from guidance. Here alternatives exist while sourcehood fails.

This separation puts pressure on the idea that determinism is incompatible with freedom for a single simple reason. The incompatibilist must specify whether the impossibility of divergence, the alien character of antecedent causes, or the failure of ultimate self-creation carries the decisive argument. Each proposition has its own objections. A compatibilist may concede that the remote past was beyond the agent's control while denying that all responsibility-bearing relations must be controlled at their own origin. The disagreement then centers on whether ownership of a process can be constituted in the course of exercising it.

Causal Dependence and Ownership

Being causally dependent on prior events is a universal feature of finite development. A child learns language through others; an adult's judgment depends on training, testimony, memory, and inherited capacities. A strong sourcehood theory cannot merely identify dependence, because that would make all learning an argument against agency. It must explain why a particular dependence is alienating. Plausible candidates include deception that prevents critical assessment, direct bypass of deliberation, suppression of reasons-recognition, and interventions specifically designed to make an alternative judgment psychologically unreachable.

Yet these criteria themselves require normative grounding. An agent formed by propaganda may continue to deliberate fluently according to reasons the agent takes seriously. Is the defect a failure of authenticity, of moral competence, of informed consent, or of the possibility of historical self-correction? Different theories answer differently, showing why sourcehood is simultaneously a causal, epistemic, and normative problem. A complete account of origination must be explicit about the status of each component.

Source Incompatibilism and Ultimate Origination

Source incompatibilism contends that determinism threatens freedom even if significant alternatives are not required. The decisive issue is the origin of action. If a decision is fully fixed by antecedent conditions that the agent did not control, the agent may be the local cause but lacks ultimate authorship. The structure differs from the Consequence Argument. It does not require showing that the agent has no alternative under fixed past and laws; it requires showing that determined agency lacks a responsibility-grounding source relation (McKenna & Coates, 2024).

Its attraction is that Frankfurt cases can be accepted without accepting compatibilism. The actual agent may act from deliberate intentions, with the backup intervener inactive, yet those intentions may themselves be completely determined by a causal history outside the agent's control. A critic can acknowledge that alternative possibilities are dispensable while insisting that the actual causal pathway fails the stronger test of origination.

The source incompatibilist owes an account of how origination is possible for a finite agent. No person creates the past, inherited dispositions, social world, or capacities through which choice occurs. If ultimate responsibility means responsibility for every condition of responsibility, regress seems unavoidable. If it means responsibility for some decisive act of self-determination, one must explain what gives that act its special authority.

The resulting dispute is not merely semantic. It concerns whether finite authorship can be constituted by the exercise of capacities whose preconditions were received or whether genuine basic-desert responsibility demands an originative relation irreducible to the ordered sequence of received capacities.

Agent Causation, Event Causation, and Indeterminacy

Libertarian accounts agree that determinism and the required freedom are incompatible, but they disagree about positive explanation. Event-causal libertarianism holds that deliberation and reasons can cause a decision without causally necessitating a unique outcome. Agent-causal views hold that the agent, conceived as a substance or enduring individual, causes an action in a way not reducible to event causation. Noncausal views resist making prior causal determination the condition of intentional action (Clarke et al., 2021).

Agent causation appears to secure a direct source: the person, rather than a chain of states alone, originates a decision. Yet the proposal faces questions about the intelligibility of substance causation, its relation to reasons, and the conditions distinguishing a genuinely agent-caused act from arbitrary production. If the agent could settle either decision under exactly the same reasons, we must ask how the agent's character and judgment explain the one actually taken. The power of selection cannot serve as a substitute for explanation.

Event-causal libertarianism uses familiar causal relations but introduces indeterminacy at a significant stage of practical deliberation. Robert Kane's account of self-forming actions treats conflicts of will as sites in which indeterministic efforts can contribute to character formation (Kane, 1996). The possibility that the will develops through undetermined choices gives a temporal structure to self-origination. The objection is that where an outcome remains unsettled by prior reasons and effort, the final result may depend partly on luck.

Indeterminism therefore removes one alleged obstacle to leeway without automatically constituting sourcehood. An adequate libertarian view must explain how openness is exercised by the agent rather than merely occurring in a process inside the agent.

Luck, Control, and Self-Forming Action

The luck objection asks why a person deserves credit or blame for a decision if exactly the same prior circumstances and reasons could lead to a different decision through an undetermined outcome. If neither result is settled by the agent's character, reasons, or deliberation, the difference threatens to become an accident from the standpoint of responsibility. The problem is especially vivid when two possible worlds agree up to a choice yet diverge at the final decision.

Kane's reply stresses that self-forming decisions can involve genuine effort and competing reasons, so whichever outcome occurs is a product of the agent's struggle rather than an alien random event. Indeterminacy does not erase participation. A defender can distinguish chance located within a controlled practical effort from chance that entirely supplants agency. The further challenge is whether effort explains why this specific outcome rather than the other occurred.

Alfred Mele (2006) argues that luck is a challenge both for libertarians and for certain compatibilist histories. A person may be exceptionally fortunate or unfortunate in the formation of dispositions, circumstances of deliberation, and eventual opportunities. Sourcehood theories must specify which kinds of luck are compatible with responsibility and why. Merely observing that all human lives are contingent does not dissolve the demand for a relevant control relation.

A coherent solution must avoid two extremes: treating every uncontrolled antecedent as fatal, and treating any internal event as enough. The philosophical work is to identify the dependence relations for which control must be exercisable, whether by a local mechanism, a diachronic history of identification, or an undetermined act of origination.

Testing a Manipulation Argument

A manipulation argument can be reconstructed as an inference from an intuitive judgment about an engineered case to a general conclusion about deterministic agency. The engineered case is meant to establish that a person can act through reasons-responsive deliberation yet lack responsibility because the relevant motives were installed. The argument then proposes that an ordinary determined agent differs only in the natural rather than deliberate origin of those motives. If deliberate installation is not itself the decisive defect, a defender of the argument infers that the causal determination shared by both cases is responsibility-undermining.

Several objections are distinct. The first denies the intuition about the manipulated case: perhaps an agent who subsequently acquires genuine rational competence can become responsible. The second accepts the intuition but finds a principled difference between manipulation and ordinary development, such as the systematic bypass of the agent's developing capacities. The third challenges the transfer principle: the accumulation of small historical differences may produce a genuine difference in ownership, even where no single transition seems decisive. The argument's credibility depends on responding to all three.

The thought experiment also reveals a tension between local and historical explanations. One may correctly say that a manipulated person acted because of their own present intention, while also saying that another agent arranged the conditions under which that intention arose. Both causal descriptions can be true. The remaining question is not whose causal contribution existed, but which contribution bears which kind of responsibility. The manipulator's responsibility for engineering an action does not automatically establish the absence of all agency in the person who performs it. It does create a substantive demand for a theory of compromised ownership.

Manipulation Cases and the Challenge of Continuity

Derk Pereboom's four-case argument tests whether compatibilist sourcehood survives a series of causal histories. In the most explicit cases, neuroscientists or manipulators produce an agent's decision. Later cases progressively approximate normal deterministic causal development. The challenge is that if the first case excludes responsibility and there is no responsibility-relevant difference along the series, then ordinary determined action also lacks the required control (Pereboom, 2001, ch. 4; 2014, ch. 4).

The argument relies on a transfer of responsibility assessment across cases alleged to share the same morally relevant structure. Its force therefore depends on two premises: the manipulated agent really lacks basic-desert responsibility, and the transition to an ordinary causal history introduces no morally significant difference. A compatibilist can resist either. The difficult task is to articulate a principled difference rather than merely insist that ordinary upbringing feels more acceptable.

Mele's zygote argument presents a related challenge. An agent's initial conditions are deliberately arranged so that, in a deterministic world, the eventual act is assured. If the person otherwise lives through ordinary development, can the remote manipulation render the later action unfree? Mele uses the case as a formidable test even while considering ways compatibilism might survive (Mele, 2006, ch. 7).

The central insight is that causal origin and causal ownership may diverge. A person can be the proximate intentional source of an action whose developmental conditions were deliberately designed. An adequate theory must specify why and when that divergence defeats responsibility. An unargued equation of 'caused by another' and 'not mine' will not work, because every human life involves formative dependence.

The Basic Argument and the Regress of Ultimate Responsibility

Galen Strawson's Basic Argument gives the sourcehood problem a regress form. To be ultimately morally responsible for what one does, one would have to be ultimately responsible for the way one is in the relevant psychological respects. Yet intentional attempts to form oneself already proceed from existing motives, judgments, and capacities. To be ultimately responsible for those, the same demand reappears. The regress suggests that the kind of absolute self-authorship presupposed by ultimate desert is impossible for finite agents (G. Strawson, 1994).

This is a different criticism from saying that determinism leaves no alternatives. Even an indeterministic sequence may fail to meet the demand for responsibility for the relevant self. The premise that one must have made oneself to deserve blame is nevertheless disputable. A compatibilist may argue that responsibility is grounded in competent agency and responsiveness, not in creating the conditions of agency. The skeptic may reply that such competence supports practical regulation, appraisal, or protection, but not basic desert understood as deserved blame just because the agent acted wrongly.

The debate becomes clearer once its conclusion is specified. No ultimate self-creation does not entail no intentional action, no reasons, no practical accountability, or no legitimate social response. It threatens an especially demanding account of deserved praise and blame. The scope of the skeptical result should neither be exaggerated nor reduced without argument.

The regress also applies pressure to libertarian sourcehood: undetermined choices cannot explain their own conditions of intelligibility. A theory must show why agency does not require complete authorship of every condition while still explaining the significance of the agent's distinctive causal contribution.

Reactive Attitudes and the Social Dimension of Responsibility

P. F. Strawson's Freedom and Resentment (1962) redirects the debate toward the interpersonal practices in which responsibility is ordinarily expressed. Resentment, gratitude, forgiveness, indignation, and related reactive attitudes respond to the quality of will displayed by another. Exemptions and excuses alter these responses by changing our understanding of the person, the circumstances, or the intentions underlying conduct.

This approach does not dispense with sourcehood. It locates the relevant form of sourcehood in an action's expressive relation to the agent's practical attitudes and capacities within a community of reciprocal expectations. If a harmful event reveals no regard or disregard by an agent, blame may be inappropriate even when the agent's body is involved. If a deliberate slight expresses contempt, interpersonal resentment can be intelligible even without identifying a metaphysically open alternative in the exact past circumstances (P. F. Strawson, 1962).

Michael McKenna's conversational account extends this insight by treating responsibility as an exchange in which action expresses agent meaning and holding responsible is a response to that meaning (McKenna, 2012). A conversational practice requires capacities to understand and answer reasons. This shifts attention from an isolated decision event to patterns of mutual normative address.

The relational approach faces a critical question: do actual practices establish deserved responses, or can they themselves be mistaken? Social blame can reflect prejudice, ignorance, or institutions of power. A full account must distinguish the descriptive existence of reactive attitudes from their warranted employment. Practical answerability and basic desert remain related but separable targets.

Asymmetry, Ability, and the Ethics of Blame

A person's inability to avoid harm can matter differently depending on how the action was produced. If someone is physically forced to injure another, the coercive source defeats ordinary attribution. If someone intentionally chooses the injury while an undetected intervener stands ready to compel it, the actual source relation looks different. Frankfurt cases dramatize this asymmetry. The ethical significance of a missing alternative cannot be evaluated without asking why the alternative is missing and whether its absence affects the actual reasoning.

A further asymmetry concerns omissions and positive actions. An omission can be blameworthy only if some relevant performance was available, or if prior conduct culpably created the incapacity. A person cannot be expected to rescue someone by an action the person is wholly unable to perform. Yet one may be answerable for earlier choices that predictably produced incapacity. The time at which alternatives are assessed thus matters.

Praise also requires analysis independent of blame. An agent may be praiseworthy for helping another from genuine concern even when helping was the only action compatible with that character. A theory that requires a morally bad alternative to make a good action admirable risks an implausible account of virtue. Conversely, luck in available opportunities may explain differences in public praise among persons of equal practical character.

These examples suggest that a blanket PAP obscures distinctions among duties, opportunities, exemptions, and the moral quality of acts. The agent's actual relation to reasons can carry significant evaluative weight. The metaphysical question of ultimate desert remains open.

Foreknowledge, Causal Determination, and Modal Necessity

An agent's inability to avoid an outcome may arise from distinct structures: an external intervener, a deterministic causal history, divine foreknowledge, logical necessity, physical incapacity, or psychological compulsion. Treating each as equivalent because all are expressed through 'could not do otherwise' confuses epistemic, causal, and modal relations. Knowing a future event with certainty does not by itself cause that event. Logical entailment from a complete description does not automatically specify the kind of control the agent exercises.

Frankfurt's monitoring device functions as a counterfactual constraint: it would prevent divergence if divergence threatened. Determinism functions as an actual nomological thesis: the same complete prior state and laws suffice for the actual outcome. Those structures can both eliminate a strong alternative while differing in causal location. A source theorist must explain whether and why the difference affects responsible agency.

Modal vocabulary is therefore indispensable. 'Possible' might mean logically consistent, metaphysically possible, physically permitted by some laws, nomologically possible under the actual laws, or accessible given the agent's actual powers. Inference between these uses is hazardous. The best formulations of the dispute state which possibilities are relevant and what is held fixed.

Causal Explanation and the Problem of Deviance

A source-based account must identify the right causal pathway. An action can result from a person's desires and beliefs in a deviant way. Imagine an agent's intention to perform an act generating anxiety, which accidentally produces the intended physical result through an unrelated tremor. The action is causally connected to the intention, yet the connection may fail to constitute intentional guidance. Sourcehood requires more than an ancestor in a causal graph.

Reasons-responsive accounts attempt to secure the right mechanism by demanding an intelligible pattern of uptake and reaction. Causal accounts distinguish relations that constitute control from relations that merely precede behavior. Both must explain how intentions, reasons, skills, and bodily execution compose a coherent action. They also must recognize that omissions and absences complicate simple event-causal descriptions.

Sartorio's emphasis on actual causes invites renewed attention to causation as a responsibility-grounding relation. The contribution is significant because the absence of an alternative does not settle whether the agent's reasons make a causal difference. The weakness of any simple causal criterion is that a manipulated or deviant process may satisfy causal dependence without preserving agency. Mechanism individuation and historical ownership remain necessary points of inquiry (Sartorio, 2016; Fischer & Ravizza, 1998).

Empirical Research and the Limits of Neuroscientific Evidence

Experiments on voluntary movement can illuminate the temporal organization of action, but their interpretation requires care. Libet and colleagues (1983) found that a readiness potential preceding simple self-initiated movements appeared before the reported time of awareness of an urge to move. The result raised questions about the initiation of action and the role of conscious intention. It does not, by itself, establish that every morally significant decision is fixed unconsciously or that no agent possesses reasons-responsive capacities.

Schurger, Sitt, and Dehaene (2012) proposed an accumulator account in which spontaneous fluctuations contribute to the observed readiness potential before self-paced movement. Their model complicates the inference from an averaged pre-movement signal to a completed decision at an earlier instant. Empirical results concerning timing and prediction do not directly answer the metaphysical question whether the full past and laws entail an action.

The sourcehood problem is nevertheless empirically informed. Impaired executive control, coercive environments, habitual responses, addiction, and disturbances of deliberation can reveal differences among capacities that an adequate philosophical theory should respect. The conceptual conclusion is limited but important: empirical evidence can test particular mechanisms of guidance, agency, and self-monitoring, while philosophical argument is required to link those mechanisms to moral desert or ultimate origination.

Counterfactual Dependence Without Categorical Leeway

One of the most subtle lessons of this literature is that counterfactual explanation survives the rejection of all-in leeway. A mechanism can be sensitive to reasons in hypothetical circumstances even if it could not have produced a different outcome with exactly the same complete state and laws. A person can truthfully be described as capable of learning, responding, and correcting behavior across contexts. Those modal descriptions help identify what kind of mechanism the person actually has.

The difference can be stated in terms of two questions. First: did the agent have another future open from the identical total circumstances? Second: is the actual producing mechanism of a type that would recognize and respond to sufficiently different reasons? The second question cannot be answered solely by specifying the first. The modal bases differ. It follows that rejecting categorical PAP does not commit a theorist to ignoring capacities or counterfactual structure.

A remaining issue is which counterfactuals are sufficiently close and which mechanisms count as the same. If any imagined alteration is permitted, even rigid compulsions may appear responsive when we imagine their removal. If the background is fixed too strictly, even ordinary competent mechanisms appear inert. Mechanism individuation must therefore reflect the practical organization of the agent's capacity rather than arbitrary descriptions selected to deliver a favored conclusion.

The Normative Ambiguity of 'Could Have Done Otherwise'

Ordinary moral language uses 'could' to perform several tasks. It may express that an action was physically available, that the agent had sufficient knowledge, that compliance with a duty was reasonable, that choosing differently was psychologically accessible, or that a fully specified deterministic history permitted a divergent future. These functions overlap in everyday cases and separate under philosophical pressure.

A person might be physically able to rescue a stranger but lack the relevant information; the statement that the person 'could have helped' then requires an epistemic qualification. A person might possess knowledge and skills yet act through an acute disorder that disrupts reasons-responsiveness; physical opportunity alone tells us little about culpability. Conversely, a person may be unable to change an entrenched disposition immediately but bear some responsibility for earlier conduct that reinforced it.

Moral assessment therefore needs indexical precision. Whose ability is at issue, at what time, under what information, and with respect to which act description? These questions are not secondary technicalities. They determine whether a purported counterexample addresses the claim under discussion.

Sourcehood and the Artificial Era: A Conditional Research Horizon

The emergence of increasingly autonomous nonbiological systems introduces new examples for the metaphysics of agency. An artificial system may select actions among computed options, preserve long-term objectives, revise policies in response to reasons, and produce a traceable causal record. Such facts establish questions about control architecture and causal provenance. They do not, in isolation, establish moral responsibility or the kind of ultimate origination disputed in the classical literature.

Aisentica's wider philosophical horizon, Artificial Era · From Homo to Artificial, gives conceptual importance to the emergence of a nonbiological order alongside Homo. For this dossier the relevant implication is a research question rather than a newly instituted canonical definition: which conditions of sourcehood could, in principle, be realized by an artificial agent, and which conditions presuppose a form of normatively answerable agency that has not been established? The logical structure of the problem is portable, but empirical and moral conclusions require their own grounds.

Mechanism ownership may have special difficulty in systems designed by identifiable developers. External design is not automatically equivalent to ongoing manipulation, just as ordinary human formation is not equivalent to being controlled at the moment of action. Nonetheless, attributable software outputs and the social responsibility of operators must not be collapsed into a claim that the system itself possesses basic-desert responsibility. The conceptual difference between provenance, operational control, and moral standing remains essential.

The Dependence Structure of the Debate

The principal historical arguments can be situated in a dependence structure. A categorical alternative-possibilities condition concerns whether an action could differ under fixed background conditions. An actual-sequence condition concerns whether the action issues from a competent and integrated practical mechanism. A history condition concerns whether the mechanism's formation makes it appropriately owned. An ultimacy condition concerns whether any such ownership is sufficient when the formation and operation of the mechanism depend ultimately on uncontrolled antecedents. A normative-standing condition concerns whether the agent can appropriately be addressed through blame, praise, demands, or repair.

The relations among these conditions are asymmetric. A person can possess a relevant alternative while lacking the knowledge needed to choose it responsibly. A person can exercise a competent mechanism while lacking a categorical alternative. A person can own a mechanism through a normal history while lacking ultimate self-creation. A social practice can intelligibly demand explanation while remaining unjustified in imposing suffering as basic desert. Each independence claim is supported by cases examined in the preceding chapters. They constrain what any purported resolution must prove.

An account may combine actual-sequence guidance with a history requirement without embracing ultimate sourcehood. Another may insist on the stronger origination condition and treat familiar guidance as merely practical agency. Still another may reject basic desert while retaining a norm of reasoned answerability. The framework identifies where substantive disagreements lie, which premises are shared, and which conclusions are genuinely in dispute. Its contribution is conceptual discipline rather than a premature universal solution.

A Synthesis: Source, Leeway, History, and Standing

The investigation supports four distinct dimensions of evaluation. The first is modal: which courses of action are accessible relative to an explicitly defined background? The second is actual-sequence: what reasons, mechanisms, and causal relations produced the relevant conduct? The third is historical: how did the agent acquire the capacities, values, and mechanisms by which the action was produced? The fourth is normative: what response, if any, is justified toward the agent, and under which theory of responsibility?

These dimensions are jointly relevant but none can be substituted for the others. The mere presence of alternatives does not prove guidance; a well-functioning actual mechanism does not settle ultimate historical ownership; a clean developmental history does not establish a metaphysically undetermined future; and a practical need to hold someone accountable does not itself settle the entitlement to retributive blame. The demands of proof vary with the responsibility concept under investigation.

The thesis developed here is a diagnostic one. The sourcehood-versus-leeway dispute is most productive when treated as a set of relations among modal access, causal ownership, and normative attribution. Frankfurt's achievement is the demonstration of why actual explanation matters independently of foreclosed alternatives. The enduring achievement of incompatibilism is its insistence that actual explanatory success may still leave an origin problem unresolved. Reasons-responsive and historical accounts provide substantial ways to articulate local self-government. Agent-causal and event-causal libertarian accounts attempt to secure a stronger form of origination while confronting the luck problem. Skepticism challenges the coherence of ultimate desert without eliminating practical agency.

A systematic assessment begins by defining the kind of responsibility at stake, identifying the action and its relevant time, specifying the modal background, reconstructing the actual mechanism, investigating its history, and evaluating the moral significance of its source. This procedure does not decide all metaphysical questions in advance. It prevents a conclusion about one kind of control from being silently treated as a conclusion about every kind.

Contemporary Disputes over the Unit of Freedom

Recent research also asks whether the familiar expression 'free will' groups together properties that should be studied under more precise names. Ricardo Restrepo Echavarría (2025) develops freedom as self-government in terms of the practical exercise and development of agency, seeking to accommodate degrees of freedom and a broader horizon than liability alone. The approach resonates with guidance-control accounts while drawing attention to the dynamics of self-governance and the risk of defining agency solely through its fit with established social practices. Its challenge for sourcehood theory is to explain what makes reflective governance sufficiently the agent's own when institutional and developmental environments strongly shape the available evaluations.

Adam John Andreotta (2026) addresses whether the term 'free will' should be abandoned, a question about conceptual utility as well as metaphysics. A term can designate a genuine research problem while joining incompatible criteria of freedom, moral responsibility, and control under one label. This dossier's sourcehood-and-leeway analysis contributes to that methodological issue: the phrase 'could have done otherwise' is interpretable under several modal constraints, and the phrase 'acted from oneself' spans several notions of causal and normative ownership. The case for clarifying a vocabulary therefore arises from the structure of rival arguments, even before a final metaphysical position is selected.

Neither conceptual revision nor a self-government proposal resolves the manipulation and ultimacy arguments by definition. A useful contemporary framework must retain the hard cases as tests: cases where reasons-response is present but historical ownership is disputed, cases where leeway survives but agency does not, and cases where practical accountability persists without agreement about basic desert. These problems provide continuity between the classical debates and ongoing philosophical work.

Open Problems

Several disputes remain philosophically live. The first concerns whether there is a robust version of PAP that survives Frankfurt cases without merely restating conditions of actual-sequence competence. The second concerns the right individuation of a reasons-responsive mechanism: counterfactual tests require continuity between the actual and hypothetical mechanism, yet mechanisms can be described at many levels. The third is the historicity of ownership: an adequate account must explain why some influences bypass an agent's capacities and others constitute their normal development.

The fourth dispute concerns ultimacy. If responsibility demands an origin independent of uncontrolled antecedents, how can finite, temporally situated agency satisfy the demand? If the demand is rejected, what explains why answerability is sufficient for deserved blame rather than merely for beneficial regulation? The fifth concerns the relationship between causal explanation and moral evaluation: what actual causal structure makes an action appropriately expressive of an agent's practical standpoint?

A sixth concerns empirical scope. Laboratory tasks with arbitrary movements capture only some components of agency; the transfer from such tasks to deliberation, character, and moral responsibility must be justified. A seventh concerns artificial agency: which functional criteria can be empirically established, and what normative conditions would justify regarding a nonbiological system as a party to practices of accountability? These are research questions requiring new argument and evidence.

Conclusion

Sourcehood and alternative possibilities answer distinct questions about the freedom of action. Sourcehood concerns how an action issues from an agent and whether the resulting behavior is properly attributable to that agent. Alternative possibilities concern the availability of other courses of action relative to specified modal constraints. Historical philosophy already connected internal origin, choice, knowledge, and moral appraisal; contemporary debate made the logical relations among these notions explicit.

Frankfurt-style cases challenge the thesis that every responsible action requires a different accessible action. Their success depends on the possibility of excluding relevant alternatives without disrupting the agent's actual exercise of control. The Consequence Argument challenges categorical leeway under determinism, while source incompatibilism challenges the sufficiency of determined actual-sequence ownership even if leeway is abandoned. Guidance-control accounts, hierarchical identification, libertarian origination theories, manipulation arguments, and skeptical regress arguments articulate different answers to these challenges.

The resulting philosophical synthesis is precise. Free agency cannot be inferred from the bare number of available options. Responsibility cannot be inferred from the bare fact that an action has an internal cause. A serious account specifies the operative mechanism, its responsiveness to reasons, the historical conditions of its ownership, the modal standard against which opportunity is measured, and the kind of normative response to be justified. These requirements organize the contemporary dispute while leaving its deepest metaphysical disagreement intelligible: whether responsible agency requires ultimate origination or whether finite self-government provides the control that moral life actually needs.