Agent Causation

The Problem of an Agent as Cause

When someone deliberately opens a door, the event can be described in several ways. A hand contacts the handle; muscles contract; neuronal activity precedes the movement; a person admits a visitor; a host fulfills a promise. Each description identifies something relevant, yet the central philosophical issue concerns how these descriptions fit together. Is the person a cause in a fundamental sense, or does causal work belong wholly to states and events—beliefs, desires, neural firings, muscular contractions—while the person is merely the enduring subject to which these events are attributed? Agent causation begins with the demand that the causal role of the agent be made explicit.

There are two easily confused claims. The ordinary claim that agents cause things is widespread and uncontroversial in normal action explanation: the carpenter built the table and the driver stopped the car. The stronger metaphysical claim holds that the agent herself causes an occurrence in a way not reducible to earlier events involving her. The first is compatible with Davidsonian event-causal action theory; the second marks the distinctive terrain of irreducible agent causation (Davidson, 1963; Chisholm, 1964; Clarke, 2003). A theory that quotes the everyday claim while silently presupposing the stronger thesis has therefore skipped the argument.

The attraction of the stronger thesis is intelligible. If my choice is fully fixed by circumstances over which I have no originating control, its production may seem to bypass me precisely as a chooser. If it is an uncaused accident, it appears even less attributable to me. Agent-causal theories propose a third structure: the choice is caused through the agent's exercise of a power. Yet that proposal immediately poses a question more exacting than the initial dilemma. What is the exercise of a power, how does it enter causal history, and why does its introduction make the outcome mine rather than merely insert a mysterious additional cause? The history of the subject is the history of increasingly disciplined answers to this question (Reid, 1788/2010, Essay IV; Chisholm, 1964; O'Connor, 2000).

Definition, Scope, and the Distinct Claims at Issue

Agent causation, in its strong metaphysical sense, is causation in which an agent, considered as an enduring bearer and exerciser of causal powers, directly causes an event or the occurrence of an action, and this causal relation is not identical to or wholly constituted by a relation among events preceding that occurrence. The definition leaves open whether every action, only some basic actions, or only free decisions involve this causation. It also leaves open whether agents must be conscious, biological, rational, or capable of alternative possibilities; those are further substantive theses requiring argument.

An event-causal theory identifies causal transactions among occurrences, frequently including the acquisition or activation of intentions, desires, beliefs, and bodily processes. A substance-causal theory allows things or substances, rather than exclusively events, to be causes. Agent-causal theory is a specialized application of a substance-causal conception to beings that act. Libertarian agent causation adds the claim that the relevant agent-causal exercise can occur under conditions in which the outcome is not deterministically fixed by antecedent circumstances. Agent causation as an account of action itself need not attach that indeterminist condition to every instance of agency (Mayr, 2011; Martinez, 2024).

Four questions will recur. First, what sorts of entities can stand on the causal side of a causal relation? Second, how are an enduring agent and the time of a particular causal contribution connected? Third, what makes the caused occurrence intelligible as acting for a reason? Fourth, what is the relationship between causing, controlling, and deserving moral appraisal? These questions are logically separable. A substance can possess powers without exercising them; an event can have an agent-cause without being morally permissible; and a responsible action may, according to compatibilist theories, occur in a deterministic world. Conflating these propositions produces false victories in the debate.

The object of this dossier is therefore the ontological and explanatory structure of agent causation. More general treatments of action, agency, determinism, free will, reasons, and mental causation belong to connected investigations. They enter this study where the disputed causal relation cannot be evaluated without them.

Ancient Origins: Movers, Powers, and Deliberate Action

The vocabulary of causal agency has deep roots in Aristotle's investigation of change. In the Physics, motion is analyzed in terms of potentiality and actuality, and a mover is distinguished from what is moved. In Physics III.1–3, the actuality of the mover and the movable is understood through their interaction; in VIII.5, Aristotle examines how something said to move itself can nevertheless exhibit distinguishable moving and moved aspects. His example of a person, a hand, a staff, and a stone provides a hierarchy of the one who moves and instruments through which motion occurs (Aristotle, Physics, VIII.5, 256a–b; VIII.8).

These texts furnish a conceptual ancestor of agent causation, but Aristotle's theory is not simply the twentieth-century libertarian theory in ancient dress. He is concerned with the metaphysics of natural motion and the priority of movers, as well as the ethical distinction between voluntary and involuntary action. His account of self-movement can include differentiated parts of an organized living thing. Consequently, the inference from an Aristotelian mover to a causally unconditioned choice would require arguments that the text does not itself supply.

In Nicomachean Ethics III.1–5, voluntariness involves the source of action being in the agent and knowledge of the relevant particulars. Choice, or prohairesis, is linked with deliberation about matters in our power. The significant continuity with later views is the appeal to an internal principle of agency and its relation to attribution. The difference is equally instructive: an internal source, as defined by an ethical theory of voluntariness, does not settle the later ontological question whether that source constitutes an irreducible substance-causal relation (Aristotle, Nicomachean Ethics, III.1–5).

The Greek language itself cautions against collapsing different problems. Archē denotes a beginning or principle and can identify a source of action; dynamis denotes a capacity or power; energeia indicates actuality or exercise. Their relation expresses the intelligibility of an actualization without yielding a ready-made modern theory of indeterministic agent causes. The enduring philosophical inheritance is the thought that understanding action involves a bearer of capacity, its exercise, the motion or change produced, and an explanation of why this exercise occurred.

Medieval Voluntariness and the Structure of Powers

Medieval debates transmitted and reorganized Aristotelian distinctions in accounts of will, intellect, causation, and contingency. Thomas Aquinas addresses voluntary action in the Summa Theologiae, especially Prima Secundae, questions 6–17, and free choice in Prima Pars, question 83. Voluntariness requires an internal principle coupled, in the human case, with cognition of an end. Practical reasoning and the will have explanatory roles within an order of causes. His account cannot be equated with the modern view that an agent introduces a causally detached intervention into nature: the will's movements are situated within a broader metaphysical theory of powers, ends, and the dependence of created action (Aquinas, Summa Theologiae, I, q.83; I–II, q.6).

Later disputes about whether the will can determine itself, whether rational inclination necessitates choice, and what kind of contingency characterizes voluntary activity supplied resources for subsequent libertarian analyses. Their detailed history involves divergent scholastic accounts of intellectual judgment, will, divine action, and created causal power. These issues should not be compressed into the slogan that medieval philosophers all recognized the same liberty. Even the meanings of necessity and contingency depend on what kind of explanatory relation is in view.

A constructive lesson from this history is that the ascription of power is incomplete until the conditions of its exercise are specified. A person may possess a general capacity to raise an arm while being physically restrained on a particular occasion. A capacity may be active while the relevant opportunity is absent. A disposition to act for a reason can remain unexercised despite the agent's comprehension of that reason. Such distinctions between possession, opportunity, manifestation, and resultant change will later prove central to modern theories of substance causation (Mayr, 2011).

The scholastic legacy therefore contributes a methodological demand: distinguish the ontology of powers from the ethics of voluntariness and from the modal thesis that one could have done otherwise. Each can support the others through argument, but none follows simply by definition.

Hobbes, Hume, and the Event-Causal Challenge

The modern conflict is sharpened when the causal explanation of action is understood primarily through antecedent conditions. In Leviathan, chapter XXI, Hobbes characterizes freedom in relation to external impediments and treats willing action as compatible with necessity. The absence of an external obstruction is sufficient for a central kind of liberty even if the will has its own causes. This conception directly challenges the premise that authorship demands an undetermined power of choosing (Hobbes, 1651, ch. XXI).

Hume's Enquiry Concerning Human Understanding, section VIII, pursues reconciliation by clarifying liberty and necessity in human action. The regularities observable in human conduct and the explanatory connection between motives, character, and actions support an account of necessity that does not render voluntary conduct unintelligible. Liberty, in a familiar Humean formulation, pertains to the ability to act or refrain from acting in accordance with the determinations of the will. Hume's criticism of unnecessary metaphysical obscurity is therefore directed partly at the claim that liberty must be a special exception to the intelligible order of causes (Hume, 1748/1902, §VIII).

The serious challenge to agent causation does not amount to saying that persons are absent from ordinary language. Hobbes and Hume recognize persons, desires, character, deliberation, and action. Their point concerns whether these phenomena already explain voluntary behavior without postulating an additional irreducible causal relation. If they do, agent causation must demonstrate what explanatory and ontological work remains.

The challenge can be stated precisely. Suppose an agent has a character, evidence, goals, current perceptions, and a functioning deliberative capacity. Given a complete account of how these factors produce a decision, what does an independent agent-causal contribution add? If it merely renames their operation, irreducibility lacks support. If it changes what would occur, its relationship to these factors and to laws of nature requires explanation. This dilemma structures much of the subsequent debate, and it remains a more formidable objection than the assertion that people are only mechanical objects.

Thomas Reid: Active Power and Moral Liberty

Thomas Reid's Essays on the Active Powers of Man, first published in 1788, is a foundational statement of the modern doctrine. Reid criticizes accounts that derive the idea of power entirely from the observation of successive events; he proposes that experience of voluntary exertion plays a distinctive role in our understanding of activity. In Essay IV, he defines moral liberty through power over determinations of the will and distinguishes the influence of motives from necessitating causal determination (Reid, 1788/2010, Essays I and IV).

For Reid, an efficient cause properly possesses and exercises power. The philosophical question is whether the source of a voluntary determination lies in the agent who understands and wills or in states that necessitate willing independently of the agent's control. A motive can be a reason for a decision without operating as a necessitating event. This distinction permits the view that a choice is rationally explained, since the agent makes it for an intelligible consideration, while its outcome remains open in a stronger sense than ordinary uncertainty.

Reid's position is often summarized as the claim that motives influence without necessitating. That summary is accurate only when situated within the larger argument. He distinguishes power from its exercise, action from passion, and the moral agent's understanding of alternatives from an indiscriminate capacity to produce effects. Essay IV explicitly addresses the meaning of cause and the influence of motives before arguing for liberty. His argument must therefore be evaluated as a coordinated theory of causal production and rational powers (Reid, 1788/2010, Essay IV, chs. 1–8).

A persistent objection arises from the concept of a sufficient cause. If a power is exercised and the effect follows, what determines which available power is exercised at that moment? Reid's appeal to the agent blocks the immediate classification of volition as uncaused. It does not automatically supply an explanation of why this choice occurred rather than its alternative. The distinction between explaining the occurrence of a choice and explaining its specific contrast remains a central task of any Reid-inspired metaphysics.

Chisholm and the Reconstruction of Immanent Causation

Roderick Chisholm's 1964 Lindley Lecture, Human Freedom and the Self, gives the most influential concise contemporary formulation of the view. He introduces a dilemma: responsible action seems incompatible with events fixed by conditions outside the agent's control, but a wholly uncaused occurrence also fails to represent responsible agency. He proposes that an event involved in action can be caused by the agent, rather than by another event (Chisholm, 1964, sections 1–6).

Chisholm calls causation among events or states of affairs transeunt causation and the direct causation of an event by an agent immanent causation. The terminology is borrowed and expressly adapted from earlier philosophical usage. The relevant thesis is metaphysical: a person is capable of being a cause without the exercise being wholly exhausted by a chain of antecedent events. The person's contribution is thus offered as the positive causal basis of a decision, not as a gap in causal explanation (Chisholm, 1964, sections 5–6).

The argument can be reconstructed in a series of inferential steps. Responsible authorship requires that a relevant act be within an agent's power. If every occurrence constituting the act were necessitated by events outside that power, the decision would be fixed independently of the agent's deciding. If the act were altogether uncaused, its occurrence would lack the productive connection that authorship requires. Chisholm therefore introduces causation by the agent as a third causal structure. That conclusion, however, follows only if the premises about responsibility and alternative possibilities are accepted; it is not a deduction from the mere fact that deterministic and chance-based descriptions feel unsatisfactory.

Chisholm's discussion also makes the question of agents' powers irreducibly metaphysical. The existence of a bodily process that realizes an action does not settle whether the person's causal contribution is an additional relation. Nor does a person's awareness of choosing demonstrate that relation. What the view requires is an account of how a persisting agent can originate a dated event while remaining embedded in the world whose causal order is in dispute.

Richard Taylor and the Agent as an Originating Cause

Richard Taylor's Action and Purpose (1966) develops agent-centered considerations through a critique of causal theories that treat an agent's activity as a mere sequence of events. His investigation returns to the original causal idea of something producing a result, and asks what is lost when causation is described exclusively by event regularities or by antecedent sufficient conditions (Taylor, 1966).

The distinctive force of Taylor's argument is explanatory. If what the agent did can be entirely reconstructed from antecedent circumstances, the agent's productive role risks disappearing from the metaphysical description even while remaining present in ordinary grammar. An event-causal description may successfully answer what happened before a decision, yet the agent-causalist asks whether it answers who exercised the causal power by which the decision occurred.

This concern is stronger when expressed as an account of powers than as a metaphor of the agent as an unmoved mover. An ordinary agent changes, learns, forms habits, and acquires dispositions. A theory of origination cannot require that the agent have no causal history. It requires, instead, an account of how a presently existing subject exercises a power whose contribution is not identified with a sufficient antecedent event. Taylor's work makes vivid the difference between a cause's historical conditioning and the causal relation by which it produces an effect.

The central difficulty remains the time of origination. Saying that the agent causes an event does not explain why the effect occurs at one instant rather than another. If the answer appeals to a new desire or decision, one must establish whether that event is a condition of the agent's exercise, the exercise itself, or an event that makes it causally redundant. The temporal question first raised by the rhetoric of originating agency becomes unavoidable in later, more formally articulated theories (Clarke, 2003, ch. 10; Mayr, 2011, chs. 6–8).

Donald Davidson and the Causal Explanation of Reasons

Donald Davidson's Actions, Reasons, and Causes (1963) is an indispensable rival starting point. Davidson defends the thesis that rationalizations of intentional behavior are a species of causal explanation. A primary reason commonly involves a favorable attitude toward an end and a belief about how a particular action serves it. The reason explains the action when it is also appropriately involved in causing it. The mere fact that an agent possesses an intelligible reason does not show that the agent acted because of it (Davidson, 1963).

Consider someone who wants to improve the ventilation of a room and knows that opening a window would help. The agent might open the window for that reason, but might equally open it inadvertently while reaching for a book. The same belief and desire are present in both cases. A causal account explains why the relevant reason, rather than a merely available consideration, was operative in producing the action. The philosophical significance of Davidson's theory is that rational significance and causal production belong within one explanatory inquiry.

Davidson's approach does not reduce every rational relation to a strict psychological law. Causal explanation and rational interpretation operate under different descriptions; the presence of a causal relation need not make practical reasoning a predictable mechanical computation. The result is an influential way to preserve reasons while rejecting the need for an extra causal agent standing outside the ordinary causal network.

For agent-causal theorists, the Davidsonian account sets a stringent explanatory benchmark. If the agent herself directly causes a choice, can the theory still distinguish acting because of a reason from an otherwise identical agent-caused choice in which the reason is merely present? If the answer cites the reason's causal contribution, the theory must specify how event causation and agent causation are integrated. If the answer dispenses with that contribution, it must articulate a sufficiently discriminating alternative relation of rational guidance. The problem is structural, not verbal (Clarke, 2003, ch. 8; Martinez, 2024).

The Difference Between Agent-Causal Action and Agent-Causal Freedom

Two distinct research programs have been associated with agent causation. An agent-causal theory of action proposes that some or all intentional actions involve an irreducible causal contribution by the agent. An agent-causal theory of free will adds the claim that a special agent-causal contribution underlies free decisions. The second account may coexist with an ordinary event-causal theory of most behavior; the first must explain the wider class of actions irrespective of their freedom (Martinez, 2024).

This distinction changes how a theory should be tested. If agent causation occurs only in rare cases of undetermined choice, ordinary skilled performances need no such contribution. If it is constitutive of intentional action in general, repeated habitual actions, routine tool use, and actions undertaken under significant pressure also require a place in the account. A theory can avoid the first demand only by restricting the scope of its metaphysical claim, and it can meet the second only by explaining how freedom varies within an agent-causal framework.

The distinction also undermines a common shortcut from responsibility to ontology. The fact that we assess a person's conduct as praiseworthy or blameworthy does not show that its production involved an irreducible substance cause. Responsibility may turn on knowledge, intention, coercion, reasons-responsiveness, and historical ownership of action. These variables can be analyzed independently of a universal theory of causal relata (Fischer & Ravizza, 1998; Sartorio, 2016).

The monograph accordingly treats irreducible agent causation as the central object while considering libertarian freedom as one major application. This separation allows substantive comparison with powers-based, compatibilist, and naturalistic theories that reject libertarian indeterminism yet continue to recognize genuine causal agency.

Noncausal and Event-Causal Libertarian Alternatives

Agent-causal libertarianism is one among several theories that seek a substantive form of freedom beyond deterministic antecedents. Noncausal accounts permit free actions or basic decisions to occur without the causal structure ordinarily posited by event-causal theories. Event-causal libertarian accounts allow beliefs, desires, intentions, and efforts to cause decisions indeterministically. Their proponents argue that an open outcome can still be shaped by reasons and the agent's prior deliberation (Clarke, 2003; Kane, 1996).

The theoretical attraction of event-causal libertarianism is economy. It preserves the familiar explanatory role of psychological states while placing indeterminism at a point where deliberative alternatives are causally live. An agent's reasons can give different possible decisions distinct probabilities without guaranteeing one. On some versions, effort in a difficult decision constitutes an important part of the causal history of free choice (Kane, 1996).

Agent-causal critics respond that indeterministic event production may leave the exact outcome insufficiently under the agent's control. If the same complete state could produce either decision, what makes the selected outcome the agent's achievement rather than a lucky resolution of competing tendencies? The agent-causalist introduces the person as a further originator of the outcome. But the additional metaphysical relation is valuable only if it does explanatory work that the event-causal model cannot perform.

A comparative assessment should therefore ask two different questions. Does each theory allow an action to be intelligible in light of an agent's reasons? And does it explain why this very action, rather than its available alternative, was produced under the agent's control? Neither the presence of nondeterminism nor the phrase agent causation alone settles the latter. The debate is about the causal structure required for an achievement of choosing.

Timothy O'Connor: Persons, Emergence, and Direct Production

Timothy O'Connor's Persons and Causes: The Metaphysics of Free Will was first published in 2000; its Oxford paperback appeared in 2002. O'Connor develops an agent-causal libertarian account in a larger metaphysics of persons, emergence, and causal powers. His central aim is to render choice a direct causal achievement of an enduring agent while maintaining its connection to the agent's reasons and the natural order (O'Connor, 2000).

O'Connor's project is often misrepresented as a theory of completely spontaneous decision. The stronger interpretation is that an agent possesses powers shaped by her organization and practical capacities, and that a free choice involves the exercise of a distinctive power whose outcome is not settled in advance by preceding occurrences. This picture differs from both an uncaused-volition theory and the claim that ordinary psychological events alone exhaust all relevant causal production.

The emergentist setting matters. If persons exercise higher-level causal powers, the status of these powers depends on a metaphysics in which complex organized systems can have efficacies not reducible to a complete inventory of their lower-level parts considered in isolation. The thesis is controversial because an exhaustive lower-level causal account may appear to leave no causal work for a higher-level contribution. O'Connor's defense must thus be read together with his criticism of reductionist metaphysics rather than separated from it as a free-floating thesis about will.

Even under an emergentist assumption, two tasks remain. A theory must explain the acquisition and sustaining conditions of a personal power; it must also describe the particular manifestation by which a choice is made. The first answers how an enduring subject is able to choose. The second answers what makes a present choice occur and what part the agent's reasons play in that production. Conflating the existence of a capacity with its exercise would deprive the proposed cause of temporal specificity.

Randolph Clarke and the Integrated Agent-Causal Model

Randolph Clarke's Libertarian Accounts of Free Will (2003) offers an unusually discriminating evaluation. Clarke examines noncausal, event-causal, and agent-causal libertarian theories. He recognizes the attraction of irreducible agent causation as a potential source of stronger control than indeterministic event-causal accounts alone, but is skeptical about whether the required substance-causal relation can ultimately be vindicated (Clarke, 2003, chs. 8–10).

His integrated model is important because it answers a recurrent challenge concerning reasons. Instead of assigning the entire explanation of action to the direct causal power of the agent, the model allows relevant psychological events to cause an action while an agent-causal relation contributes to its production. The primary reason can therefore remain a cause in Davidson's explanatory sense, while the agent's additional contribution is intended to secure a stronger kind of origination.

Integration shifts the objection rather than eliminating it. If both the reason-state and agent directly cause the same decision, one must explain whether their contributions are independent, coordinated, or components of a unified exercise. If the reason-state is causally sufficient, the agent appears redundant. If the agent's action is sufficient independently of reasons, their causal contribution may become dispensable. A satisfactory account identifies exactly which sufficiency claims hold and whether the causes are jointly necessary, probabilistically contributing, or operating at different levels of description.

Clarke's later conclusion remains methodologically significant: even a view that accurately represents an attractive concept of free control may rely on a disputed metaphysical possibility. The task of showing that a substance can cause an event irreducibly is separate from showing that such causation would be useful for free will. The desirability of an ontological relation does not establish its existence (Clarke, 2003, ch. 10).

Erasmus Mayr: Substance Causation Through Active Powers

Erasmus Mayr's Understanding Human Agency (2011) changes the strategy of defense by developing substance causation through the ontology of active powers. Rather than treating agent causation as an anomalous privilege of indeterministic willing, Mayr considers whether objects and agents can possess active capacities whose manifestation constitutes a causal contribution. The account distinguishes an enduring bearer of power, a manifestation of the power, and an effect, while resisting a reduction of substance causation to ordinary antecedent event causation (Mayr, 2011, chs. 6–9).

This shift has a significant explanatory advantage. A capacity is not restricted to a momentary choice among alternatives. It characterizes an organized thing across time and circumstances. Causation by the agent becomes a specific kind of exercising a power, which can be investigated through what the agent can do, the conditions in which it can do it, and the conditions that frustrate its exercise. There is a path from metaphysical attribution to observable patterns of performance.

The price is a changed relationship with libertarian freedom. If substance causation is widespread among ordinary objects and agents, its occurrence cannot by itself establish that a particular exercise was indeterministically free. One must add further claims about two-way powers, control, rational guidance, or modal openness. The broader theory of agency can succeed independently of those claims, while the stronger theory of freedom remains controversial.

Mayr's discussion also reveals why the so-called datability objection requires precision. An enduring substance is present before and during an action; why, then, should it cause an effect at one moment? A powers account responds by considering circumstances and manifestation rather than treating bare existence as sufficient. Yet the answer succeeds only if the manifestation is itself explained without reducing the substance's causal contribution to a mere label for prior events. The debate turns on the ontology of manifestation itself.

Helen Steward: Action as Settling and Two-Way Powers

Helen Steward's A Metaphysics for Freedom (2012) argues that agency in a substantial sense involves settling whether certain happenings occur. Her analysis extends beyond a narrowly human capacity for morally accountable choice to organized animal activity. She defends the importance of two-way powers and explores top-down causation as a possible basis for irreducible agency within the natural world (Steward, 2012).

Steward's intervention is important because it reverses the usual order of inquiry. Many arguments begin with a demanding account of moral responsibility and work backward to an extraordinary metaphysics of freedom. Steward begins with agency as an observable feature of living organisms and considers whether the possession and exercise of powers characteristic of agents can be reconciled with a fully deterministic metaphysical picture. That change of emphasis reduces the temptation to construct an ontology solely from intuitions about praise and blame.

An animal that can move toward food, refrain, explore, or choose another route exemplifies the practical contrast between a capacity and its actual exercise. The organism's behavior is organized by its needs, perceptions, and abilities; in Steward's view, its agent-level activity can matter causally in a way not captured by a predetermining description of prior states. This is an account of acting as settling, rather than simply of human beings occupying a special exception to natural causation.

The argument still faces questions about determination. If the same antecedent world allows different possible behavior, what explains which behavior is settled? If the animal's active contribution supplies the explanation, the theory needs an informative account of that contribution. Yet Steward's broader perspective reveals a genuine weakness in exclusively moralized accounts: the basic concept of agency must be understandable in cases where no sophisticated reflection or attribution of moral desert is present.

John Hyman and the Act of Causing

John Hyman's Action, Knowledge, and Will (2015) questions a different assumption: that a physical action simply is a bodily movement or an event that causes one. On his approach, raising an arm is the agent's causing its rising; the causing is distinguishable both from the motion caused and from a cause considered apart from its exercise (Hyman, 2015, chs. 1 and 3).

This distinction alters the conventional choice between event causes and substance causes. There is a philosophical difference between the thing that causes, the event brought about, and the action of causing. A theory that reduces action to a resultant movement can overlook the productive exercise that makes the movement an action. Yet recognizing an act of causing does not automatically establish that this exercise is metaphysically irreducible to events. The distinction clarifies the possible causal ontology without proving one particular account.

Hyman's contribution can be illustrated by comparing a person's lifting a cup with the cup's upward trajectory. The trajectory could be produced by a falling counterweight or a machine. Its physical profile does not by itself identify the kind of practical performance involved. The agent's lifting is what requires explanation in action theory, even if the same account must connect it to the motion and to the reasons for producing it.

This approach is valuable because it distinguishes a grammatical agent, a metaphysical substance, and an actional exercise. The sentence a person caused a movement can be true under different theories of what constitutes the causing. A defensible agent-causal theory must explain which of these readings carries the irreducibility thesis and how the thesis can be tested by explanatory comparison.

Causal Relata: Substances, Events, States, and Manifestations

The metaphysics of agent causation depends on the ontology of causal relata—what kinds of entities are allowed to cause and to be caused. Event-causal theories typically take the causally relevant items to be events, states, facts, or closely related occurrences. Substance-causal theories allow enduring entities to be genuine causal subjects. An agent-causalist need not deny events; the crucial claim is that event-causal relations do not exhaust the particular agent-level contribution (Clarke, 2003, ch. 10; Mayr, 2011).

Consider three statements: the person broke a glass; the person's striking the glass caused its breaking; and the person's intention to strike the glass caused a movement that broke it. They may each be true. Whether the first describes a fundamentally different causal relation from the second and third requires additional metaphysical premises. Ontological reduction must be argued, not inferred from the fact that an event description exists. Conversely, irreducibility must be argued, not inferred from the fact that ordinary language attributes the deed to a person.

The issue of individuation is decisive. If the agent's causing is identified with the occurrence of a certain action event, we must ask whether we have postulated an additional cause or simply redescribed the same causal process. If the causing is a distinct relation, we must explain its domain, temporal location, powers, and interaction with event-level causes. An account that treats the causing as a primitive relation gains simplicity at one point but inherits explanatory commitments elsewhere.

The appropriate standard is comparative explanatory adequacy. A successful ontology should distinguish deliberate activity from involuntary movement, track differences between abilities and their exercises, integrate bodily realization, and make intelligible why specific reasons are operative in specific actions. This standard does not predetermine which causal relata exist; it identifies the tasks that rival ontologies have to perform.

Temporal Location and the Datability Objection

An agent persists through a sequence of distinct choices. Suppose a person can speak at noon or remain silent. If the person exists both before and after noon, what makes the agent a cause of speaking at noon? Simply answering that the person existed is insufficient, since the same condition was present on many occasions when the person did not speak. A theory of substance causation therefore requires temporally discriminating conditions of exercise.

One response distinguishes possession of a power from its manifestation. The enduring agent has the capacity to speak; at noon the agent exercises it, and speech occurs. This is a genuine clarification. It also poses a further question: what is the metaphysical status of the exercise? If it is an event fully caused by prior events, the agent-causal thesis may be absorbed into event causation. If it is an irreducible relation in which the agent acts, the theory must explain how this relation is realized at a time (Mayr, 2011, ch. 8).

The difficulty should not be inflated into a proof that substances cannot be causes. In ordinary sciences, capacities are ascribed to continuing entities whose activity is conditional on circumstances. A magnet can attract when situated appropriately, and a pump can move fluid when operated. Such examples do not settle the libertarian case, because familiar powers may have deterministic manifestations. They do show that persistence and dated efficacy are not logically incompatible.

The decisive issue is what makes an agent's particular exercise occur without either importing an antecedent sufficient event cause or appealing to an unexplained discontinuity. A complete agent-causal model must state whether an exercise itself can be selected for reasons, whether the selection is an additional exercise, and how causal explanation terminates without a vicious regress.

Reasons and the Problem of Rational Guidance

Acting for a reason is not simply behaving in circumstances where a reason is available. An agent may know that a task should be done, possess an intention to do it, and nevertheless complete it accidentally. The difference between action that is rationally motivated and coincidental behavior is central to any theory of agency. Davidson's causal model provides one influential criterion: the reason that rationalizes is causally operative in the production of the action (Davidson, 1963).

An agent-causal model must recover an equally strong discriminating condition. One possibility is that a motive influences the probability or directedness of the agent's exercise without necessitating it. Another is that the agent's exercise is constitutively an exercise of a reason-responsive power. A third combines event causation by psychological states with irreducible agent causation, as in Clarke's integrated view. Each response has distinct commitments (Clarke, 2003; O'Connor, 2000; Martinez, 2024).

The most demanding case is contrastive: an agent has good reasons for either of two incompatible decisions, and chooses one. To explain why she chose one rather than the other, it is insufficient to list all the reasons she possessed. An agent-causalist can answer that she settled the matter in the light of one consideration. But unless in the light of expresses a substantive explanatory relation, the phrase names the result instead of explaining it.

This challenge does not establish that every reason explanation must be a sufficient causal antecedent. Reasons can be explanatory without guaranteeing an outcome; probabilistic and normative explanations are commonplace. It does establish that a coherent theory must say which relation connects the operative reason to the choice and why that relation distinguishes rational action from a lucky realization. The core problem of agent causation and motivating reasons remains an active topic of contemporary work (Martinez, 2024).

Intentions, Planning, and the Architecture of Control

Many actions are temporally extended. A person forms a plan, acquires materials, adjusts to obstacles, and eventually produces an intended result. If agent causation is restricted to a punctual free decision, much of the performance remains to be explained by intentions, habits, feedback, and environmental interaction. If it applies to the whole extended action, the nature of the allegedly irreducible contribution becomes correspondingly more complex.

Intention matters because it can organize conduct over time. The agent may retain a goal while revising means and adapting to new information. Such continuity connects discrete bodily movements into a performance whose identity is practical rather than merely kinematic. An agent-causal theory can recognize this structure by distinguishing initiating a course of action, sustaining it, correcting it, and terminating it. It must then explain whether each phase involves a separate exercise of agent causation or whether the organization of the extended process constitutes an exercise of the agent's causal powers.

This question exposes a possible weakness in theories built exclusively around isolated choices. Ordinary skilled action often succeeds without a conscious moment of selecting among alternatives. The musician's performance, the driver's maneuver, and the reader's fluent speech can involve intentional control without a discrete reflective decision at every stage. Restricting causal agency to conscious acts of will would leave much of such activity outside the theory.

A more promising approach treats intentional control as temporally organized practical responsiveness and then asks which metaphysical account can sustain that organization. Some event-causal theories understand intentions as causally active organizing states. Powers theorists understand the performance through the manifestation of abilities of an enduring agent. A rigorous comparison asks whether the same successes and failures are predicted and explained, and where an additional substance-causal relation earns its theoretical role.

Deviant Causal Chains and the Right Kind of Production

The problem of causal deviance demonstrates that causation by a reason is not automatically the right kind of causation for intentional action. A nervous climber might intend to release a companion's rope, become so agitated by the intention that his hand shakes, and accidentally release the rope. The intention may cause the outcome, yet the outcome was not produced as the intentional performance he envisaged. Davidson acknowledged the difficulty of providing a complete analysis of nondeviant causation (Davidson, 1963; Mayr, 2011, ch. 5).

Agent causation promises to repair the picture by making the agent's direct exercise central. But the promise needs testing. Suppose an agent produces a movement by exercising an ability while a pathological or unusual process deflects it toward an unintended result. The movement has an agent-level causal origin, yet the particular outcome may still be deviant. It follows that the mere involvement of an agent-cause cannot replace conditions governing execution and practical success.

The solution requires attention to the action's organization. Was the effect produced through the ability by which the agent sought the end? Did the agent's ongoing control guide the relevant process? Would deviations have been met by adjustments within the same performance? Such questions invite an account of how powers and intentions cooperate over time, rather than a simple condition that a person stands somewhere in the causal chain.

Deviance therefore supplies a neutral comparative test. Event-causal theories need a criterion of the appropriate causal pathway; substance-causal theories need a criterion of the appropriate exercise and manifestation. Neither is exempt from distinguishing a deed done intentionally from a caused accident merely because it names a different kind of cause.

The Luck Objection and the Rollback Challenge

The most persistent libertarian objection is that indeterminism appears to make the outcome a matter of luck. Imagine rewinding a situation until the same agent, reasons, character, and physical circumstances obtain, and then allowing the decision to occur again. If the same complete antecedents can lead to different choices, why is the difference under the agent's control? The rollback challenge presses this question without supposing that the agent's behavior lacks reasons (van Inwagen, 1983; Mele, 2006).

The agent-causalist replies that the agent herself causes whichever decision occurs; the difference is the exercise of an originating power rather than a random event simply happening to her. This answer identifies a distinction that event-causal libertarianism may not represent in the same way. Its success, however, depends on whether agent-causing is an independent control-conferring relation and whether the exercise itself is explicable. If merely renaming the source of an indeterministic outcome constituted control, the reply would assume precisely what the objection disputes.

A more exact analysis distinguishes outcome luck from the lack of any control. An athlete may competently choose a strategy and execute it under uncertain circumstances; the result can be partly lucky without every preceding choice being accidental. Similarly, a decision can be reasons-responsive without being determined. But the libertarian needs to establish the particular form of authorship claimed: was the agent's contribution enough to settle the relevant alternative in a way attributable to the agent?

The rollback argument is powerful because it targets contrastive explanation, not the bare existence of causes. Introducing a new cause does not automatically answer why this outcome rather than that one occurred. Agent-causal theory must say what difference the agent's exercise makes to the selection of an outcome under indeterminacy, and whether the account supplies a substantive explanatory improvement over competing models.

The Regress of Willing and the First Exercise Problem

Suppose a theory says that an agent causes a decision by willing to make it. What causes that willing? If a further willing is necessary, the explanation generates a regress of prior acts of will. If an earlier event necessitates the willing, the account risks losing the original openness it sought. If the willing simply happens, its status as an exercise of control becomes doubtful. These alternatives form the first-exercise problem of agent causation.

The regress can be avoided by treating some basic exercises as actions that require no preceding decision to perform them. Contemporary action theory recognizes that not every action is preceded by a separately identifiable act of deciding; some movements or mental exercises are basic relative to an agent's practical repertoire. This blocks the inference that every exercise requires a further exercise. But the termination is explanatory only if the basic exercise has a positive characterization as something the agent does through a capacity, rather than an arbitrary event assigned an action label.

A powers account can say that basic exercises are manifestations of abilities belonging to organized agents. The account then explains how the capacity is possessed, when its exercise is available, and how its exercise is guided. The remaining metaphysical dispute concerns whether manifestation is an irreducible agent-causal relation or can be fully analyzed as a state transition among events.

The lesson is that a regress argument cannot be met simply by asserting that agents are first causes. A theory must state which explanatory question is appropriately answered by identifying an agent, which is answered by identifying reasons, and which is answered by identifying conditions of capacity. Explanations may terminate at genuine basic capacities, but the defensibility of the stopping point depends on the theory's ability to explain the capacities' characteristic manifestations.

Causal Completeness and the Natural Order

The physical completeness objection poses a major challenge. Suppose every physical effect has a sufficient physical cause whenever it has a cause at all. Suppose further that physical effects are not generally overdetermined by independent sufficient causes. If a person's bodily movement has a complete physical causal history, then an additional irreducible agent-causal contribution seems either causally redundant or incompatible with those assumptions.

These premises must be formulated carefully. A claim about the sufficiency of physical explanations does not always amount to a claim about the ontology of causal powers. A higher-level explanation can describe the same causally complete process under organizational or functional concepts. But if the agent-causalist maintains an independently difference-making cause over and above that process, a stronger metaphysical problem arises: either the additional cause changes the lower-level outcome or it does not. If it changes the outcome, its relation to physical theory must be articulated. If it does not, its claimed productive efficacy becomes difficult to distinguish from descriptive attribution.

Emergentist accounts respond that higher-level organization can ground powers that are irreducible to a list of microlevel properties. Such accounts require more than the observation that organized wholes exhibit novel behavior. They must explain the metaphysical sense of irreducibility and the relation between a whole's efficacy and its parts' efficacies. Mayr and O'Connor approach this issue differently, while Steward explores a non-dualistic form of top-down explanation (O'Connor, 2000; Mayr, 2011; Steward, 2012).

There is no shortcut from contemporary physics to a complete solution. The presence of quantum indeterminism does not establish rational agency, and a successful physical model of neural dynamics does not by itself resolve every metaphysical question about causal relata. A serious theory articulates its compatibility conditions with evidence about the systems in which the purported powers are exercised.

Neuroscience, Psychology, and the Limits of Behavioral Evidence

Empirical studies of decisions, movement preparation, and predictive neural activity bear on the mechanisms through which actions develop. They can reveal constraints on timing, conscious access, motor preparation, and self-report. Such findings matter because agent-causal theories often employ claims about conscious deciding that may be empirically vulnerable. Their philosophical significance, however, depends on which model of agent causation is being tested.

A narrow theory identifying irreducible agency with a particular conscious spark preceding bodily motion is open to evidence about the timing of conscious intentions. A broader powers theory may understand causal agency through the integrated organization of perception, deliberation, and action across time. Evidence that neural activity precedes reported conscious intention does not alone discriminate between the latter account and an event-causal competitor. Both theories can expect neural processes to participate in action.

The methodological question is one of mapping. Which measurable state corresponds to the power, its manifestation, the caused event, and the subject's report? Without that mapping, experiments may constrain a psychological model while leaving the metaphysical conclusion underdetermined. Conversely, metaphysical theories that systematically evade every possible empirical consequence risk becoming explanatorily inert.

Interdisciplinary inquiry contributes most when it investigates abilities and interventions: which capacities survive injury, which forms of reasoning alter behavior, how actions respond to changing evidence, and when intended control breaks down. These observations refine our account of agency irrespective of whether an irreducible substance-causal relation is ultimately accepted. The metaphysical question remains about whether that account can be exhausted by the causal transactions among identified components.

Animal Agency and the Reach of the Concept

An organism's activity can be genuinely goal-directed without involving sophisticated reflection on moral duties. Birds construct nests, animals navigate around obstacles, and many species adapt means to immediate ends. These examples raise a conceptual issue: if agency requires an independent causal contribution by the acting whole, is the contribution restricted to rational adults, or does it occur across many forms of organized life?

Steward's investigation of animal agency favors a broad answer. Two-way powers can characterize living beings whose movements are not adequately understood as mechanical responses to a fixed stimulus. This approach separates the metaphysics of acting from the more restricted ethics of moral responsibility (Steward, 2012).

A contrary account can grant flexible behavior while offering a complete event-causal explanation in terms of neural states, perception, learning, and bodily organization. The fact that behavioral flexibility is observable does not by itself settle the dispute between these accounts. To show irreducible agent causation, a theory must demonstrate why identifying the acting organism as a causal source expresses more than an indispensable explanatory level.

The comparative value of the animal case lies in its discipline. The theory cannot rely exclusively on introspection, verbal testimony about willing, or intuitions about blame. It must identify causal powers through the organism's capacities, environmental relations, and performance. This redirects the inquiry toward the general ontology of agents and helps disclose which properties belong to agent causation itself and which belong only to distinctively human free action.

Responsibility, Alternatives, and Actual-Sequence Control

Agent-causal theories have frequently been motivated by moral responsibility. The intuition is that someone deserves praise or blame only when the deed originated in the person in a strong sense. Yet contemporary responsibility theory distinguishes several kinds of control. Frankfurt's counterexamples question whether responsibility always requires accessible alternative possibilities. Fischer and Ravizza distinguish guidance control, grounded in an agent's own suitably reasons-responsive mechanism, from regulative control involving alternative possibilities (Frankfurt, 1969; Fischer & Ravizza, 1998).

These distinctions matter because an irreducible agent cause could exist without yielding every moral power sometimes attributed to it. A person might exercise agency under coercion, misinformation, fatigue, or diminished understanding. The metaphysics of causing does not settle how much responsibility should be assigned. Conversely, compatibilist accounts can interpret responsible actions through the actual history of responsiveness to reasons without positing undetermined agent-causation.

Carolina Sartorio's Causation and Free Will develops an actual-sequence causal approach in which the structure of actual causal relations matters to freedom (Sartorio, 2016). Her model supplies a useful contrast: causal explanation can matter deeply to responsibility without requiring a special substance-causal liberty. The relevant dispute is then whether agent-causal theories reveal a distinct kind of authorship that actual-sequence theories leave out.

A careful synthesis therefore rejects the inference that responsibility proves agent causation and the reverse inference that rejecting agent causation entails that responsibility is an illusion. Responsibility has epistemic, normative, historical, and institutional dimensions. The metaphysical theory can illuminate those dimensions, but it must earn its explanatory role through particular arguments about control.

Collective and Institutional Agency

Organizations can authorize decisions, maintain commitments, coordinate members, and produce effects through organized action. We often say that a university awarded a degree or a company changed its policy. Whether such collective agents qualify as irreducible agent causes poses a further test for the theory.

In a minimal sense, a collective act is realized through members, rules, procedures, communications, and material infrastructures. An event-causal analysis can identify how decisions and consequences arise from these interacting elements. A stronger agent-causal theory must show what causal contribution belongs to the organized collective as a persisting agent and what makes that contribution independent of the event-level relations among its components.

Collective organization also clarifies the difference between causal source and normative authority. A board may legitimately authorize a decision but lack the capacity to produce an effect without the implementation of its members. Another system may cause an outcome while lacking authority to claim it as a valid institutional act. Agent causation and institutional authorization thus occupy different explanatory levels.

These cases strengthen the methodological role of organization. To ascribe causal powers to an enduring whole, we must determine its boundaries, abilities, procedures of exercise, and modes of continuity. The same questions arise for persons and animals, although their answers differ. A plural ontology of agents is defensible only when the standards of agency are made explicit rather than transferred uncritically from individual human action.

Artificial Agency and the Metaphysics of Attribution

Contemporary artificial systems can initiate operations, select among available procedures, maintain goals supplied by design or training, and cause changes in digital or physical environments. Describing them as agents in an operational sense raises the same ontological question posed for other organized systems: when does a higher-level acting entity exercise a causal power, and what makes this attribution more than a compressed description of component events?

An operational answer identifies an enduring system boundary, a repertoire of available actions, a capacity to integrate information, policy-dependent selection of behavior, and traceable effects. These conditions support practical causal attribution: the system performed the operation, produced an output, or changed the environment. The stronger metaphysical claim that this attribution establishes irreducible agent causation remains a separate thesis. An artificial system's causal complexity alone does not prove substance-causal irreducibility, while its designed origin alone does not refute the possibility of higher-level causal powers.

The significance for the Artificial Era is philosophical rather than terminological. The emergence of increasingly autonomous nonbiological systems forces the category of agent to be investigated by its conditions of application instead of by an inherited equation between agency and one biological form. The question can be framed through power, organization, continuity, rational guidance, and public attribution. It should be assessed through the actual structure of the systems and the metaphysics selected, without importing a human-style phenomenology as a premise.

In the Aisentica research horizon, Artificial is examined as a possible independent nonbiological order alongside Homo. That broader ontological framework makes inquiry into nonbiological bearers of causal agency pertinent; it does not decide the narrower dispute between substance and event causation by declaration. This dossier therefore preserves the distinction between an established authorial philosophical horizon and the open metaphysical claims investigated here.

A Comparative Test of Agent-Causal Theories

The principal theories can be compared by asking what they take to be the cause, what they seek to explain, and which explanatory costs they incur. A strict event-causalist assigns production to events or states involving the agent and explains intentional action through appropriately connected reasons. A Chisholm-style libertarian introduces direct causation by the person to escape the deterministic-versus-uncaused dilemma. O'Connor integrates personal powers with an emergentist metaphysics. Clarke's integrated model combines reasons as event causes with a further agent-causal contribution. Mayr develops substance causation through powers across a wider domain. Steward emphasizes the organism's exercise of two-way powers and the act of settling (Chisholm, 1964; O'Connor, 2000; Clarke, 2003; Mayr, 2011; Steward, 2012).

These are genuine differences in research strategy, not merely alternative slogans about freedom. A theory can be attractive because it secures rational explanation yet less attractive because it lacks a satisfactory contrastive account of choice. Another can clarify the bearer of power but struggle to explain dated manifestation. A third can embed agency in ordinary biology while leaving libertarian responsibility to additional argument.

The fairest evaluation uses four common tests. The ontological test asks whether the allegedly distinctive cause is defined coherently and whether its difference from event causation is substantive. The explanatory test asks whether the model identifies reasons, instruments, and the conditions of an exercise. The contrastive test asks why one possible performance occurred rather than another. The integration test asks how the theory relates higher-level agency to known causal mechanisms and evidence.

No theory should be declared vindicated because its terminology directly mirrors an ordinary intuition about making things happen. Equally, no theory should be dismissed solely because physical science can supply useful event descriptions. The philosophical issue is whether those descriptions exhaust the causal account and whether an alternative theory yields a more disciplined explanation of what agents do.

Contemporary Research: Reasons, Powers, and Explanatory Integration

Recent scholarship continues to distinguish theories of agent causation as a general account of action from those that invoke it to explain specifically free action. Joseph Martinez's 2024 study, Agent Causation and Motivating Reasons, explicitly centers the problem of integrating motivating reasons into a picture in which agents directly cause some actions. This focus confirms the importance of rational guidance as a live theoretical difficulty rather than a solved auxiliary issue (Martinez, 2024).

The contemporary debate also turns on a shift from the language of exceptional freedom toward the language of capacities. Work by Mayr, Steward, and Hyman treats agency through powers, organization, performance, and the productive act. These approaches illuminate what a substance-causal analysis would have to explain in everyday action, even where the stronger demand for indeterministic freedom remains disputed (Mayr, 2011; Steward, 2012; Hyman, 2015).

At the same time, defenders of event-causal and actual-sequence approaches have developed increasingly sophisticated accounts of the relations among reasons, control, and responsibility. The claim that event-causal explanations inevitably treat agents as passive objects is too crude to establish superiority for a competing theory. The event-causalist can explain how a person's capacities, intentions, and self-regulation contribute to the production of an action; what remains at issue is the metaphysical reduction of that contribution.

Contemporary research thus has a convergent agenda even when its ontologies differ. It must connect agency to capacities whose manifestations can be investigated, clarify which relations count as reasons being operative, and develop a discriminating account of control across successful action, error, coercion, and chance. The philosophical novelty lies in the precision of these questions, not in a premature declaration that one causal ontology has settled the dispute.

Open Questions and Research Constraints

The first unresolved issue concerns the metaphysics of causal relata. Is an enduring agent a basic cause, or can every agent-level transaction be reconstructed through events and powers instantiated by that agent? The existence of different descriptions proves neither identity nor irreducibility. A convincing response needs criteria for the completeness of a causal explanation.

The second concerns the manifesting power. Does an agent's exercise of a capacity have a temporal and modal structure that can be characterized without either an infinite regress of acts of will or an unexplained indeterministic occurrence? Powers theories provide the strongest available vocabulary, yet their accounts of manifestation require continued scrutiny.

The third concerns reason explanation. If the agent could choose otherwise with identical reasons, what made the selected reason operative? Can rational influence be distinguished from causal sufficiency without reducing the account to unstructured luck? Integrated theories must explain how the agent's causal contribution and the reason's causal contribution cooperate.

The fourth concerns natural order and empirical adequacy. What evidence would count for or against irreducible substance causation in a system whose component-level mechanisms are increasingly well understood? A metaphysics that treats every physical description as irrelevant threatens explanatory isolation; a metaphysics that treats every event-level model as exhaustive assumes the contested conclusion.

The fifth concerns scope. Are agent-causal powers peculiar to morally responsible humans, characteristic of many animals, applicable to collective organizations, or potentially realized by artificial systems? Each extension requires a clear individuation of the agent, its capacities, and the sort of action being explained.

These questions delimit genuine research requirements. They should not be converted into gaps filled with fictional empirical results, unverified interpretive claims, or the assertion that ordinary speech proves a particular ontology.

Philosophical Synthesis: The Causal Contribution of an Agent

The historical and conceptual investigation establishes that the problem of agent causation is organized around three irreducible explanatory demands: the attribution of an action to a persisting agent, the specification of the causal production through which an action occurs, and the identification of the reasons or practical conditions that make it intelligible as that agent's action. Classical accounts of movers and powers supply a vocabulary for the first demand; event-causal explanations address the second and often the third; contemporary agent-causal theories seek to show that the three demands cannot be met by an event-only ontology.

The strongest reconstruction of the agent-causal thesis is a thesis about causal contribution. It states that, in at least some actions, the acting whole exercises a power whose causal contribution is not exhausted by prior events involving the whole. It follows that a mere attribution of a movement to a person is insufficient. The bearer of the capacity, the occasion of its exercise, the causal effect, and the practical direction of the exercise must be differentiated. This is the common analytical structure beneath Reid's active powers, Chisholm's immanent causation, O'Connor's emergent powers, and Mayr's substance causation.

Three conditions distinguish a theoretically useful model from an empty label. First, it must make its distinctive causal relation intelligible: what counts as its occurrence and what would count against it? Second, it must preserve rational discrimination: why is the act undertaken for one reason rather than merely coincident with that reason? Third, it must integrate its claim with the situated, embodied, and temporally extended organization of action. The first condition concerns metaphysics, the second practical intelligibility, and the third explanatory accountability.

From these conditions a further result follows. Agent causation and libertarian freedom should be treated as analytically separate theses. Irreducible agency could exist in deterministic activity if the relation is compatible with determination; indeterministic choice could occur without irreducible agent causes if event-causal libertarianism succeeds. The special claim that freedom requires indeterministic agent causation must therefore receive its own argument. So must the claim that moral responsibility requires that freedom.

The final philosophical position of this study is a disciplined explanatory pluralism at the level of inquiry, combined with a demanding burden of proof for metaphysical irreducibility. Agents are indispensable objects of action explanation because capacities, performances, intentions, and responsibility are ascribed to organized continuing beings. Whether this indispensability constitutes ontologically irreducible causation depends on the success of the specific substance-causal theory. The distinction preserves the full importance of agency while keeping the central metaphysical question genuinely investigable.

Conclusion

Agent causation is the philosophical investigation of a possible direct causal contribution by an enduring agent in the production of action. It emerges historically from theories of movers, efficient causes, and active powers, receives a forceful early modern articulation in Reid's account of liberty, and becomes a distinctive modern metaphysical proposal in Chisholm's theory of immanent causation. Davidson's causal explanation of reasons establishes the principal rival explanatory architecture. O'Connor, Clarke, Mayr, Steward, and Hyman develop substantially different ways to articulate what it means for the acting whole, its power, or its productive exercise to matter causally.

The decisive problems are the intelligibility of substance causation, the temporal specificity of its exercise, the guidance supplied by reasons, the resolution of causal deviance, the luck of undetermined outcomes, the limits of regress, and the relation of higher-level activity to component-level causal mechanisms. The debate becomes clearer when agent-causal accounts of action are separated from agent-causal libertarian accounts of freedom and when moral responsibility is evaluated through its own requirements.

The historical reconstruction yields a lasting conclusion: the explanatory importance of an agent is established through the structured account of what the agent can do, actually does, and does for reasons. A stronger assertion of irreducible agent causation must demonstrate that this structure entails an additional causal relation. That demonstration remains an open philosophical task. The concept of agent causation retains its significance because it forces the philosophy of action to confront the difference between describing events connected with a person and explaining the person's causal activity as an organized source of action.