Causal Theory of Action

The Explanandum: What a Causal Theory Must Explain

An account of intentional action must explain how events occurring in a body or in an environment become intelligible as things someone does. The same movement of a hand can be a greeting, a warning, the beginning of a musical performance, an attempt to get attention, or an involuntary spasm. A description of the movement's trajectory cannot establish which of these descriptions is correct. The causal theory proposes that the determining difference lies in how the performance arises from the agent's practical psychology. The belief that a visitor has arrived, a desire to greet the visitor, and an intention to make a recognizable gesture may form a pattern whose causal realization explains why the hand moved as it did.

This claim has several forms. A thesis about explanation says that reasons sometimes explain conduct by causing it. A thesis about intentionality says that action is intentional under a description because of the appropriate causal role played by reasons or intentions. A more ambitious reduction says that every fact about acting intentionally can be derived from appropriately specified mental causes and resulting bodily events. These propositions are distinguishable. Davidson's defense of causal explanation, for example, does not by itself settle every necessary and sufficient condition for intentional action, while later accounts of motor control may endorse causal production yet refuse a simple belief-desire reduction (Davidson, 1963; Shepherd, 2021).

The question of the relevant cause must also be separated from the question of the agent's causal power. Event-causal views characteristically cite occurrences or states such as intending and believing. Agent-causal accounts assign a more primitive causal role to the agent. The explanatory contrast is important even when both positions acknowledge physical processes, practical reasons, and causal relations. Our focal problem is the event-causal understanding of intentionally doing something, together with arguments about its adequacy. Agent causation requires an independent dossier because its central claims concern the metaphysics of the originating cause rather than the internal conditions of Davidsonian reason explanation.

The thesis guiding this investigation is that a satisfactory causalist must answer three different demands: why this particular reason mattered, how the resulting performance was organized, and why that organization qualifies the performance as the agent's action. Each demand is prompted by cases in which the others can be satisfied without successful agency. The distinction gives the theory its real historical depth and identifies its surviving contemporary research program (Davidson, 1963; Frankfurt, 1978; Bratman, 1987; Shepherd, 2021).

The Concept of Action and the Limits of a Bodily-Movement Model

The simplest causal model begins with an event in the mind and ends with a bodily movement. It appears attractive because bodily movements have causes, mental states can be causally relevant, and ordinary action explanations connect the two. Yet actions have varying temporal and material forms. Deliberately remaining still, speaking a sentence, preparing a meal, supervising a group, and finishing a long project cannot all be identified straightforwardly with one discrete muscular contraction. Even in the simple case, the action may be moving an arm, signaling a friend, or sending a warning by moving an arm. The vocabulary of action depends on reasons, descriptions, and consequences (Davidson, 1963; Piñeros Glasscock & Tenenbaum, 2023).

There is a separate question whether the action is the bodily event itself, an event plus a relation to intention, or an episode organized by an agent's guidance. If events are individuated by their causes and effects, one theory of individuation may place turning the switch and turning on the light in a single event. If actions are individuated more finely by descriptions or intentions, the same physical occurrence may correspond to several things done. Disagreements about individuation therefore influence what a causal theory is asked to explain. A theory of intentional performance should specify whether it treats actions as individual events, processes extended through time, or purposively structured performances with multiple realizations.

A causal theory also must not infer intentionality simply from the presence of a cause. A reflex, a panic response, and a carefully executed choice all have causes. The contrast concerns the relevance and organization of those causes: whether the antecedent practical states are involved as reasons and whether their influence is expressed through agentive control. The concept of action accordingly imposes a selective task on causation, not the impossible task of locating a causeless exception to the natural order.

Historical Foundations: Aristotle and the Origin of Movement in the Agent

Aristotle supplies an enduring vocabulary for the conditions of responsible doing. In the Nicomachean Ethics, Book III, the voluntary is characterized by the origin or moving principle being in the agent and by knowledge of the relevant particulars (III.1, 1110a–1111a). Choice, prohairesis, is distinguished from voluntary movement in general and connected to deliberation about what lies within the agent's power (III.2–3, 1111b–1113a). The distinction matters because merely wanting an end differs from selecting what to do to achieve it. In Aristotle's discussion, action is already linked to the agent's practical standpoint and to a temporally situated initiation of movement.

The Nicomachean account is historically related to Aristotle's discussions of animal motion and practical inference, but it is misleading to present Aristotle as a twentieth-century event-causalist. The Aristotelian archē, or initiating principle, belongs to a wider explanation of activity, living capacities, desires, and ends. Modern talk of mental events causing bodily motions abstracts one question from that network. The continuity lies in a problem rather than in a ready-made identical answer: an agent's knowledge, wanting, and deliberative orientation make sense of a movement as something undertaken.

Aristotle's distinction also provides a resource for assessing simplistic causalism. A forced movement can have an ordinary physical cause while lacking the voluntary origin in the agent. A voluntary act can occur without deliberate choice, as Aristotle's examples of sudden acts illustrate (Nicomachean Ethics III.2, 1111b). Thus a complete account cannot collapse action, voluntariness, deliberation, and rational choice into one definition. It must identify which feature of action the proposed causal explanation is intended to recover.

This inheritance extends through medieval discussions of will, intention, choice, and command. Aquinas distinguishes voluntary acts, interior acts of the will, and the execution of outward activity in the Summa Theologiae, especially Prima Secundae questions 6, 12, 13, and 17. These investigations concern different relations between an end, the selection of means, and a performance. They are relevant as conceptual ancestors of the modern explanation of intentional action, while operating within a distinct metaphysics of powers, ends, and human nature.

From Practical Reason to Early Modern Motivation

The early modern debate changes the setting by emphasizing causal psychology and the relations among passions, beliefs, volitions, and bodily behavior. Hume's Treatise of Human Nature, Book II, Part III, especially section 3, argues that reason considered alone does not supply the motivational impulse of action and that passions have a central practical role. The position helps explain why later theories treat desire-like states as candidates for the causes of action. It does not settle the question whether the desire itself, a decision, or a continuing intention must be the immediately relevant cause.

A critical distinction follows. The claim that motivational states have causal force is a psychological and metaphysical thesis; the claim that the language of reasons always includes a desire is a separate thesis about the structure of rationalizing explanation. A further claim that normative reasons depend on wants is yet another thesis. The enduring Humean influence partly consists in bringing these claims into close proximity, but they need not stand or fall together. Smith's analysis of the Humean theory of motivation develops precisely this separation between the causal status of reason explanation and the composition of motivational reasons (Smith, 1987).

Kant contributes another contrast. The Groundwork of the Metaphysics of Morals places action under the conception of laws and develops a conception of practical rationality structured by maxims and principles (Akademie pagination 4:412–421). His question concerns how rational agency and moral obligation are possible. That inquiry bears on the causal theory when someone asks whether normative authority is exhausted by a psychological causal history. It does not license the conclusion that Kant's practical reason is simply an alternative event-causal mechanism. The modern debate inherits both the psychology of motivation and the normative structure of acting for reasons; a defensible account must preserve their difference.

Anscombe and the Grammar of Intentional Action

G. E. M. Anscombe's Intention (first published in 1957) reorganized the philosophy of action around the question why something was done. Her famous methodological inquiry examines the uses of intention as directed toward the future, intention with which an act is performed, and intentional action itself. These are closely related applications of one philosophical problem rather than three independent psychological compartments. In Anscombe's account, the question why receives a special application when an agent can supply a reason that places what she is doing within practical knowledge and purposive understanding (Anscombe, 1957/2000, especially §§5–8, 19–21, 32).

Anscombe's illustration of someone engaged in a sequence of activities for increasingly remote purposes, as in pumping water in the context of a larger undertaking, makes action intelligible through a structure of means and ends. The person's knowledge of what she is doing has practical character; it is implicated in the execution rather than merely arrived at by looking at completed movements. This position resists a picture in which intentions are private mental episodes added to movements from outside. It also illuminates why the same physical behavior is describable at several intentional levels.

The disagreement between Anscombe and Davidson is often compressed into a simple opposition between reasons and causes. That compression loses what was at issue. Anscombe's investigation describes the conceptual relations among doing, intending, knowing, and answering why; Davidson accepts much of the importance of intentional descriptions and reason-giving, then argues that the occurrence of doing something for a reason requires a causal connection. Whether practical knowledge can be made compatible with that connection remains open. Both philosophers face the same explanatory fact: merely satisfying an observer's plausible description of why someone might act does not establish what this agent actually did for that reason.

Davidson's Intervention: Reasons as Causes

Donald Davidson's 1963 article Actions, Reasons, and Causes is a central text of the modern causal theory. His thesis is that rationalization, explanation by the agent's reason, belongs to ordinary causal explanation. The argument emerged against approaches that treated the rational relation between action and reason as evidence of an unbridgeable difference from causation. Davidson does not deny that reasons rationalize. He maintains that their rationalizing role and their causal role answer different but compatible questions (Davidson, 1963, pp. 685–700).

His formulation must be understood carefully. When someone acts for a reason, the reason helps show what the person found attractive, appropriate, necessary, or worthwhile in a type of action. A belief identifies the prospective performance as having the desirable characteristic. In this form the reason provides an intelligible perspective from which to interpret the action. But a person can possess many such outlooks simultaneously. The explanation of why one of those outlooks actually issued in conduct therefore requires a connection stronger than the availability of a rationalization.

Davidson's text makes two linked proposals. First, primary reasons supply a fundamental model for how a reason rationalizes. Second, the primary reason for an action is its cause. The second proposal is philosophically substantive. The word because in the claim that a person acted because she believed something must select an operative explanatory relation. A reason that the agent possessed but that played no role in producing the action may support an appraisal of the act, yet it will fail as the explanation of that person's performance. This marks the difference between an intelligible possible explanation and the true causal history of what happened.

The Primary Reason and Rationalization

A primary reason combines a pro-attitude toward actions of a kind and a belief that the actual action belongs to that kind. Pro-attitudes are broader than immediate desire: they include valuing, feeling obligated, wishing, caring, and favorable practical orientation. A belief may concern means, opportunities, rules, or circumstances. For example, an agent may value keeping a promise and believe that sending a message now will fulfill that promise. If she sends it on that basis, the belief and commitment together rationalize the act (Davidson, 1963, pp. 685–689).

This structure is useful because it is description-sensitive. If one and the same movement sends a message and inadvertently interrupts another person's concentration, the pro-attitude toward keeping a promise explains it as messaging, while it need not explain it as interruption. The agent's reason identifies an intelligible aspect of the action. A causalist does not have to claim that the reason caused an abstract description. The cause is related to an event, while the description identifies which aspect is rationalized by it. The explanatory grammar of the action sentence and the ontology of the causal relata thus require separate treatment.

The primary reason is a conceptual model, not a claim that the agent consciously rehearses a two-premise syllogism before each movement. Everyday skilled action frequently occurs without explicit conscious deliberation. Nor does the model imply that every motivation is fully sincere, coherent, or self-transparent. It identifies minimal practical elements in terms of which acting for a reason becomes intelligible. Subsequent theories expand the model by recognizing intentions, plans, hierarchical commitments, skill, and temporally extended control.

The Argument from Acting for One Reason Rather Than Another

The decisive explanatory challenge can be expressed through a counterfactual comparison. Suppose an investigator believes a document should be sent because it will help a colleague and also believes sending it will improve her professional reputation. She sends it. The available facts about her outlook provide two rationalizations. Yet she may have acted from concern for her colleague, from ambition, or from both. A theory treating possession of a rationalizing reason and conformity of behavior to it as sufficient cannot discriminate these cases.

Davidson's answer assigns an explanatory role to the reason that actually causes the action. The argument begins with the possibility of an agent having an unused reason, adds the occurrence of behavior consistent with that reason, and asks what further fact distinguishes acting for the reason from mere coincidence. A causal connection supplies a principled candidate for that further fact. It identifies a difference-making relation within the actual history of the performance (Davidson, 1963, especially pp. 691–694).

This is an argument for the explanatory relevance of causation, though its stronger metaphysical conclusion remains contested. A noncausal theorist may argue that the operative relation is a form of rational guidance, practical knowledge, or intelligible fit that is not itself causal. The causalist can respond that genuine guidance requires the reason to make an appropriate difference to what occurs. Yet the causal response must then specify appropriateness. That requirement is not a minor footnote; it creates the deviance problem. The most persuasive argument for causation therefore also generates the strongest obligation to analyze how causation operates.

Action Under a Description

Davidson's light-switch example supplies a durable model of the complexity of action description. An agent flips a switch, turns on a light, illuminates a room, and unknowingly alerts an intruder. On a coarse event-individuation view these descriptions may designate one action-event. Yet the agent acts intentionally in flipping the switch, perhaps in illuminating the room, and not in alerting the intruder. The reason explains one description without making all descriptions intended (Davidson, 1963, pp. 686–688).

The case implies a distinction between the truth of a causal claim and the success of a reason explanation under a chosen description. Causes can be referred to using many descriptions, but a rationalizing interpretation depends on the agent's practical conception of what she was doing. One cannot substitute any co-referential description into a reason explanation while preserving its explanatory sense. What physically happened may be a single causal history; what the agent intentionally did depends on how the action was represented in that history.

Some philosophers respond by assigning a finer identity to intentional actions, while others retain an event ontology with description-relative intentionality. The causal theory needs no dogmatic commitment to one universal granularity so long as it explains why the reason bears on the action under the relevant description. The example also establishes that actual consequences and intended effects are not equivalent. A planned signal, an unplanned injury, and an anticipated side effect can belong to one worldly sequence without sharing one intention.

Events, Causes, and the Ontology of Action

Event-causal action theory is closely connected to Davidson's larger metaphysics of events and mind. In Mental Events (1970), Davidson advanced anomalous monism: mental events may be identical with physical events although strict psychophysical laws connecting mental descriptions and physical descriptions are unavailable. The underlying idea allows the same occurrence to be described as a psychological state or event and to participate in physical causal relations. This offers a way of rejecting the inference that meaningful reason descriptions must be causally powerless because they differ from natural-scientific descriptions (Davidson, 1970/2001).

The distinction between an event and its description matters here. When a reason explains an action, the explanation employs intentional terms that place the event within a space of interpretation. When the event enters a causal relation, its causal identity need not be determined by the vocabulary used to describe it. Critics dispute whether such a framework adequately secures the causal relevance of mental properties, rather than merely the causal efficacy of events describable mentally. That challenge is part of the broader mental causation problem and must not be silently resolved by the phrase reasons are causes.

A different metaphysics treats actions as processes with phases of initiation, continuation, correction, and completion. On this view, the causes of initiation and the causes of sustained performance may have distinct explanatory roles. The issue is especially acute for writing a letter, playing a sonata, or crossing an unfamiliar street. A theory centered solely on the first movement risks losing the fact that agents keep acting and revise what they do as their situations change. These cases motivate a move from punctual antecedent causation to organized causal processes.

Belief, Desire, Pro-Attitude, and Humean Motivation

A standard presentation of causalism states that desires motivate, beliefs represent means, and their conjunction causes action. This schematic account captures one common practical pattern. An agent wants water, believes that turning a tap will supply water, and turns it. Yet simple conjunction is inadequate. A person can desire incompatible outcomes, retain outdated beliefs, or intentionally act against a currently stronger desire. The theory must determine which motivational configuration governs the behavior and why (Davidson, 1963; Mele, 1992).

Humean motivation and causalism should be kept conceptually separate. Michael Smith argues that a causal conception of reason explanation does not by itself privilege a desire-based account of the motivating state: one could adopt causal explanation while disputing whether desires alone have a distinctive role (Smith, 1987). This matters because a familiar historical shortcut infers a Humean theory from acceptance of causal reasons. The inference is invalid without an additional premise about motivational psychology. Davidson's primary reason includes pro-attitudes, but the categories of pro-attitude and motivating reason are broad enough to preserve this dispute.

Another distinction separates the reason that explains what the agent did from a reason that justified doing it. A false belief may figure in a genuine explanation. A cruel desire may causally explain an act without giving the agent an adequate justification. Conversely, a compelling normative reason may have no influence at all on what a particular person actually does. A workable causalism therefore recognizes explanatory reasons as psychologically operative while allowing independent assessment of their truth, rationality, and normative force.

Intentions as Causes and as Commitments

Intentions enter action theory because desires and beliefs often leave the practical transition unresolved. People can desire a distant end without deciding to pursue it, or believe a course is desirable while continuing to deliberate. An intention represents a practical settling of an issue: the agent has adopted a plan or commitment that organizes later reasoning and performance. Bratman's Intention, Plans, and Practical Reason (1987) develops this insight by treating intentions as elements in partial plans that coordinate activity over time and between agents.

A causal account can assign intentions three roles. They may arise through deliberation; they may persist as organizing states that filter alternatives and guide subsequent decisions; and they may help produce the actions that implement them. These roles are related without being identical. A person may sincerely intend to travel tomorrow yet fail to travel because conditions change. A person may perform the intended act through an accident rather than through the operation of the plan. And a person can carry out many unplanned details while pursuing a general intention.

The explanatory advantage of plans is their temporal structure. The agent's conduct is not produced once and for all by an antecedent desire; it is shaped by continuing commitments, background policies, and sensitivity to new information. This improves on a thin stimulus-response model. It also introduces new standards of failure: an action may be caused by an intention while bypassing the plan's required means, or may unfold in a way incompatible with its intended character. The causalist must decide whether persistence, feedback, and correct implementation are constitutive of acting intentionally, or merely typical features.

Prior Intention and Intention in Action

John Searle's article The Intentionality of Intention and Action, published in Cognitive Science in 1980, and the later chapter on intention in his 1983 Intentionality distinguish an intention formed prior to performance from the intention operative in acting. Searle develops an account on which intentions have causally self-referential conditions of satisfaction: it is part of their success conditions that the represented action occur because of the relevant intention. He connects this structure to experience of acting and the problem of deviant causal chains (Searle, 1980, pp. 47–70; 1983, ch. 3).

A prior intention is appropriate to projects prepared in advance. An intention in action is meant to account for the intentional organization of present doing, including action lacking a distinct earlier decision. The distinction answers an evident difficulty for theories that require a preceding choice: many skilled and spontaneous performances are intentional without conscious advance planning. It also clarifies why the source of the movement and the agent's practical orientation during the movement may be philosophically different.

Searle's proposal is more than a vocabulary change. By writing the relevant causal relation into the intention's conditions of satisfaction, it seeks to distinguish successful intended performance from an outcome merely resembling what was intended. The critical difficulty is whether causal self-reference already specifies a nondeviant route. An intention might cause the desired bodily event through panic or accident while still satisfying a thin causal clause. Searle's account therefore raises the question of how conditions of satisfaction relate to an independently specified action-generating mechanism.

The Fundamental Problem of Causal Deviance

The problem of causal deviance concerns a case where a suitable mental state causes the expected behavior or outcome, yet does so through a route that makes the behavior unintentional under the relevant description. A naive biconditional states that a bodily event is an intentional action whenever it is caused by the right mental states. Deviance cases refute that formula by separating the existence of a cause from the character of the connection between cause and performance. The difficulty is widely treated as a central unresolved test of event-causal theories (Davidson, 1973/2001; Frankfurt, 1978; Piñeros Glasscock & Tenenbaum, 2023).

Two methodological cautions are essential. First, a counterexample to a simple reductive definition is not automatically a counterexample to the thesis that reasons cause actions in genuine cases. Second, not all deviance concerns the same explanatory failure. Some chains interrupt the initiation of action, while others begin with an intentional movement but reach the intended end through an unforeseen mechanism. A convincing solution needs a route-sensitive account rather than the stipulation that the cause must operate in the right way.

The problem is generated by the causal theory's own strength. The theory explains why the agent did this rather than merely having a reason that would support doing it. To succeed, it must discriminate cases in which the reason really produces the performance as an exercise of the agent's agency from cases in which the reason merely contributes to a disaster, symptom, or coincidence. This discriminating task makes causal deviance a point at which metaphysics, psychology, and the phenomenology of action intersect.

Primary Deviance: A Reason That Produces an Accident

Davidson's mountain-climber example is the classic primary-deviance case. A climber wants to free himself from holding a fellow climber and believes relaxing his grip would achieve that goal. His thought of doing so unnerves him, causing his hand to loosen involuntarily. The appropriate belief and desire are causally relevant to the movement; nevertheless, he did not intentionally release his companion. The mechanism passes through disturbance rather than the execution of the contemplated act (Davidson, 1973/2001, “Freedom to Act”; see also Piñeros Glasscock & Tenenbaum, 2023).

The argument can be reconstructed as follows. A mental-state pair satisfies the rationalization condition for an intentional release; the pair causes a release; but the relevant causal sequence produces the movement as a symptom of distress. Hence rationalizing content plus an actual causal connection is insufficient. The missing element is a form of action-producing organization. It cannot be supplied merely by identifying the total causal chain, since both ordinary intention-driven movement and nervous malfunction are causal chains.

There are several candidate repairs. The cause might have to function through an intention that directly produces the movement; the agent might have to retain the capacity to adjust the movement in response to changed circumstances; or the reason might have to participate in a skill-sensitive control system. Each repair faces a further question about necessary conditions. Some intentional actions are hasty, clumsy, or performed despite poor monitoring. An account so demanding that every successful movement requires reflective control would exclude ordinary cases of impulsive or automatic yet intentional action. The aim is to specify enough guidance to distinguish acting from merely suffering the causal effects of thought.

Secondary Deviance: When the Outcome Outruns the Plan

Secondary deviance arises when the agent intentionally initiates an act but the intended outcome occurs by an abnormal causal route. Davidson describes an assailant whose shot misses the intended victim, yet causes a disturbance that sets other events in motion and results in the victim's death. The assailant may have intentionally fired and intentionally tried to kill. It is doubtful that the eventual killing is intentional in the same sense when it occurs through the unforeseen intervening route (Davidson, 1973/2001; Piñeros Glasscock & Tenenbaum, 2023).

This case forces a distinction between intending an outcome, bringing it about causally, and bringing it about in executing one's intention. Intentions can be fulfilled extensionally in surprising ways, but intentional action is not exhausted by the favorable coincidence of outcome and goal. A plan, however incomplete, constrains some aspects of the route to success. That constraint need not require detailed prediction: someone can intentionally win a game without anticipating every move, or intentionally wake a friend without knowing the exact acoustic mechanism. The relevant question is which causal variations remain within the agent's practical control.

The challenge is particularly sharp for indirect action, where much of the causal work occurs beyond the agent's immediate bodily control. Sending a message, setting a device, or initiating a legal procedure can intentionally bring about outcomes through other people and institutions. A criterion restricted to predictable immediate movements would be too narrow. A criterion admitting every remote consequence of an intentional first act would be too broad. The causal theory needs an account of planned, authorized, or otherwise intentionally adopted routes to consequences.

Frankfurt's Guidance Objection

Harry Frankfurt's The Problem of Action (1978) argues that an account focusing exclusively on what causes movement before it begins cannot capture the difference between the activity of an agent and a bodily process that merely happens within the agent. What matters is whether behavior is guided by the agent while it occurs. Frankfurt's intervention shifts attention from causal ancestry to the agent's ongoing participation in performance (Frankfurt, 1978, pp. 157–162).

Guidance is not identical with conscious surveillance. A trained pianist can regulate timing and pressure without inspecting each muscular contraction. A driver can make a series of sensitive adjustments while concentrating on traffic rather than rehearsing an explicit reason for every movement. Such cases indicate that practical governance can be distributed among intentions, learned abilities, perceptual information, and automatic subroutines. The question is whether these processes are organized as the agent's doing.

Frankfurt's challenge need not force abandonment of causal explanation. The relation that sustains and regulates a movement may itself be causal. A strengthened causalism can include both antecedent initiating causes and contemporaneous control relations. This repair, however, changes what the theory claims to reduce: if guidance is explained by invoking agency itself, the account risks circularity; if it is explained by a purely physiological feedback loop, it risks mistaking any self-regulating mechanism for an intentional agent. The central task is to connect ongoing causal regulation with the practical significance of what the agent is undertaking.

Mele's Proximal Causes and the Production of Action

Alfred Mele's Springs of Action (1992) investigates the roles of belief, desire, reason, intention, and action etiology while defending the explanatory significance of intentional mental states. A promising modification of early causalism is to require appropriately proximate causes rather than any distant causal influence. On such an approach, an operative intention must contribute to producing the action through a sufficiently direct and relevant pathway (Mele, 1992).

Proximity improves discriminatory precision. A thought that produces anxiety, which then produces a spasm, is not related to movement in the same way as an intention that mobilizes an action-executing system. Yet physical or temporal closeness alone cannot determine proximity in the relevant philosophical sense. A series of ordinary neural and motor intermediaries may separate intention from movement. Conversely, a panic mechanism can be physically immediate while remaining agentively inappropriate. The relevant proximity is functional, characterized by how an intention is integrated into mechanisms that organize performance.

Mele's work also illuminates an explanatory tension. Highly general causal laws may be unnecessary for understanding a particular agent's action, while an unstructured invocation of the person's intention says too little about the causal route. The philosophical point is to locate the psychological states in a credible etiology of intentional behavior. Proximity, guidance, and reason responsiveness are different attempts to articulate that etiology; none can be accepted as sufficient without testing difficult cases.

Planning, Persistence, and Temporally Extended Agency

Much human action is extended across time. Someone undertakes research, builds a house, learns a language, or repairs a relationship. Such activity involves pauses, subplans, shifting information, and local correction. A simple formula relating an antecedent desire to a final movement loses the structure of the performance. Bratman's planning theory treats intentions as stable though revisable commitments with roles in coordinating practical reasoning and action (Bratman, 1987).

Plans make the causes of action temporally organized. An intention to complete an experiment may constrain which opportunities the agent considers, provide criteria for deciding between methods, and remain operative through sequences of subordinate movements. The persistent intention need not be a continuously conscious event. Its causal relevance can manifest through dispositions, decisions, expectations, and the selective organization of activity across intervals. Such a picture also helps explain how a person can act intentionally in circumstances where a specific bodily trajectory was never represented in advance.

A difficulty remains concerning success conditions. A plan can be abandoned rationally, modified prudently, or disrupted unintentionally. Whether a particular deviation destroys the intentional character of an action depends on which end and which means-description are at stake. Planning theory thereby enriches causalism with practical organization, while retaining the need for a theory of how an adopted plan is realized in a performance rather than merely present among the agent's mental states.

Velleman's Challenge: Where Is the Agent?

J. David Velleman's What Happens When Someone Acts? (1992) argues that the familiar sequence from desire and belief through intention to bodily movement leaves the agent's role insufficiently accounted for. It is possible to tell a causal story about mental states and movements while still feeling that the person has disappeared from the explanation. Velleman's challenge is not a demand to discard causation; it is a demand to explain how the causal history expresses the agent's activity rather than treating mental states as interacting parts of a machine (Velleman, 1992, pp. 461–481).

One way of sharpening the point is to ask whether intentional states act in accordance with an agent's practical grasp of what is occurring. The agent is not identical to a single desire, and competing motives may move the body in ways the person neither endorses nor recognizes as their practical pursuit. On the other hand, identifying agency wholly with reflective endorsement threatens to exclude habitual and unreflective but intentional action. The critique therefore compels a distinction between the possession of mental states and their integration into the person's exercise of capacity.

Velleman's work invites a general lesson for causalism. Successful explanation requires a relation between motives and performance that is attributable to the agent as an organized practical system. This attribution need not entail a metaphysical agent-cause separate from all events. It does require an account of how intentional organization emerges at the level of the acting person. Event-causal explanations become philosophically substantial when they show that organization, rather than merely naming the mental events that precede bodily motion.

Teleological and Noncausal Alternatives

A teleological explanation presents behavior in light of an end. Someone makes a shelter to obtain protection, visits a friend to provide assistance, or enters a room to retrieve a book. Such explanation is familiar in ordinary understanding and has a historical inheritance from Aristotle and Anscombe. The crucial issue is whether its intelligibility depends on a causal relation between the end-directed mental states and the performance. Scott Sehon's Teleological Realism (2005) defends a nonreductive, noncausal conception of common-sense psychological explanation organized through purposes and reasons.

The noncausal position has a genuine attraction. It emphasizes that reasons are understood in a normative and interpretive framework; their adequacy is not settled by detecting a physical pathway. Practical explanation is sensitive to context, standards of coherence, and the agent's grasp of an end. It can explain intelligibly even when the microscopic causal mechanism remains unknown. But the causalist's question returns: how does an explanation distinguish an end genuinely pursued from an end that simply could have rationalized the behavior? The rival must provide its own account of the actuality of acting for that end.

Another alternative attributes a primitive causal role to the agent, rather than taking reasons and intentions to be the complete immediate causes. Event-causalists can accept the importance of the person while denying that this requires an irreducible additional cause. The comparative question is which view gives the more adequate account of practical explanation, control, and responsibility. Neither the claim that every explanation must be causal nor the claim that purposive meaning makes causation irrelevant is established by the mere existence of successful everyday reason explanations.

The Logical Connection Argument Reconsidered

A historically influential objection claimed that a reason and an intentional action are logically or conceptually linked, whereas causes and effects are distinct occurrences related contingently. Therefore, the objection continued, a reason cannot be a cause of the action it rationalizes. Mid-twentieth-century discussions associated with Melden and related critics gave this challenge prominence, and Davidson's 1963 article directly contests the inference (Melden, 1961; Davidson, 1963).

The argument contains an important observation and an invalid transition. If someone wants shelter and believes entering a building will supply it, the rationalization of entering appeals to a conceptual connection between the descriptions of desire, belief, and action. But that relation among descriptions is compatible with a causal connection between concrete psychological and bodily events. Conceptual intelligibility concerns what makes the explanation apt; causation concerns whether the cited reason actually made the relevant difference. Two kinds of relation can characterize one explanatory episode.

This response does not establish that all causal explanations are rationalizations. A loud noise may cause someone to flinch without providing a reason for doing so. Nor does it settle whether reasons understood as facts, rather than attitudes, can literally be causal relata. The significance of Davidson's answer is limited but decisive: conceptual connections do not by themselves exclude causal relevance. A more sophisticated opposition must contest the explanatory necessity, ontological placement, or proposed sufficiency of those causal relations, rather than relying on a bare incompatibility.

Causal Explanation, Normative Reasons, and Justification

The modern philosophy of reasons distinguishes motivating explanations, normative justifications, and the epistemic grounds for believing that a person has a reason. Maria Alvarez's Kinds of Reasons (2010) develops an extensive analysis of this terrain. The distinctions affect causalism directly. What counts as an agent's motivating reason may be characterized through beliefs, apparent facts, or recognized circumstances; what justifies acting may be a fact that the agent did not grasp; what caused a movement may include processes that are neither a motive nor a justification (Alvarez, 2010).

Consider someone who opens a door believing that a colleague is in danger. The belief can explain an intentional rescue attempt even if there is no danger. If danger is genuinely present but the agent is unaware of it, the fact may support the act's objective appropriateness without being the operative motive. If a flash of light triggers the arm movement that opens the door, the resulting event may causally save the colleague without constituting an intentional rescue. Each description has a separate truth condition.

A strong causalist should defend a thesis about why an action occurred, not assert that causal success makes its motives good. Rationalization makes action intelligible from a standpoint, while justification evaluates it against reasons that can exceed the agent's standpoint. Confusing the two would obscure responsibility: a reprehensible act can be intentional and caused by the agent's reasons; a virtuous outcome can occur accidentally without the corresponding intention. The causal theory belongs within a broader practical philosophy that includes norms, knowledge, criticism, and character.

Phenomenology, Practical Knowledge, and the Experience of Doing

The experience of agency concerns how acting presents itself to the agent: as trying, controlling, exerting, adjusting, or achieving something. That experience can be related to causal processes, but it is not a transparent record of them. An actor may feel fully in command while overlooking a crucial environmental factor, or may perform a skilled action with little explicit awareness of its component movements. The philosophical question is what role first-person awareness plays in distinguishing intentional doing from passive occurrence.

Anscombe's notion of practical knowledge offers a direct challenge to purely observational accounts. Someone intentionally writing a sentence typically knows what she is doing through her participation in writing, even before examining the finished page. The knowledge is embedded in purposive activity and may be imperfect in detail. Searle's account of intention in action links such first-person phenomena to the causal self-reference of intentional states, though his framework is philosophically different from Anscombe's (Anscombe, 1957/2000; Searle, 1983).

Elisabeth Pacherie's work on the phenomenology of action develops a layered perspective, relating distal planning, more proximal action specification, and motor-level processes to the experience of agency (Pacherie, 2008). This suggests that the phenomenology of doing has several temporal scales rather than a single inner signal. The result for causalism is methodological: phenomenological distinctions can help locate plausible control structures, while neither their presence nor their absence provides a complete independent proof of the causal route.

Cognitive Science, Hierarchical Control, and Motor Agency

Empirically informed philosophy complicates the assumption that intentional action is directly produced by one explicit propositional state. Planning, motor preparation, sensory prediction, error correction, and attention can operate on different timescales. The connection between a general intention to pick up an object and the actual trajectory of a hand may pass through multiple representations and feedback loops. These processes provide plausible candidates for the causal organization that simpler theories left unspecified (Pacherie, 2008; Shepherd, 2021).

A layered model can distinguish a distal goal, a proximal selection of an act under current circumstances, and motor-level regulation of its execution. The distinctions should not be interpreted as fixed compartments present in every action. Their explanatory value lies in recognizing that achieving the same goal under changed conditions may require responsive alterations of movement. Such responsiveness gives substance to the idea of guidance: if the cup moves, the reaching hand can adjust; if the route closes, the walker can find another route while preserving the plan.

Joshua Shepherd's The Shape of Agency (2021) joins philosophical analyses of control, nondeviant causation, intentional action, and skill. Its importance for causalism is that it offers an alternative to treating the deviance problem as a demand for a magic single link from intention to movement. The relevant organization may involve competence and flexible control across many interacting causal components. This does not prove that cognitive-scientific mechanisms themselves settle intentionality. Their relevance depends on bridging a descriptive mechanism with the content of the agent's practical aim.

Collective, Institutional, and Artificially Mediated Action

Actions can be distributed through institutions and technical systems. A person authorizes a transfer, instructs a team, submits a legal filing, or commissions a machine to execute a task. The causal chain then includes other decision-makers, software, rules, and material events. A narrow bodily-movement account misses what is intentionally done through these structures. The causal theory must distinguish the initiating agent's act, the effects of the act, and the conditions under which downstream events are attributable to the initiating plan.

In group activity, a participant's reasons may causally explain participation while collective organization explains the outcome. Mere causal contribution is too weak to determine what the group intended or which individuals intentionally brought about the result. Authorization, role, coordination, and expectations govern attribution. These issues connect causalism to social ontology and collective agency, though they demand separate theoretical analysis beyond the basic event-causal model.

Artificially mediated processes sharpen the distinction without deciding the ontological status of artificial agency. A user may intentionally initiate an automated procedure; the system may then generate steps and outcomes whose particular details were never represented by the user. The relevant causal questions concern delegated goals, policy specification, opportunities for intervention, and the causal history of output. A general account of action should be capable of examining nonbiological systems through clearly established capacities and institutional roles, while leaving contested claims about artificial intentionality to their own arguments and canonical contexts. Within the philosophy of the Artificial Era, this is a genuine research extension of the action problem, not a retroactive reinterpretation of Davidson or Aristotle.

A Critical Reconstruction of the Strongest Causalist Position

The strongest available event-causal thesis is more specific than the claim that mental states happen to precede movements. It says that an action is intentional under a description when a practical representation relevant to that description contributes to the performance through a causally organized route that constitutes or realizes the agent's purposive control. Such a thesis gives causation a difference-making role while incorporating the explanatory lessons of planning, guidance, and deviance. It remains a reconstruction of the debate rather than a reported proposition adopted word for word by any one historical philosopher.

We can state the core argumentative progression without pretending it is a deductive proof. First, a person may have a reason that would rationalize a behavior without acting for that reason. Second, an accurate motivating explanation must identify why the actual episode occurred as it did. Third, appropriately operative practical states provide causally relevant differences between genuine action and coincident behavior. Fourth, mere causal relevance fails to distinguish deviant chains. Fifth, an adequate account must locate the cause within organized guidance, planning, or reason-sensitive performance. The conclusion is a qualified but robust causalism: action explanation requires both the causal selection of the operative motive and the agentive organization of the route.

Three levels of evaluation follow. At the interpretive level, an explanation identifies the description under which the agent acted for a reason. At the etiological level, it shows that the relevant practical states were causally operative. At the control level, it shows that their operation belonged to an organized process of doing rather than to a symptom or coincidence. These levels are analytically distinguishable, though actual action may integrate them. The framework reveals why both a purely intentional vocabulary and a purely mechanistic causal vocabulary can leave part of the phenomenon unexplained.

This reconstruction is strongest when it allows fallibility. An agent can act intentionally on a mistaken belief, perform badly, or fail to achieve an intended end. Control is not guaranteed success; it is the structured opportunity for an agent's practical orientation to guide performance within available capacities and circumstances. The claim that a particular movement was intentional must therefore be supported by relevant evidence about the agent and the process, rather than inferred merely from either favorable outcome or prior mental content.

Open Problems and the Conditions of Explanatory Success

Several fundamental problems remain open. There is no universally accepted noncircular specification of nondeviant causation that covers spontaneous movements, skilled execution, extended undertakings, omissions, and indirect consequences. Guidance conditions often rely on intuitions about what the agent would have done under counterfactual changes, but the relevant counterfactual range is itself contested. A precise control theory must say which changes count as tests of agency and why (Frankfurt, 1978; Shepherd, 2021; Piñeros Glasscock & Tenenbaum, 2023).

The ontology of motivating reasons also remains disputed. Are operative reasons psychological states, facts the agent recognizes, intentional attitudes characterized by contents, or complexes spanning several kinds of relata? Davidson's account privileges mental states and events, while reasons-first approaches question whether an agent's explanation should be translated into a story about internal attitudes. Causalists need to clarify the relation between what a person takes to be a reason and the entities that exercise causal influence.

Another problem is the scope of the theory. Some bodily control is intentional without explicit deliberation; some deliberation never produces action; some intentionally initiated projects contain autonomous subroutines and other people's interventions. The causal theory must explain action across these scales without reducing all examples to the most convenient one. There is also an empirical question how cognitive and motor models could establish that a hypothesized mechanism is causally responsible for performance in a particular case. Philosophical criteria and laboratory evidence support one another, but neither substitutes automatically for the other.

Finally, agency raises questions about responsibility and justification that transcend causal etiology. The mere fact that motives cause an action does not determine whether the agent could have acted otherwise, whether reasons are normatively adequate, or whether the act is blameworthy. The causal theory of action can be a foundational component of an account of responsible agency without becoming a complete ethics or theory of free will.

Conclusion

The causal theory of action emerged from a historical effort to explain the relation between practical understanding and the events people bring about. Aristotle identified the agent as the origin of voluntary movement and differentiated choice from the wider class of voluntary conduct. Early modern philosophy sharpened the motivational role of passions and practical beliefs. Anscombe organized modern action theory around intentional description, reasons, and practical knowledge. Davidson's decisive contribution was to insist that, when an agent acts for a reason, the reason must do more than make a hypothetical action intelligible: it must explain the actual occurrence of that agent's performance through a causal relation.

The subsequent debate exposed the limits of interpreting that causal relation as a simple antecedent mental state followed by a bodily event. Causal deviance demonstrates that practical states can cause predicted outcomes without making those outcomes intentional. Frankfurt directed attention toward the guidance of ongoing performance. Mele clarified the structure of action etiology; Searle developed the causal self-reference of intention; Bratman supplied the temporally persistent architecture of planning; Velleman challenged explanations in which the agent disappears behind interacting states; work on phenomenology and control provided resources for integrating the practical, cognitive, and motor aspects of action.

The philosophical synthesis is a threefold explanatory requirement. A reason must be operative rather than merely available; its causal operation must be connected to an intelligible intentional description; and its realization must belong to an adequately integrated episode of agentive production or guidance. These are related dimensions of one explanation, and a theory succeeds to the extent that it specifies their relation across ordinary and difficult cases. The remaining dispute concerns whether guidance can be fully analyzed in event-causal terms, how nondeviance is identified, and what counts as the agent's practical contribution to an extended chain of events. The lasting achievement of causalism is to turn the truth of an action explanation into a serious philosophical question about why this action actually happened as the action of this agent.