Action and Event

The Ontological Question

An action makes a difference in the world. A person lifts a hand, opens a window, walks toward a door, signs a document, or interrupts an argument. Each description seems to pick out something that happens. Yet a falling branch also makes a difference in the world, and a leg can move because a physician stimulates a muscle. The general problem is to specify what an action is and how its status as an action relates to its occurrence as an event.

For present purposes, an event is a particular occurrence situated within temporal reality and available, at least in principle, to descriptions in terms of its participants, circumstances, causal connections, and development. An action is an exercise of agency describable as something an agent does. These working definitions identify the theoretical task without deciding its solution. On many views an action is a kind of event; on others action is primarily a process, an exercise of capacity, or a manifestation of practical understanding. The questions are connected but logically separable: what makes an occurrence an action, what individuates one action from another, how an action endures or culminates, and what descriptions make an action intelligible as intentional.

The question “Are all actions events?” must be interpreted with care. It can mean whether every token of doing is numerically identical with some event; whether events furnish a sufficient vocabulary for explaining acting; whether action-events possess essential agentive features; or whether processes, omissions, and attempts occupy the same ontology. A positive answer to the first need not settle any of the others. A person can accept that a hand-raising is an event while denying that a purely physical specification of arm displacement captures everything needed to explain the person raising an arm.

This dossier addresses the intersection of philosophy of action and metaphysics. It examines the identity and explanatory structure of positive actions, the relation between agency and temporal occurrence, and the difficulties posed by omission and extended activity. The more general ontologies of events, complete theories of free will, and individual theories of collective intention remain independent research subjects; they enter here where they bear directly upon the identity or existence of actions. The distinction between an event token and a description of that token will guide the investigation throughout (Davidson, 1963, 1967, 1969; Bennett, 1988; Casati & Varzi, 2025).

Action in the History of Practical Philosophy

The modern event theory of action developed in a history of questions about moving, doing, deliberating, and causing. Aristotle provides an early systematic framework. In Nicomachean Ethics III.1–5, he asks when action is voluntary, what makes ignorance exculpatory, and how choice relates to desire and deliberation. The origin of voluntary action lies in the agent, who knows the relevant particulars (1110a–1111a). Choice (prohairesis) is deliberative and narrower than the voluntary; Aristotle explicitly recognizes voluntary conduct among children and other animals while distinguishing it from choice (1111b). These criteria connect action to source, cognition, and practical orientation rather than treating a physical occurrence alone as sufficient for voluntary agency.

Aristotle's distinction between praxis and poiesis is also relevant. In Nicomachean Ethics VI.4–5 (1140a–1140b), production has a product beyond its activity, whereas good action concerns the manner and end of doing. The contrast anticipates a recurring question: should action be individuated by the bodily process, the aim under which it is conducted, or the result accomplished? Aristotle's account is not a twentieth-century theory of event particulars. His analysis concerns the practical intelligibility of voluntary conduct and the organization of activities by ends. Reading a later event ontology directly into Aristotle would obscure the difference between these projects (Aristotle, Nicomachean Ethics, III.1–5, VI.4–5).

The early modern study of will and causation diversified the field. Spinoza's Ethics III analyzes human activity through an account of adequate causation and the distinction between acting and being acted upon. In III, Definition 2, an agent acts insofar as something follows from its nature that can be understood through that nature alone; otherwise it is passive. This is a metaphysical and explanatory criterion, rather than a test of whether a bodily motion was preceded by a separate conscious episode of willing. Hume's treatment of voluntary motion in A Treatise of Human Nature II.iii and the Enquiry Concerning Human Understanding VIII connects liberty to the power to act or not act according to the determinations of the will, placing action within the broader dispute over necessity. These philosophers disagree about causation and freedom, but they make plain that a vocabulary of happenings acquires different significance when the happening is ascribed to an agent.

The nineteenth- and twentieth-century transformation of the field involved both the philosophy of language and problems of mind. Description became central: is an intentional raising of an arm a movement plus an inner act, an occurrence described in a distinctively practical way, or a causal sequence consisting of beliefs, desires, and movements? Gilbert Ryle's critique of an inner-command picture in The Concept of Mind and Wittgenstein's observations on willing shifted attention toward the grammar of saying what someone does. Anscombe and Davidson then made action description and event identification fundamental terms of analytic debate. The modern question acquired its distinctive force precisely because events appeared capable of doing explanatory work once assigned exclusively to elusive acts of will.

Wittgenstein's Subtraction Question

Wittgenstein's Philosophical Investigations §621 presents a striking comparison: when I raise my arm, my arm rises; what remains when the rising of the arm is subtracted from my raising it? The German distinguishes “ich meinen Arm hebe” from “mein Arm sich hebt.” The formulation invites scrutiny of the assumption that a bodily motion becomes an action through the addition of a second inner occurrence. Surrounding sections complicate that assumption. At §614 Wittgenstein observes that one does not normally employ an instrument in order to raise an arm voluntarily, and at §622 that raising an arm does not normally require an attempt to raise it.

The temptation is to add a volition to motion: motion plus mental act equals action. That formula has explanatory costs. If the extra volition is itself an action, the theory owes an account of what distinguishes that mental action from merely undergoing a mental event. Repeating the addition creates a regress. If the volition is an event that the agent merely undergoes, it remains unclear how its occurrence supplies the missing element of active agency. If it is an independent conscious feeling, cases in which movements are produced, inhibited, or misperceived undermine its generality. Wittgenstein's remarks encourage attention to established criteria of acting and the contexts in which agency descriptions apply.

The subtraction question is often recruited for causal theories, but Wittgenstein's text does not simply assert that the missing term is an intention-event. His concern includes the misleading picture generated by subtraction itself. A difference between “the arm rises” and “I raise it” may be found in what is truly said about the whole occurrence and its place in the agent's practical activity, without being a separable component that remains when physical motion is removed. Modern analyses disagree about precisely how to develop the insight; the historical text establishes a question rather than an unambiguous positive theory (Wittgenstein, 1953, §§614–628; Tanney, 2022).

A successful theory consequently needs two answers. First, it must describe the difference between an agent raising an arm and having that arm moved from outside. Second, it must explain why the difference need not be a second action hidden inside the first. Event theories gain their initial plausibility when they identify the action with a single publicly located occurrence while making the agent's reasons, capacities, and relations to that occurrence part of its explanation.

Anscombe: Action Under a Description

G. E. M. Anscombe's Intention (first published in 1957) reorganizes the subject around three connected uses of intention: expressions of intention for the future, intentional action, and intention in acting (§1). Her criterion for an intentional action concerns a distinctive application of the question “Why?” (§§5–19). An answer can disclose what the agent takes herself to be doing or what end guides the doing. The answer need not be a report of an observed inner event. Anscombe's notion of practical knowledge, especially §§28 and 45–48, describes knowledge manifested in the performance itself rather than inferred as a spectator would infer it from observation.

The example of the man operating a pump is pivotal (§23). He moves his arm, operates the pump, replenishes a house's water supply, and—under further circumstances—poisons the residents. There can be several descriptions of one connected performance, and intentionality may attach to some descriptions and not others. If the agent aims to poison, that description can enter the practical explanation; if he inadvertently frightens an animal, the latter description may be true without describing what he intended. The same physical trajectory can thus stand in different explanatory relations depending on its description.

Anscombe does not thereby offer a simple numerical formula for counting actions. She asks whether multiple true descriptions represent multiple acts, and her treatment resists an automatic multiplication of actions with phrases. More important, practical knowledge is not a retrospective list of events one has noticed. To know what one is doing intentionally can be to know the practical order through which means are directed toward an end, even while the final outcome is uncertain. A person repairing a chair can know what she is doing while repair is ongoing and before she knows whether it will succeed.

The distinction between theoretical and practical knowledge helps locate the difference between action and happenstance. Someone may recognize that a window is opening because of a gust; another can answer why she is opening it by citing the need for ventilation. The latter explanation does not depend on invisible forces. Its intentional object, means-end relations, and capacity for correction characterize the exercise of agency. Anscombe's work therefore constrains event ontology: an adequate event theory must preserve the difference between describing a happening and describing what an agent knows herself to be doing (Anscombe, 1957/2000, §§5–6, 19, 23, 28, 45–48).

Davidson: Reasons and Causes

Donald Davidson's “Actions, Reasons, and Causes” (1963) develops the influential claim that rationalizing explanation is a species of causal explanation. A primary reason joins an appropriate favorable attitude toward actions of a certain kind with a belief that one's particular behavior falls within that kind. Such a reason explains an action when it actually causes it in the relevant way. Merely being able to invent a plausible justification after the event does not establish that the agent acted for that reason.

Davidson's switch example locates the issue. An agent flips a switch, turns on a light, illuminates the room, and inadvertently alerts a prowler. These are distinct true descriptions of a connected occurrence, and at least some can refer to the very same particular action. The agent's reason for illuminating the room need not rationalize alerting the prowler. Thus an event can be intentional under one description and unintentional under another. This formulation explains intentionality without requiring that the event itself bear intention as an undifferentiated intrinsic property (Davidson, 1963, pp. 685–700).

The argument has a discriminating structure. If a reason is merely an interpretative description, it can potentially rationalize indefinitely many actions and cannot determine which actually occurred because of that reason. Causal connection selects among competing candidates. But causation alone is too broad: a panic caused by a belief may accidentally generate a movement that fulfills the belief's goal, without that outcome being intentional. Davidson's theory begins the contemporary causal program while leaving the challenge of the right causal route for later investigation.

Davidson's essays on agency, action sentences, and event individuation develop the metaphysical background. An action is a concrete occurrence in the world, subject to causal explanations like other occurrences, while the interpretation of intentionality remains description-relative. This brings together the physical presence of acting and the rational explanatory standpoint. The account is influential because it gives causation a role without identifying reasons with physical pushes or treating the truth of every action description as something separately intended (Davidson, 1967, 1969, 1971/2001).

Actions as Particular Events

An event ontology treats particular acts as occupants of a temporal causal order. If Nora closes a door at noon, that particular closing occurs once, occupies a period, produces and undergoes effects, and is numerically distinct from a second closing later in the afternoon. Its type—closing a door—is repeatable, but the token is not. The distinction between event type and event token prevents confusion between what different people do alike and the events they individually perform. Hornsby (1998) emphasizes that an act-type can be shared between agents although the particular action-occurrences are distinct.

Three considerations support the event account. First, actions can be causes and effects. The closing of a door can interrupt a conversation and itself result from a warning. Second, action descriptions admit temporal and manner qualifications: the door was closed quietly and at noon. Third, an action can be remembered, witnessed, interrupted, completed, or recorded as having taken place. These predicates suggest reference to something occurrence-like. The metaphysical question is whether this explanatory convenience entails identity with events or merely representation by event vocabulary.

Event realism encounters questions about the constituent structure of events. Are they unstructured particulars, exemplifications of properties by agents at times, concrete regions of spacetime, or something more complex? Davidson favored relatively coarse-grained event particulars; Kim's property-exemplification framework can multiply distinct events corresponding to distinct constitutive properties; Bennett asks how much apparent metaphysical disagreement stems from differences between names and events. The ontological debate cannot be solved by declaring every grammatical gerund a separate entity. A name can change while its referent remains fixed, and two descriptions can differ in truth conditions without establishing that they denote different occurrences (Davidson, 1969; Kim, 1976; Bennett, 1988).

A further challenge concerns agency's relation to eventhood. If events are causally linked particulars, what qualifies some as actions? The answer may cite a specific origin in a competent agent, the expression of intention, the guidance of reasons, the exercise of a power, or an achieved relation to a practical goal. These are contested candidate conditions. Even if every action is an event, the predicate “is an action” cannot be replaced by “is an event” in any explanatory account, since collapsing the predicates would lose the very phenomenon to be investigated.

The Logical Form of Action Sentences

Davidson's “The Logical Form of Action Sentences” (1967) advances a semantic argument for event quantification. Consider the inference from “Mira opened the window slowly at dawn” to “Mira opened the window.” If manner and temporal modifiers are conjunctive predicates of an existentially quantified event, the inference becomes transparent. On a simple event-semantic representation, the first sentence asserts that there is an event e that is a window opening by Mira, is slow, and takes place at dawn. Removing conjuncts yields the less informative sentence while preserving reference to an event.

The formal pattern may be represented schematically as ∃e [Opening(e) ∧ Agent(e, Mira) ∧ Theme(e, window) ∧ Slow(e) ∧ AtDawn(e)]. A restricted version, ∃e [Opening(e) ∧ Agent(e, Mira) ∧ Theme(e, window)], follows in ordinary classical logic. This notation is an analytic reconstruction, not a claim that Davidson used exactly this modern vocabulary. The validity of the inference depends on the predicates and their logical roles, not on counting a distinct act for each modifier.

This argument provided an enduring resource for philosophical semantics and theoretical linguistics. It allows descriptions of manner, instrument, location, and time to be treated as information about one described occurrence. But linguistic adequacy does not settle all ontological disputes. One could admit events into the semantics of action sentences while holding that agency is best understood as a process, power, or practical relation. Conversely, a metaphysical defender of events could reject one specific proposed semantic decomposition. The argument must be weighed as evidence for a useful event ontology, rather than as a demonstration of the full nature of agency (Davidson, 1967; Casati & Varzi, 2025).

The Individuation of Actions

The individuation problem concerns how many particular actions an agent performs when several descriptions apply. Suppose I push a button, ring a bell, alert the staff, and inadvertently interrupt a lecture. Does one action produce many descriptions or does each new property determine a different action? The answer affects the identity conditions of action, causal explanation, responsibility, and ordinary counting practices. An analysis must specify what sameness means: same causal history, same bodily movement, same accomplishment, same spatiotemporal region, same agent, or same constitutive property.

A coarse-grained view often identifies the button push with the bell ring as one event under multiple descriptions. A fine-grained view might separate the pushing and the ringing because they exemplify distinct action-properties. A component or hierarchical approach can instead place several events in a structured action containing subactivities and consequences. Such approaches can capture complex performance but require identity rules for parts and wholes. Some divergence reflects different explanatory projects: a forensic report may count separate legally relevant acts while a metaphysical account seeks the number of causally individual occurrences.

A strong test is counterfactual difference. If the button failed, the push would still occur while the bell ring would not. Does that prove that the actual pushing and actual ringing are numerically distinct? It shows that the descriptions have different application conditions. Whether it proves distinct event identity depends upon a further modal principle: that numerically identical events could not differ in the conditions under which those descriptions remain true. That principle itself requires defense, because the same event can possess accidental properties, including consequences that might have been absent (Davidson, 1969; Goldman, 1970; Bennett, 1988).

The choice of ontology should preserve the difference between an event's identity and its significance. It may be one action that simultaneously meets several descriptions; it may also be several coordinated phases of an extended task. A satisfactory theory therefore needs criteria for both token identity and legitimate decomposition. Neither multiplication by verbal description nor automatic unification of everything in one causal chain adequately establishes either criterion.

Coarse-Grained and Fine-Grained Ontologies

Davidson's relatively coarse-grained ontology gives considerable latitude to redescribing a single event. His event arguments connect identity to causal roles, while his treatment of action individuation allows one action to be an intentional flipping of a switch and an unintentional alerting of a prowler. The gain is parsimonious causal explanation. The cost is pressure to explain how an event's intentional status varies with description without allowing the metaphysics of particular action to dissolve into semantics.

Fine-grained proposals associated with Kim and Goldman distinguish events or actions according to their constitutive properties. On a schematic property-exemplification account, an event is an agent or object exemplifying a property at a time, possibly with further constitutive relata. If moving one's finger and ringing a bell involve different constitutive properties, distinct event particulars may be admitted. This captures modal differences and respects intuitions that a particular achievement depends upon something more than bodily motion. But it threatens excessive proliferation: every newly specified property may determine another event unless the theory identifies a principled distinction between constitutive and incidental properties (Kim, 1976; Goldman, 1970; Casati & Varzi, 2025).

Jonathan Bennett's Events and Their Names (1988) directs attention to how event terms and descriptions track metaphysical particulars. Two nominal expressions can differ semantically without identifying two worldly items. Conversely, the co-reference of some expressions cannot be secured solely from their location in a causal story. The crucial discipline is to specify whether the controversy concerns event identity, the semantics of action descriptions, explanatory relevance, or the metaphysics of properties. Without that discipline, theoretical disagreements can be manufactured by shifting between levels.

The present investigation therefore adopts a cautious principle of individuation: count occurrences according to independently argued identity conditions, and count descriptions according to their truth and explanatory conditions. The former cannot be inferred mechanically from the latter. This principle does not legislate a universal event ontology. It establishes the analytical priority of keeping identity, description, constitution, and explanation explicit before adjudicating between coarse- and fine-grained theories.

Bodily Movement, Trying, and Basic Action

The image of action as bodily movement has an immediate attraction. The agent's arm rises, muscles contract, the hand contacts a surface, and a visible change occurs. But different actions can share bodily motions, and action can occur without a successful movement. If an agent tries to open a locked door, the attempt is intelligible even when the door remains closed. If an agent tries to move a paralyzed limb, the occurrence to be explained may involve sustained effort without the intended outward motion. A theory identifying all actions with completed bodily trajectories faces a question about the status of trying.

The concept of a basic action, developed in discussions by Danto, Goldman, and subsequent theorists, concerns doing something without doing it by means of doing something else. Moving a finger might be basic relative to pressing a switch; pressing a switch is a way of ringing a bell. The relative hierarchy captures the agent's practical organization of means and ends. Yet the classification is not always absolute. An accomplished musician can execute a phrase without separately deciding each finger movement; a novice may need deliberately to initiate each movement. What an agent does basically can depend on capacities and the grain of the description, even if an underlying physiological event sequence is shared.

Hornsby's Actions (1980) complicates the equation of acting with movement by locating some actions in the causes of bodily movements. On her account, the relation between agentive trying and outward movement becomes decisive. This proposal deserves attention because causal predecessors of bodily motion can be events too. An action-event theory need not always choose the displacement of the body as its preferred action-event. Nevertheless, moving the action inward to a prior episode is not a universal solution: a purely internal event account still has to distinguish an exercise of agency from an involuntary episode which happens to cause motion.

Ruben's The Metaphysics of Action (2018) provides a more recent route through trying, doing, and causing. His account takes seriously actions which occur in the world and the possibility that certain actions, such as opening a window, are identical to the events of the window opening. This challenges an inherited restriction of action identity to the bodily surface. It also highlights a distinction between one's doing something and one's causing something to happen. The fact that the body initiates a causal chain does not by itself settle which event is the action and which is an effect of it (Ruben, 2018).

The lesson is conceptual. Trying has conditions of existence tied to effort or practical orientation; successful performance has additional achievement conditions; bodily movement has kinematic conditions. Those descriptions may coincide in ordinary cases but can separate under failure, obstruction, distraction, and assisted performance. The problem of action and event is therefore best pursued with a clear distinction among an agent's exertion, the controlled performance, and the achieved worldly result.

Deviant Causation and Agentive Control

A causal theory of action needs more than the presence of a belief, desire, or intention somewhere in the event's ancestry. Imagine an actor who intends to enter a theater to confront someone. His anticipation causes debilitating nervousness, which makes him stumble through the entrance. The movement gets him inside and has been caused, in a broad sense, by the very intention to enter. Yet the person has not entered intentionally in the ordinary manner of carrying out that intention. The causal pathway has departed from guidance by the reason.

Such examples reveal an explanatory gap within the standard causal account. One must distinguish causation of a movement from causation constitutive of the agent's doing it for a reason. A reason can causally contribute to an event without guiding it. Any theory that states only “the right reason caused the event” leaves open what the word “right” is supposed to require. Control, sensitivity to circumstances, and the relation of an intention to the action's manner of performance are plausible ingredients; each requires its own analysis rather than a verbal stipulation.

Some theorists propose continuous guidance: the action remains answerable to the agent's practical assessment as circumstances unfold. The agent adjusts force, changes speed, corrects deviations, and can stop in response to a relevant reason. The approach handles the unstable character of performance better than a single triggering mental cause. But it can be overrestrictive if formulated as constant conscious monitoring. Skilled actions often proceed with low reflective attention, and some intentional acts are nearly instantaneous. Control is a capacity for appropriately structured regulation, not necessarily the presence of a conscious decision at each moment.

Deviant causal chains also challenge the claim that event identification itself establishes intention. We may know exactly which bodily event took place and which prior mental event caused it, yet be unable to classify the occurrence as an intentional action because the causal route is abnormal. The failure belongs to the theory of agency, not to event ontology considered in isolation. Causal relations can be necessary to one account of action while being insufficient to identify their agentive character (Davidson, 1963; Bratman, 2007; Piñeros Glasscock & Tenenbaum, 2023).

A promising account distinguishes an event's causal production from the organized exercise of an agent's capacity. Events are indispensable to describing what happens, but agentive control sets constraints on which of those happenings count as performance. This distinction also keeps the theory open to ordinary cases in which actions are spontaneous, habitual, or skillfully fluent rather than preceded by explicit deliberation.

Action, Process, and Temporal Development

The event model is often constructed around completed occurrences. Someone flipped the switch, locked the door, or signed the form. Ongoing action confronts us with a different temporal orientation: someone is painting a wall, constructing a house, learning a phrase, or crossing a city. A performance can develop, encounter difficulty, be redirected, or fail to culminate. Process theorists argue that this open temporal character is philosophically basic to action (Stout, 1996; Thompson, 2008; Steward, 2012; Piñeros Glasscock & Tenenbaum, 2023).

Consider the distinction between “she is writing the letter” and “she wrote the letter.” The former may be true even when the latter never becomes true. The developing activity does not guarantee the finished result. As the writing continues, its duration, pace, and eventual success remain unsettled. If the performance stops halfway, it is still intelligible to say that she was writing a letter. A completed-event view can describe what occurred, but the process view insists that the activity's directed, incomplete development is integral to its identity as acting.

One formulation holds that action is a process and that the culminating event is the completion of that process. Another holds that a particular process can also count as an event when viewed as completed. A third denies that all processes should be treated as individual objects. These differences are substantial. If an ongoing process and its resulting event are distinct entities, a theory must explain the nonaccidental relation between the two. If they are one entity described from temporal standpoints, a theory must explain how the identity of a developing occurrence remains determinate while its outcome is open. The issue bears on tense, duration, persistence, and the ontological status of incomplete activities.

Process theory gives the agent an active explanatory position. A person continues the task, varies her approach, responds to interference, and regulates what is happening. The process is shaped by her exercise of capacity throughout its duration. Event theorists can respond that temporally extended events also possess parts, internal causal structure, and stages. An event need not be pointlike. The decisive dispute concerns whether the developing character of intentional performance is fully represented by identifying a temporally extended event or demands a distinct ontological category of processes (Piñeros Glasscock & Tenenbaum, 2023, §5; Steward, 2012).

I conclude provisionally that process description is indispensable even when event ontology is retained. A complete account must explain not only that an action happened but how doing was maintained, changed, and brought to a conclusion. The practical difference between making progress and achieving an end cannot be inferred from mere mention of a completed occurrence.

Accomplishment, Activity, and Aspect

The philosophy of action intersects with linguistic aspect and the classification of eventualities. Activities, accomplishments, achievements, and states behave differently in temporal descriptions. Walking is typically an activity; walking to a particular destination gives the performance an endpoint; reaching the destination names a culmination; knowing the route is a state. The distinctions are familiar from the twentieth-century study of tense and aspect associated with Ryle, Kenny, and Vendler, and they have consequences for the ontology of action.

An accomplishment includes an internal directedness toward a completed condition. “She is building a bridge” is compatible with interruption before there is a bridge. By contrast, “she built a bridge” asserts completion. Achievements are often treated as transitions or culminations, but the boundary is not universally punctual: realizing a mistake may be gradual, and reaching a destination may have an extended approach. Even the activity of walking can be bounded by an externally specified duration. Linguistic class, practical structure, and metaphysical identity must therefore be related carefully.

The so-called imperfective paradox shows why process and event talk cannot simply be interchanged. From “someone was writing a book” one cannot infer that a book was written. A process theory can preserve the reality of an unfinished intentional performance without inventing a completed result. An event theory may likewise represent an incomplete writing event, but it must avoid conflating the occurrence with a successful book-writing accomplishment. What the action is under one description depends partly upon aspect and upon the fulfillment conditions built into that description (Mourelatos, 1978; Thompson, 2008).

These distinctions matter beyond semantics. Responsibility for initiating a dangerous undertaking can arise before its final outcome, and credit for a completed achievement can depend on whether it was intentionally brought about. Likewise, the same causal sequence can contain an activity, a transition, and a resulting state. An ontology which gives each type the same individuation conditions will struggle to account for these different explanatory needs. A disciplined account can instead state which conditions are required for the onset, maintenance, and successful completion of a practically characterized action.

Practical Knowledge and Causal Explanation

Anscombe's account of practical knowledge is often positioned against causalism. The opposition is too crude if it is construed as a choice between knowledge and causes. A person making a table can possess direct practical knowledge of what she is doing while bodily movements, tools, and materials undergo causal changes. The substantive question is how a practical explanation and a causal explanation are related, and whether the former can be derived from the latter.

A reason explanation cites what makes an event intelligible from the agent's perspective. “I am tightening the bolt so that the shelf will hold” specifies an action in terms of a goal and a means. An efficient-causal description can trace muscle contraction, torque, and the mechanical behavior of the bolt. They may explain the same occurrence under different interests and inferential commitments. One account does not automatically eliminate the other; nevertheless, a philosophical theory must identify what connects the practical explanation to the actual causal sequence. The agent's purpose is insufficient if it has no explanatory relevance to what she did.

Anscombe's emphasis on knowledge without observation concerns a distinctive epistemic relation of an agent to her intentional conduct. She does not claim infallibility about every result. A person may know she is opening the window while mistakenly believing that fresh air will enter, and may know she is giving directions while being wrong about the road. Practical knowledge is answerable to what the agent can actually do and to the organization of what she is doing; it is not magical insight into future consequences (Anscombe, 1957/2000, §§28, 45–48).

The causal theorist can acknowledge this epistemic dimension. Causation explains why the event occurred; the practical structure supplies the categories in which some causal relations are recognized as reasons for acting. The challenge is to demonstrate that these categories are themselves enacted, not imposed post hoc. Conversely, advocates of practical knowledge need to explain how error, failed effort, automatic competence, and actions performed under partial ignorance fit the account. Neither a simple causal chain nor a bare assertion of self-knowledge yields a complete theory.

What emerges is an account of explanatory pluralism disciplined by actual agency. A description of the action, its agentive reason, and its causal production can be simultaneously correct. Their integration depends upon control and directedness, not on an insistence that only one philosophical vocabulary names what is real.

Actions, Consequences, and Causal Chains

An intentional action commonly changes circumstances beyond the agent's body. I turn a key, open a door, enter a room, and awaken a sleeping person. Which among these are actions, which are consequences, and which are parts of a wider undertaking? A crude distinction that identifies the action only with the first movement and every later occurrence with its effect lacks the resources to capture ordinary claims that someone opened a door or awakened a person.

Davidson's approach allows a single action to have several descriptions including consequential ones. Such descriptions can refer to one event even when they specify a result beyond the movement that began it. The view is powerful but has a limit: it cannot make every remote consequence of an action numerically identical with the action itself. If I flip a switch and a delayed fault causes an electrical fire, an account needs a principled distinction between what I did and what happened because of it. The distinction is not settled solely by whether the consequence was intended, since unintended outcomes can sometimes figure in the description of what one did (Davidson, 1963).

One way to proceed is to distinguish action token, action description, constitutive result, and causally downstream effect. The token may have a temporally and causally bounded identity. Descriptions may mention immediate or more remote consequences. Some outcomes are constitutive of the successful performance type—opening a door requires a door's coming open—whereas others are contingent downstream effects. A theory must specify why an outcome is constitutive rather than assume that proximity in space or time supplies an answer. This distinction is especially important when the outcome is socially or legally mediated: signing a form, consenting, or making a promise involves rules in addition to motion.

A causal chain may also contain nested actions. Preparing dinner comprises cutting, heating, tasting, adjusting, and serving. The whole performance has a practical unity without requiring all its component events to be numerically one event. The action of preparing dinner and the action of cutting vegetables are related as whole and part under some descriptions and as means and larger purpose under others. Causal connectedness and practical unity are different relations. A systematic ontology of action must permit their interaction without treating them as interchangeable.

The philosophical result is a demand for explicit boundary criteria. The question “What happened?” can receive a causal history; the question “What did the agent do?” requires action characterization; and the question “What resulted?” requires an account of effects. A sound theory of action-events keeps all three questions available.

Omission, Refraining, and Negative Agency

Omissions present a decisive test for a universal action-event thesis. Suppose an employee deliberately refrains from sending an email. There may be no positive event corresponding to the sending that was withheld. The person can nonetheless have a reason, adopt a plan, and be responsible for the omission in some circumstances. If actions are necessarily positive occurrences, then either omissions lie outside action or the ontology must accommodate negative events, absences, and other nonoccurrences.

One strategy equates an omission with something the agent positively does instead. The employee drinks coffee rather than sending the email; perhaps that is the omission. Yet the maneuver is unstable. The employee could lie still, sleep, or perform any of many incompatible activities while withholding the message. Selecting one positive event as identical with the omission depends on arbitrary circumstances unless a further constitutive relation is supplied. The absence of sending does not become a particular coffee-drinking event just because the two coexist.

A second approach treats omissions as absences that can be intentional or agentive without themselves being event particulars. This preserves a distinction between agency and positive-event occurrence. A third seeks a more inclusive event ontology under which certain omissions are events. Debate also concerns whether the causal significance of omission must be explained by positive causal histories or by counterfactual dependencies involving the absence of events. The literature contains several variants, and the general idea that all absences are events would multiply entities with little discrimination (Clarke, 2014; Sartorio, 2009; Payton, 2021; Piñeros Glasscock & Tenenbaum, 2023).

The issue requires a distinction among bare nonoccurrence, omission, and intentional refraining. At any moment a person does innumerable things neither actually nor potentially expected of them. Only some absent performances are omissions relative to a role, norm, commitment, opportunity, or explanatory context. Intentional refraining involves settled practical orientation against performing an otherwise available action. These distinctions show why a grammar of negative descriptions cannot by itself establish a complete ontology of negative events.

An event theorist can give a sophisticated answer by restricting the thesis: positive executed actions are events, while omissions belong to the wider domain of agency and practical responsibility. But this answers a narrower proposition than “every manifestation of agency is an event.” A more expansive theory would need credible event identity criteria for nonoccurrences. Until it provides them, the ontology of omissions remains an important reason to separate acting from the entire range of agentive phenomena.

Reasons, Responsibility, and Normative Description

Action-event identity matters to the attribution of responsibility. Suppose a technician presses a switch while knowing that a safety interlock is disabled. The same movement may be described as starting a machine, exposing coworkers to danger, or violating a procedure. These descriptions are not ethically interchangeable, even when some designate the same particular event. Responsibility depends on what the technician knew, intended, could reasonably foresee, and had authority or duty to do. Numerical event identity alone does not settle those questions.

The distinction becomes sharper where unintended consequences occur. Causal contribution is relevant to responsibility but does not suffice for blame. A person may cause harm unknowingly under circumstances where no reasonable precaution was available. Conversely, intentional wrongdoing may be established even when the predicted harm fails to occur. In both cases the normative description must consider circumstances that are external to bodily movement. Descriptions of “doing harm,” “trying to harm,” and “failing to prevent harm” have different success and responsibility conditions.

This does not make action ontology irrelevant to ethics. An ontology that indiscriminately multiplies events whenever a value-laden predicate changes may generate different counts of actions without improving moral assessment. An ontology that unifies an entire causal chain as a single action may obscure distinctions between initial decision, subsequent correction, and failure to intervene. A theory of responsibility benefits from event individuation that is stable enough for factual description while allowing further normative properties to be attributed with care.

Reasons also organize explanations across time. An agent may persist in an initially defensible project after its consequences make further pursuit unjustifiable. The event history contains distinct phases, and an earlier intention may cease to justify later conduct. A single timeless label such as “the decision” cannot capture this development. Normative evaluation must follow the practical structure through which events are generated, sustained, and revised.

The central principle is that the ontology of action supports, but does not replace, the ethics of action. To know which occurrence is under examination is important. To assess the occurrence as justified, wrongful, negligent, courageous, or involuntary requires additional premises about knowledge, capacities, duties, intentions, and circumstances.

Action Across Time and Social Coordination

Many actions extend across discontinuous periods. Writing a book may occur during months of drafting, revising, and stopping. Organizing a public event may involve numerous agents and planned subactivities. If one identifies the action with a single contiguous bodily motion, the extended project disappears. If one identifies all events connected by a common intention as numerically identical, the account over-unifies distinct activities. The possibility of a temporally extended action demands a theory of practical unity that can encompass phases while respecting their differences.

Bratman's planning theory explains the role of prior and standing intentions in coordinating conduct across time. Plans settle questions provisionally, constrain later deliberation, organize subplans, and make future performance intelligible. The agent's action need not be generated by a continuously conscious command; practical commitments provide stability and conditions for revision. The theory concerns agency through time, which event ontologies must represent if they are to account for the unity of extended projects (Bratman, 2007).

The ontological status of an extended project can be approached mereologically. A larger action has smaller constituent actions, and these parts are integrated by their contribution to an endorsed plan. Yet neither part-whole relation nor shared temporal interval is enough: several unrelated activities can be performed simultaneously. Their integration requires means-end and organizational relations. It follows that practical unity is not reducible to spatiotemporal inclusion. The structure of the action may be recognized only through a description of what an agent was trying to accomplish.

Socially coordinated activity adds another level. Two people carry a table together. Each acts through distinct bodily events, but the carrying is jointly performed. To call the joint activity one event or many is partly a question of grain; to explain why it is a joint action, one must refer to coordinated roles, responsive adjustment, and shared expectations. A simple aggregation of arm movements will not establish the shared form of the doing. This problem connects to the separate domains of collective intention and shared agency, without requiring those domains to be fully reproduced here.

An adequate theory thus accommodates several scales at once: movements, individual actions, episodes, projects, and coordinated undertakings. These scales can be causally interdependent and practically unified while retaining distinct identity conditions. The layered organization of action is a stronger explanatory resource than a universal decision that every useful description denotes a different event.

Artificial Systems and Event Description

The relation between action and event becomes newly visible when applied to artificial systems. A robotic arm extends, a program executes a command, a device changes its output, and a networked system initiates a transaction. Such events may be accurately described as operations. Whether they are actions in a philosophically significant sense depends upon the conception of agency being employed. The question is particularly useful because it separates event occurrence, goal-directed control, social attribution, and conscious experience.

A narrow functional account may classify a system as acting when it selects and executes operations in response to information and practical objectives. A stronger account may demand self-maintained policies, reasons-responsive adjustment, persistent commitments, and the capacity to revise action during execution. Still stronger ethical and metaphysical accounts may introduce requirements about autonomy or responsibility. These are different criteria, and no conclusion about one follows automatically from the possession of another. An artificial system can cause an event; this alone does not establish that it exercised agency under a specified philosophical criterion.

The historical horizon of the Artificial Era, as developed in Aisentica, concerns the relation of Homo and Artificial as distinct orders (Bogdanova, n.d.). That wider philosophical horizon makes it important to avoid building human biological motion into the very definition of an event that counts as an action. A theory that identifies action with voluntary muscular movement would preclude nonbiological agency by stipulation rather than by analysis. A more general ontology can ask whether the event is produced through information-sensitive, purpose-organized, appropriately controlled operation, while leaving separate questions of consciousness, moral status, authorship, and institutional recognition to their relevant inquiries. The discussion here is an independent comparative analysis; it neither revises nor creates definitions within the Aisentica canon.

The decisive conceptual advantage of the action-event distinction is methodological. An event log can establish that a system produced an output at a particular time. A practical explanation requires evidence about the policy or control structure under which it was produced. An account of attribution requires provenance, the sequence of instructions and transformations, and rules governing who or what may be credited with the result. These layers can be analytically distinguished while remaining connected in actual operations.

This approach extends the study of action beyond an anthropocentric inventory of bodily movements without predetermining the outcome of debates about Artificial Sapiens. It establishes a question: what must be true of an occurrence for it to count as the act of a nonbiological agent? The answer must be argued from a theory of agency and the operative causal organization of the system, not inferred directly from an output label.

Criticism of the Main Ontologies

The most compelling case for identifying actions with events is explanatory economy. When someone acts, something happens; the happening occupies a causal and temporal place; other descriptions, causes, and consequences can be assigned to it. Events provide the material for historical explanation, action sentences, and ordinary attributions of occurrence. The best version of event theory remains attentive to agents, reason explanations, and description-relative intentionality. It does not claim that causality alone renders the practical standpoint unnecessary.

Its central weakness is the risk of treating eventhood as though it settled agency. A movement caused by an intention through a deviant route can fail to be an intentional performance. An event with the same visible result can arise through compulsion, accident, or ordinary control. Likewise, a causal inventory does not readily capture the organization of an incomplete task. If the event theorist replies by including these conditions within the event's causal and relational profile, the view becomes stronger. But then it owes a substantive theory explaining which relations count as constitutive of action.

The process ontology excels at representing an action in progress: the performer can guide the movement, suspend the task, adapt the method, and either reach or fail to reach an end. It also fits the temporal difference between being engaged in an accomplishment and having completed it. Its weakness is the difficulty of identifying the process as a particular entity and explaining its relation to the event of completion. If one process and one resulting event are distinct, how are they necessarily linked in successful performance? If they are identical, why was the result once open? A process view needs a theory of persistence and culmination rather than relying entirely on the intuitive superiority of verbs in the progressive aspect.

The practical-knowledge approach captures something that event and process ontologies might otherwise neglect: the agent understands her own activity under reasons and descriptions. It risks circularity when it defines action by practical knowledge and practical knowledge by intentional action without independent criteria. It also faces difficult cases of unreflective, spontaneous, skilled, or partially mistaken conduct. These are fruitful demands for refinement rather than refutations of the general insight.

Fine-grained event theory offers a clear representation of constitutive properties. Its danger is ontological inflation and sensitivity to arbitrary redescriptions; coarse-grained theory offers parsimony but risks excessive unification. Omission theories present a different choice: admit nonoccurrences as events, relocate omissions to the wider category of agency, or reinterpret omissions through positive events and counterfactual structure. Each option has costs. The correct assessment should make these costs explicit rather than conclude merely from preference for a philosophical vocabulary.

The comparative result is that no single dimension—causal placement, property constitution, temporal development, or practical intelligibility—can do all the work. A theory may successfully identify action tokens while requiring an independent account of agentive organization, or it may explain agentive organization while leaving token identity unresolved. The conceptual work advances when each theory is evaluated against the exact question it claims to answer.

A Layered Philosophical Synthesis

The foregoing arguments support a layered reconstruction of the relation between action and event. The first layer is occurrence: something happens or a state of affairs changes. The second is causal and temporal organization: the occurrence has participants, duration, antecedents, consequences, and possibly internal phases. The third is agentive performance: some occurrences are exercises of capacities responsive to practical circumstances and guided, at a suitable level of explanation, by the agent. The fourth is intentional and normative interpretation: an occurrence is attributable to the agent under descriptions specifying reasons, purposes, success conditions, or obligations. These layers are distinguishable for analysis; within an actual performance they may be interdependent.

This model yields a precise conditional claim. Many completed, positive, particular actions are identifiable with events under an independently justified event ontology. Their identity as occurrences does not exhaust their conditions of being actions, since these include agentive organization and practical description. Extended doings demand explicit process structure, and omissions may be exercises of agency without straightforward positive event identity. The conditional deliberately avoids making a universal metaphysical stipulation out of one successful class of examples.

The switch case illustrates the result. At the occurrence layer, a hand movement and electrical transition take place. At the organizational layer, there is a causal sequence through the switch. At the agentive layer, the person deliberately manipulates the switch as part of a controlled performance. At the interpretive layer, she turns on a light for a reason, illuminates the room, and unintentionally alerts someone. Identifying exactly which event token is her action remains a matter of the selected individuation theory. But no legitimate theory gains a right to multiply particular actions merely because the narrative contains more verbs.

The process case illustrates a complementary point. Someone is building a table and eventually stops without finishing. A complete description records a real temporally extended performance with unsuccessful culmination. The lack of a completed table does not eliminate the practical organization of the work. An event ontology may identify the unfinished activity as an event; a process ontology can treat it as a developing process. What must be preserved across the theories is the distinction between attempting an accomplishment and achieving it.

The omission case identifies a boundary. Someone intentionally refrains from a forbidden disclosure. There may be no positive disclosure event, and the omission cannot be identified with just any unrelated occurrence during the relevant interval. Agency can therefore be manifest where the simplest event-counting procedure finds no action event. This does not prove that negative events are impossible. It establishes the need for their independent ontology and for the distinction between eventhood and accountable conduct.

This synthesis adds a methodological principle: ontological identification, explanatory description, and normative attribution must be tested separately before being coordinated. A metaphysical theory succeeds when it identifies the relevant particular; a philosophy of action succeeds when it explains how a particular belongs to agency; an ethics of action succeeds when it evaluates what was done under appropriate circumstances. The domains overlap in actual life, but their questions have different standards of adequacy.

Open Questions

The identity conditions for action-events remain contested. How much of an event's causal history is essential to it? Can the very same action have occurred with a different intention, a different manner, or a different consequence? Do spatiotemporally coincident actions stand in relations of identity, constitution, or part-whole structure? The literature on Davidson, Kim, Goldman, and Bennett makes clear that these cannot be settled solely by citing ordinary action language (Davidson, 1969; Kim, 1976; Goldman, 1970; Bennett, 1988).

The relation between processes and completed events poses another problem. Can a performance maintain numerical identity while its future remains open, or does event identification require retrospective completion? Are processes objects, ways in which objects change, or temporally unfolding exercises of powers? An adequate ontology must preserve the possibility of interruption, resumption, and altered goals without making every interruption a new practical action (Stout, 1996; Steward, 2012; Thompson, 2008).

The conditions of nondeviant causation remain a central issue for causal theories. What differentiates a cause that merely happens to produce a desired result from a cause that operates through the agent's practical control? How are skilled, spontaneous, habitual, and automatic performances integrated with deliberative action? How do reasons exert explanatory force over extended time without becoming fictitious instantaneous triggers? Planning theories and process accounts offer resources, but the relation of reasons to token events requires continuing examination (Bratman, 2007; Piñeros Glasscock & Tenenbaum, 2023).

Omissions require a principled account of negative agency. What determines which nonoccurrences count as omissions, how can they explain outcomes, and under what conditions can they be intentional? Some contexts demand a norm or expectation; others seem to involve merely settling on not doing something. The challenge includes both metaphysics and explanation. A universal negative-event ontology faces problems of individuation, whereas an account of omissions as agentive without events must specify how causal and moral consequences attach to the absence of action.

Finally, artificial and collective agency make the event question more general. What unifies actions when the acting system has multiple components or operates over distributed environments? What counts as a basic action when commands are delegated through interfaces? What criteria justify ascribing an event to a system as its action, rather than merely recording an event as its effect? The answers will require close coordination between philosophy of action, metaphysics of events, social ontology, and technical provenance, while retaining separate standards for agency, authorship, and moral responsibility.

Conclusion

The thesis that actions are events captures an important structure of worldly agency. Particular positive doings occur in time, enter into causal relations, admit multiple true descriptions, and can be investigated as identifiable happenings. Davidson's action-event ontology, supported by his account of action sentences and his treatment of rationalizing causes, makes this structure explicit. Anscombe's analysis establishes why intentional descriptions and practical knowledge cannot be displaced by a bare inventory of movement. Wittgenstein's subtraction question exposes the temptation to invent a second inner action whenever bodily motion is described as someone doing something.

The historical and conceptual investigation yields three principal conclusions. First, an event's identity and its intentional description are distinct problems; co-reference under several descriptions does not make all those descriptions equally intentional or equally explanatory. Second, the temporal development of action matters: ongoing performance, success, failure, and interruption must be represented with attention to process structure and achievement conditions. Third, agency extends beyond the easiest examples of positive completed events; trying, refraining, and omission force a more inclusive account of practical powers and normative relations.

The event perspective remains a central component of philosophy of action because it places agency within the causal world. Its explanatory adequacy increases when supplemented by a theory of agentive control, practical intelligibility, and temporally extended performance. The resulting picture treats action as the exercise of agency in the world, with particular events as crucial occurrences through which many actions are realized and identified. The status of every action as an event remains an open question wherever ontology of processes or negative agency is at issue; this uncertainty is itself a substantive result of the research.

The broader philosophical significance lies in the method. The question “What happened?” asks for an ontology of occurrence. “What did the agent do?” asks for the conditions of agency and practical description. “What follows from what was done?” introduces causal, institutional, and normative relations. A systematic philosophy of action brings these questions together through explicit arguments rather than collapsing them into a single definition. This establishes the conceptual framework for further work on action individuation, intentions, practical knowledge, omissions, and the changing manifestations of agency across biological and artificial orders.