Reasons, Intentions, and Action

An action is not merely a bodily movement. Raising an arm can be voting, signalling, stretching, or an involuntary spasm depending on how it is produced and understood. Philosophy of action asks what distinguishes action from happening, how reasons explain what agents do, what intentions contribute, and how bodily movements remain guided through time.

The dominant causal theory says actions are events appropriately caused by the agent's reasons and intentions. It explains why an action is attributable to a person rather than merely occurring in their body. Deviant causal chains show that causation is not enough: reasons must cause behaviour in the right way. Non-causal and agent-causal theories reject or supplement the event-causal picture.

This page concerns action as a mental-causal achievement. Whether such action is free, whether alternatives were possible, and whether the agent is morally responsible remain in the Free Will section.

Action versus mere behaviour

Some bodily events are clearly not actions: a reflex, seizure, or externally moved arm. Other cases are intermediate: habitual movement, automatic skilled response, or action under coercion. A theory needs conditions beyond the movement's physical shape.

Candidates include:

• the movement is caused by an appropriate desire and belief;
• it is guided by an intention or plan;
• it is under the agent's control and responsive to reasons;
• the agent knows what they are doing without observation;
• the movement expresses or is attributable to the agent.

No one condition is obviously sufficient. Reasons can cause accidentally, intentions can fail, control comes in degrees, and subjects can act intentionally under descriptions they do not know.

Reasons explanations

A normative reason counts in favour of an action. A motivating reason is the consideration for which the agent acts. An explanatory reason helps explain why the action occurred. These roles can diverge: a bad reason may motivate and explain without normatively favouring the action; a good reason may exist without motivating.

Reasons explanations have a rational form. Wanting to stay dry and believing the umbrella will help make taking it intelligible. Yet rational intelligibility alone does not show that these attitudes caused the action. A person may have the reason but act from habit, fear, or another consideration.

This creates a connection requirement: acting for a reason requires that the reason figure appropriately in the production or guidance of action.

Davidson's causal theory

Davidson's primary reason combines a pro-attitude toward actions with a belief that the action has the relevant feature. The reason rationalises the action and causes it.

The theory explains the difference between:

• having a reason for doing $A$;
• doing $A$ while possessing that reason;
• doing $A$ because of that reason.

Only the third requires the causal link. The account also integrates action into the causal world without reducing rational explanation to a law. As Anomalous Monism argues, singular reasons can cause even when mental types enter no strict psychophysical laws.

The simple analysis is:

$$Reason+Belief\text{-}DesirePair\stackrel{appropriatecausation}{\to }Action.$$

The word "appropriate" carries the remaining problem.

Deviant causal chains

A reason can cause the intended outcome in the wrong way. A climber intends to release a companion's rope; nervousness caused by that intention makes the climber's hand shake, accidentally releasing it. The intention causes the movement, but the release is not performed intentionally in the relevant way.

In another case, an assassin's intention causes agitation, which startles the victim into a fatal accident. The intended death occurs because of the intention, but not through the planned action.

These deviant causal chains show that action is not simply behaviour caused by a reason. The causal route must involve guidance, skill, and sensitivity to success. A theory that defines the right route as "non-deviant" merely labels the problem.

Guidance and control

Actions unfold through feedback. An agent monitors progress, corrects error, and adjusts movement as conditions change. Guidance control is not a single initiating event but an organised process responsive to intention and sensory information.

This helps distinguish action from accidental causation. The climber's trembling does not implement a control policy aimed at release; a deliberate hand movement does. The intention sets success conditions under which ongoing processes are corrected.

Guidance can be subpersonal and automatic. A skilled pianist need not consciously intend every finger movement. High-level intention structures lower-level motor control without representing each detail. This is a clear case of realised higher-level causation.

Intention

Intentions have several roles:

• prospective: they commit an agent to future action;
• planning: they coordinate means, timing, and other intentions;
• settling: they end deliberation provisionally and stabilise behaviour;
• guidance: intentions-in-action organise execution;
• social: expressed intentions support reliance and coordination.

An intention is not simply a strong desire. One can desire incompatible outcomes while intending one, and intention creates pressure for consistency and means-end coherence. Nor is it merely a prediction about one's future behaviour; intending includes practical commitment.

Bratman's planning theory emphasises partial plans that constrain later deliberation without specifying every step. This fits bounded agents whose actions extend over time.

Prior intention and intention-in-action

A prior intention exists before action: intending tonight to run tomorrow. An intention-in-action is the guiding structure present during intentional performance. Spontaneous basic actions may have the latter without a prior plan.

The distinction avoids an infinite regress. If every action required a prior act of intending, that act would require another intention. Intention-in-action can be a feature of the action's organisation rather than a separate initiating action.

It also explains imperfect execution. An action can be intentional under a broad description - writing a name - while particular pen movements are not separately intended.

Basic action

A basic action is something an agent does not do by doing something else. One turns on a light by flipping a switch and flips the switch by moving a finger. Is moving the finger basic, or does the agent act by contracting muscles or firing neurons?

Agents ordinarily know and control movements at a practical level without selecting neural details. The basic level is therefore not necessarily the lowest physical event. It may be relative to skill and description: a pianist plays a chord directly where a novice performs laborious component movements.

Basic action blocks a regress of means. It also supports the thought that agency is embodied know-how rather than command from an inner executive.

Trying

Trying appears common to successful and failed action. A paralysed subject may try to raise an arm without movement; an agent may try and fail because of external resistance. This suggests the action cannot simply be identical with the bodily outcome.

One view treats trying as the most basic mental action that normally causes movement. Another treats failed attempts and successful actions as different events sharing an intention and control organisation. A third analyses trying through commitment and effort rather than an inner act.

Making trying a separate mental cause risks recreating the interaction problem. A physicalist can identify trying with motor preparation and control states while allowing its practical description to explain the attempted action.

Knowledge in action

Agents often know what they are intentionally doing without observation. One need not watch one's hand to know one is writing, although observation can reveal that the pen has run dry. Anscombe treats this practical knowledge as distinctive: action is known under the description that guides it.

Non-observational knowledge is fallible. A subject can intend to draw a straight line while unknowingly drawing a curve. They know what they are doing intentionally in one sense and may be wrong about success.

The phenomenon constrains causal theories. An intention is not merely a hidden cause; it structures the action and the agent's knowledge of it. Guidance and practical knowledge may be two aspects of one control process.

Causalism and non-causalism

Causalists hold that appropriate reason causation is necessary for action. Their advantages are integration with natural science and a clear distinction between acting for a reason and coincidental fit.

Non-causalists argue that reason explanation is normative or teleological rather than causal. An agent acts in order to achieve an end; forcing this relation into efficient causation invites deviant chains and turns agency into something that happens to the person.

Non-causalism must explain how reasons connect to bodily events and why actions occur when they do. Causalism must explain appropriateness without circular appeal to action. Guidance-based theories offer a middle path: causal processes constitute action when organised by reason-responsive control.

Agent causation

Agent-causal theories hold that the agent as a substance causes an action, not merely one event causing another. This is intended to secure authorship and avoid reducing the person to a causal chain.

Agent causation requires an account of substance causation, its relation to event causation, and how reasons make one exercise intelligible rather than arbitrary. Its main contemporary role is in libertarian theories, developed in Libertarianism. The present issue is narrower: ordinary intentional action does not by itself show that irreducible agent causation occurs.

Rationalising and mechanistic explanation

A complete action explanation can contain several levels:

1. rationalising: the agent believed $p$, wanted $q$, and saw the action as a means;
2. computational: information was integrated into a policy or decision;
3. mechanistic: neural and bodily control produced movement;
4. social: norms and roles made certain reasons salient.

The levels need not compete. Rational explanation identifies content and practical structure; mechanism explains implementation. Conflict arises if reasons are claimed to possess independent powers beyond the mechanism or if mechanism is taken to eliminate the rational pattern.

The higher-level causation framework treats a reason as causal when it is a robust, proportional variable implemented through the mechanism.

Relation to freedom and responsibility

An action can be caused by the agent's reasons and still be determined. It can be reason-responsive and still result from manipulation or compulsion. Conversely, indeterminism can disrupt guidance rather than enhance it.

Philosophy of action therefore supplies the event that free-will theories assess but does not settle the assessment. The Free Will Problem adds alternative possibilities and sourcehood; Compatibilism develops reason-responsiveness; Libertarianism considers agent-causal and non-causal alternatives.

Assessment

Reasons explain action only when connected to production and guidance. Simple event causation is too weak because causal chains can deviate; purely normative fit is too weak because an agent can act for another reason. Intentions contribute commitment, planning, and online control, while basic action prevents regress.

The most promising causal accounts identify action with a reason-responsive control process rather than a movement preceded by a mental trigger. They leave open the further metaphysical and moral questions that define free will.

Selected references

• Anscombe, G. E. M. Intention (1957).
• Bratman, Michael E. Intention, Plans, and Practical Reason (1987).
• Davidson, Donald. "Actions, Reasons, and Causes" (1963).
• Frankfurt, Harry G. "The Problem of Action" (1978).
• Mele, Alfred R. Springs of Action (1992).