Choice, Intention, and Agency

We often speak as if action follows a simple sequence:

I choose.

I intend.

I act.

Real agency is more complicated.

Choice, intention, and agency are related.

They are not identical.

Choice

A choice is selection among alternatives.

Examples:

• tea or coffee,
• stay or leave.

Not every choice produces action.

You can choose and later change your mind.

Intention

An intention is a commitment or plan to act.

It organizes future behavior.

Intentions can persist through time.

Agency

Agency is broader.

It is the capacity to:

• form goals,
• make choices,
• execute actions,
• monitor outcomes.

Agency is a system-level property.

Choice Without Action

You may decide:

I will exercise tonight.

Then fail to act.

The choice existed.

Execution failed.

This is weakness of will or implementation failure.

Action Without Explicit Choice

You can:

• catch a falling cup,
• adjust posture

without consciously choosing among explicit alternatives.

Agency can operate automatically.

Intention Without Immediate Action

You may intend:

submit a paper next month.

The intention persists.

It guides future planning.

Future-Directed Intentions

Philosopher Michael Bratman emphasized that intentions help coordinate:

• future action,
• present planning.

They reduce the need to reconsider every option constantly.

Intentions as Commitments

Once you intend:

take the train,

you begin organizing:

• departure time,
• ticket,
• route.

Intentions constrain later choices.

Planning Agency

Humans are planning agents.

We do not select every action independently.

We create hierarchical structures of goals.

Goal Hierarchy

Example:

Goal: finish degree.

Subgoal: pass course.

Subgoal: study tonight.

Action: open textbook.

Agency operates across levels.

Means–End Reasoning

An intention toward an end generates pressure to adopt necessary means.

If I intend to attend the conference, I should arrange transportation.

This is practical reasoning.

Intention Stability

Intentions should be stable enough to guide action.

But not rigid.

New evidence can justify revision.

Good agency balances:

• persistence,
• flexibility.

Akrasia

Akrasia is acting against one’s better judgment.

You judge:

I should not eat the cake.

Then eat it.

This reveals conflict among:

• judgment,
• desire,
• control.

Weakness of Will

Akrasia challenges simplistic models where:

strongest reason automatically causes action.

Human agency contains competing systems.

Desire vs Intention

A desire is:

wanting an outcome.

An intention is stronger:

committing to pursue it.

You can desire:

to be fit

without intending:

to exercise today.

Belief–Desire Model

Classical action theory often explains action through:

• belief,
• desire.

You want water.

You believe the glass contains water.

You reach.

This is useful but incomplete.

Reasons for Action

A reason can be:

• normative: what counts in favor of action,
• motivating: what actually moves the person.

These can diverge.

Normative vs Motivating Reason

You may have strong normative reason to:

save money.

But motivating desire to:

buy something expensive.

Agency involves conflict between reason types.

Intention and Causation

Does an intention cause action?

In many theories:

yes.

The intention is a higher-level causal state realized neurally.

Davidson

Donald Davidson argued that reasons can be causes of actions.

This helped integrate rational explanation with causal explanation.

Reasons do not float outside nature.

Action for a Reason

An action is not merely caused by a belief and desire.

It should be caused in the right way.

Otherwise deviant causal chains arise.

Deviant Causal Chain

Suppose a climber intends to loosen another climber’s rope.

The nervousness caused by the intention makes his hand shake, accidentally loosening it.

His intention caused the outcome.

But he did not intentionally loosen the rope in the ordinary sense.

The Right Causal Relation

Intentional action requires more than causal connection.

The action must be appropriately guided by the intention.

This is a control problem.

Motor Control

Once a high-level intention forms, motor systems translate it into:

• trajectories,
• muscle activation.

Conscious thought does not specify every contraction.

Hierarchical Control

Agency delegates.

High level:

pick up cup.

Lower level:

calculate grip, adjust force, correct error.

This hierarchy explains how intentions control complex bodies.

Feedback

Actions are monitored through:

• vision,
• proprioception,
• touch.

Errors trigger corrections.

Agency is closed-loop control.

Comparator Models

Motor-control theories compare:

predicted sensory consequences

with:

actual feedback.

Matching signals contribute to sense of agency.

Efference Copy

The brain generates internal copies of motor commands.

These help predict:

what movement should feel like.

This aids self/other distinction.

Tickling Yourself

Self-generated touch is less surprising because the brain predicts it.

This is why self-tickling is difficult.

Prediction alters perception.

Sense of Agency

The sense of agency is the feeling:

I caused this action.

It depends on:

• prediction,
• intention,
• feedback.

It can be manipulated.

Agency Without Sense

Some actions may genuinely be self-caused even if the person lacks normal sense of agency.

Neurological disorders reveal dissociation.

Sense Without Agency

Conversely, people can feel responsible for outcomes they did not cause.

Agency experience is an inference.

It is not infallible.

Intentional Binding

When people voluntarily act and an effect follows, they often perceive the action and effect as temporally closer.

This is called intentional binding.

It has been studied as a marker of agency.

Interpretation Caution

Intentional binding is not a direct consciousness meter.

Its relation to agency depends on task and interpretation.

No single psychophysical effect defines agency.

Choice Architecture

Decisions depend on how options are presented.

This is choice architecture.

Framing can influence selection.

Framing Effects

People may choose differently depending on whether outcomes are described as:

• gains,
• losses.

This challenges the idea of a perfectly stable chooser.

Default Effects

People often accept default options.

This can reflect:

• convenience,
• implied recommendation.

Agency is context-sensitive.

Does Influence Eliminate Choice?

No.

All choices occur in environments.

The relevant question is whether influence becomes:

• manipulative,
• coercive,
• bypassing.

Degrees matter.

Nudges

A nudge changes choice architecture without eliminating options.

Nudges raise ethical questions about:

• autonomy,
• paternalism.

Agency depends on both internal control and external design.

Manipulation

If another agent intentionally shapes your decision through hidden psychological techniques, authorship may be reduced.

This links agency to authenticity.

Autonomy

Autonomy means self-governance.

It is related to agency but stronger.

An autonomous agent acts according to principles or goals it can in some sense endorse.

Kantian Autonomy

For Kant, autonomy is tied to rational self-legislation.

Freedom is not doing whatever one desires.

It is acting under rationally endorsed principles.

Contemporary Autonomy

Modern theories often emphasize:

• reflection,
• authenticity,
• absence of controlling manipulation.

Autonomy has ethical and political importance.

Reasons Responsiveness

An agent demonstrates control when behavior changes appropriately with reasons.

This is central to compatibilism.

Counterfactual Sensitivity

Ask:

If the evidence changed, would the decision change?

If yes, the system exhibits counterfactual responsiveness.

This is a hallmark of genuine control.

Habitual Agency

Habits reduce deliberation.

But habits can still be self-endorsed.

A person can intentionally build a habit to automate good behavior.

Delegated Agency

You may let:

• calendar,
• navigation system,
• assistant

choose details for you.

Delegation does not necessarily remove agency if the delegation itself is intentional and revocable.

Extended Agency

Tools can extend decision-making.

Notes, software, institutions

become parts of practical reasoning.

Agency can be distributed beyond the skull.

Collective Agency

Organizations can have:

• goals,
• procedures,
• decisions.

Do corporations or teams count as agents?

At a functional level, often yes.

Moral responsibility becomes more complicated.

Machine Agency

Artificial agents can:

• represent goals,
• select plans,
• execute actions.

This is functional agency.

Whether they possess:

• autonomy,
• responsibility

depends on architecture and social role.

Goal Assignment

A machine may pursue a goal assigned by a human.

That makes it an agent operationally.

It does not make it the ultimate author of the goal.

Self-Generated Goals

Stronger agency would involve:

• forming,
• revising

goals internally.

This begins to resemble autonomy.

Persistent Agency

Long-horizon agents require:

• memory,
• state,
• commitment.

Without persistence, agency remains shallow.

Choice and Consciousness

Some choices occur consciously.

Others do not.

Consciousness may be especially useful when:

• conflict exists,
• novelty appears,
• long-term consequences matter.

Intentions Can Be Unconscious?

Some goal-directed processes operate without explicit conscious awareness.

This suggests intention-like control may have unconscious components.

Terminology becomes theory-dependent.

The Self as Control System

The self may be understood partly as a system integrating:

• goals,
• memories,
• action policies.

Agency is then not a little inner commander.

It is organized control.

Agency Is Temporal

An action may begin years before execution through:

• planning,
• habit formation,
• value development.

Focusing on one millisecond misses this temporal depth.

Distributed Responsibility

If decisions arise from many neural and social causes, can the person still be responsible?

A compatibilist says:

yes, because the person is the organized system through which those causes operate.

The Philosophical Lesson

Choice, intention, and agency form a hierarchy.

Choice selects.

Intention commits.

Agency coordinates the whole process of goal-directed control.

Free will, if it exists, is not one mysterious instant.

It is a property of this organized system.

The Next Question

Once we understand agency, the moral question becomes unavoidable.

When should an agent deserve:

• praise,
• blame,
• punishment?

The next essay is:

Free Will and Moral Responsibility.