Sense of Agency

When an action feels like yours

You reach for a cup. Your hand moves towards it, your fingers close around it and the expected weight arrives as you lift it.

The action does not usually require investigation. It simply feels like something you did.

Now imagine moving a cursor that responds too late or travels in a different direction than expected. The movement of your hand may still feel self-generated, while your control over the cursor becomes less certain.

That disruption reveals a process that normally remains unnoticed: the sense of agency.

What produces the experience that you are the one initiating or controlling an action?

What sense of agency means

The sense of agency is the experience of being the initiator or controller of an action and, sometimes, of the consequences that follow.

Feeling that I initiated a movement is related to, but not identical with, feeling that I produced an external result.

Several questions must therefore remain separate:

• Did you perform the movement?
• How did the movement contribute to the outcome?
• Did you experience yourself as controlling it?
• Did you later judge yourself to be the author?

These questions often receive similar answers, but they need not.

A person can contribute to an outcome without feeling much control. They can also feel that they produced something that would have occurred anyway. Subjective agency and objective causal control frequently correspond, but they are not identical.

As The Many Processes Called the Self explains, agency is one process contributing to selfhood. It connects actions and outcomes with oneself, but it is not memory, identity, bodily ownership or the whole self.

Agency is also distinct from executive control. Executive Functions Coordinate Behaviour describes processes involved in maintaining goals, selecting responses and regulating behaviour. Those processes can operate with little conscious monitoring, while a strong feeling of control does not prove that behaviour was coordinated effectively.

Control processes and the experience of control are related but different.

Intention, prediction and preparation

Intention contributes to agency. A goal can select an action, prepare movement and establish expectations about what will happen.

Some actions also follow valuation and decision processes, as explored in Decisions Compare Values. But intention alone is not proof of authorship.

People can intend to act and fail to do so. They can act routinely with little conscious deliberation. An intended movement can produce an unintended consequence. A desired outcome can occur for reasons unrelated to the person's action.

Processes involved in preparing and controlling movement can also generate predictions. Before and during action, they can anticipate how the body will move and what sensory consequences are likely to follow.

One influential account proposes that:

• an action is prepared;
• its sensory consequences are predicted;
• feedback becomes available;
• predicted and observed information are compared.

Correspondence can support the experience that an action was self-generated. Mismatch can weaken agency or make it uncertain.

This comparator logic explains important findings, but it is not a complete theory. Agency can concern distant outcomes, depend on contextual beliefs and remain present despite imperfect sensory correspondence.

Neither intention nor prediction proves authorship.

Feedback, outcomes and retrospective interpretation

Agency is also shaped by what happens during and after action.

Relevant information can include:

• visual, tactile and proprioceptive feedback;
• timing;
• correspondence between action and outcome;
• whether the result was expected;
• whether another plausible cause was present.

Interoceptive signals can contribute information about the body's current condition, but the strongest evidence about agency concerns motor, sensory and contextual cues. Interoception describes that embodied contribution.

A mismatch does not automatically eliminate agency. Its effect depends on its size, reliability and timing, as well as on expectations and other available information.

People also interpret outcomes retrospectively. If an expected event follows closely after an intentional action and no alternative cause seems plausible, the person may infer that they produced it. Prior beliefs, causal plausibility and knowledge of intention can all influence that judgement.

This means agency can sometimes be experienced over an outcome the person did not objectively control. It can also feel reduced when the person did make a causal contribution.

Retrospective influence does not make all agency a story invented after the event. Prospective intention and motor prediction also contribute.

Different agency judgements appear to draw on different combinations of information available before, during and after action. Research has not established one universal weighting rule.

Ownership and agency can come apart

Bodily ownership and agency ordinarily work together, but they are not the same.

Bodily ownership is the experience that a body or movement is mine.

Agency is the experience that I initiated or controlled the movement.

An externally guided movement can still feel connected with your body without feeling self-initiated. Experiments that alter visual and movement feedback likewise show that ownership and agency can change separately.

A movement can belong to my body without feeling fully caused or controlled by me.

The wider relationship between action and control is connected with D8.1. But terms such as action ownership or an action feeling like mine should not obscure the distinction between mineness and authorship.

Why agency is difficult to measure

Researchers can ask directly:

• Did you cause the outcome?
• How much control did you experience?
• Did the action feel self-generated?

These judgements provide information about conscious experience, but their answers depend partly on wording, expectations and retrospective interpretation. "Cause," "control" and "authorship" do not ask exactly the same question.

Researchers also use indirect measures.

Intentional binding refers to perceived compression of time between a voluntary action and its consequence.

Sensory attenuation refers to reduced processing or perceived intensity of some predicted, self-generated effects.

Both have been used as agency-related measures, but neither is a pure readout of agency.

Intentional binding can be affected by causal expectations, attention, learning and task design. Sensory attenuation may reflect sensory prediction rather than a conscious feeling of control. These measures do not always correspond with explicit reports.

Calling a measure implicit means it assesses agency indirectly. It does not mean the measure is theory-free or definitive.

No single task provides direct access to the complete sense of agency.

Agency can be accurate, mistaken or uncertain

Agency often tracks real control because the available information usually agrees. Intention precedes movement, movement matches prediction and the expected consequence occurs without another plausible cause.

But agency can be mistaken.

An outcome may coincidentally follow an intention. Feedback may be delayed or distorted. Another cause may be overlooked. Several people may contribute to an outcome while each misjudges their share of control.

Uncertainty can therefore be appropriate. When information conflicts or several causes remain plausible, reduced confidence may reflect the evidence more accurately than a strong claim of authorship.

Feeling agency does not prove causal control. Reduced agency does not prove that the person made no causal contribution.

The fact that agency can be manipulated or misattributed in experiments also does not make ordinary agency meaningless. A system can be fallible while remaining informative.

Constructed, informative and fallible

The sense of agency is constructed from multiple sources of information. That does not make it false.

By helping organise which actions and outcomes are attributed to oneself, agency-related processing can inform action monitoring, learning and adjustment.

But agency remains one part of selfhood and one source of information about control. A strong feeling of agency does not prove effective executive coordination, complete conscious causation or moral responsibility. A weak feeling of agency does not independently remove responsibility.

Agency research can constrain theories about human action, but it does not by itself settle whether free will exists.

Agency is not supplied by intention alone, and it is not invalidated by being constructed. It is a fallible but functional experience through which actions and outcomes come to feel like something I am doing.