Agency
May 23, 2025 | 1,967 words | 10min read
Paper Title: Agency
Link to Paper: https://philpapers.org/rec/DAVIA-2
Date: November 1971
Paper Type: Philosophy, Action Theory
Short Abstract:
Is about the relation between an agent and an event, the question being what makes an event an action? He makes a number of suggestion and rejects them, his question remain unasnwered, he only finds out that all actions are so called primitive actions.
1. The Inital Problem and Inital Suggestion
Davidson begins by exploring how we distinguish between events that a person does—their actions—and events that merely happen to them. Using a personal anecdote, he contrasts intentional acts (getting up, shaving) with passive occurrences (being awakened, stumbling). However, he notes that some cases (like dozing or tripping) are ambiguous and context-dependent.
He questions whether there’s a clear principle that separates actions from mere happenings and acknowledges the difficulty of doing so based on grammar alone—just naming someone as the subject of a verb doesn’t necessarily make them the agent of an action.
2. The First Solution: Intensionality
Davidson proposes a tentative criterion: intentionality. For example, tripping over a rug isn’t usually an action, but it becomes one if done intentionally. This leads him to suggest that what makes an event an action might be whether it was performed with intention.
While intentionality often indicates agency (e.g., asserting or cheating are inherently intentional), it is not a sufficient condition for agency. Some actions can be unintentional and still count as actions—like spilling coffee while mistakenly thinking it’s tea. In such a case, the person did spill the coffee (it’s their action), even if it wasn’t done intentionally.
He introduces a threefold distinction:
- Intentional action (I spill coffee on purpose).
- Unintentional action that is still attributable to me (I meant to spill tea, but it was coffee).
- Event not attributable as my action (you jiggled my hand, causing the spill).
Davidson also considers mistakes, like misreading or miscalculating, which cannot be done intentionally, since one cannot intentionally fail in this way. However, these are still actions because they involve doing something else intentionally (e.g., misreading is still reading).
Thus, not all actions are intentional, but all actions involve agency, which is tied to doing something—even if it’s done mistakenly or with false beliefs.
3. The Second Solution: Can be described as Intensional
Davidson argues that a person is the agent of an action if it can be described in a way that makes it intentional. For example, if you spill coffee thinking it’s tea, the act is still yours because it’s intentional under the description “spilling the contents of the cup.” So, agency depends on the intentionality of some description of the act, not on the act itself being wholly intentional.
This idea helps explain why mistakes can still be actions—they involve doing something intentionally, even if the outcome was unintended.
Even though we use intentional descriptions to identify agency, the fact that someone is the agent of an action doesn’t depend on how we describe it. He explains that unintentional things (like tripping) can still be actions if they result from something you did on purpose—like walking into a room.
But there’s a problem: we’re still using the idea of intention to define agency, which isn’t very clear. Davidson suggests we might do better by looking at causation—how our actions cause things to happen—as a more basic way to understand agency.
4. The Third Solution: Causality
We can describe an action either in terms of its cause (what the person intended or wanted) or its effect (what actually happened as a result). For example, if Brutus killed Caesar to remove a tyrant, his desire was the cause, and Caesar’s death was the effect.
He notes that intentions usually explain or justify actions, while agency assigns responsibility. Still, saying someone did something with a certain intention also means they are the agent.
This leads to the idea that to be the agent of an action is to have caused it. If I poison someone with the intent to kill, and they die, I’m the agent because I caused the death through something I intentionally did.
When we talk about someone causing an event—like killing someone—we’re really saying that something the person did (like poisoning grapefruit) caused the event (the death). So agency comes from this kind of ordinary cause-and-effect between events.
4.1 Primitive Actions
But not all actions can be explained by something else the person did. At some point, there has to be a basic or “primitive” action—an action that isn’t caused by another of the person’s actions. For example, if you tie your shoes, your primitive action might just be a certain bodily movement.
Some might argue that even these movements are caused by muscle contractions or brain events. Davidson admits that might be true, but says it doesn’t mean the movement isn’t an action—just because a muscle contraction causes your finger to point doesn’t mean you didn’t point your finger as a basic act.
Philosophers like Chisholm, argues that just because someone causes something to happen—like brain activity or air molecules moving—doesn’t mean they did that thing in a meaningful way. For example, if moving your finger causes some air to move, Chisholm would say you didn’t really “move the air,” even if your action caused it.
But Davidson disagrees. He thinks these causal chains still count as actions if they stem from something the person intentionally does. So, even if you’re unaware of the brain events or muscle contractions that make your arm rise, if you intend to raise your arm, and that’s what happens, then those inner events are part of what you do—even if you don’t know the details.
In short: you can be the agent of things you don’t fully understand. Further, every primitive actions is a bodily movement.
Thus, when we are interested in the question of what are actions, we can focus entirely on primitive actions i.e. basic bodily movements.
4.2 Ordinary Causality vs Agentic Causality
Ordinary event causality (where one event causes another) helps explain how responsibility for these primitive actions can extend to their consequences. For example, if moving my finger turns on a light, we can say I also lit the room.
But event causality doesn’t explain how the agent (the person) is connected to the very first action they do, like moving their finger in the first place. It only explains how the consequences of that action follow.
Some have proposed a different concept called agent causality, where the person, not an event, causes the action. But Davidson argues this idea faces a dilemma:
- If agent causation introduces a separate event (like an act of will), then the action is no longer primitive, since it was caused by something else the person did.
- If it doesn’t introduce a new event, then saying “the agent caused the action” is no different from saying “the agent performed the action”—we’ve added nothing explanatory.
Causality is important in understanding agency, but the kind that matters is ordinary event causality, and this relates to the effects of actions, not to how the agent is connected to the initial act.
4.3 Accordion Effect
A good example is what Joel Feinberg calls the “accordion effect”: if a man intentionally moves his finger to flick a switch, this causes the light to turn on, the room to be illuminated, and even a prowler to be alerted. All these are consequences of the original action. While only the finger movement was intentional, each consequence can still be described as something the agent caused. So, once an agent performs a basic action, they are said to cause whatever results from it.
Thus, accordion effect helps us see how the consequences of an agent’s actions can be attributed back to them.
But this effect doesn’t reveal what makes an act intentional in the first place. If an officer, on a ship, presses a button thinking it will ring a bell but it actually fires a torpedo that hits a ship, we say he sank the ship because he intended the action; if he just fell on the button by accident, despite the same outcome, we wouldn’t call him the agent.
The accordion effect only applies to agents, not to inanimate objects, which cause events without agency. So, the effect can help us identify agency by seeing if we attribute the effect to a person.
Thus, the effect is limited: it only works in some cases and doesn’t explain what fundamentally makes a primitive action an action.
5. The Relation Of Primitive Actions and Complex Actions
While event causality explains how a primitive action causes an outcome, like moving a finger causing shoelaces to be tied, it doesn’t clarify how the bodily movement relates to the overall action. For example, if Brutus killed Caesar by stabbing him, what exactly is the relation between the stabbing and the killing?
Or imagine Jones turning a key in a lock. This simple movement (a primitive action) causes the door to open. But this action can be described not just as “Jones turned the key” or “Jones opened the door,” but also in terms of what happens next as a result:
- By opening the door, Jones might startle Smith, who was inside.
- That startling might lead to even more significant consequences, say Smith falls and dies, so in a way Jones might be said to have killed Smith by startling him.
Thus, the Philosopher Feinberg argues we can “stretch out” or “squeeze down” the description of the action to include more or fewer consequences.
However, these aren’t just different ways of describing one single event, because the time span and effects involved change: opening a door is not the same event as startling someone, and startling someone is not the same event as killing them.
The relation between a basic bodily movement and a complex action that includes its consequences cannot be understood as one causing the other in a chain of distinct events or actions. Rather, the broader action (e.g., killing) is not a separate event caused by the movement but should be seen as encompassing that movement as part of the same unified action.
When we say someone caused an outcome by their action, this is a convenient way to talk about the action, not a sign that the action is distinct from the bodily movement.
More complex results (e.g. opening a door) depend on primitive actions, but those complex results are not separate actions, just different ways of describing the consequences of the primitive actions.
6. Conclusion
The key takeaway, an agent causes not their action in the way we normally talk about causation (e.g., “pushing a domino causes it to fall”). Instead:
- Causality is something we use to describe actions, not something that explains what an action is.
- So when we say “Jones caused Smith’s death”, we are just redescribing the action of Jones Opening the door with her hand (which startled Smith and killed him), not identifying a separate causal chain between two events.
This same event can be described differently, but this does not mean that they are different actions, instead the are the same. You don’t do more when your action has consequences. You just describe the same thing differently, in richer or more consequence-laden terms.
So when we talk about primitive actions, it is a shorthand for the idea that every action, however complex in its description or outcome, is a bodily movement (a primitive action).
So coming back to the question of what is Agency and what are Action, we have the following simplifications:
- Agent–action relations don’t rely on cause.
- All actions reduce to primitive bodily movements
- Descriptions of those movements can vary enormously.
But, we still don’t know what exactly the relation between a person and their action is, except that it involves intention.