Aristotle on Agency

My focus in this essay is on a simple but basic question: what is the relation, in Aristotle’s account, between an agent and his or her action when the agent acts “voluntarily”¹?

In addressing this question, I shall begin by considering Aristotle’s answers to two prior questions:

We need answers to these questions before we can determine what, in Aristotle’s view, is the relation between the agent and the “voluntary” action he or she performs.

Let us begin with a specific example: Iannis throws some part of the cargo overboard in a storm to save himself and other members of his crew. Aristotle describes a case like this in the Nicomachean Ethics (NE 1110a9–11). The action, in this case, is Iannis’s throwing the cargo overboard (at a specific time) in a storm. He acts as he does for the sake of his goal: safety.

To make progress, we need to establish Aristotle’s answers to the following questions:

and so find his answer to the basic question:

Although answers to these questions have important implications for Aristotle’s discussions of virtuous action, acrasia, and human flourishing, I shall leave them aside in this essay. My concern is exclusively with Aristotle’s contribution to issues in the philosophy of action, a subject he initiated.² My goal is to sketch an account, in some areas highly controversial, of his views in this restricted area, noting gaps and problems as they arise. Much remains to be done on these issues.

Aristotle would describe Iannis’s action of throwing the cargo overboard as a *process* (kinēsis). Iannis moves his limbs and the cargo (NE 1110a15; see also Eudemian Ethics [EE] 1220b28–30). What does he mean by a *process*? ³ I shall argue that his *processes* (kinēseis) are best understood as processes rather than as events.

Aristotle describes some *processes* as continuous (Physics [Phys] 229a1, 267a20), others as quick or slow (228b29), as lasting from one time to another (239b1ff), as even or uneven (228b18f, 228b27ff, 238a22, 265b11ff), as complete or incomplete (1174a28ff), as interrupted or uninterrupted (255b7, 199b16ff), as accelerated or retarded (238a6ff), and as forced or in line with nature (215a1ff). Still others are said to go on forever (261a30ff).

How to understand Aristotle’s claims? I shall consider three aspects of his *processes*: their connections with time, space, and possible interruption.

Aristotle thinks of *processes* as follows:

*Processes* are like enduring substances in continuing through time. Substances change; *processes* are the changes that occur. More precisely, *processes* are not things that change but are the changings themselves. For a *process* to exist is for a changing to occur. Some *processes*are now under way, some were once under way but have now been completed, some have not yet begun.

Some processes are movings through space with their own direction. This is why two processes can be described as being in different directions (Phys 264a28, b14ff).

*Processes* can be interrupted or incomplete: some particular *processes* could have gone on longer and been completed had they not been interrupted. These *processes* have goals built into them which, on occasion, they fail to meet. Particular *processes* have (what I shall call) modal depth: the same particular process could have developed differently.

These features of Aristotle’s discussion of *processes* parallel our own descriptions of processes. We say:

My conjecture, on this basis, is that Aristotle’s *processes* are processes. In these respects, they differ from events (as understood in today’s philosophical literature).⁴

We cannot talk of an individual event as speeding up or slowing down. If we talk of events as accelerating or slowing down, we mean that the rate (or frequency) of the occurrence of a type of event speeded up or slowed down: there were more or fewer of them. There is not one event which itself was (underivatively) initially slow and is now fast. An event (as understood in the philosophical literature) may have parts which are slow and quick. But the event itself does not itself (underivatively) possess these different properties at different times. It is quick and slow in virtue of having parts that are quick and slow.

There is a further, related difference. We cannot say when an event is ongoing that “it is occurring at a constant or regular speed.” If we talk of constant or regular events, we think of several events of the same type occurring one after another. We are not thinking of one particular event itself as occurring in a constant or regular manner.

To see this, consider the sentence: “Yesterday afternoon, I was swimming constantly and regularly from A to B.” If we think of processes, the attributes “constantly” and “regularly” qualify the manner in which I was swimming the one swim I did: throughout the afternoon I was swimming at a regular and constant pace from A to B. By contrast, when we apply these attributes to events, we mean that I swam from A to B several times, making several different swimming journeys (swims) at regular intervals.

Whereas processes themselves develop, possessing different properties at different times, events develop only in virtue of their parts (subevents) having differing properties at different times. In this respect, processes (like enduring substances) exist without qualification at each time they exist, whereas only parts of events do. Processes and enduring substances can (in this way) be described as “wholly present” at each time during their existence, whereas events cannot. While an event may be partly present in virtue of one of its parts being wholly present, a part of a process is present in virtue of the whole presence of the process of which it is a part. In sum: while processes (so understood) persist by enduring through time, events, in contemporary philosophical terminology, persist by “perduring.”⁵

Events do not move through space. While parts of an event (subevents) occur at different places, the event itself does not move from place to place. It is a mistake to say:

“The chemical event began here but continued far out into the bay.”

“The chemical event is moving in that direction, but the biological event is moving in the other direction.”

Whereas events are occurrences that happen at (and between) places, they are not movings from one place to another. They do not have spatial directions. While an event is a spatial movement that occurs, it is not itself the moving from one place to another. This is why events are sometimes thought of as static: their properties, once acquired, do not change. Processes, by contrast, are changings with differing properties, such as different speeds, at different times.

Individual events, unlike processes, are not themselves interrupted or incomplete. Considered as changes, they are simply what occur. A change is complete when it is finished. This is true even if it is a member of a type of event which normally goes on longer. If the same process had gone on longer, there would have been a different event.

These considerations confirm my original conjecture:

However, before proceeding, we should consider a possible objection. It may be suggested that while processes are not individual events, they are sequences of events. Is this how Aristotle thought of *processes*? It does not seem that it is.

A sequence of events unfolds as one subevent happens after another subevent. It continues as long as the subevents that compose it occur. A sequence may speed up (or slow down) derivatively in virtue of the subevents that compose it occurring at different speeds: the ones at the beginning may occur at a slower speed than the ones at the end. Perhaps a sequence may be said to be incomplete if it does not contain the number of subevents it normally does or might be hoped to have. However, the sequence itself does not speed up or slow down, even if its subevents occur at different times or speeds. There is no one thing which itself (underivatively) slows down or speeds up or alters by having different properties at different times. What is lost (when we think of a sequence of events) is Aristotle’s idea of one continuous unfolding. This idea is replaced by the quite different thought of a series of subevents, one happening after another. Since he talks of *processes* themselves as continuous, we can conclude that his *processes* are processes. (From now on, I shall speak of Aristotle’s processes.)

Clarity on this issue is important. Philosophers and interpreters often talk of events without distinguishing them from processes. Others talk of processes but without marking out their distinctive features.⁶ However, if we fail to draw the relevant distinctions, we will not (I shall suggest) properly understand Aristotle’s account of agency or see what is distinctive about it.

According to Aristotle, when A moves B, B is moved by A. In the terminology of Physics III.3, the first is a making (poiesis), the second a suffering (pathesis). Both occur in the same place at the same time. Aristotle describes them as “the same but different in being,” apparently trying out several ways to formulate this claim.⁷ In one, there are two different processes that necessarily co-occur. In another, there is one process with different descriptions. In neither model is A’s moving B the cause of B’s being moved by A. In the latter, there is just one process. A (or some feature of A) is the cause equally of A’s moving B and of B’s being moved by A. In the former, although A’s moving B to a given place might be A’s causing B to be in a different place, A’s moving B is not itself the cause of B’s being moved to that place. If one asks what is the cause of the latter process, the answer would again be A (or some feature of A). A’s moving B, although the exercise of A’s power (or ability to move), does not itself have any power (or ability) to move anything. Only substances, or substances in certain conditions, have causal powers to act in this way.

Does Aristotle define A’s moving B in terms of A’s causing B to be in a different place: the one to which A’s relevant causal power is directed?⁸ According to this suggestion, although many cases of A’s moving B are ones in which A causes B to be at the position aimed at, not all are. In fact, the very same process might have caused a wide variety of states depending on how far A succeeded in achieving his aim. That is, the individual process of A’s moving B may cause B to be at p₁ or p₂, … … p_n depending on how far A succeeds in moving B. If A tries to push B to p₁ but fails to do so, the individual process in which A engages is incomplete in that it does not reach p₁.

There is an alternative proposal: Aristotle might have defined A’s moving B in terms of A’s realizing its power to move B to the place that is A’s goal.⁹ Here, too, the individual process will remain the same process even if A does not, in fact, bring about the state aimed at. The process will only be one of causing B to be at the place aimed at when it is successful. Instead it will be defined as the realization of A’s power to move B to a (favored) end state(p_w), not as the realization of its power to cause B to be at that end state. On this view, so far from the process of A’s moving B being defined in terms of A’s efficiently causing B to be in the resultant state, the latter will be defined in terms of the former.

The alternative has advantages. Some of Aristotle’s processes are not easily understood as the agent’s producing a resultant end state. Take the case of flute-playing: when S plays the flute, the flute is being played (a suffering of the flute). In Aristotle’s terminology, S is the cause of his playing the flute (the doing) and of the flute being played by him (the suffering). He could, no doubt, have introduced a further state: the flute in question having been played by S, the state of the flute once S has finished playing it. But there is no sign that he did so or thought that a state of this type had any role to play in his account of flute-playing. Furthermore, even had he done so, the relevant end state would, as we just noted, have been defined as the one that results from S’s playing the flute. If so, it could not have been used to define flute-playing.¹⁰

There is a second advantage. A has the power to cause her hand to be at some place other than by moving it. She can, for example, order someone else to move it for her (while she did her best to resist) or put it in a stream of water that moved it to the place at issue. She has the power to jump to p_w while keeping her arm firmly by her side.¹¹ (There are, as the recent discussions of deviant causal connections have made clear, very many differing ways of causing one’s hand to be at a different place other than by exercising one’s ability to move one’s hand.) What is important, if A is to move her hand, is that A causes her hand to be in another place by moving it. This is not, of course, to deny that A’s moving her hand is generally A’s causing her hand to be at the place to which A (or A’s power) is directed. The central claim is that the former should not be defined in terms of the latter. While A’s moving her hand will, if all goes well, cause her hand to be in the specific different state aimed at, her moving her hand is not itself to be defined in terms of her causing it to be in that state.¹² Nor will it help to suggest that the relevant causal power is to cause her hand to be in that state by moving it. Once again, the agent’s relevant power to cause her hand to be at that place will depend on, and not constitute, her power to move her hand to that place.

It may be suggested that Aristotle could, nonetheless, have maintained a causal analysis of processes and actions had he introduced a separate event: the movement of B (the flute being played or the hand being moved) and taken this as the causal result of A’s moving B. Perhaps A caused the movement of B—the event in question—by moving B (the process). Indeed, it is just this thought that has led some contemporary theorists to think of processes (and actions) as the causings of events. However, there is no evidence that Aristotle thought in this way. In his discussion in Physics III.3 he talks only of B’s being moved by A and A’s moving B (processes), without any reference to a separate (intransitive) event of the movement of B.¹³ The results of his processes (when successful) are standardly objects (such as houses) or (on occasion) resultant states (such as being in good condition), not events. Nor did he engage with the problems an event-based account must encounter. Even if one agrees that A causes the movement of her hand (event) by moving her hand, one need not think that the relation between the process of moving her hand and the resulting event (the movement of her hand) is itself a causal one. How, in fact, does a process causally generate the relevant event? Aristotle does not address this issue.

Indeed, given his account as just sketched, had he spoken of events (or changes) at all, they would have been understood as processes under their perfective aspect (seen as completed). They would not have composed an added level of entity resulting from, or generated by, processes. Furthermore, even had he introduced such a level of entity and regarded resultant events as (in some way) the products of processes, he would not have defined the process in question as the causing of such an event. Instead, he would have defined the causing of events in terms of processes. Which event was actually caused would have depended on where the relevant process ended, how successful it was, when it unfolded, and the like.

We have the beginning of an answer to our first question: what occurs when the agent moves his hand? In the basic case, what occurs is a process, not to be defined either as an event, the causing of an event, or the producing of a resultant state. More positively, the process in question is the realization of one of A’s goal-directed capacities, such as the capacity to move her hand to the relevant goal. Whereas A’s moving her hand is A’s causing her hand to be in a different place, her action is not to be defined as the causing of her hand to be at that place.

This is, however, only the beginning of an answer. We need to know next where the relevant process occurs, how it is individuated, when it is an action, and what role (in general) the agent plays in its production. I shall seek to address these issues, which raise important questions for Aristotle’s general theory of processes, capacities, and causation, only in so far as they are relevant to his account of action.

Aristotle, it seems, is prepared to regard growing old or dying (Common Ethics [CE] 1135b1) as well as breathing, falling asleep, or waking (De Partibus Animalium 645b33ff) as things we do (prattein, praxeis) even though they are not, in general, the products of our perception or desire. Nor does he exclude from this category processes that originate in us without being selected (haireton) or “ordered” by our imagination or desire, such as being sexually aroused (De Motu Animalium [DMA] 703b6ff). Things we do could also include blushing, frowning, raising one’s eyebrow, or crying (which are sometimes described as “subintentional”¹⁴), even when they are not selected by an agent. Aristotle does not give close attention to the question: what is to count as an action?¹⁵ His focus is rather on a more specific issue: under what conditions is an action “voluntary”? In his discussion of these actions, Aristotle considers what the agent does to achieve a goal the agent has set him- or herself: doing A for the sake of a goal G (as Iannis acts “voluntarily” in throwing the goods overboard for the sake of safety). In Aristotle’s view, Iannis acts “voluntarily” when the starting point for moving his limbs lies in him, more specifically in his selection of what to do (NE 1110a15ff). His “voluntary” action is the process Iannis knowingly selects to achieve his goal.

Aristotle notes that actions need not be confined to the agent’s body: moving a stick (DMA 702b6f) can be treated as something the agent does to achieve his or her goal, at least in cases where one is able to regard the stick as “a detached limb,” something within one’s immediate control. The skilled weaver’s moving the loom across the wool would be something the weaver does to achieve the goal, in this case without further deliberation about means.¹⁶

There are several aspects of the model suggested by Aristotle’s process and capacity-based model that deserve comment. While I shall sketch a view of (what I take to be) his view and its distinctive commitments, each separate claim requires, and will repay, further scrutiny.

Aristotle takes some actions to extend to include object affected. When you teach someone, your teaching (in his view) occurs in the learner “although it is not cut off” from the agent (Phys 202b6ff; Metaphysics [Meta] 1050a28–32). If Iannis throws the cargo into the sea, his action of putting the cargo into the sea will occur (at least in part) where the sea is, although this, too, will not be “cut off” from Iannis. It is certainly not the case that all he does is to move his body, leaving the rest being up to nature.¹⁷ Agents, in the cases under discussion, can act in places other than the ones they (and even their detached limbs) occupy.

What precisely is Aristotle’s view? His remarks appear to be “of a piece” with his discussion of processes: the process that is the action begins with Iannis’s picking up the cargo, continues while he takes it to the side of the ship and drops it overboard, and ends only when the cargo enters the water. The whole action is that of his throwing the cargo overboard, a process that unfolds through several stages. On this account, putting the cargo into the sea, which occurs where the sea is, will be the final stage of a process that began when Iannis picked up the boxes with this goal in mind. He controls the whole process as it develops up to its final stage, which occurs when and where the cargo hits the sea.¹⁸

In the case of teaching, there is, on this model, one process that begins with the teacher’s saying something or writing it on the board (with the aim of the pupil’s learning) and ends with the learner learning. If the term “teaching” is applied only to what happens in the pupil, it will refer to the last stage in a process that began with the teacher’s talking or writing something on the board. There will be one process that includes the teacher speaking with the aim of teaching the pupil and, if successful, also includes the effect the action has on the pupil. In unsuccessful cases, this process will not have this result. However, even then there will be the same process, the realization of the same capacity to teach as is exercised in successful cases.¹⁹ The capacity exercised is not simply the ability to say something or write it on the board but rather to do so in a way guided by the teacher’s skill and its characteristic goals (of teaching the learner).

Aristotle sometimes talks in ways that suggest that the action begins with the agent’s desire (or choice) to do something. This is, in his terms, the starting point of the action, the place from which the process begins (see, for example CE 1139a31f). On this model, the processes initiated by desire, including those in the connate pneuma which move the limbs, will be part of the unfolding action that culminates in the cargo hitting the water. In Iannis’s case, there is one action that issues from his skill, beginning inside his body and extending beyond its periphery into the world.

Aristotle notes in De Anima (DA) 408b11–16 that:

to say that the soul becomes angry is like saying the soul weaves or builds. Indeed it would be better not to say that the soul pities or learns or thinks but that the human being does so with respect to the soul.

It should now be clear why it would be infelicitous to say that the soul weaves or builds. The processes involved in these cases involve the agent moving his or her hands, the loom, and wool. The agent (as a psycho-physical entity) controls and directs these processes using manual skills and physical abilities to do so. These abilities belong to the agent as a whole and not to his or her soul. The agent needs to have information (feedback) about how the process is developing and the skills required to guide it. Even if such processes are initiated by desire (and the desiderative soul), action extends to the rest of the body and the world beyond. The soul, by itself, without the rest of the body, lacks the ability to control and guide actions of this type as they develop. This is why Aristotle insists that it is the unified human being and not the soul that (properly speaking) weaves and builds houses.

The answer to this question depends, as much else, on how the relevant capacities or skills are individuated. In Aristotle’s view, are they (or some of them) defined by their goals? If so, how specific are the goals in question? Are they the goals achieved in optimal conditions by a perfect practitioner?²⁴ Does the imperfect practitioner in suboptimal conditions exercise the same skill? No doubt, some capacities are defined, in his account, in terms of the good of the organism whose capacities they are. But is this true of all? What are the conditions relevant to their exercise? Can capacities of this kind be used by agents for purposes that are not, in reality, good for them?

These are important questions about Aristotle’s theory of skills and capacities. He addressed some but not all of them. Perhaps he was sufficiently confident of his basic approach that he was happy to leave these issues of detail for his successors to tackle? In any event, although there is much for Aristotelian scholars and neo-Aristotelian philosophers to do to clarify these issues further, I shall not attempt to pursue them further here.

That said, in general terms, it seems clear that the actions under discussion are, in Aristotle’s account, individuated by the skills (or capacities) from which they flow, the material changes involved in doing them, and the goals for which they are performed: their efficient, material, and final causes. All three features are relevant to determining what the agent does. Iannis’s action is a body-involving throwing of cargo overboard resulting from his skill as a sailor for the sake of safety (whether or not safety is his aim qua sailor). ²⁵ In this characterization, three Aristotelian causes are used to fix the identity of the process. Nor should this surprise us: actions like weaving or throwing the cargo overboard fall within the category of “enmattered formulae,” which—like anger—cannot be defined without reference to the body involved in the process, its goal, and its efficient cause.²⁶ In his terms, their definition will be unclear if one does not refer to the relevant cause (or causes). Actions cannot be defined without mentioning what suffers (e.g., the agent or his body), their goal, and efficient cause (Meta H.4, 1044b10ff). Without these, the nature (and identity) of the action will not be clear. (I shall return to some related questions in Section VI.)

My focus in what follows is on what makes actions, understood as processes of the type indicated in this section, “voluntary.”

In NE III.1²⁷ Aristotle uses two ideas in summing up his account of “voluntary” action: (A) the starting point (archê) of the action lies in the agent, and (B) the agent knows the particulars involved in the action.

He writes in 1111a22–3: “the “voluntary” would seem to be that where the starting point is in the agent who knows the particulars involved in the action.”

In such cases, the agent acts “voluntarily” because “the starting point of moving the limbs in such actions lies in the agent” (1110a15; see also 1110b4). When agents act “voluntarily,” their action is chosen or selected (haireton: 1110a12). The starting point of the action is a choice (or selection: hairesis, a taking by the agent). Aristotle focuses on what the agent chooses (or would choose: 1110a18) and suggests that this (e.g., the action) will be “voluntary” if it is chosen (even if what is chosen is not worthy of choice: see 1110a30, b7: haireteon). This is because he is concerned to find an internal cause of the resulting action (as required by [A]).²⁸

It is important to note that when the agent chooses to act, his or her choice need not be a preferential choice (sometimes called a “decision”)²⁹ because the latter requires rational deliberation (see 1112a15ff).³⁰ Children and animals act “voluntarily” even though they lack preferential choice (1111a26, b9). While preferential choices may be a subset of choices, there are choices that are not preferential choices. Children and animals do not need to consider whether it is best to act or to forbear from acting (let alone whether it is better to do something else; DA 431a7ff). In some way, they say “yes” to the action in question.³¹ Their choice may consist simply in their selecting (and going for) what seems pleasant to them. By contrast, preferential choice is based on a consideration of what is best to do (NE 1111b15f, 1113a3ff). Elsewhere Aristotle specifies the type of choice (or selection) on which children and animals act as sensual desire (epithumia: 1111b12–3). This is one type of choice or selection (1104b30f: hairesis). Other types are based on what is seen as noble, what is judged best, or what is useful. Choice (hairesis) is a generic term of which there can be several species (one of which is set out in the following discussion of preferential choice in 1111b11ff).³²

Aristotle formulates the knowledge condition, (B), in terms of the undemanding term (eidenai), felicitously translated as “grasp.” B does not require the agent to have scientific knowledge (epistēmē) of the relevant factual claim. Children and animals lack scientific knowledge even though they act “voluntarily.” In a similar way, an acratic can act “voluntarily,” grasping that this is a cake, even when at the time of action he or she lacks scientific knowledge that this is so (1147a8).

But why did Aristotle add a knowledge condition? Isn’t such knowledge simply a consequence of the agent’s choosing to act? Shouldn’t he have favored an account of “voluntary” action solely in terms of choice? This possibility cannot be excluded since Aristotle does not seem concerned in NE III.1–5 to give a definition of the “voluntary” (or of preferential choice) that meets his own conditions on definition, as set out in the Analytics. It may be that, had he developed his account further, he would have arrived at a unified definition involving choice alone. But it is not obvious that he would have done so. There are cases in which the knowledge condition appears to play an independent role. Suppose, for example, Iannis chose to act in a given way, but his choice made him so nervous that he did not know while acting what he was doing. Is his resulting action “voluntary”? Since Aristotle does not explicitly take a view on such cases, it is not clear that he would (or could) have dispensed with the knowledge condition and built his account of the “voluntary” solely on the idea of what is chosen.³³

Choice, when combined with knowledge of what one is doing, is sufficient to make the action done “voluntary” (hekousion).³⁴ In the case of mixed actions, Iannis’s action (of throwing the cargo overboard) is done “voluntarily” (NE 1110a15, see a18), even though no one would choose to throw the cargo overboard for its own sake (or with no further end: see CE 1151b1–5). ³⁵ Such actions are sometimes praised, sometimes censured. Even when one acts under coercion because of factors no one could withstand, the resulting actions may still be “voluntary” provided that one chooses to act (even if the factors go beyond human nature to resist and are pardoned: 1110a25). These “voluntary” actions will typically be pardoned or excused, not praised or blamed.³⁶ What do these remarks show about “voluntary” actions?

Actions done under extreme coercion of the type Aristotle envisages are typically classified (in English) as intentional, not as voluntary.³⁷ Since in NE III.1 the Greek term “hekousion” (“voluntary”) is applied also to these coerced actions, it does not apply solely to voluntary actions.³⁸ Instead, the extension of this term seems to be intentional rather than voluntary actions—where voluntary actions are a subset of intentional ones. Aristotle is, it seems, talking of intentional actions when he talks of “voluntary” ones. (From now on, I shall take him to be offering an account of intentional actions in NE III.1–5.³⁹)

In Aristotle’s discussion, if the starting point of action (archê) is the agent’s choice, it is up to him or her to do or not do the action (see NE 1110a17ff). The order of explanation is as follows:

Claim 1 is the basis of Aristotle’s account. If the starting point is in us, it is up to us to do or forbear from doing the action. We are (in this way) controllers of what we do. (Aristotle employs a similar order of concepts in 1113b21, 1114a19, 1114b30ff.)

Consider the connection between 1 and 2: in Aristotle’s view, the agent’s choice, in the case of voluntary actions, determines what happens. This requires that (i) if he or she had not chosen to do the action, it would not have happened, and (ii) the action is the causal consequence of his (or her) choice. In these two ways, the agent’s choice plays an ineliminable role among the antecedents of voluntary action. These claims will remain true even if the agent could not have rationally chosen differently in the circumstances. For, even in those cases, it is still the case that had he or she not chosen to act, the action would not have happened.

It is still up to the agent to do or forbear from doing the action. This will be true if no one could have endured the pain of acting differently.⁴⁰ Even in such extreme conditions, it is the agent’s choice that determines whether he will act or forbear from acting. Without that choice, the action would not have occurred. Given this dependence on the agent’s choice, doing or forbearing from doing the action rests with the agent (it is up to them to do or forbear from doing the action; 1110a17). What actually happens results from the agent’s choice. Had the agent chosen differently, something else would have happened.

Although these features of the Nicomachean account are relatively clear, there are further issues to consider: one concerns action individuation, an issue raised earlier; the other the connection between an action’s being intentional and praise or blame (sometimes associated with issues of legal or “moral” responsibility).

(Issue 1) When Oedipus—in the famous incident at the crossroads—hit an old man intentionally but hit his father unintentionally, what precisely is intentional, what unintentional (non-“voluntary”), what counterintentional (“involuntary”)? Are there two separate actions (hitting the old man and hitting his father), one of which is intentional, the other unintentional? Or is one and the same action intentional under one specification, unintentional under another? In NE III.1–5, Aristotle does not engage with these issues.

(Issue 2) What is the connection between an action being intentional and praise and blame? An individual action’s being intentional is clearly not sufficient for it to be open to praise and blame. In Aristotle’s view, the actions of wild animals are intentional (1111a26: see the case of the lion in 1118a20ff), even though they are not blamed for doing them. Furthermore, some mixed actions that are intentional are not blamed but pardoned (1110a23ff). Is an action being intentional a necessary condition of its being praised or blamed? This issue is not easily resolved in NE III.1–5. Actions done in ignorance through passion (such as drunkenness) are classified as counterintentional (“involuntary”) when they are regretted (1111a2). However, later in NE III.5, Aristotle comments favorably on Pittacus’s law, which imposed double punishments in cases of actions done in ignorance induced by drunkenness (1113b32). Are the latter cases of unintentional action that is punished? Or does the fact that they are punished make them intentional, on the assumption that actions can only be praised or blamed if they are intentional? (If so, should we doubt our original view that Aristotle’s “voluntary” actions are intentional?) Although Aristotle does not rule explicitly on this question in NE III.5, he does suggest that the primary focus of punishment is ignorance induced by drunkenness (1113b30). This remark might imply that we can reasonably punish actions that result from our ignorance (as is our practice) when we have done something to induce that ignorance—even though the action itself need not be intentional (“voluntary”). If so, we can (in Aristotle’s account) be punished for an unintentional action if there was something intentional (or “voluntary”) about what happened in the “lead up” to the action (such as getting drunk and so causing ignorance).⁴¹ However, while this line of interpretation is attractive and may be suggested by some remarks at the beginning of NE III.1, Aristotle does not make his position clear in NE III.1–5.

Let us consider each issue in turn:

(Issue 1) In the Eudemian Ethics, Aristotle considers in some detail the issue of action individuation while discussing the actions of the acratic and self-controlled. He comments, in his most elaborate formulation: “it is impossible for one to do the same act ‘voluntarily’ and ‘involuntarily’ at the same time and in respect of the same feature of the act” (EE 1223b22ff). The addition of the phrase “in respect of the same feature of the act” is suggestive. It seems to allow that one can do the same act “voluntarily” with respect to some feature and “involuntarily” with respect to another. But what are the features involved?

In CE 1135a26f, Aristotle comments that “the agent must act with regard to the features mentioned not accidentally” (1135a26) if he is to act “voluntarily.” His point is exemplified by a case described as follows: “the person struck could be one’s father, and one grasps that he is a bystander or a man but not be aware that he is your father” (1135a27ff). Oedipus (for it is clearly he!) may hit the bystander “voluntarily” and nonaccidentally but hit his father “involuntarily” and accidentally. Aristotle’s use of the phrases “accidental” and “nonaccidental” is revealing: he is employing the same terminology that he uses elsewhere to describe one thing with distinct features, such as one agent who is nonaccidentally a builder but accidentally a musician (Phys 195b2ff.) or one shade of white that happens to be Simmias’s favorite color (Phys 224b15ff).⁴² In the latter case, there is one whitening (the realization of one capacity) that ends at one shade of color, the one that Simmias likes. There is no suggestion that Aristotle is thinking of two processes, one to a given shade, another to Simmias’s preferred color. If Aristotle is (as it appears) applying the same model to Oedipus’s action, there will be one process of hitting a man (the realization, perhaps, of his capacity as a warrior), who happens to be his father. This action can be described, as Aristotle does, as nonaccidentally hitting the old man (because this captures Oedipus’s goal) while hitting one’s father. When we see Aristotle’s actions as processes that realize and are controlled by an agent’s goal-directed capacities, there is no temptation to see two separate hittings in this case.⁴³

(Issue 2) In CE 1136a5–8, Aristotle classifies actions done in ignorance through a passion “which is neither part of our nature nor common to all men” as “involuntary” (akousia) but not pardonable. Given this addition: (i) actions can be “involuntary” if done in ignorance (and not through ignorance: contrast EE 1225b8); (ii) cases of negligent action (ameleia) are “involuntary” but not pardonable; and (iii) cases of actions done in ignorance brought on by pain/pleasure/negligence are “involuntary” but not pardonable

Assuming (as seems to be the case!) that drunkenness is a passion that is specific to particular agents and not common to all humans, an action done in ignorance resulting from drunkenness will be “involuntary” but not pardonable. Aristotle is focusing, in this discussion, on questions of justice, punishment, and acting unjustly. His remarks in 1136a5–8 flow from this perspective: some actions, although “involuntary,” are not pardonable because the agent came to do them because of his or her own specific nature. The agents themselves were the cause of the resulting actions even if the latter were not done knowingly. In commenting on these cases, Aristotle shows that an action’s being “voluntary” is not (in his view) a necessary condition for its being praised or blamed. In CE 1136a5–8, Aristotle clarifies the connections between the “voluntary” and praise and blame that were left (somewhat) unclear in the NE III.⁴⁴

The positive account of “voluntary” action offered in the NE and common books (CE) differs in certain respects from that offered in the EE. In this regard, the NE may seem less reliant than the EE on the use (and coherence) of concepts drawn from Aristotle’s theoretical philosophy.⁴⁵ One can, in the NE account, act intentionally even if one acts because of a pain no one could withstand or for reasons that arise from the difficulty of one’s circumstances, as when one’s action is coerced. In such cases, one does not act in the way one would naturally, if unconstrained by external conditions. One’s motivations need not be ones that are (in the terminology of the EE) “natural” to one. According to the NE account, one can act intentionally even though one is not the complete controller of one’s actions in the way required in the EE. Aristotle, in the NE, accepts that we can act intentionally (and be praised and blamed) when doing actions that do not flow from our natures, even when we cannot bear the pain of not acting. This is because, in such cases, it is our choice that determines what happens. In this respect, the NE account of the “voluntary” applies to a wider range of cases than the one developed in the EE.

The theory in the NE, despite being more wide-ranging than that in the EE, offers a more unified account of the antecedents of “voluntary” action. In the NE there is, it seems, one general type of cause that leads to action: choice (hairesis) of which there can be different types (preferential choice, choice in anger, for sensual pleasure, etc.; see 1104b31). Since Aristotle uses the notion of choosing (haireisthai) frequently in NE I (1094a20, 1097a26, b3–6), it is no surprise to see it (and its derivatives: choice (hairesis) and what is chosen (haireton) used in crucial contexts in NE II and III. Whereas in the EE Aristotle suggests that one can act on distinct types of motivation (reason, desire), in the NE he proposes choice (hairesis) as a type of motivational state common to the whole range of cases he considers: animals, acratics, self-controlled, virtuous. Armed with this starting point, he can allow animals and children to act intentionally (“voluntarily”), even though they lack the type of intellect (dianoia) demanded of “voluntary” agents in the EE. Indeed, the general notion of choice (hairesis), which is not confined to rationally based choice, is important in that it offers the basis for a unitary type of starting point common to all “voluntary” actions. It is striking that the EE does not use the verb “choosing” (haireisthai) and does not give “choice” (hairesis) an important theoretical role.⁴⁶ The NE account, by contrast, in taking the idea of what is chosen (haireton) as fundamental, avoids a difficulty implicit in the EE, nature-based account.

The EE, as already noted, uses the idea of “nature” in two ways: action, if it is to be “voluntary,” should follow from one’s natural desires or reason and be such that one’s nature can bear the pain of acting differently. But why impose both conditions? Why should an action that one can forbear not doing be made involuntary by the fact that its motivations are not natural to one (e.g., in cases of mild coercion)? Conversely, why, if motivations spring from one’s nature, should one also require (for “voluntary” action) that one’s nature could bear not doing the action (as might happen when one’s nature has become so crystallized that one cannot bear acting differently; see 1114a16ff)? It is not obvious why both conditions are required or how they cohere. In this respect, the NE rests on less controversial, perhaps more defensible, concepts than those employed in the EE.

What accounts for these theoretical differences? Although this question requires further study, I shall conclude with a tentative hypothesis. In the EE, the “voluntary” does not extend to include actions that are coerced or “unnatural” (out of character). In the NE, by contrast, Aristotle is attempting, I have suggested, to formulate a concept of intentional (“voluntary”) action, which is not confined to character-revealing actions but can be used in legal contexts and for praise and punishment more generally (1109b32ff).⁴⁷ Here, he is developing a more inclusive account of intentional action, resting on the idea of choice as the appropriate starting point, which (when spelled out) can be applied to children, animals, and mature adult humans. Even if ideas of natural character (or ethical nature) are useful in characterizing actions that reveal our characters, they are not required in an account that applies to all intentional actions. Aristotle, in the NE, is, in effect, separating the issue of what is an intentional action from a quite different one, apparently to the fore in the EE: what needs to be present if an action is to reveal one’s character as virtuous or vicious.⁴⁸

Aristotle’s answer to this question (at least in the Nicomachean Ethics) has several aspects: (i) the action, understood as a process, is the realization of one of the agent’s capacities; (ii) the agent chooses (or selects) to do the action, where his or her choice is the efficient cause of the action; and (iii) the agent knows while acting what he or she is doing (throwing the cargo overboard to save the crew). In this account, the agent plays an ineliminable role: it is not just that his or her choice is the efficient cause of the action. In addition, the action itself is the realization of one of his or her capacities and, as a result, must be understood as his or her action. Iannis is the mover, and what happens is the realization of his capacity to move. Furthermore, in acting voluntarily, he knows what he is doing and controls and guides his action—the ongoing process—in the light of the goal he has set himself and his knowledge of how his action is progressing. Iannis has the ability to act and will do so, if he chooses. One reason that Aristotle talks of the agent as the cause and controller of his action is that the agent plays several essential roles in the action: he or she is the one unifying locus of the relevant capacity, the choice that leads to action, and the knowledge that guides it. Since the relevant cause of “voluntary” action must play all these roles, Aristotle specifies the cause as the agent, the starting point and knowing controller of the relevant process, itself the realization of one of his or her capacities. It would be a mistake to specify the cause of this “voluntary” action simply as Iannis’s choice to act since that ignores the essential role played by his capacity to act and his knowledge in guiding the action in question. In Aristotle’s account, the action is (in the cases we have considered) a process, the exercise of a capacity of the agent directed toward one of his goals. Iannis’s goal and his skill as a sailor both play some role in determining what he throws overboard, when and where he does so, and the manner in which he acts.⁴⁹

It might seem that there are too many efficient causes in Aristotle’s picture. The mover is Iannis. He moves as he does because he is a sailor, guided by his skill as a sailor. He also selects the action which he does as a result of his deliberation about how best to achieve his goal of saving himself and the crew. His skill as a sailor also plays a role in his deliberation about what to do in a storm when his ship is in danger of going down. However, his selection of which action to do is the immediate cause of his action. It might seem that there are three efficient causes of the same process: Iannis, his choice, and his skill as a sailor. How do they fit together in Aristotle’s account of “voluntary” action? Or is the action (in some way) overdetermined by distinct, independent causes?

The three features mentioned cause different aspects of what happens. Iannis’s selection or choice to throw the cargo overboard here and now is the immediate efficient cause of his acting when and where he did. He also selects his goal: to save the crew and himself rather than to go down with the ship. Iannis himself is the mover, the substance with the physical abilities to drag the cargo to the side of the ship and throw it into the water. He initiates and carries through the action. His skill as a sailor is what causes him to act in the way he does (gathering the cargo together while, e.g., turning the ship into the wind). It also guides and informs his deliberation about when and how to act, leading him (given that his goal is safety) to choose to act as he did. But how do all these features fit together into one unified story?

In Aristotle’s account, Iannis has an overriding goal (the safety of the crew in situations of this type) and then works out—using his skill as a sailor—how to achieve it. His deliberation leads to his choice as to what to do here and now. In implementing this choice, he uses his skills as a sailor and his physical strength to throw the cargo overboard in the way he does. With a less skilful person, the action would have developed differently.

Processes may be, once again, important for this account. There is, it seems, one unfolding process guided throughout by Iannis.⁵⁰ His deliberation, selection of goal, choice of action, and subsequent action are all parts of this one process which begins with his working out what to do in a storm and ends in his action of throwing the cargo overboard when and where he does. His skill and physical strength are used in this process. What occurs is not properly described as a sequence of events occurring one after another: Iannis’s perception of danger, his selection of a goal, his choice of an action, his movement of his limbs. If one thinks in the latter way, one loses sight of the unity of (i) the goal-directed process that begins with Iannis’s practical thinking and ends in his action and (ii) Iannis’s role as the initiator and guide of this process (in the light of his skill, character, etc.). The three causal features mentioned earlier are unified in that they have distinct roles to play in this one agent-controlled process whose final stage is his acting.

Aristotle’s account can be contrasted with action as it is generally understood in contemporary philosophy. Many modern accounts, inspired by Davidson’s seminal work, begin by construing my moving of my hand as follows: ⁵¹

Most theorists are reluctant to accept that there is a brute causal relation between the agent and the event and so (reasonably enough) invoke the agent’s wants, choices, or intentions as causes. Thus, they add to A:

Modern action theorists often add two more ingredients to their account:

As we have seen, Aristotle explicitly rejects this picture: actions can extend beyond the body of the agent.

The desire in question is not, in Aristotle’s account, an event of this type. Instead, it is best presented, when active, as a process: Iannis’s being actively drawn or attracted to eating an apple now (DA 433b18). What is causally important, on this view, is not the onset (let alone the “onslaught”) of the relevant desire, but the agent being drawn to the object in question. Iannis’s being attracted to the apple (perhaps taking pleasure in the prospect of eating it: DA 431a8–12) is part of a process that ends with his acting as he does (picking up the apple and biting into it).

Aristotle’s picture, in these respects, differs fundamentally from those modern accounts of agency that present action as the last event in a sequence of causally connected events. His is, through and through, a process-based view in which the causal role of the agent is central and ineliminable.

Much more remains to be investigated. While we have considered certain aspects of Aristotle’s account, we have not examined the role of practical thinking (or the practical syllogism) in leading to an agent’s choice. It seems, at least in De Motu and De Anima, that such thinking is a type of process (or perhaps activity) which the agent engages in prior to action. On occasion, an agent can arrive at a conclusion of practical thinking and still not act (as in the case of weak acrasia: CE 1149b9–12). However, it lies outside the scope of this essay to consider the precise connections, in Aristotle’s account, between practical thinking, choosing, and acting. ⁵³

Despite these and other major gaps, enough may have been said to show that, in several important respects, Aristotle developed an interesting and challenging way of thinking about action and agency with much to commend it. He offers a radically distinct and distinctly promising alternative to contemporary orthodoxy in action theory.⁵⁴