2.1. Aristotelian two-way powers are based in, and consist in, systematic knowledge

Aristotle never used the term ‘two-way power’.Footnote ¹ In the relevant main text, Metaphysics Book Theta, Chapters 2 and 5, Aristotle’s term is dunamis meta logou, or ‘potentiality with an account’. This is often translated as ‘rational power’: an elegant but potentially misleading contraction. I prefer the uglier ‘account-based power’ to ‘rational power’, for reasons I explain below.

Aristotle gives arts like medicine and building as examples of account-based powers. Account-based powers, unlike other powers, are directed to contraries. Hot stones have a power to heat. Hot stones don’t, as such, have a power to cool. The expert doctor, by contrast, can produce health by applying her medical knowledge, but she can also produce illness by applying that knowledge if she wants. The reason account-based powers are distinctively capable of contraries, unlike all other powers involved in motion or change, is that they’re based in systematic knowledge. Aristotle (Metaphysics Book Theta, 1046b7-10) takes pains to explain that account-based powers are unlike non-account-based powers because ‘knowledge is an account, and the same account clarifies both the thing and the privation, though not in the same way’. For example, the medical art includes an account of health, which clarifies health and also the privation of health, illness. ‘Systematic knowledge’ is my term for this kind of account. It’s because they depend on this kind of account that I call them ‘account-based powers’.

Although Aristotle didn’t use the term, there is a relatively clear reason to call account-based powers ‘two-way powers’, because they’re directed to contraries, like health and illness, in a distinctive way.

‘Systematic knowledge’ is my term, not Aristotle’s. Why think that this knowledge is systematic? Couldn’t an ‘account of health’ be wholly given in a solitary abstract formula: as the good condition of a living thing’s body, say? No. Aristotle (Physics, 194a20-25) says ‘it is the part of the same discipline to know the form and the matter up to a point (e.g., the doctor has a knowledge of health and also of bile and phlegm, in which health is realized and the builder both of the form of the house and of the matter, namely that it is bricks and beams, and so forth)’. So even if an account of health is exhausted by an abstract formula, in the medical art this formula must be connected to material conditions. Besides, Aristotle’s illustration of how an account of health can be put to work also indicates that it is not exhausted by a mere solitary abstract formula, but involves rich conceptual connections:

Here we see a detailed, systematic account of health put to work in the service of desired change. At its core, the account answers the question ‘What is health?’. The answer to this question is not the empty abstract formula ‘Whatever is the good condition of a living body’ because it follows from the account of health (‘…since this is health…’) that various consequent conditional claims hold (‘if the subject is to be healthy…’, etc.). In other words, the relevant account of health already includes rich connections to other ideas, like uniformity, change, heat, and so on. I call this ‘systematic knowledge’ to indicate that it is detailed and involves many contentful connections organized around central ideas.

Why not just say that two-way powers are based in ‘scientific knowledge’? This is true, but it risks missing something important: namely, that it is through systematic connections between theory and practice that we make sense of the power making it no accident that something relevant is produced. Two-way powers consist in productive systematic knowledge that contains a core of theoretical systematic knowledge. For Aristotle, scientific knowledge (epistêmê), strictly speaking, is theoretical, not practical or productive, and indeed is a kind of pure theory, because it’s of what’s universal, necessary, and eternal.Footnote ² For the medical art, part of the theoretical core presumably includes universal, necessary, eternal truths like ‘Uniformity of the body must be present for a subject to be healthy.’ But such theory is not, as such, concerned with practicalities. A mere theoretician might know all about the nature of health and still not know how to put this theory to productive use, because they are at a loss when dealing practically with the contingent. No body of mere theory is a productive science, and it is the medical art that is an example of a two-way power, not merely its (purely) theoretical core. The sketched example of medical practical reasoning illustrates this. At later steps in the reasoning, we depart from theoretical considerations about health, uniformity, heat, etc., and instead consider premises that concern practically-significant aspects of the situation (e.g., ‘This patient is cold’), and what the physician can do here and now (e.g., ‘Rubbing would warm this patient.’). The reasoning traces a path from the purely theoretical core of the medical art, which is scientific knowledge, strictly speaking, to its productive outer layers, which are not. The medical art too is ‘systematic knowledge’ in my sense; it’s a suitably loose term.Footnote ³

Why then is ‘rational power’ a potentially misleading translation of dunamis meta logou? If one simply followed this translation, it would be natural to assume that a rational power is any capacity, disposition, habit, etc., that is somehow assignable to the rational part of the soul, such as the power to know, or knowing Greek, or perhaps flute-playing, or even the capacity to obey one’s master. One might then think that what Aristotle says about two-way powers generalizes to these powers. The account of two-way powers doesn’t generalize in this way. Aristotle never provides clear examples of a dunamis meta logou apart from medicine and building. Flute playing is explicitly given as an example of a power that does not involve learning, although it does involve practice (Aristotle, Metaphysics, 1047b30–35): in the text, it functions as a contrast to building and medicine, rather than as another example of the same kind of power. Aristotle is also clear, early in the Metaphysics, that skills acquired merely through habituation and experience are like the one-way powers of inanimate things:

Even if the power of a manual labourer to respond promptly and competently to the order ‘Make a wall of bricks here’ is assignable to the rational part of the soul, it is not like the master-builder deploying her account-based power. The habit of the manual labourer, however well-developed, is not a two-way power: it is instead like fire’s one-way power to burn. This is quite plausible. Someone may be very well habituated to assemble walls of bricks at command while having little or no idea how to take the bricks apart again.Footnote ⁴ Their habit of assembling walls doesn’t become a two-way power by being coupled with another habit that deals with orders like ‘Demolish that wall’. The manual labourer’s habit would rather need the systematic articulation involved in knowing what a wall is; why it is as it is when serving its function well, etc.; how those relate to bricks, and to what one can do here and now; etc. – in other words, all that’s involved in the systematic knowledge of building – to count as a two-way power. The same goes for knowing Greek and similar cases. It is one thing to efficiently and habitually conduct linguistic exchanges in Greek – a kind of ‘manual labouring’ at language – and quite another to be a master-linguist who knows grammar and the ‘whys’ of linguistic production.

2.2. Systematic knowledge is the only explanation of how two-way powers are possible

Aristotle’s only explanation of how two-way powers are possible, given that every other principle of change is ‘one-way’, turns on the fact that an account of something like health is also, in a way, an account of the privation of that thing. He also flirts with an objection that two-way powers are impossible, because they are powers for contraries, and one might want to produce both at once, but nothing can have both contraries produced in it at once (Aristotle, Metaphysics, 1048a15–25). He clearly thinks that the possibility of two-way powers is puzzling enough to be worth explaining and defending.

If there were Aristotelian two-way powers that didn’t consist in systematic knowledge, we would expect a different explanation of their possibility, or at least acknowledgement that this is a puzzling unresolved question. For example, if non-human animals or young children or uneducated manual labourers or natural slaves had two-way powers that didn’t consist in systematic knowledge, we would expect Aristotle to try to explain how these powers are possible, even though they are quite different from medicine and building. But there is nothing like that – not even a hint – in the entire corpus.

Aristotle says many times that self-moving animals have in themselves a principle of motion and rest. For example, discussing the self-motion of living things, Aristotle (Physics 255a5–10) says that ‘… if e.g. a thing can cause itself to walk it can also cause itself not to walk’. Is the self-motion of a giraffe a two-way power because it concerns motion and rest? No. It’s not puzzling. There is no need for that sort of power to invoke knowledge, or something analogous, to explain how the power is even possible, in contrast to all the other ‘one way’ powers that there are.Footnote ⁵ In fact, Aristotle (Metaphysics 1051a5-11) says that all potentialities are in some sense for contraries:

That again is not puzzling (at least for Aristotle). Apart from the power to build, these others are not examples of two-way powers in the special puzzling sense whose possibility, in contrast to one-way powers, requires explanation by appeal to knowledge.

What’s the difference? Aristotle (Metaphysics, 1050b30–35) says that ‘that which can move another in this way can also move it not in this way, i.e. if it acts according to [an account]. But the same [non-account-based powers] can produce opposite results only by their presence or absence’. It’s a little unclear how to understand ‘their presence or absence’ here.Footnote ⁶ However we understand it, the passage marks a contrast in how account-based and non-account-based powers work to relevant contraries. The doctor’s account-based power produces health or illness through its actualization. But a power to be heated, even if it is in some sense the same as the power to be cooled, does not produce being cooled through its actualization. If the power to be heated is actualized, because something hot is present, the bearer of the power is heated, not cooled. It is only on condition of the absence of being heated – absence of that actualization or ‘result’, for one reason or another – that the bearer is cooled.

We should also note that when Aristotle (in translation) says that ‘one and the same’ potentiality is a potentiality for health and illness, we shouldn’t take him too seriously. Elsewhere, Aristotle (Physics 201a30–201b5) says that ‘[to] be capable of health and to be capable of illness are not the same; for if they were there would be no difference between being ill and being well. Yet the subject both of health and of sickness – whether it is humour or blood – is one and the same’. So, he was probably speaking loosely when saying that one and the same potentiality is for opposites.

2.3. Aristotelian two-way powers are determined to exercise by separate desire or choice

For Aristotle, non-account-based powers have conditions of exercise that are necessary and sufficient for exercise. If a hot rock meets a suitable heatable thing, in suitable conditions, it heats it. But account-based powers have conditions of exercise that are merely necessary for exercise. If a doctor meets a patient who is in suitable conditions to be healed or sickened through applying medical knowledge, it is not yet determined what will happen. For this reason, Aristotle (Metaphysics, 1048a9) says that ‘[that] which decides [which way the power is exercised], then, must be something else; I mean by this, desire [orexis] or choice [prohairesis]’. Without determination by desire or choice, considered as alien to the two-way power itself, the two-way power won’t be exercised at all. For example, an indifferent doctor might play golf instead of expertly healing or sickening her patient: in that case, her two-way power is not exercised one way or the other. Whatever change there is in the patient’s condition in such a case is not explained by application of the doctor’s knowledge (one way or the other).Footnote ⁷ The case is different with deliberate expert poisoning. There the doctor wants to sicken, and their knowledge does have a distinctive positive role to play in explaining the relevant change.Footnote ⁸

Many aspects of rationality don’t involve determination to exercise by separate desire or choice. Understanding something brief, said loudly, in a language I understand, when conditions are ripe for understanding it (i.e., I am not lost in thought, asleep, etc.), doesn’t depend on desire or choice the way that concentrating on a boring conversation does. I just understand what is said, whether I like it or not. Similarly for many aspects of theoretical knowledge. It is not as though, knowing health, one only knows illness by desiring or choosing to know it. Rather, in knowing health one already knows illness (albeit in a different way, as the privation of health). One can’t help but know it. Similarly, for ‘moral virtue’. The virtuous person does not have their virtue further determined to exercise by separate desire or choice: rather it already includes some desires and not others. If someone virtuous (or better: someone we thought virtuous) does something vicious, that is not their virtue at work in the ‘opposite’ way; but if a master-builder demolishes their house, that is their building art at work in the ‘opposite’ way. Similarly for practical wisdom. Aristotle (Nicomachean Ethics, 1144b30) is very clear that one cannot have practical wisdom without having moral virtue and vice versa. So as with true moral virtue, if someone practically wise (or better: someone we thought practically wise) does something vicious, that is not practical wisdom at work in the ‘opposite’ way. Practical wisdom is not a two-way power.

Kenny (Reference Kenny1975, pp. 52–53) says that a dog has a two-way ability to come when called, or not, and that this ability ‘interested Aristotle, who drew a sharp distinction between rational powers, such as the ability to speak Greek, and natural powers like the power of fire to burn.’ He goes on to criticize Aristotle for conflating two-way powers and rational powers, because, for example, understanding things said in a language I know isn’t normally determined by desire or choice. We can already see that Kenny is wrong about Aristotle in several ways. Aristotle doesn’t think that any non-rational animal has any two-way power. Metaphysically, he wasn’t much interested in the powers of non-rational animals, except insofar as they are powers of self-motion, which is not, as such, a two-way power, because it’s not based in systematic knowledge. Aristotle doesn’t think that merely habitually speaking Greek is a two-way power either; that is ‘manual labouring at language’. One needs systematic knowledge to inform one’s speech (in order to produce something) before a two-way power is really in view.

2.4. Aristotelian two-way powers are principles of change

At the opening of Metaphysics Book Theta, Aristotle says that he will first explain potentiality in ‘the strictest sense’, which concerns motion and change, even though this is not most useful overall. (This is an example of Aristotle proceeding from what is most intelligible to us to what is most intelligible as such.) It is only in Chapter 6 of that Book that he turns to discuss potentiality in other senses, understood by analogy with potentialities that are principles of change. The entire discussion of account-based powers is within the scope of this restriction. There is no attempt to describe analogous two-way powers that are not principles of change, as there is discussion of potentially being a statue as analogous to potentially heating. We may conclude that two-way powers are principles of change.

Aristotle had restrictive views about change. He argues that apart from generation and corruption, there are only three kinds of change: change in place, change in quality (in the sense of being acted upon), or change in quantity (Aristotle, Physics, 226a15–b10). He explicitly denies that to acquire virtue of character or an intellectual virtue like scientific knowledge is to change in quality: rather it is a perfection (Aristotle, Physics, 246b5-15). So, if we don’t see the possessor of a two-way power making a difference in place, quality (in the sense of being acted upon), or quantity, then no change is, strictly speaking, at work, and neither is any two-way power. For example, if to judge or to choose is not to change oneself, in Aristotle’s sense of ‘change’, it is not the exercise of an Aristotelian two-way power.

Aristotle’s claims about change are softened a touch in De Anima, but not by much. There he says that ‘what possesses knowledge becomes an actual knower by a transition which is either not an alteration of it at all (being in reality a development into [the same state and into] actuality) or at least an alteration in a quite different sense’ (Aristotle, On the Soul, 417b5–10). Later he says that ‘alteration’ (i.e., change in quality) is an ‘incorrect term’ to use for further actualization of potentialities, but also that we can’t help using the term (Aristotle, On the Soul, 418a1–5). The best we can do is acknowledge that change in this ‘further actualization’ sense is quite unlike change in the sense of being acted upon by another.Footnote ⁹

One might think that Aristotle left it to the reader to work out analogous two-way powers which are not principles of change, and that Chapter 6 of Metaphysics Book Theta sketches examples of how to carry out this exercise. But none of the examples given there (such as waking; being a statue; seeing; etc.) suggest two opposed ways of actualizing one and the same power. There is a good reason why Aristotle restricts his discussion of account-based powers to principles of change (proper). Consider the doctor’s activity in abstraction from the specific effect produced (whether health or illness). Whatever the case, the doctor is doing the same thing: applying her medical knowledge. There is no opposition between exercises at this level of abstraction. If we bring the patient into our thought of this activity, but abstract from the specific change effected, we still have no opposition between exercises: whatever the case, the doctor is applying medical knowledge to effect a specific change in the patient. We need to consider not only the patient involved, but the specific contrary effected in the patient, before we bring two opposed ways of exercising one and the same power into view: healing and sickening. That is the basic reason why two-way powers are all principles of change, and not merely analogous to principles of change. It is only with principles of change that two opposed ways of exercising one and the same power emerge, such as the way towards health, and the opposite way towards illness.

Consider how this point applies to many forms of further actualizing a capacity to know, such as reflecting on what one knows, or attending to what one knows, etc. A power to know health is surely a ‘rational power’ in some sense. To reflect is not to change oneself, in Aristotle’s restrictive sense of ‘change’, but let us suppose for a moment (‘helplessly’) that such further actualizations of a power to know health are closely analogous to genuine change. One might then think that even merely theoretical (i.e., non-productive) knowledge is an excellent candidate to be a two-way power, because it is knowledge, and the same account clarifies the thing and its privation, etc. It’s true that knowing illness is not opposed to knowing health, in that to know health just is to know illness too. But when one reflects on what one knows, aren’t there two opposed ways of reflecting: one for reflecting on health and one for reflecting on illness? No. There are (at least) two varieties of reflecting on what one knows, which are defined by objects that are contraries. But the varieties themselves are not opposed in a way that makes for genuine ‘two-ness’. We only understand illness through health, as its privation: there is no way to reflect on illness without reflecting thereby on that of which it is the privation. So, reflecting on illness is not opposed to reflecting on health in a way that makes sense of ‘two ways’. The same goes for similar acts, like attending to illness: thereby one attends to health, of which it is the privation.

2.5. Aristotelian two-way powers are principles of change in another, or in oneself as other

Two-way powers are principles of change. Can we conclude something stronger? Aristotle (Metaphysics, 1019a15–20) says, ‘[c]apacity [dunamis] … is the source, in general, of change or movement in another thing or in the same thing qua other’. (Here he is defining the primary sense of dunamis, before moving on to analogous senses in Chapter 6 of Metaphysics Book Theta.) If the doctor heals herself, what she produces is health, not the medical art, and it is as patient, not as doctor, that she is healed. Perhaps we may conclude that two-way powers are all principles of change in another or in oneself as other. That is: all two-way powers involve alienation (real, or at least notional) from what they act on or produce.

What about capacities for activities like flute-playing, or dancing? Such skills are not aimed at a product distinct from the activity, but they are surely principles of change or motion.Footnote ¹⁰ When exercising such skills, one moves oneself, changing one’s location or the position of one’s limbs, etc. Here we seem to deal with aspects of ‘second nature’, where nature is a principle of motion or change in oneself.Footnote ¹¹ Such exercises seem determined to exercise by separate desire or choice. An indifferent flute player won’t play the flute; she must want or choose to if she is to play. Can’t such aspects of second nature provide examples of two-way powers that don’t involve alienation from what they act on or produce?

I don’t think such skills are two-way powers, and I will return to that issue in the next section. For now, supposing that they are (somehow) two-way powers, it is enough to note that this doesn’t challenge the claim about alienation. Skill in flute-playing is a principle of change in the air and (thereby) what’s audible and perhaps even (thereby) the audience. None of these are the flute-player, just as such: alienation is involved. Even the relevant changes in the flautist’s lungs, lips, and limbs involve alienation. Aristotle (Physics, 257a1–15) is relatively clear that self-motion is only possible in something that has parts that are moved, and that are not identical with the mover, nor with the whole, if that is different. This is not to say that Aristotle doesn’t really believe in self-motion; he clearly does. It is just to say that the flute player is not her lungs, and neither is her skill. Aristotelian self-motion is only possible through a specific kind of other-motion: the kind involved in moving one’s parts.

2.6. Aristotelian two-way powers are all arts (i.e. productive forms of knowledge)

Aristotle (Metaphysics, 1046b1–5) says that ‘all arts, i.e. all productive forms of knowledge, are potentialities; they are principles of change in another thing or in the artist himself considered as other’. Building and medicine are the only examples Aristotle gives of account-based powers. Aristotle is not shy of varying examples when doing so illustrates important conceptual points. May we conclude that Aristotelian two-way powers are all arts, i.e., productive forms of knowledge?

What of skills like flute-playing or dancing? Although such skills are principles of many relevant changes, they are not, as such, aimed at production of an effect distinct from the activity itself. Rather the end of flute-playing lies in playing the flute (well); the end of dancing likes in dancing (well); etc. These skills seem dependent on desire or choice for their exercise. Why aren’t they two-way powers?

The main reason is that such skills have no properly opposed contrary actualization. The medical art has healing and sickening as properly opposed contrary actualizations; neither is a determination of the other. The building art has building and demolishing as properly opposed actualizations. But what is the contrary exercise of the power to play the flute? Banjo-playing is not the contrary. No other instrument is the opposite of the flute, and one does not exercise one’s flute playing skill in banjo playing. Refraining from playing the flute is not the contrary exercise of the power either. It is not the role of the art of flute-playing to decide when to refrain from flute playing, except where rests in music are concerned, and even there, one is still playing the flute in a broad sense. Resting in flute music is not opposed to flute-playing: it is part of it, much as pausing between breaths in CPR is part of the medical art. One might think that playing the flute badly – making many mistakes, playing in an ugly way, etc. – is the relevant contrary to playing the flute. But that doesn’t make any sense. Being contrary is a kind of opposition. Playing the flute badly is not opposed to playing the flute. Indeed, if one is playing the flute badly one must be playing the flute: the former is a determination of the latter.

Playing the flute badly is opposed to playing it well. But if someone who plays music well deliberately plays ‘badly’ they are, in a qualified sense (i.e. given a special aim), still playing well. It just might not sound that way. For example, I used to be in a choir. Once the chorister next to me mocked an unskilled singer who tended to sing flat by deliberately singing precisely a semi-tone flat the whole way through. This is very difficult! As an exercise in transposition, it was masterful (if mildly cruel): it displayed great skill and knowledge. The morality of mocking aside, it is only when considering the music as product – as faithful production of what’s in the score, say – that we can say that it was bad; as musical mockery it was good. So we must put scare quotes around ‘badly’ for such cases. Aristotle agrees. In his commentary on deliberate limping in Plato’s Hippias Minor, Aristotle (Metaphysics, 1025a5-15) explicitly describes deliberate limping as pretending to limp. Similarly, for playing the flute ‘badly’. There isn’t real opposition between expertly pretending to play badly and expertly playing the flute, because the former is a determination of the latter, albeit one that messes about with appearance and reality.

Can’t an expert flautist play badly (without scare quotes) while drunk, or while deceived, or while pursuing an extraneous end that distracts from, or overcomes concern with, the end of flute playing? Yes, of course; they’re only human. But in those cases, there is some special distorting factor: it is an accident that they play badly, given that they have the power to play well. If we are to model playing the flute badly on sickening through application of medical knowledge, we need to think of cases where it is no accident that the relevant contrary is realized, relative to the power in question. As noted in Section 2.2, two-way powers produce contraries through their actualization. To find a good analogy to a productive two-way power, the analogous power must be fully at work to its contrary. But there just isn’t something like that available for powers that are not productive forms of knowledge.

Aristotle seems to think of skills like flute-playing as mere knacks, based mainly in repetition, habituation, and experience, rather than as arts, informed by knowledge of causes. Such skills are not based in systematic knowledge, so they aren’t two-way powers (see §2.1). Flute-playing, so conceived, is a principle of change, but like fire’s power to burn, or uneducated bricklaying, it is a ‘one-way’ principle of change that has only one way of being actualized. That said, the view that flute-playing in particular is a mere knack is clearly not central to Aristotle’s philosophy. Let us suppose that for conservatory-educated flautists, their flute playing is at least partially informed by their knowledge of music theory. (Compare thoroughly uneducated buskers or young children who merely imitate.) Such flautists are not ‘manual labourers at flute-playing’; they have some science informing what they expertly do. Is ‘scientific flute-playing’ an example of a two-way power at work?

What have we imagined here? A pure musicologist might know all about music theory without those universal, necessary, eternal truths about melody, etc., informing what they do when they pick up a flute. If we conceive of scientific flute-playing as producing musical effects with a flute, then it is not the mere musicologist but the master-composer who is the one who knows the ‘whys’ of music in a way that is relevant to producing musical effects. Here we think of a composer as a ‘builder of music’, whether they work at building notation (as a builder might draw up a floor plan) or directly with the materials (i.e., the sounds) themselves. Conceived this way, the imagined art of ‘scientific’ flute-playing is a productive form of knowledge: it is just like building. Just as a master-builder might build an imposing lintel, a master-composer might produce a searing high C on the flute. The master-composer also knows how to produce the privation of the same; a two-way power is at work. (Compare the uneducated ‘manual labourer’ at flute playing who has no idea why this is a searing high C, etc.)

This way of conceiving of flute-playing doesn’t seem very Aristotelian, because it makes the end of such flute playing lie in something distinct from the activity (namely, the effect). (It seems alright as an account of ‘musical building’, or composition, as we call it.) But if we suppose that ‘scientific flute-playing’ is instead a power for activity as such, and is not a productive form of knowledge, then we can’t identify an appropriate contrary exercise of the relevant power, for the same reasons as before: no relevant exercise is properly opposed to flute-playing. Absent some other model for a non-productive two-way power, we may conclude that all Aristotelian two-way powers are arts, i.e., productive forms of knowledge. That’s why Aristotle only gave medicine and building as examples.

2.7. Two-way powers explain both specific contrary effects as no accident, relative to the power

We have ruled out skills that aren’t (productive) arts as examples of two-way powers. We have also ruled out some further actualizations of powers, such as reflection on what one knows, as examples of exercises of two-way powers that are (merely, ‘helplessly’) analogous to principles of change.

What of powers of theoretical judgment and choice, considered as topic-general two-way powers? One can believe something or deny that very same thing: aren’t these clearly contrary exercises of one and the same theoretical power to judge? One can choose to do something or choose not to do that very same thing: aren’t these clearly contrary exercises of one and the same practical power to choose?

Both belief and choice, as powers, have exercises defined by contrary objects, and this seems part of Aristotle’s conception of such things. But I think they are not two-way powers nor closely analogous.

Earlier I noted that in order to find plausible analogous two-way powers which are not principles of change in something other, we must identify contrary exercises where it is no accident that the contrary is realized, relative to the power. Another way to put this: if an analogous two-way power is at work, then the systematic knowledge in which it consists must be fully at work in both contrary exercises. If an expert doctor heals you, it is no accident that you are healed, because they know what they’re doing; likewise for being sickened. Any candidate analogous two-way powers must be like that: they must explain both contrary exercises or results as no accident, relative to the power.

One’s knowledge can explain a judgment as no accident. If the evidence – what I know, considered collectively – points to the butler, that can explain my judgment that the butler did it as no accident, and perhaps even explain why I hit the truth rather than the false as no accident too, depending on what sort of evidence is involved and how much. But in that kind of case, the ‘contrary judgment’ must be an accident, relative to the power. If my evidence points to the butler, and I deny that they did it, then my denial is an accident relative to what I know; something else, like a desire not to cast aspersions, must explain my denial, and it will be an accident if I happen to be right. Similarly on the practical side. If my evidence – what I know about how to live well – suggests that drinking is a bad idea, that can explain my choice not to drink as no accident, and perhaps even explain why I hit the practical truth (where this is ‘truth in agreement with right desire’).Footnote ¹² A weak-willed person might well drink in that situation. That choice and action is an accident with respect to what they know. They don’t drink because they know what to do; they drink for some other reason. A given body of knowledge cannot explain both judgments on either side of a question – such as whether (to do) A or not – as no accident. Either one side would be an accident relative to the body of knowledge or they both would.

What about cases of equal and conflicting evidence? Sometimes we may face difficult situations where the evidence is (or at least, seems) equally plausible on both sides, and yet we have to make a judgment anyway. When we do, responsible judges (for example, in courts of law) often try to cite grounds that justify the judgment they make. But in doing so, they cite a subset of their relevant body of evidence. Relative to the cited body, the judgment may be no accident. But relative to the whole, any judgment is an accident, because ex hypothesi the whole does not favour it. Another way to put this: sometimes (in courts of law, say) we pretend, for good reasons, that we know less than we do.

I have focused on what’s informed by knowledge: ‘educated guesses’ and ‘informed choices’ as we say. What about a power of judgment conceived in abstraction from knowledge, as some non-epistemic ‘deciding factor’? Isn’t that a two-way power? Certainly not in Aristotle’s sense: it precisely doesn’t consist in systematic knowledge. But I don’t think this should count as a two-way power in anyone else’s view either. Powers explain their exercises – their actualizations – as no accident. A primitive non-epistemic power of judgment explains that something was judged as no accident. But it never explains that one determinate thing was judged, and not the other, as no accident. So, it does not have two opposed actualizations. It merely has many determinations of one way of being actualized. It’s like purely guessing that way. If I am told to guess which integer you are randomly thinking of, my power to guess can explain that I guess something, but it cannot explain that I guess 509 as no accident. Either something else explains that I guess 509 (perhaps it flashed in my imagination just now) or nothing in particular does. And my power to guess can never explain my guessing correctly as no accident.

Another way to put this: a non-epistemic power of judgment is not really a power to discriminate between alternatives, even if it is a power to determine oneself to some alternative. That is one reason why some libertarian views of free will have been criticized for mistaking freedom for slavery to chance.

Recall that two-way powers consist in systematic knowledge like the arts of medicine or building (see §2.1) and that they are determined to exercise by separate desire or choice (see §2.3). On the theoretical side, it is usually only in pathological cases, like wishful thinking, that separate desire or choice could plausibly play a role in determining what one judges. There it is not knowledge but mere desire or choice which explains why one judges something. Otherwise, one usually is – and ought to be – pretty helpless in the face of the evidence. Where evidence is absent, or unclear, or equal on both sides, one is usually helpless to do anything but admit ignorance, unless desire or choice interferes and determines one to believe in a guess, or to imagine that there is more or less evidence than there is, or similar. We are of course often under pressure to appear poised to judge well when we aren’t. There may also be special cases, like those James (Reference James2005, p. 4) describes in ‘The Will to Believe’, where determination to believe by desire or choice need not be pathological. But if so, they are fairly rare: the questions in such cases are, as James puts it, living, forced, and momentous. Most questions aren’t momentous in the way that, say, religious belief is momentous. On the practical side, if choice were determined to exercise by separate desire or choice, a regress would threaten. Even if the regress didn’t threaten, it would be the separate desire or choice that explains the relevant determined choice, not the chooser’s knowledge. These are further reasons to deny that topic-general powers of judgment or choice are two-way powers.

2.8. Aristotle’s view of two-way powers is pluralistic

Some medievals think the will is a (or the) two-way power: there is willing or volition, and there is ‘nilling’ or nolition, and these are opposed exercises of this topic-general power. What I have said already speaks against attributing anything like this view to Aristotle. Aristotle never talks about the will as a topic-general two-way power, or anything like it, such as boulêsis in general or prohaeresis in general. If the will were an Aristotelian two-way power, there would be a regress, for it would need separate desire or choice to determine its exercise. In the ethical works, Aristotle talks of the man or animal as the source of voluntary action, not ‘the will’. So, if by ‘two-way’ power you mean ‘the will’, conceived as a topic-general power, Aristotle doesn’t believe in it. As Oscar Wilde might say ‘it is not one of his words’. If he did believe in it, his account of the will should dominate his practical philosophy, as it does dominate the practical philosophy of many medievals. But it doesn’t, so, he didn’t.Footnote ¹³

There is a slightly more general point in the area. If by ‘two-way’ power you mean ‘the will’, then there is pressure to say that there is really only one two-way power: namely, the will. I think that’s how it is for Lowe’s (Reference Lowe2008) view of two-way powers. He’s almost ecstatic about how special such a power is, in contrast to every other power, even the spontaneous ones. That makes sense if you want the very idea of a two-way power to do some metaphysical heavy lifting, such as securing freedom in a way that is compatible with, say, God’s presence in and grounding of all things. Something similar goes for those who mean ‘agency’ by ‘two-way power’.Footnote ¹⁴ This pressure pushes directly against any kind of pluralism about two-way powers. Aristotle’s view is clearly pluralistic. For Aristotle, there are many two-way powers, appropriate to explaining many diverse specific changes. (Medicine is not building, etc.) His is not a view on which the only two-way power is the will, or agency as such, or knowledge in general, or similar.

2.9. The capacity for voluntary action is not an Aristotelian two-way power

While discussing what is voluntary, Aristotle (Nicomachean Ethics 1110a15–20) says that when a captain is throwing his cargo overboard to save himself in a storm, ‘the principle that moves the instrumental parts of the body in such actions is in him, and the things of which the moving principle is in a man himself are in his power to do or not to do’. One might think this is a clear reference to the two-way powers discussed in Metaphysics Book Theta. Alvarez (Reference Alvarez2013, p. 102, fn. 2), for example, after rehearsing the idea that Aristotle’s two-way powers have conditions of exercise which are not sufficient for exercise, in contrast to one-way powers – a point that comes directly from Metaphysics Book Theta – immediately cites Nicomachean Ethics and Eudemian Ethics as places where Aristotle discusses these powers. She doesn’t say where to look, but one natural place to look is in the account of voluntary action. (What follows is not a direct criticism of Alvarez, but of a view that might be hers.)

The discussion of ‘what is in one’s power’ in the Nicomachean Ethics and Eudemian Ethics has nothing much to do with the discussion of account-based powers in Metaphysics Book Theta. This can be easy to miss because of translation issues. The English word ‘power’ occurs in translations of passages in both texts, which might mislead us into thinking there is a connection. But the term dunamis (meta logou) doesn’t appear in Nicomachean Ethics 1110a15–20. Instead, one finds talk of what is ‘en autō’ and ‘ep’ autō’: what is ‘in’ a person and what is ‘up to’ them. Something similar applies to similar passages in the Eudemian Ethics, such as 1225b5–10, where the voluntary is again linked (in English translation) to ‘what is in one’s power’. Dunamis (meta logou) doesn’t appear there either.

In English, when one says things like ‘It is in my power to A’, one often means that one has a relevant ability and one has opportunity, and perhaps authority, available too. (Think of ‘It is in my power to banish you’.) For these cases, nothing stands in the way, or if there are obstacles (such as an official proclamation to be made), they are trivially overcome, so whether you do what it is in your power to do is up to you or on you or in you, not something else. I think that is a point Aristotle is rehearsing at Nicomachean Ethics 1110a15–20 in speaking of what is ‘in one’s power’; a basic and familiar one, which is why he doesn’t even argue for it. He doesn’t think this needs explanation. It’s just a way of clarifying what it means to be the source of an action. (He then applies this in clarifying what ‘voluntary’ means. But that is not my topic here.) In contrast to the point about what is in one’s power coming from the ethical works, Aristotle plainly does think that the non-canonical exercise of a dunamis meta logou needs explanation (see §2.2). There is no suggestion anywhere that young children or non-rational animals have the systematic knowledge in which two-way powers consist, and which explains their non-canonical exercises: that requires learning of a special kind. In the ethical works, Aristotle is very clear that both children and non-rational animals have voluntary action. So, if Aristotle has a view about two-way powers, it is an exclusive one, on which only mature humans have them. Relatedly, the discussion of voluntary action in the ethical works is a quite different topic.

One might grant my main point here about terms and texts, but then say that Aristotle really has two conceptions of two-way powers: one proper to the Metaphysics – call it the ‘epistemic’ conception, because of the role of knowledge – and the other proper to voluntary action – call it the ‘voluntary’ conception. I think Steward (Reference Steward2021) is trying for something close to this idea; she says there is a conception of two-way power based on scientific knowledge, and one based on the special explanatory role of desire as a ‘deciding factor’, and that only the former is explicit in the Metaphysics. (I am unsure if what follows tracks Steward’s reading precisely, so again, treat it as a view that might be hers.)

What would this ‘voluntary’ conception of two-way powers be like? Consider a wolf caught in a trap that chews its leg off in a way that is directly analogous to the ship captain voluntarily throwing cargo overboard in a storm. We might say of this wolf that chewing its leg off is voluntary in the circumstances, but generally, involuntary: no one chews off their leg for fun, including wolves. (This is what Aristotle says about the ship captain’s action of throwing cargo overboard in a storm.)

Could we then also say that ‘the things of which the moving principle is in a wolf themselves are in the wolf’s power to do or not to do’? Isn’t that enough to ground a relevant Aristotelian ‘voluntary conception’ of a two-way power: a power to do or not do, as people sometimes say in this literature?

I think this is also wrong. Aristotle never clearly describes non-human animals as acting at their discretion, as if they could reason about what is best, and so oppose what appetite presents as good; rather they tend to be presented as ruled by their appetites. This idea is relatively clearly expressed in the Eudemian Ethics, where Aristotle notes that non-rational animals share a ‘simplicity of moving principle’ with inanimate objects, because they don’t have different forms of principle at work in them:

So, when the wolf chews off its leg, it is not analogous to the ship captain throwing stuff overboard; it is more like what happens when a ship is blown by two different breezes at once. The ship captain calculates: rationality moves him to do what he does, and as soon as he’s calculated, he acts. The wolf may look like it has calculated, because it may take a while before it finally chews, but what is really going on is that a desire to be free grows stronger until it overcomes the pain of moving, much as a ship may be blown in one direction until a growing counter-breeze pushes it in another.

I think we can understand this view of non-rational animals in a way analogous to the discussion of judgment above. A given body of affections of the soul, prompted by appetites, emotions, perceptions, memories, imaginings, etc., in the relevant circumstances, can explain an action as no accident. For example: a startled tiger usually freezes; a really startled tiger jumps; a tired tiger rests; an exhausted or thoroughly uninterested tiger sleeps. If a tiger is hungry but also more scared of a nearby elephant than hungry, that can explain its action of remaining still, lashing its tail in frustration. Relative to that body of affections of the soul, including their strengths and so on, it would be an accident if the tiger did something else, such as sleeping or charging at the elephant. In unusual cases, either something else would explain the action, or nothing would. (Unlike the case of rationality opposing sensory appetite, we are hard pressed to find such alien explanations of action in non-human animals; if the tiger were startled by a snake, that would alter the relevant body of affections of the soul.) There can be cases where a given body of affections of the soul is equally strong in favour of two alternatives: then either would be an accident relative to that body, given the lack of dominance of principle. Consider the conflicted tiger on the verge of fleeing or attacking; probably its imagination will tip the balance soon if nothing else does. There can also be cases where a given body of affections doesn’t favour anything much; then doing something specific would be an accident relative to that body. As with analogous powers of judgment based on knowledge, there isn’t a two-way power at work here, because this power for movement and rest based on (summing over) affections of the soul doesn’t have two opposed actualizations which are both explained as no accident, relative to the power.

The tiger’s principle of movement is a sensory one – it is determined by a body of affections traceable to memory, perception, and imagination. The rational human has another form of principle – an epistemic one, determined by what they know and conclude on the basis of deliberation. For us, in particular cases, these bodies might favour opposed actions. But if one hasn’t grown up to love what’s reasonable, then sensory appetite will tend to win such conflicts. Some of us are still slaves to appetite.Footnote ¹⁵

If I am right, then there is an illicit reification possible from ‘It is in an animal’s power to do X, or not’ to ‘The animal has a two-way power to do X or not’. We shouldn’t attribute this to Aristotle. If we did attribute it, he ought to have said that there is as much of a puzzle about animal self-motion as there is about medicine and building, and he would have needed to haul in explanatory tools that are like systematic knowledge, without being systematic knowledge, to explain how animals could possibly have powers directed to contraries. The best explanation of why he didn’t do that is that he didn’t reify in the way described. The topic of what it is one’s power to do is different from that of two-way powers.

2.10. Contrary exercises of two-way powers are not, as such, defective, and two-way powers don’t guide ethics

Kern (Reference Kern2017, p. 162), discussing Aristotle’s view of rational capacities in Metaphysics Book Theta says ‘rational capacities are the sorts of capacities that cannot only be used but also misused’. Later, she says that the contrary exercises of rational capacities are normatively defective:

As an interpretation of Aristotle, this is quite wrong. Kern includes skills like skiing. We have seen why that is wrong in Section 2.6. Skiing has no opposed contrary exercise. A skier who crashes accidentally is not using her skill to make it no accident that there is an opposite change in something other. What she does is an accident relative to her power to ski, such as it is. If she produces a crash deliberately, that is pretending to crash, which is not in opposition to skiing (well).

More importantly, and more generally, Aristotelian two-way powers do not guide ethics. No two-way power is addressed to how to live as such (whether individually, or collectively.) It is ultimately politics, not medicine, which decides which arts should be pursued, and in what way, and to what degree. A good politician might dictate war, and a good general in that war might dictate poisoning the opposing forces, and a good doctor might carry out orders to use their knowledge to sicken the enemy, secure in the knowledge that this is what true virtue and practical wisdom require. What Kern would call ‘deviant misuse’ of the power to heal – sickening the enemy – is rather doctoring at its best (in the circumstances). Aristotelian two-way powers do not ‘guide’ in the sense that Kern (Reference Kern2017, pp. 174–175) describes. If they did, doctors would know when to heal (always) or sicken (never) without knowing how to live.

This is what we should expect, given that Aristotelian two-way powers are exhausted by principles of change in something other, like medicine or building, and are determined to exercise by separate desire or choice. Being a good doctor doesn’t make you a good person. It might mean you must have some natural virtues developed to some degree: learning typically requires some discipline. But the evil can be disciplined too in their evil, and the more disciplined, the more terrible they are.