1. Introduction
The focus of this paper will be two-way powers, construed as powers to act in some way (to ɸ) or to refrain from acting in that way (to refrain from ɸ-ing). And the question to be addressed here is whether a two-way power to ɸ or refrain from ɸ-ing could be a causal power. That is, the question is whether the power to ɸ or refrain from ɸ-ing could amount to the power to cause an effect of some kind, E, or refrain from causing an E-effect.Footnote 1 Such a causal conception of two-way powers is derivative of a causal conception of acting: a conception of the action of ɸ-ing by an agent, A, as a matter of A causing an effect of some kind, E. The question, then, really comes to this: is the positing of two-way powers compatible with the causal conception of acting?
Those who take the exercise of human agency to be the exercise of two-way powers often suppose that acting is a matter of the agent causing some outcome, such that human agency is a causal power. Maria Alvarez, for example, holds that human agency is ‘a particular kind of causal power: a ‘“two-way” causal power’ (Reference Alvarez2013, p. 101). Helen Steward (Reference Steward2012) too holds and defends this claim. Steward (Reference Steward2012), moreover, holds that the exercise of this two-way causal power integral to agency is not compatible with determinism. She has, in other words, an incompatibilist conception of two-way powers, and she takes these powers to be both integral to agency, and causal in nature.
I do not intend to assess the claim that human agency is a two-way power, but I do intend to assess the view that agency, as a two-way power, on the incompatibilist conception of such powers, is a causal power. Indeed, I will focus here primarily on the question of whether a two-way power, as conceived of by the incompatibilist, could be a causal power.
My focus is primarily on the incompatibilist conception of two-way powers not because I endorse this conception over the compatibilist conception, but because the ‘two-wayness’ of two-way powers on the compatibilist conception presents no particular difficulty for the view that those powers are causal powers. I elaborate on this suggestion in §5. By contrast, I will argue at length that the ‘two-wayness’ of two-way powers on the incompatibilist conception does present a particular difficulty for the view that these powers are causal powers.
I build here on work from Thomas Pink (Reference Pink2016, Reference Pink2019), who has previously argued that two-way powers, on the incompatibilist conception of them, cannot be causal powers. Pink appeals to a claimed certain feature of incompatibilist two-way powers, proposing that causal powers cannot have such a feature. I maintain that Pink is right that causal powers cannot have the claimed feature of incompatibilist two-way powers, but also that no powers, including non-causal powers, can have such a feature. The conception of that feature is in fact simply incoherent. As such, the basis for opposition to the causality of incompatibilist two-way powers tells against Pink’s own alternative, non-causal, account of incompatibilist two-way powers too. I then propose that the only available option for the incompatibilist is another non-causal account. This account, I suggest, avoids the problems facing both incompatibilist causal accounts and Pink’s incompatibilist non-causal account, while retaining the central aspects that make those accounts attractive.
I begin here with some clarifications regarding some prior issues pertaining to the nature of two-way powers, so as to smooth the way for directly tackling the central question of this paper (§1). I then briefly discuss the incompatibilist v. the compatibilist conception of two-way powers (§2) and outline the attractions of the causal conception of two-way powers (§3). Having done so, I directly assess the question of whether two-way powers, on the incompatibilist conception of them, are causal powers, arguing that they are not (§§4–5). I end by proposing that while the incompatibilist about two-way powers may not hold that actions that are exercises of those powers involve causation by the agent, they may hold that those actions instead involve non-causal determination by the agent (§6).
2. Preliminaries
2.1 Powers
I’ll begin here by saying something about the background notion of powers. A power is a kind of capacity, and, in particular, when we are discussing agentive powers – i.e., powers of animate agents to do something or to refrain from doing something – as we generally are here, the term ‘power’ is interchangeable with the term ‘ability’.Footnote 2 And, just as the term ‘ability’ has both a more general and a more specific sense (see e.g., Mele, Reference Mele2003, Maier, Reference Maier2015, Jaster, Reference Jaster2020), so too does the term ‘power’.Footnote 3 Abilities/powers in the general sense concern an agent’s capacities across a wide range of circumstances, whereas abilities/powers in the very specific sense concern what is available to the agent in the specific circumstances they are in at the relevant time. One would use the general sense in stating that someone has the power/ability to score a goal in football, even while they are not playing football. And one would use a more specific sense in stating that one only has the power/ability to score a goal in football when one is positioned correctly as a participant in a football match.Footnote 4 The relevance of the contrast between more general and specific sense of the term ‘power’ will be particularly important here in the comparison of the compatibilist v. the incompatibilist conception of two-way powers (§2).
There are several further questions regarding the nature of two-way powers in particular, which I will briefly address here. I lay out my stance on these questions in order to smooth the way for directly tackling the central topic of this paper. These questions are the following: (a) What is the relationship between refraining from ɸ-ing and merely not ɸ-ing? (b) Is refraining from ɸ-ing an exercise of the two-way power to ɸ or refrain from ɸ-ing? And (c) Is having the two-way power to ɸ or refrain from ɸ-ing just a matter of having both the power to ɸ and the power to refrain from ɸ-ing?
2.2 What is the relationship between refraining from ɸ-ing and merely not ɸ-ing?
We should suppose that refraining from ɸ-ing at least involves not ɸ-ing, though it is more than merely not ɸ-ing. One might not ɸ without refraining from ɸ-ing. Indeed, this happens all the time: I am currently neither flying at light speed, nor keeping perfectly still, but these are not things that I am refraining from doing. Refraining is always agentive – an exercise of agency – and requires the power to do that which one refrains from. Not doing something (an omission) need not be agentive, and it does not require the power to do that which one does not do. Refraining from ɸ-ing is, more specifically, something like not ɸ-ing (omitting to ɸ) as a choice over ɸ-ing, when ɸ-ing is a salient option to the agent.Footnote 5, Footnote 6 This seems to be in line with how two-way powers theorists have conceived of refraining from ɸ-ing: it is something that can constitute a choice to not ɸ, but where this need not amount to a purely mental act of choosing to ɸ, and does not require any distinct cognitively demanding mental state of intending not to ɸ. In this way, many non-human animals, as well as us human agents, can be said to be able to refrain from doing things.Footnote 7 A dog might be trained to refrain from eating a treat until they are given a signal, for example.
2.3 Is refraining from ɸ-ing an exercise of the two-way power to ɸ or refrain from ɸ-ing?
Different positions on this question are taken by different proponents of two-way powers. Alvarez (Reference Alvarez2013) and Frost (Reference Frost2020), for example, both hold that what makes a power two-way is that it can be exercised in two ways: either by the agent’s doing something, ɸ-ing, or by their refraining from doing that thing, ɸ-ing. By contrast, Steward (Reference Steward2013, Reference Steward2020) has suggested that two-way powers can in fact only be exercised one way, by doing something, ɸ-ing.Footnote 8
What makes a power two-way, for Steward, is not the two ways in which it can be exercised, but rather, the two ways in which things can proceed at a given time for an agent who has the two-way power to ɸ at that time (see especially Steward, Reference Steward2020, p. 348). One way that things can go is that the agent exercises their power, and thereby ɸ-s, but another way things can go is that the agent doesn’t exercise their power, instead refraining from ɸ-ing. And, for Steward, it is crucially up to the agent whether they exercise their two-way power or not.
As a result, on Steward’s view, though refraining from ɸ-ing is not considered to be an exercise of the two-way power to ɸ or refrain from ɸ-ing, that it is up to an agent whether to ɸ or refrain from ɸ-ing at time t is what it means for the agent to have the two-way power to ɸ or refrain from ɸ-ing at t. But this is in fact also what it ultimately means for an agent to have a two-way power to ɸ or refrain from ɸ-ing at t for a two-way powers theorist who holds that refraining from ɸ-ing at t is a way of exercising that two-way power. For such a two-way powers theorist, this is simply considered to be a matter of it being up to the agent how they exercise their two-way power, rather than whether to exercise their two-way power.
I will say more in response to (c), about the idea of it being up to the agent whether to ɸ or refrain from ɸ-ing at t, which is crucial to what distinguishes having the two-way power to ɸ or refrain from ɸ-ing at t from having both the power to ɸ at t and the power to refrain from ɸ-ing at t. But it can be noted for now that there seems to be no real contrast of substance, rather than just of labelling, between Steward’s view that refraining from ɸ-ing is not a way of exercising of the two-way power to ɸ or refrain from ɸ-ing, and the view that it is. And so, for simplicity, I will here talk of refraining from ɸ-ing as one of the two ways in which the two-way power to ɸ or refrain from ɸ-ing can be exercised.
2.4 Is having the two-way power to ɸ or refrain from ɸ-ing just a matter of having both the power to ɸ and the power to refrain from ɸ-ing?
I suppose that, for any two-powers theorist, genuinely having a two-way power to ɸ or refrain from ɸ-ing must be distinctive from having both the power to ɸ and the power to refrain from ɸ-ing.Footnote 9 But how should this distinction be drawn?
The distinction should not, I think, be drawn in terms of the distinct necessary conditions for exercising the two-way power v. the distinct necessary conditions for either exercising the power to ɸ or exercising the power to refrain from ɸ-ing. The distinct necessary conditions for an agent’s exercising the two-way power to ɸ or refrain from ɸ-ing will be just the same as the distinct necessary conditions for them either exercising the power to ɸ or exercising the power to refrain from ɸ-ing.Footnote 10 That is, it is open to an agent to exercise the two-way power to ɸ or refrain from ɸ-ing just when it is open to them to either ɸ or refrain from ɸ-ing.
The distinction, instead, must amount to what it is to exercise the two-way power to ɸ or refrain from ɸ-ing, as opposed to either exercising the one-way power to ɸ or exercising the one-way power to refrain from ɸ-ing. The exercise of the two-way power to ɸ or refrain from ɸ-ing must be a matter of the agent choosing between these two options when the two options are both available.Footnote 11 That is not to say that the exercises of the two-way power to ɸ or refrain from ɸ-ing are really acts of choice rather than ɸ-ing or refraining from ɸ-ing; the choice of ɸ-ing over refraining from ɸ-ing is constituted by the action of ɸ-ing that is the exercise of the two-way power.Footnote 12 And similarly, the choice of refraining from ɸ-ing is constituted by the refraining from ɸ-ing that is the exercise of the two-way power.
In exercising the two-way power to ɸ or refrain from ɸ-ing, the agent opts for one of these options over the other, when it was up to them which option to take. By contrast, in exercising either the one-way power to ɸ or the one-way power to refrain from ɸ-ing, even when it is open to one to do one or other of these things – in the sense that it is possible to either ɸ or refrain from ɸ-ing – one need not be opting for, or choosing between, the exercise of one one-way power rather than the other. For, in exercising their one-way power to ɸ, rather than their one-way power to refrain from ɸ-ing, the agent may simply ɸ (perhaps, not even intentionally), where their ɸ-ing means that they do not refrain from ɸ-ing, since ɸ-ing is incompatible with refraining from ɸ-ing. In this way, though there is a sense in which it is open to the agent to exercise either one of their one-way powers, this will be a matter of mere possibility of the exercise of one or other power. The exercise of one or other power is not open to the agent in the sense of being genuinely up to them which power is exercised, such that the exercise constitutes a choice to exercise that power rather than the other.
Someone may, for example, have both the power to wrinkle their nose at time t and to refrain from wrinkling their nose at t, and they find themselves wrinkling their nose at t, not as a choice of doing this over refraining from doing so. The wrinkling of the nose may rather simply be something that the agent finds themselves spontaneously doing.Footnote 13
3. Incompatibilist v. Compatibilist Two-Way Powers
In this section, I outline the incompatibilist and the compatibilist conceptions of two-way powers, seeking to make clear the contrast between them. The central contrast between the two conceptions is in terms of their views with regards to the determination of exercises of two-way powers, which I outline briefly before identifying further contrasting features of the two conceptions.
The notion of determination, in terms of which the central contrast between the two conceptions of two-way powers is to be drawn, is that of guaranteeing, or making necessary, such that any alternative is impossible. It is a notion for which I take the terms ‘fixing’ and ‘settling’ to be interchangeable.Footnote 14 And, according to an incompatibilist conception of two-way powers, whether and how a two-way power is exercised cannot be determined by any distinct occurrences, or states. The exercise is not determined by anything distinct, except perhaps the agent themselves, according to some incompatibilists.Footnote 15 On this conception, the exercise of two-way powers is clearly incompatible with determinism, the thesis that each occurrence and state is determined by the laws of nature and prior circumstances. According to a compatibilist conception of two-way powers, by contrast, whether and how a two-way power is exercised may be determined by distinct occurrences and states, and these determining occurrences and states may themselves be determined by the laws of nature and prior circumstances. So, on this conception, the exercise of two-way powers is perfectly compatible with determinism. In what follows I outline some further features of the two contrasting conceptions of two-way powers.
3.1 The compatibilist conception of two-way powers
We can characterise what is distinctive about the compatibilist conception of two-way powers in terms of the contrast between general and specific powers. For the compatibilist, who supposes that an agent’s having the two-way power to ɸ or refrain from ɸ-ing is unaffected by whether or not the agent’s exercises of agency are determined by distinct occurrences and states, an agent may only have the two-way power to ɸ or refrain from ɸ-ing in some less-than-maximally-specific sense.Footnote 16 That is, it is only non-maximally-specific two-way powers that can be had by an agent whose actions and refrainings are determined by distinct occurrences and states. After all, if an agent’s broader circumstances determine that they will ɸ, that agent cannot be said to have, in those circumstances, the maximally specific power to refrain from ɸ-ing. That agent may be said to have the power to refrain from ɸ-ing, and the two-way power to ɸ or refrain from ɸ-ing, in only some more general sense (abstracting from all the precise details of their circumstances).Footnote 17
Compatibilists may differ with regards to exactly in which more general sense it is that an agent can be said to have the power to either ɸ or refrain from ɸ-ing at t. Some may suppose that, even if an agent’s current circumstances at t (excluding whether they ɸ or not) are fully specified, insofar as the world history up until t is not fully specified, it is not fixed either that the agent ɸ-s or that they do not ɸ at t. Others, however, may suppose that when the agent’s current circumstances at t are fully specified (excluding whether they ɸ or not, but including all the agent’s mental states at t ) it is fixed either that the agent ɸ-s or that they do not ɸ at t. In particular, some compatibilists have held that desire (or rather, predominant desire: a desire to do something that is stronger than any desire to do anything incompatible with that)Footnote 18 is the key determining factor for whether and how an agent exercises a two-way power.Footnote 19 That is, it is supposed that when an agent with the general two-way power to ɸ or refrain from ɸ-ing is in otherwise congenial circumstances for either ɸ-ing or refraining from ɸ-ing, a predominant desire to ɸ (or refrain from ɸ-ing) determines that they exercise their two-way power by ɸ-ing (or refraining from ɸ-ing).
Taking the determining factor to be a predominant desire is part of the compatibilist’s attempt to make sense of the idea that the exercise of a two-way power is distinctively a matter of choosing between two available options and opting for one over the other. And as was indicated in §1, this idea identifies what is distinctive of having a two-way power, as opposed to having two one-way powers. I won’t here review the highly contested question of whether the compatibilist succeeds in making sense of the exercise of a two-way power as a matter of choice between two available options.Footnote 20 The concern here is not to defend either incompatibilism over compatibilism or vice versa. I have outlined the compatibilist position here just so as to provide a contrast to the incompatibilist picture, where that contrast should help clarify that incompatibilist picture. Let us now consider, then, the contrasting incompatibilist conception of two-way powers.
3.2 The incompatibilist conception of two-way powers
On the incompatibilist conception of two-way powers, whether and how an agent exercises their two-way power is not determined by any distinct occurrences, states, or sets thereof. Thus, an agent can be said to have the two-way power to ɸ or refrain from ɸ-ing at t in the maximally specific sense. That is, however fully the circumstances of the agent at t, and the history up until t, are specified (excluding whether the agent ɸ-s or not), it is left open whether the agent ɸ-s or refrains from ɸ-ing.
It is important for the incompatibilist regarding two-way powers that, although the occurrence and nature of the agent’s exercise of their two-way power – their choice – is not determined by any distinct occurrences or states, it is not simply random, in the way that the indeterministic occurrence of the emission of a particular alpha particle from a particular radioactive atom at a particular time is random. And the distinction here lies precisely in the fact that the agent, unlike the radioactive atom (or the emitted alpha particle), exercises a two-way power. The agent either chooses to act, when it had been open to them to refrain from acting, or they choose to refrain from acting, when it had been open to them to act. By contrast, if the alpha particle is emitted, then while it had been an open possibility that it would not in fact be emitted, its emission from the atom is not a choice of either the atom or the particle, and it is not an action from which either could have refrained. It had been neither open to the atom to refrain from emitting the particle, nor open to the particle to refrain from its own emission.
Now, for the incompatibilist, a predominant desire (or, equally, an intention, or judgment about what is best to do) is something that is not sufficient (together with other circumstances) for the exercise of the two-way power to ɸ or refrain from ɸ-ing one way or the other. For the incompatibilist, it is open to an agent with the two-way power to ɸ or refrain from ɸ-ing at t to refrain from ɸ-ing at t even if the circumstances are all congenial to either ɸ-ing or refraining from ɸ-ing, and the agent has a predominant desire (or intention, or judgment that it would be best) to ɸ at t. Similarly, it is held to be open to the agent to ɸ at t, given similarly congenial circumstances and a predominant desire (or intention, or judgment that it would be best) to refrain from ɸ-ing at t. In other words, for the incompatibilist, the agent can exercise their two-way powers irrationally, and akratically against their predominant desires, intentions and judgments about what is best to do.Footnote 21
4. The Causal Conception of Two-Way Powers
In this section, I say more about the causal conception of two-way powers, identifying why one might suppose that two-way powers are causal powers. Recall that the question of whether two-way powers are causal powers is to be understood as the question of whether having the two-way power to ɸ or refrain from ɸ is a matter of having the power to cause an effect of some kind, E, or refrain from causing an E-effect. So, on a causal conception of two-way powers, to act is to cause an effect of some kind. Why might one suppose this?
It is supposed by the proponent of the causal conception that to act is to make something happen, or make the world be some way, and moreover that ‘making happen’ here must be, and can only be, understood as causing.Footnote 22 What the agent is said to make happen are the events that are the results of their actions: hereafter, result-events, where ‘result-events’ can be defined precisely as follows:
RESULT-EVENTS: x is a result-event of an actually occurring action, y, iff x occurs because y occurs, and x is a token event of a non-action type, X, and y is a token action of an action type, Y, such that a token of Y occurs only if a token of X occurs.
To illustrate, the result-event of an action of cracking some ice is the event of the ice cracking: the event comes about because the action of cracking some ice occurred, and a cracking of ice only occurs if some ice cracks. But now, importantly, the ‘because’ in the above definition of result-events expresses an explanatory order between an action and its result-events, leaving open precisely what the underlying metaphysical relationship between the action and its result-event is. The relationship may be a causal one, but the above definition leaves open that it is not.Footnote 23 This is left open as long as we allow, as we should, that there can be explanatory relationships between entities that are not causally related.
But, for the proponent of the causal conception, the explanatory relationship is grounded in a causal relationship, between the agent and the result-event of their action. The ice’s cracking is explained by the agent’s action of cracking the ice because that action is a matter of the agent’s causing the ice’s cracking.Footnote 24 And, for the incompatibilist about two-way powers, the agent determines the otherwise undetermined result-events of their actions that are exercises of their two-way powers. In exercising a two-way power by ɸ-ing, which necessitates an event of a certain type (the type that is a result of ɸ-ing actions), the agent determines that an event of such a type occurs.Footnote 25 The incompatibilist who is drawn to the causal conception of action then supposes that the determination involved in exercising a two-way power by acting is specifically causal determination.Footnote 26
Whether, and how, the causal account of two-way powers is appropriately opposed turns on whether one is an incompatibilist or compatibilist. For the compatibilist, I suggest that it is unproblematic to suppose that an agent’s exercising a two-way power by acting is a matter of them causing the action’s result-event. That is, this supposition is unproblematic if a compatibilist allows that the agent’s causing the result-event takes place in virtue of some state of theirs – their predominant desire, intention, or judgment about what is best to do, perhaps – causing the result-event. I will indicate in §5 why this offers a way for the compatibilist to avoid the problem that dogs the incompatibilist.
5. Incompatibilist Two-Way Powers and Contingent Determination
In this section, I begin to lay out my opposition to the causal conception of incompatibilist two-way powers. I focus here on developing an objection to a causal conception of incompatibilist two-way powers put forward by Thomas Pink (Reference Pink2016, Reference Pink2019). This objection concerns the role of ‘contingent determination’ and the relation between this and causation.
5.1 Contingency of determination
Pink (Reference Pink2016, Reference Pink2019) holds that free action is the exercise of incompatibilist two-way powers where each exercise is either an action or a refraining from acting in some way.Footnote 27 Yet, Pink maintains that such powers cannot be causal powers. The basis of Pink’s opposition to conceiving of incompatibilist two-way powers as causal powers is the idea that the exercise of these powers in action must involve determination of a certain kind: contingent determination. Moreover, Pink argues that causal determination does not and cannot involve contingency in the relevant sense. I think Pink is right here, but there is more to be said as to why causal determination does not and cannot involve contingency. And what more there is to say in fact indicates how determination in general, rather than merely causal determination, does not and cannot involve contingency.
Let us begin, however, by considering just what contingency of determination is. Pink articulates the notion of contingency of determination as follows:
[Contingency of determination] leaves it open whether an agent with the power to determine a given outcome will so exercise his power. The agent’s exercise of his power to produce a given outcome is not … guaranteed by factors for which he need not be responsible – by his circumstances and by the very nature of his power. But nor … is the outcome simply left open by the limited nature of the power, so that the outcome depends on mere chance. The agent has the power to determine that the outcome happens; but the outcome’s occurrence is contingent on whether the agent actually exercises this power – something not guaranteed by the agent’s mere possession of it.
...
Contingency of determination means that an agent can be present with the power then to determine that he does A, along with all the conditions required for the power's successful exercise – and not exercise it. (Pink, Reference Pink2016, p. 160)
So, contingent determination is determination of an outcome by an agent, such that it was open whether the agent, with their present circumstance and state, determines the outcome or not, though whether determination takes place or not is not merely random.Footnote 28 Pink in fact holds that contingency of determination is a feature that any incompatibilist who takes two-way powers to be determinative must appeal to in accounting for the distinctive nature of two-way powers (see Pink, Reference Pink2019, p. 21).Footnote 29 This is a claim that I will come to dispute. But let us first consider Pink’s basis for claiming that causal determination may not be contingent.
5.2 Causation and contingency of determination
Pink maintains that causal determination does not and cannot have the feature of contingency of determination. If both this, and Pink’s claim that the determination involved in the exercise of two-way powers must have the feature of contingency of determination, are true, then two-way powers cannot be causal powers. Pink’s contention regarding causation can be boiled down to the idea that insofar as a cause determines its effect, it does not do so contingently.Footnote 30
A cause may exercise (or manifest) some causal power, and yet fail to determine its effect, however.Footnote 31 Such a cause only raises the probability of an effect of its kind occurring when it does.Footnote 32 In this kind of case, there is a contingency to an effect of that kind’s occurring, despite the presence of the cause. The presence of the cause leaves it undetermined that an effect of that kind does occur. But this contingency is not a contingency of determination. Here the causation is simply non-deterministic (see Pink, Reference Pink2019, p. 20).
A cause either determines its effect or is such that its effect’s occurrence is contingent despite the presence of the cause; both cannot be true of a single cause of an effect, qua cause of that effect. And, since Pink holds that the exercise of a two-way power in action is a matter of contingent determination by the agent, he proposes that it cannot be a matter of causation by the agent: ‘[c]ontingency of determination distinguishes a free agent from any cause – including a probabilistic cause.’ (ibid., p. 21). For Pink, in exercising a two-way power by acting, an agent determines the coming about of some outcome, but that the agent determines the coming about of that outcome, rather than another, is not determined by the presence of the agent, or any of the conditions or circumstances of the agent. This is what it is for two-way powers to have the feature of contingency of determination, which cannot be had by causal powers.
Since Pink holds that the exercise of a two-way power in action is a matter of contingent determination, Pink asserts that this determination is non-causal determination. For Pink, then, non-causal determination can involve the contingent determination of outcomes, while causal determination cannot. Yet, from what has been said so far, it remains unclear why this should be so. We may well question why it is that non-causal determination might be such that it can involve contingent determination of outcomes while causal determination cannot. If there is such a thing as contingent determination, why should we not suppose that it makes for just another form of causation? That is, why should we not suppose that causes do at least sometimes contingently, but nonetheless causally, determine their effects? This question is pressing insofar as it is accepted that contingent determination is possible.
But I would like to suggest that there are good reasons for supposing that contingent determination as such is not possible. In that case, while Pink is correct in suggesting that two-way powers could not be causal powers whose exercise has the feature of contingent determination, he is wrong in supposing that the exercise of a two-way power could have the feature of contingent determination at all. In the following subsection I argue that determination as such – whether it is causal determination or non-causal determination – cannot be contingent.
5.3 Against the possibility of contingent determination
For an agent to exercise a power with the feature of contingency of determination is for the agent to determine a given outcome, but where that the agent determines that outcome is not guaranteed by the agent’s circumstances, or state, but is similarly not simply a matter of chance. This, I want to suggest, is not a coherent option. And it is on account of the incoherency of this option that causal powers cannot have the feature of contingency of determination. Nothing can determine an outcome contingently, since insofar as anything determines some outcome, it does not do so contingently.
The basis of my suggestion is the following. If an entity, x, determines an outcome, o, then the presence of x, qua an entity of a certain type, T1, with some certain feature(s), F1 (, …, Fn), at some time, t, must be such that the occurrence of an outcome of some type, T2, is necessary at t, or sometime thereafter, where o is an outcome of type T2. That is, it is not possible for there to be x, as it is (a T1 entity with features F1, …, Fn), at some time, without there also being the occurrence of a T2 outcome then or thereafter.Footnote 33, Footnote 34 This is simply what it must amount to for an entity to determine an outcome, in the sense of guaranteeing it, or making it necessary, such that any alternative is impossible. And this is similarly what must be involved in fixing or settling matters.
But, in that case, insofar as any entity – including an agent – determines an outcome, it cannot do so contingently: an entity cannot determine an outcome without some circumstance, or state, of the entity guaranteeing that outcome. There is simply no scope, then, for contingent determination. And, insofar as there is contingency of an O outcome occurring, despite the presence of entity x, then x cannot be said to determine all by itself the O outcome. And so, x cannot be said to contingently determine the O outcome, in Pink’s sense. Perhaps, in this case, x is only part of some conjunction of factors that do determine the O outcome, but this is not compatible with x contingently determining the O outcome.
Two-way powers cannot, then, be said to have the feature of contingency of determination. No power has this feature, since the very notion of contingent determination is incoherent. A fortiori, no power has the feature of contingency of causal determination. Where does that leave us, with regards to the question of whether two-way powers are causal powers? Could a two-way power be a causal power without having the feature of contingency of determination? In the following section, I argue that, for the incompatibilist about two-way powers, at least, it cannot.
6. Incompatibilism and Non-Contingent Causal Determination
In this section, I argue that two-way powers, as conceived of by incompatibilists, cannot be causal powers, where causation does not involve contingent determination. I note also that compatibilists do not face this same problem.
Insofar as an incompatibilist about two-way powers takes these powers to be causal powers, they must suppose that the causation involved in the exercise of such a power in action is a matter of causal determination rather than merely contingent, probabilistic causation. This is because, as noted, for the incompatibilist, the agent determines the result-events of their actions that are exercises of their two-way powers. And so, insofar as the incompatibilist takes two-way powers to be causal powers, such that the agent causes the result-events of their actions that are exercises of two-way powers, the incompatibilist must hold that the causation here is a matter of causal determination. So, what we need to consider is whether there is scope for an incompatibilist conception of two-way powers as exercised in action through causal determination. I propose that there is not, since the incompatibilist encounters a dilemma here.
The dilemma can be spelled out as follows. Causal determination of an outcome by an agent is a matter of the presence of the agent, as an object with a certain set of features, causally determining that outcome. If we suppose, then, that an agent, in acting, causally determines the result-event of their action, there seem to be two options available with regards to what this amounts to. The set of features of the agent that figure in the causal determination of the result-event may either be: (i) some feature(s) that the agent has that is (are) distinct from their actions; or (ii) they may simply amount to the feature of the agent acting in the way that they do.Footnote 35
Option (i) is clearly not available to the incompatibilist, who supposes that an agent’s exercise of their two-way power is not determined by any states or occurrences distinct from the exercise. This supposition is contravened on option (i), since insofar as the result-event of an action is causally determined by some feature(s) of the agent that is (are) distinct from the action, it is causally determined that the agent acts in that way. We should note here, however, that this option is available to the compatibilist two-way powers theorist. The compatibilist may hold, for example, that the agent’s having a certain constellation of mental states, including certain predominant desires, in the broader circumstances that that they are in, causally determines the action’s result-event. For this reason, the dilemma facing the incompatibilist about two-way powers is not similarly faced by the compatibilist.
For the incompatibilist, the unavailability of option (i) presents a serious problem, since option (ii) is not attractive for several reasons. The problems with option (ii), however, are not ones that are specific to incompatibilism about two-way powers. The problems are generated by the fact that to suppose that the result-event of an action is always causally determined by the agent’s acting as they do, is to suppose that the result-event is always caused by and distinct from the agent’s action.
The first problem that is generated is that, as noted earlier, on the causal conception of action, an action is a matter of the agent’s causing the action’s result-event. As such the action is not itself a cause of the result-event. So, option (ii) requires an amendment of the causal conception of action. The further problems with option (ii) are not unrelated, indicating reasons to favour the conception of an action as the agent’s causing of the action’s result-event rather than as itself a cause of its result-event.
One issue is that to suppose that when an agent exercises a two-way power in action, their action causes its result-event is to take up an odd and alienating picture of an agent’s relationship with the result-events of certain of their actions. The actions in question are those that are a matter of directly changing oneself – what we might call basic self-change actions – where one changes one’s body or mind, without doing so by means of doing anything else (apart from simply by means of doing one part of that action and then another, until it is complete).Footnote 36 When one raises one’s arm directly, for example, one does so without raising it by means of doing something else (apart from simply raising it bit by bit, until one is done), and this can be contrasted with cases where one raises one arm by means of lifting it with the other. And moreover, when one raises one’s arm directly, it seems that the result-event of one’s action – the arm’s rising – should not be conceived of as a distinct and subsequent effect of the action of arm-raising. To suppose otherwise would be to think of the action as something one does in order to get one’s arm to then subsequently react in a certain way.
This is, more or less, the picture of action that was defended by Hornsby (Reference Hornsby1980), though her focus was specifically on basic bodily actions, which are argued to be distinct and prior causes of their bodily result-events. This picture was challenged initially by Watson (Reference Watson1982), with the challenge then taken up by Haddock (Reference Haddock2005). As Haddock explains:
Our bodies are pictured as entities whose powers are wholly distinct from our powers of agency as entities that we can (at best) only cause to move – and in this respect they are the same as any other worldly object. Jane moves her body just as she moves (say) a glass of water – by exercising her agentive powers in an act … that, if she is lucky, has the consequence of causing the relevant object to move. Her body is no different to the glass of water, or any other object, in this respect. (Haddock Reference Haddock2005, p. 161)
The body is thereby presented, on Hornsby’s picture, contrary to our intuitions and phenomenological experience, as a mechanism that one manipulates by acting in such a way as causes the right kinds of effects in that mechanism.
And similarly, if we suppose that our basic mental actions are distinct and prior causes of their mental result-events, we picture ourselves as standing in an odd relationship to our own minds. Our minds are presented also as mechanisms that we manipulate by acting in such a way as causes the right kinds of effect in these mechanisms. According to the view under consideration, when one turns one’s attention to some stimulus, for example, one must do something that then causes the attention to shift appropriately. Such a view seems not only counter-intuitive, but also raises serious questions about the metaphysics and ontology of the acting agent who seems to stand apart from both their mind and body.
These issues present a serious concern – even if they do not amount to a knock-down objection – for the idea that in exercising a two-way power through action, the agent’s action causally determines its own result-event, insofar as it is supposed that basic self-change actions can be exercises of two-way powers. And indeed, basic self-change actions seem apt to be considered the primary cases of actions that are the exercises of two-way powers, as Steward (Reference Steward2012) has especially emphasised. After all, all other action we do by means of directly changing ourselves – a point that I will return to in the next section.
But now, there is a further concern with regards to the picture of basic bodily self-change actions as causes of their bodily result-events. This is a concern that has been highlighted by several of those who favour the conception of action as the agent’s causing of result-events. The focus here is on one’s relationship with the basic bodily self-change actions of others, rather than one’s own. The concern is that it is hard to see how the basic bodily self-change actions of others might be straightforwardly observable, if they are the prior causes of the bodily goings-on that are their result-events (see e.g., Alvarez & Hyman, Reference Maria and Hyman1998, p. 230, and Steward, Reference Steward2001). And we generally do take ourselves to straightforwardly observe not only the bodily goings-on that are the result-events of basic bodily self-change actions (e.g., someone’s hand rising) but also the basic bodily self-change actions themselves (the person raising their hand). We do not seem to observe the bodily goings-on and then infer that there was a bodily action causally prior. But nor do we seem to see both some bodily goings-on and some extra thing that is the agent’s bodily action which caused it.
None of the above concerns amount to knock-down objections to the view that, in exercising a two-way power through action, the agent’s acting in the manner they do causes their action’s result-events. They are, however, at least significant warning signs that the view embraces an unattractive picture of action. What is perhaps a bigger problem is the fact that the causality involved in the picture of action is a matter of causal determination. The problem is that it is not plausible that one event might cause another without the possibility of some intervention preventing the occurrence of the effect, despite the occurrence of the cause.Footnote 37 As Haddock’s characterisation of the challenge to Hornsby’s (Reference Hornsby1980) account indicates, on this account, the effect of Jane’s bodily action only comes about ‘if she is lucky’. That is, insofar as we suppose that an action and its result-event are distinct and causally related, it seems that we should deny that the action determines its result-event. It is just not plausible then that any action causally determines its result-event.
Altogether, it looks like the incompatibilist about two-way powers really should avoid, if possible, endorsing option (ii), that when an agent exercises a two-way power by acting, their acting in the way that they do causally determines their action’s result-event. And, so, for the incompatibilist about two-way powers, there is no good option on which an action that is the exercise of such a power involves the agent causing the result-event of their action.
7. Incompatibilism and Non-Contingent Non-Causal Determination
I have argued so far that the incompatibilist about two-way powers should not suppose that an action that is the exercise of such a power involves either (a) contingent determination of the action’s result-event by the agent; (b) non-determinative (probabilistic, contingent) causation of the result-event by the agent; or (c) non-contingent causal determination of the result-event by the agent. I want to suggest in this section that what is available to the incompatibilist about two-way powers is instead a view on which an action that is the exercise of such a power involves non-contingent non-causal determination by the agent of their action’s result-event.
This suggestion builds on the problems facing each of the two forks of the dilemma spelt out for the incompatibilist picture of two-way powers as causal in the previous section (§5). For, if we suppose that an agent, in acting, non-causally determines the result of their action, there seem to be two options available with regards to what this amounts to, which are analogous to those options identified in §5: The set of features of the agent that figure in the non-causal determination of the result-event may either be: (i) some feature(s) that the agent has prior to their actions; or (ii) they may simply amount to the feature of the agent acting in the way that they do.
Now, option (i) is, again, not available to the incompatibilist about two-way powers. The incompatibilist after all denies that an action that is an exercise of a two-way power, and its result-event, are determined by anything prior to exercise itself. But option (ii), by contrast, is available. For, while there are problems facing the idea that actions that are exercises of two-way powers are causes of their result-events, which are thereby conceived of as distinct and subsequent to the actions, there is no analogous problem facing the idea that actions that are exercises of two-way powers non-causally determine their result-events. An action may non-causally determine a result-event that is not distinct, but rather either a part of, constitutive of, or identifiable with the action.
It may seem like an odd proposal that an action may determine a result-event that is part of/constitutive of/identifiable with itself. It may seem especially odd to suppose that an event may determine itself. But the sense of oddness should dissipate when it is appreciated that determination of an outcome by an entity, as I have suggested it should be understood, is a matter of necessity of an outcome of some type, given the presence of the determining entity qua an entity of certain type, with certain features. In this way, determination is really a relation between particular entities qua types. What it is for an action, a, to determine an event, e, where e is part of/constitutive of/identifiable with a is for the occurrence of an action of type ɸ-ing, with certain features, at t, to be such that an event of type E, is necessary at t, where a is an action of type ɸ-ing and e is an event of type E. We might say then that it is really a qua action of type ɸ-ing that determines e qua event of type E. And so determinative statements, such as ‘Agent A determined event e’ or ‘A’s action of ɸ-ing determined event e’, are intensional: designations of entities in these statements cannot be substituted for any other designations of those entities salva veritate.
And indeed, non-causal determination by an event, qua some type, of some part, some constituting event, or the same event qua some other type is plausibly common, regardless of whether we accept that at least some actions non-causally determine their result-events.Footnote 38 Consider, for example, that my parking my car on a yellow line might determine my breaking of the law (given our laws, the occurrence of an action of my parking my car on a yellow line is such that the occurrence of an event of my breaking the law is necessary).
And the determination would seem to be non-causal here. My parking my car on a yellow line plausibly simply is what amounts to my breaking of the law, such that there is plausibly just one event here described in two ways. Alternatively, we might suppose that my parking my car on a yellow line is an event that constitutes the event of my breaking of the law.Footnote 39 Similarly, my car’s movement from one side of a road to the other would seem to non-causally determine my car’s crossing the middle of the road (an occurrence of an event of my car’s moving from one side of a road to the other is such that the occurrence of an event of my car’s crossing the middle of the road is necessary), where the car’s crossing the middle of the road is an event that forms part of the larger event of the car’s moving from one side of the road to the other.
As a result, we may suppose that an agent’s basic self-change action non-causally determines its result-event, allowing that the result-event is just part of, constitutive of, or identifiable with the basic self-change action.Footnote 40 An agent’s basic self-change action of raising their arm, for example, may non-causally determine their arm’s rising, where the arm’s rising is just part of, constitutive of, or identifiable with the agent’s raising their arm. And it seems like we ought to suppose that there is some such relationship between an agent’s basic self-change actions and their result-events, given the alienating picture that is the causal alternative.
In this way, we may also suppose that the agent themselves non-causally determines the result-event of their basic self-change action, insofar as the presence of the agent with the feature of changing themselves in the way that they do renders necessary the occurrence of an event of the relevant type. An agent with the feature of raising their arm, for example, renders necessary the occurrence of an event of their arm rising.
But what, then, to say of actions that are exercises of two-way powers, though they are not basic self-change actions? That is, mightn’t actions like the pulling of a trigger, the kicking of a ball, and the opening of a door amount to exercises of two-way powers, though they are not basic self-change actions? These are actions that are interactions with external objects, whereby the agent changes those objects that are interacted with. The result-events of these interactions are changes in those external objects. As such, these interactions are at least not merely actions of self-change. Similarly, they are not basic actions, since the agent always interacts with an external object by means of changing themselves in some way. As O’Brien has put it, ‘we change objects distinct from ourselves only by changing ourselves. I move the kettle to the stove, the sandwich to my lips, the note to my friend, by, myself, moving’ (forthcoming). So, while an agent may often change objects other than themselves when they act, they do so only by means of changing themselves. As such, it seems that all basic action is self-change action, and all action involves basic self-change action.
The incompatibilist about two-way powers may then hold that, in every action that is an exercise of a two-way power, the agent non-causally determines the result-event(s) of some basic self-change action(s). The result-events of the agent’s non-basic interactions with external objects – changes in those external objects – are, by contrast, most plausibly merely probabilistically (and contingently) caused by the agent and their basic action of self-change. When it comes to changes in external objects, it seems appropriate to suppose that we bring these about causally, by changing ourselves in such a way that we hope, if we are lucky, will bring about the desired kind of change. Here, there is always the possibility of some intervention that prevents the change that we seek to bring about through our action. This is a key part of the contrast between how we relate to ourselves as agents, and how we relate to the wider world.Footnote 41
But, in that case, it is not obvious that the incompatibilist about two-way powers should hold that any of our non-basic interactions with external objects are themselves further exercises of two-way powers. After all, can it really be said that it is wholly up-to-us whether we pull the trigger of the gun, if we only succeed in pulling the trigger of the gun when nothing intervenes between our basic self-change action and the trigger’s moving? My suggestion here, for the incompatibilist two-way powers theorist, is that they take only our basic self-change actions to be actions that are exercises of two-way powers. Our two-way powers then concern only whether and how we change ourselves. This is, in fact, not far off from Steward’s (Reference Steward2012) picture of incompatibilist powers, which she characterises as fundamentally powers of self-movement.
8. Conclusion
The incompatibilist regarding two-way powers should not conceive of two-way powers as causal powers. They should not suppose that the power to ɸ or refrain from ɸ-ing is a power to cause an F event or refrain from causing an F event. But the incompatibilist can hold onto the idea that in exercising a two-way power in acting, the agent determines that which occurs as a result of their basic actions of self-change. And while this determination should be recognised as non-causal, this picture is one that retains what is most fundamental to the incompatibilist’s causal conception of two-way powers: when an agent exercises their two-way powers in acting, they determine what happens in an act of choice, where it is up to them how things proceed. The error of the causal conception is just to suppose that this determination is causal.
Open access
