How to Respond to the Problem of Deviant Formal Causation
- First Online:
- Cite this article as:
- Davey, S. Philosophia (2013) 41: 703. doi:10.1007/s11406-012-9398-x
- 124 Views
Recently, a new problem has arisen for an Anscombean conception of intentional action. The claim is that the Anscombean’s emphasis on the formally causal character of practical knowledge precludes distinguishing between an aim and a merely foreseen side effect. I propose a solution to this problem: the difference between aim and side effect should be understood in terms of the familiar Anscombean distinction between acting intentionally and the intention with which one acts. I also argue that this solution has advantages over an alternative that has already been endorsed in the literature: it is a better fit for the Anscombean theory, and it naturally accommodates intuitions about the moral significance of aiming vs. merely foreseeing.
KeywordsActionAnscombeDeviant formal causationDouble effectIntentionReasons
The Problem of Deviant Formal Causation
Recently, a new challenge has been raised for a certain view in the philosophy of action that is typically associated with G.E.M. Anscombe. Roughly, the Anscombean view is that an agent’s practical knowledge determines the set of descriptions according to which her action is intentional – makes it the intentional action that it is. Paul (2011) argues that this core feature of the Anscombean view precludes it from drawing an important distinction concerning the structure of action: the distinction between an aim and a merely foreseen side effect. She offers a case where a description of an action that the agent seems to know in just the right sort of way to play this formally causal role is a description which we intuitively think of as picking out a mere side effect. But if it is known in this way, then the Anscombean view treats it as an intention, and so (we might reasonably expect) an aim, and thus blurs the distinction.
This is a significant challenge in its own right, but it also has important implications in moral philosophy.1 Any theory unable to account for this distinction is at a loss for how to explain the intuition that sometimes one is less blameworthy for bringing about a bad consequence as a merely foreseen side effect than one would be if bringing about the bad consequence were one’s aim. Setting aside the difficult question of how precisely to formulate this so-called doctrine of double effect, most – Anscombe (1958, 1961, 1982) included – would agree that something in the core idea has strong appeal. Not being able to accommodate this intuition would be a significant strike against the theory.
Most of the focus in the discussion so far – in Paul’s article, and in a reply from Van Miltenburg (2012) – has been on whether the Anscombean is indeed committed to saying that the agent has practical (formal-causally efficacious) knowledge of her action under the description that mentions the side effect. If there is no such commitment, then there is no problem; in virtue of not being known in the right way, the description does not mention an intention of the agent. For ease of discussion, I’ll call that view incidentalism.2 But I want to suggest that this is not how the most promising Anscombean solution proceeds. The most promising response for the Anscombean is not to deny that the agent has practical knowledge of her action under that description, but rather to deny that including side effects in the characterization of the intentional action precludes distinguishing side effects from aims.3 A description of an action may properly be understood as mentioning a merely foreseen side effect, and not an aim, even if it is a description of the action as intentional.
I will call that strategy intentionalism. Intentionalism has several advantages. First, it does not lead to the sort of tension with the Anscombean theory that results from the most promising version of incidentalism. Yet, second, it preserves a desirable feature of that strategy: that it attributes a special status to the descriptions of an action which correspond with the agent’s reasons for acting. Third, it coheres nicely with intuitions about double effect. I will offer a solution along these lines in section 3 and defend it against objections in section 4. But first, in section 2, I will describe the case that Paul offers and discuss various ways in which one might attempt to develop an incidentalist solution. It seems to me that careful attention to how the distinctions are drawn here will go a long way toward clearing up much of the current debate.
The case is drawn from Anscombe (2000, p. 37). A man is operating a water pump by moving his arm up and down while grasping the handle. He is doing this in order to fill a cistern and replenish the house water supply with water that he knows to be poisoned. And he is doing this because the inhabitants of the house are despots, and killing them off will pave the way for a noble group of leaders to come to power. When asked why he is poisoning the inhabitants, this man replies, “in order to polish off that bad lot and get the good lot in.” He poisons the inhabitants, by pumping the water, by operating the pump, by moving his arm up and down.
Now consider a second man who engages in all the same motions and does so with the same prior knowledge of the water’s being poisoned. When asked why he is poisoning the inhabitants, the second man (sincerely) replies, “I don’t care about that, I was just earning my pay, and that means replenishing the water, and this was the only water source.” Call the first Murderous Gardener and the second Indifferent Gardener. The difference in the responses that the men give intuitively corresponds to the difference between aim and foreseen side effect. Murderous Gardener does, and Indifferent Gardener does not, aim to poison the inhabitants; for Indifferent Gardener, their being poisoned is a side effect which he merely foresees.
Now, one might expect this difference to be understood in terms the gardeners’ intentions. But Paul’s challenge is that this is not available to an Anscombean theory. That theory conceives of practical knowledge as a special sort of non-observational knowledge of what one is doing. But the Anscombean should allow that the two gardeners’ non-observational knowledge is identical. In particular, the Anscombean should allow that both gardeners know without observation that what they are doing is a poisoning. And it seems to me that Paul is right about at least this much. This is important because if we could make the case that Indifferent Gardner’s knowledge that he is poisoning the inhabitants is not (while Murderous Gardener’s knowledge is) knowledge without observation, this would provide us with an explanation of why Indifferent Gardener merely foresees that he will (while Murderous Gardener aims to) poison the inhabitants by pumping the water: Indifferent Gardener’s knowledge that he is poisoning the inhabitants is not of the right sort to formally cause his action, so his action is not intentional under that description. However, it is difficult to imagine how we could make this case.
Indifferent Gardener’s knowledge could not, for example, be deemed observational simply because he must observe evidence that the water is poisoned in order to know that he poisons the inhabitants by pumping it. While this is true, it is also true for Murderous Gardener, but it is important that his knowledge count as non-observational if we are to draw the distinction. This is why Paul stipulates that the cases are identical with respect to the information that each gardener has before acting. Furthermore, if no antecedent observational knowledge were allowed – if knowledge were deemed observational just in virtue of being tainted by observation in this way – there would be no such thing as knowledge without observation. Practical reasoning of any kind requires beliefs about one’s environment and about the candidate ways of achieving one’s ends.
Nor is it the case that it must be pointed out to Indifferent Gardener that in pumping the water he also lays the poison. This is the clearest type of case that Anscombe considers as a contrast to non-observational knowledge: an agent is not aware that his action fits a certain description until he looks to see that his motions have had a certain result. But Indifferent Gardener cannot (sincerely) answer, “Why are you poisoning the inhabitants?” by insisting, “I was not aware I was doing that.” He knows before he begins to act that his pumping the water will be his poisoning the inhabitants.
Anscombe (p. 13) also characterizes knowledge without observation as lacking the “separately describable sensation” upon which observational knowledge is based. But there is no separately describable sensation that shows Indifferent Gardener that he is poisoning.4 For what could the sensation possibly be? Whatever it is, it would have to be something not shared by Murderous Gardener, but nothing in the case suggests that they have different sensations.
These are the ways that Anscombe characterizes the observational/non-observational distinction, and none of them makes Indifferent Gardener’s knowledge that he is poisoning the inhabitants seem observational. However, even on Anscombe’s original account, there was more to practical knowledge than its being non-observational. It might be that Indifferent Gardener’s knowledge that he is poisoning the inhabitants is knowledge without observation, but not practical knowledge (in the sense that it does not formally cause the intentional action). The strategy here would still be to show that Indifferent Gardener does not intentionally poison the inhabitants because his knowledge is not of the right sort to formally cause his action. But on this account, the feature that makes it the wrong sort of knowledge is a feature that it has despite being non-observational.
Our enquiries into the question ‘Why?’ enable us to narrow down our consideration of the descriptions of what he is doing to a range covering all and only his intentional actions. ‘He is X-ing’ is a description of an intentional action if (a) it is true and (b) there is such a thing as an answer in the range that I have defined to the question ‘Why are you X-ing?’. (pp. 37–8)5
If we can show that Indifferent Gardener’s answer to the question “Why are you poisoning the inhabitants?” refuses it application in its special sense, then we have what we need to characterize his poisoning as a side effect rather than an aim. Unfortunately, Anscombe does not give us what we are looking for. She considers two types of answer – those that reveal the action to be involuntary, and those that give a mental cause – and both, I think, can quickly be dismissed as irrelevant to Indifferent Gardener’s poisoning the inhabitants. Involuntary motions, according to Anscombe (pp. 15, 25), are known to the agent without observation, but the special sense of the ‘Why?’ question is still denied application because there is “no such thing as a cause known without observation.” The example that she gives (p. 13) is “the odd sort of jerk or jump that one’s body sometimes gives when one is falling asleep.” But Indifferent Gardener’s poisoning the inhabitants is not like that; moving his arm up and down while grasping the pump handle is certainly something he does voluntarily. Nor, it would seem, is Indifferent Gardener’s reply to the ‘Why?’ question one that denies it application by citing a mental cause: a cause that is known, qua cause, without observation, but which is to be distinguished from a motive or intention (pp. 16–24). The answer that Paul (p. 14) envisions – plausibly, I think – to “Why are you poisoning the inhabitants?” is “In order to earn my pay; I have to fill the cistern, and this is the only water there is.” This reply is not causal, but purposive.
So if there is a way of showing that Indifferent Gardener’s knowledge that he is poisoning the inhabitants is not practical knowledge, despite its being non-observational, it will have to be a way that Anscombe does not envision. For all Anscombe says, we should agree with Paul that the special sense of the ‘Why?’ question is granted application in Indifferent Gardener’s case.
[N]on-observational knowledge is practical only when it plays a specific role in the coming about of the action that it represents. But if this is so then, even though both gardeners might know the same things non-observationally, the two gardeners do not have the same practical knowledge.
But there seems to be some disagreement between Paul and van Miltenburg over what this type of strategy is entitled to suppose. Van Miltenbrug argues that Indifferent Gardener’s knowledge that he is poisoning is not practically efficacious and therefore not practical knowledge. But Paul’s argument in anticipation of this line is that the Anscombean must beg some question or other in order to support the claim that Indifferent Gardener’s knowledge that he is poisoning is not practically efficacious.
I think Paul and van Miltenbrg are talking past one another here, and I attribute this to a blurring of two important distinctions. The importance of these distinctions will become clear as we evaluate the corresponding versions of incidentalism, which will be our task for the remainder of the section. First, we must be clear about just what sort of difference this strategy purports to reveal in our two gardener’s deliberations: a difference in their ends, a difference in what they view as means to their ends, or a difference in what they take to be reasons for their actions. Second, with respect to drawing the distinction in terms of their reasons, we must be clear about where the description “poisoning the inhabitants” appears: in the specification of the reason, or in the specification of the action done for a reason.
take poisoning the inhabitants as his end.
take poisoning the inhabitants to be a (his chosen) means to his end.
take the fact that there is poison in the water as his reason to pump it.
take himself to have a reason to poison the inhabitants.
I want now to suggest that, in the face of Paul’s arguments, van Miltenburg’s response supports only version (C), but that the success of incidentalism depends upon providing support for version (D). And this is because the truth of (D) is what would show that Indifferent Gardener refuses the question “Why are you poisoning the inhabitants?” application.
While version (A) might be an accurate characterization of the case, it would be of no use against Paul’s challenge; the notion of an aim is just too similar to the notion of an end. As Paul points out (p. 13), the suggestion that only Murderous Gardener has the aim of poisoning the inhabitants because only he takes poisoning them as his end, while perhaps true, would not be a solution to the problem; it would be a stipulation that the problem is solved.
But version (B) is somewhat trickier, and both authors seem to conflate it with either version (C) or version (D). Paul (p.13) first describes the strategy as an “appeal to how the agents conceive of their reasons for action,” but then explains it in terms of means to ends. Van Miltenburg first explains his solution (pp. 2–3) – and then insists that Indifferent Gardener’s knowledge is not practically efficacious (pp. 3–4) – in terms of means to ends, but then responds to Paul’s charge of question-begging by drawing the distinction in terms of reasons (p. 4). Having made these distinctions, we can try to pull their comments apart and assess each version on its own.
Paul cautions against thinking of means and ends in terms of types. Indifferent Gardener’s act of replenishing the water supply is an act of poisoning the inhabitants, whether or not replenishings are typically poisonings. Insofar as his replenishing is a means to earning his pay, it must also be correct to say that his poisoning is. Furthermore, Indifferent Gardener knows this – he knows that certain steps that he might take toward earning his salary would be, among other things, poisoning the inhabitants, and he opts to take those steps. The fact that his act’s being a poisoning is not the feature in virtue of which it is a means to earning his pay does not change the fact that it is a means to earning his pay – indeed, the only one.
This is a strike against version (B); it insists that both gardeners choose an act of poisoning as a means to their respective ends. And this argument is effective against some of what van Miltenburg has to say. In laying out his solution (p. 3), he says, “The Anscombean idea…is that the agent brings about the sequence of events only because she views them as means to the action she knows she is performing.” But the implication that Indifferent Gardener does not view the act of poisoning as a means to his end conflates that which serves as a means – the act which is both a poisoning and a replenishing – with the feature in virtue of which it serves as a means – its being a replenishing.
If Indifferent Gardener had been ignorant of the fact that the water was poisoned and thus hadn’t known that his replenishing would bring about the poisoning, he would still have replenished the water supply. Replenishing, after all, is a fine means to earning his salary.
But the fact that Indifferent Gardener might, in different circumstances, select pumping (what he took to be) perfectly ordinary water as a means to earning his salary does nothing to show that he does not, in his actual circumstances, select an act of poisoning as his means. Again, this is to conflate the act which is a means with the feature in virtue of which it is a means. If this subjunctive conditional test shows anything, it is that Indifferent Gardener’s belief that the water is poisoned does not efficiently cause his action. But efficient causation it not what is at issue here.
For [Anscombe], practical reasoning is the deduction of means from ends with a view to action. An Anscombean would therefore say that the indifference of our gardener consists in his failure to treat his knowledge that the water is poisoned as a reason to refrain from pumping.
This is not quite right, but it is in the vicinity of (C). If the action under consideration were refraining from pumping, his failing to treat the presence of the poison as a reason would not serve to distinguish him from Murderous Gardener.6 That detail aside, if the question is whether, or in what way, the presence of the poison is treated as a reason, then Paul’s arguments do not seem to get a hold. Murderous Gardener views the presence of the poison as a reason to do what he does, and Indifferent Gardener does not. This is a real difference between their cases, and it presupposes neither a difference in aims (as version (A) does), nor a type relation between means and ends (as version (B) does). Furthermore, in saying this, I am not, as Paul anticipates (p. 14), simply denying that the presence of the poison has a practical relevance for Indifferent Gardener. That is, I am not merely insisting that it plays a theoretical, rather than a practical role in his deliberations. On the contrary, I am happy to stipulate that Indifferent Gardener views the presence of the poison as a reason. But he views it as a reason not to do what he does. Indifferent Gardener does not view the presence of the poison as a reason to do any of the things that he does, and the same is not true of Murderous Gardener. This is a difference in the role that the poison plays in their respective deliberations, but it is not a difference in practical vs. theoretical roles.
Thus, so far as version (C) goes, van Miltenburg is right: the presence of the poison plays a different role in Indifferent Gardener’s deliberations, and it is a difference that is fair game for the Anscombean. But, I want to suggest, it is not the right difference to account for the distinction between aim and side effect on an incidentalist account. The relevant question here is whether Indifferent Gardener has practical knowledge of his action as a poisoning (and, therefore, whether he poisons intentionally). For Anscombe, the answer to that question is revealed by the applicability of the question ‘Why?’ asked in a special sense which queries the agent’s reasons for acting. And the appropriateness of that question is what shows it to be intentional under the description queried. So if we want to know whether Indifferent Gardener poisons intentionally, the question to ask is not whether the presence of the poison shows up as a reason for acting in an answer to the question ‘Why?’, but whether there is an answer of the appropriate sort to the question “Why are you poisoning?” To say that there is not, in order to distinguish Indifferent Gardener from Murderous Gardener, is to endorse version (D).7
But we have already seen that Indifferent Gardener’s answer does seem to admit the question application. It is available to him to offer a reason for which he poisons the inhabitants, namely, that doing so is the only readily available way to earn his salary.8 And the availability of this answer shows that his action, under the description “poisoning”, is of a sort for which it is appropriate to query reasons. That is, he poisons the inhabitants intentionally. No argument to the effect that Indifferent Gardener does not take the presence of the poison as a reason to act detracts from this fact. And it seems to me that this is what one would likely have in mind if one were to resist appearances in this case and insist that Indifferent Gardener could not grant the ‘Why?’ question application. That line would be motivated by the suggestion that Indifferent Gardener’s answer reveals his lack of interest in the inhabitants’ being poisoned – that he only cares about the fact that he is doing his job and earning his pay. But that would be an appeal to version (C), not version (D). The fact that Indifferent Gardener does not care about some feature of his action shows that he does not take that feature to be a reason for doing it, but it does not show that he does not take himself to have some other reason for doing an action with that feature.9
Now I think there is a very simple explanation for why it might have seemed as though version (C) could provide the right sort of difference to distinguish between our two gardeners: the difference in how the presence of the poison features as a reason is indeed part of what accounts for the difference between aim and side effect. But that is not to say that it does so via a difference in the intentional status of the action. In the next section, I will suggest another way in which it might do so.
The arguments of this section may not be decisive; there may yet be reasons for denying that “In order to earn my pay; I have to fill the cistern, and this is the only water there is” is an answer that admits the ‘Why?’ question application by giving a reason for poisoning the inhabitants. But we do now have a fairly strong prima facie reason to be skeptical of incidentalism. We ought to look for a more promising option in intentionalism. In the next section, I will argue that the difference that version (C) emphasizes can be put to use without our needing to deny that Indifferent Gardener intentionally poisons the inhabitants. Furthermore, acknowledging that we can describe actions as intentional by citing merely foreseen side effects is a natural way to accommodate intuitions about their moral significance.
If what I have said so far makes it seem doubtful that the Anscombean can account for the difference between aim and foreseen side effect in terms of the intentional status of the action, one might therefore doubt the Anscombean account. After all, what one aims at in acting seems prima facie awfully similar to what one intends in acting. It would be quite bizarre to say, “My intention in acting was to X” but to deny, “In so acting, I aimed to X.” But we ought not to make the mistake of presupposing that it follows, from the fact that X is something I do intentionally, that I aim to X. It does not sound nearly so bizarre to say “I admit that I Xed intentionally, but my aim in so acting was really not to X, but to Y.” One might choose a course of action despite its being an Xing, and in such a case, it is plausible to think of X as a description of something done intentionally, though not aimed at. It is one thing to intentionally X, and another thing to act with the intention of Xing.10 This is the natural starting point for intentionalism, and there is a very simple way of understanding the difference between aim and side effect along these lines once we recognize a certain connection between the intention with which an agent acts and her reasons for acting.
In general, the intention with which an agent acts can be understood as a redescription of the reason for which the agent acts. Giving the reason for which the agent acts reveals what the agent sees in so acting, and giving the intention with which the agent acts depicts the action as being done in the service ofthat. Murderous Gardener sees the opportunity to pave the way for the noble leaders as a (decisive) reason to pump the water, and he acts on that reason; he acts with the intention of paving their way. Indifferent Gardener does not. He sees the fact that it is the only (readily available) way to earn his salary as a (decisive) reason to pump the water, and he acts on that reason; he acts with the intention of earning his salary.11 And this can be said for any of the other intentional descriptions of their actions; Murderous Gardener Xs with the intention of paving the way for the noble leaders, and Indifferent Gardener Xs with the intention of earning his salary, whether Xing is poisoning, or pumping, or moving his arm up and down.
The narrower the description of the action – that is, the earlier it appears in the succession of ‘Why?’ questions – the more there is to say about the agent’s reason for doing it, and so the more there is to say about the intention with which it is done. Murderous Gardener’s reasons for moving his arm up and down while grasping the pump handle include its being a way of pumping and of poisoning the inhabitants, and so he can also be said to act with the intention of pumping the water and of poisoning the inhabitants. Indifferent Gardener’s reasons for moving his arm up and down while grasping the pump handle include its being a way of pumping and of replenishing the water supply, and so he can also be said to act with the intention of pumping the water and of replenishing the water supply. And here is where the difference we observed in considering version (C) of the incidentalist strategy has an impact. Unlike Murderous Gardener, Indifferent Gardener does not take the presence of the poison to be a reason for his action under any description, so there is nothing that Indifferent Gardener does that he does with the intention of poisoning. While both gardeners intentionally poison the inhabitants, only Murderous Gardener acts with the intention of poisoning them, and this is how the Anscombean can understand the difference between aim and side effect.
Recall that this difference – in the reasons for which the gardeners act – is fair game; it creates none of the problems of versions (A) and (B) of the incidentalist strategy, and it draws no illicit lines between the practical and the theoretical. Furthermore, in how it makes use of this difference, intentionalism is a better fit for the Anscombean view because it recognizes that Indifferent Gardener poisons the inhabitants intentionally (and so grants the “Why?” question application) whether or not he takes the presence of the poison as a reason to pump the water. But I want now to add that this is not just an advantage in terms of cohesion in the theory; it also provides a natural explanation of certain intuitions about the moral status of merely foreseen side effects.
While there is often something to be preferred in merely foreseeing a morally objectionable result rather than aiming to produce it, there is nonetheless a strong inclination to place blame on any agent who knowingly brings it about. Or if we do not place blame all things considered (perhaps we agree that the benefits outweigh the costs), we at least think of the agent as accountable for the result. As Paul (p. 15) points out, “there is a certain kind of unity present in one’s acceptance of all one foresees one will bring about in implementing a candidate action plan.” An advocate of intentionalism can say that this unity is a matter of the agent’s bringing about the objectionable result intentionally, whether she aims to bring it about or merely foresees that she will. No version of incidentalism has the same explanation of this unity ready to hand. An incidentalist might want instead to say that it is a matter of choice, or acceptance, or something along these lines. But then we will want to know something more about what choice or acceptance amounts to such that one can choose or accept an outcome without intending to bring it about. Of course these questions may well have answers, but they are not even demanded of intentionalism, and that should be counted in its favor.
Furthermore, the added moral significance of aiming to produce an objectionable result that underlies the doctrine of double effect has a straightforward and natural explanation on this account; it is to be understood in terms of the agent’s treating the wrong sorts of things as reasons to act. If we think that the agent who aims is more blameworthy than the agent who merely foresees, it is because only the former is compelled to act by the wrong-making features of the action or the bad-making features of the outcome, and being compelled by wrong-making or bad-making features is itself objectionable.12
The distinction that I am leaning on here – between the intentional action and the intention with which one acts – is not without controversy, even among Anscombeans. Thompson (2008, pp. 107–112), for example, has argued that it follows from an agent’s Xing intentionally that she acts with the intention of Xing. Similarly, Moran and Stone (2011, pp. 57–58) have argued that being a description which admits the ‘Why?’ question application implies being an answer to some “serially related” ‘Why?’ question. Both arguments suggest that there is a unity implicit in the underlying form of the series of descriptions that iteration of the ‘Why?’ question elicits (what Anscombe (p. 41) calls their A-B-C-D form).13 I agree that this is an illuminating point of focus, and one of the keys to understanding Anscombe’s project. But both arguments overestimate the significance of this unity precisely when it comes to the relation between intentional action and the intention with which one acts. I suspect that in both cases the overestimation is due to a rigidly linear conception of the A-B-C-D form. Cases like Indifferent Gardener’s should lead us to acknowledge that the form of descriptions of intentional actions sometimes branches off in a pattern that mirrors the difference in the two gardeners’ reasons for action
To a certain extent the three divisions of the subject made in §1, are simply equivalent. That is to say, where the answers ‘I am going to fetch my camera’, ‘I am fetching my camera’ and ‘in order to fetch my camera’ are interchangeable as answers to the question ‘Why?’ asked when I go upstairs.
This observation might lead one to view these three forms as more or less interchangeable across the board, and indeed this seems to be the case for Moran and Stone. They immediately infer14 that “the unity of the divisions lies both in the applicability of the question ‘Why?’ to the material of each, and in the suitability of each to itself rationalize action – i.e., to figure in an agent’s answer to the question ‘Why?’ asked about something else he is doing” (p. 59). But Anscombe’s comments are qualified in a way that makes them irrelevant to the current discussion. The most that follows from them is that these three forms are interchangeable where the description canbe offered as aresponse to a ‘Why?’question. But we are interested in precisely those cases where it cannot. The question is whether an intentional description of Indifferent Gardener’s action (his poisoning the inhabitants) might not be available as an answer to the ‘Why?’ question asked of his action under some other description. We cannot settle that question by appeal to how things generally are when such an answer is available. So nothing here establishes the claim that all descriptions of intentional actions are also descriptions of intentions with which.
They commit a similar (or perhaps the same) non sequitur on the next page. From the fact that “a positive answer to the question ‘Why?’ is itself the description of an intentional action, and, as such, subject to that question,” they conclude that “a reversal is implicit in the chainlike structure of what Anscombe calls ‘the ABCD form’” such that “an action, suitably described, is something which can be an agent’s answer to the question ‘Why?’” (p. 59, their italics). But again, this simply does not follow. Even granting that all answers are themselves subject to the question, it remains a possibility that not everything subject to the question is itself an answer.
Thompson’s argument is that since doing things takes time, for everything I do intentionally there will be a prior stage in the doing of it for which I might be asked “Why are you doing that?” in Anscombe’s sense of ‘Why?’. An answer to that question will mention something further along in the A-B-C-D series, and since ex hypothesi the description we start with is further along in the same course of action, it should be eligible as an answer.15 He concludes that “Xs doing A is an intentional action (proper) under that description just in case the agent can be said, truly, to have done something else because he or shewas doing A” (p. 112, his italics), in a sense of “because” that implies that X does it with the intention of Aing. But it does not follow from one description’s being satisfied sooner than another description – or even from its being a way of satisfying another description – that the former is prior to the latter in the sense relevant to the structure of Anscombe’s A-B-C-D series – in the sense, that is, that the latter could be an answer to the ‘Why?’ question asked of the former. And if that notion of priority is in play in the argument, then the first premise plainly begs the question; it says that every intentional description of one’s action will be a suitable answer to the ‘Why?’ question asked of some other description, but that is just to say that it will be the intention with which one acts under that description.
While there are intentional descriptions of the action that precede the description of it as a poisoning in the sense that they come first and are a way doing it, there are none that precede it as ‘Why?’ questions which might take “In order to poison the inhabitants” as an answer. And this is just another way of saying that Indifferent Gardener does not take the presence of the poison as his reason for doing any of the things that he does.
But what is it to take something as one’s reason to act? The version of intentionalism that I have been developing so far seems to rest entirely on this notion, and one might worry that we will not have gained anything if we leave it unexplained. However this question, while interesting and worth discussing in its own right, does not demand a complete answer here, nor will I give it one. My task in this essay is not to establish the truth of the Anscombean theory, nor some specific version of intentionalism; it is merely to show that the Anscombean has a reasonably plausible option other than incidentalism for responding to Paul’s challenge and to point out some of the virtues of this option. If I have succeeded thus far, I hope I will be allowed this admittedly large promissory note. Of course, I will not even have achieved this much if the Anscombean view in general, or intentionalism in particular, is in an obviously poor position going forward to elaborate on what it is to take something as one’s reason. So I will end by saying two things in anticipation of this complaint.
First, whatever explanatory work remains, it remains on both sides. It is not a point of contention that Murderous Gardener does, and Indifferent Gardener does not, take its being a way of poisoning the inhabitants as his reason for pumping the water; this is very nearly a stipulation of the case. So any theory in the business of distinguishing what is going on with the two of them should have something to say about what taking something as one’s reason amounts to, whether or not the theory makes use of this notion as explicitly and directly as I have. Paul (p. 19), for example, appeals to a difference in how each gardener “represents the action to himself,” and how these representations are “factored into [his] act of decision.” This is how she begins to explain the difference in the gardener’s practical commitments, which in turn explains the difference in their aims. Perhaps there are various ways of fleshing this out, but it sounds, at least initially, very much like taking something as one’s reason, and it is doing just as much work in Paul’s theory as it is in mine.
Second, there is no reason to think that an Anscombean theory has especially limited resources in this regard. Paul suggests that it does in wedding it to a form of cognitivism; she thinks the sort of story she tells is unavailable to the Anscombean because the Anscombean is committed to the intentional action’s being determined by cognitive states with mind-to-world directions of fit.18 But first, as van Miltenburg (p. 6) correctly notes, this is a misrepresentation of the view; practical knowledge, for Anscombe, has the world-to-mind direction of fit that Paul (p. 18) takes to be of “primary importance” in solving the problem.19 And second, since intentionalism separates the characterization of the intentional action from the question of aim vs. side effect, the Anscombean’s commitments concerning the former need not restrict her understanding of the latter.20 In particular, it is available to the Anscombean – not just her opponent – to appeal to something like a practical decision in answering the latter question whether or not such an appeal is off the table concerning the former.21
I hope to have shown two things: first, that there is a solution to the problem of deviant formal causation along intentionalist lines, and second, that this strategy is preferable to the rival incidentalist one. That this option has been overlooked is, I think, due to a blurring of certain distinctions that I tried to tease apart in section 2. Having done so, we see that not only can the Anscombean account for the difference between an aim and a merely foreseen side effect; she can do so in a way that straightforwardly accommodates our intuitions about the moral significance of that distinction.22
Paul (p. 11) rightly insists that the difference is”at bottom a metaphysical one.” While the implications for the doctrine of double effect make the problem especially poignant, there would be an interesting challenge here even if we had very good reason to reject any moral significance in the difference between aim and side effect.
Of course, nothing at all turns on the labels here. I choose ‘incidentalism’ for this view because it treats merely foreseen side effects as incidental – they accompany the intentional action, but they do not mark it as intentional. I choose ‘intentionalism’ for the view I endorse because it is the view that descriptions of foreseen side effects do mark the action as intentional.
Paul does not ignore this option, but she only considers one (somewhat implausible) version. I will propose a different version in section 3.
There is a separately describable sensation that tells him there is poison in the water – the sensation of seeing someone else poor the contents of a vial into the well, say – but (i) this is different, and (ii) the same goes for Murderous Gardener.
Cf. Paul (p. 7) “More precisely, what happens will amount to an intentional action under all and only those further descriptions the agent is in a position to offer in answer to the ‘Why?’ question without appeal to observation of what he is doing.” This reading is mistaken in at least two ways: (1) the relevant set of descriptions are those that feature in the questions, not in the answers (and, as I will argue in sections 3 and 4, these sets are not identical); (2) as I mention in the text, citing a mental cause does not involve an appeal to observation, but neither does it qualify as admitting the ‘Why?’ question application.
I would also prefer to say that it is the presence of the poison, rather then each gardener’s knowledge of that state of affairs, that is, or is not, treated as a reason. But this correction is not important for my purposes here.
Though Paul (p. 14) emphasizes the fact that Indifferent Gardener has an answer to “Why are you poisoning,” she also seems to slide back into version (C) from time to time. For example, she considers (p. 13), on the Anscombean’s behalf, whether the presence of the poison is among the considerations that provide Indifferent Gardener with “…a sufficient basis for operating the pump handle.” See also, supra note 4.
If the reader still doubts that this answer gives a reason for acting, I encourage her to consider that (1) no one would doubt that “In order to earn my salary” gives a reason if the question is “Why are you refilling the cistern?” and (2) describing the action as a poisoning rather than a refilling does not change the fact that it is a means to earning his pay (as I argued in considering version (B)).
Admittedly, matters are complicated here by the fact that Anscombe herself, in considering the very same case (p. 42), denies that the ‘Why?’ question applies. But this, it seems to me, is a mistake by her own lights. She says it would be “incorrect, by our criteria, to say that this act of replenishing the house water supply with poisoned water was intentional.” But, as I have been arguing, her criteria in fact suggest just the opposite, and she does not explain herself further. Later (p. 89), she characterizes actions like Indifferent Gardener’s poisoning the inhabitants as voluntary rather than intentional, again insisting that the ‘Why?’ question is refused application. But again, these comments get no support from her earlier discussion. The case for the ‘Why?’ question’s not applying to Indifferent Gardener’s poisoning the inhabitants ought to be made in the initial characterization of the special sense of the question, but it is not. In the absence of some special explanation then, the move seems at best ad hoc. The most natural explanation to read into the text here is the one I reject above, but even if we ignore my arguments, this line would not be available to Anscombe since she is very clear that “for no reason” is not an answer that refuses the question application (p. 25). For these reasons, I do not think that what she says in these few brief excerpts is compatible with the rest of her theory. I think she should have said what I say in the next section, and I think there are plenty of resources in her theory to make it work. However, I am most interested in defending the core of the Anscombean view from Paul’s charge, and if it turns out that the surviving view is one that Anscombe herself would not have wholly endorsed, I can live with that.
In light of my quick rejection of incidentalism version (A), one might expect this strategy to likewise insist upon a solution rather than provide one. But in what follows, I do more than merely insist that Murderous Gardener does, and Indifference Gardener does not, act with the intention of poisoning the inhabitants; I give an account (in terms of the notion of taking something as one’s reason for acting) that is compatible with the Anscombean theory.
I use “the fact that” loosely here. What is important is that the agent takes it to be the case, not that it is in fact the case.
I take this comment to be in keeping with the distinction that T.M. Scanlon draws between the “deliberative” and “critical” applications of moral principles in his (2008, pp. 8–36), though he frames his discussion as a rejection of double effect.
Or perhaps simply interpret this to mean…
I have taken some license in recreating the argument this way, but this was unavoidable where Thompson uses locutions such as “…it is hard to see why we shouldn’t say…” and “Why not?” (p. 108). In the following few pages, Thomson adds some detail to the argument in anticipation of a skeptical reply, but those comments are irrelevant for the purposes of my discussion.
I do not wish to weigh in here on whether there are such things as basic actions, nor on how they are best understood. For some interesting discussion, see Alvarez and Hyman (1998); Davidson (2001); McCann (1974). However, I do want to mention that one ought not to be mislead by the fact that these diagrams have left-most descriptions into thinking that this aspect of the Anscombean view comes with commitments one way or the other.
Anscombe (p. 47) also considers a variation on the linear form, but not the one I consider here. We might say that hers branches to the right, while mine branches to the left.
She characterizes this commitment at (pp. 5–6, 9) and the restrictions that she thinks follow at (p. 20).
This is a slight overstatement. I still take it to be a requirement of the Anscombean view that an agent have practical knowledge of the intention with which she acts. But there are no restrictions of the sort that would make Paul’s point.
Thus, Paul’s argument (p. 12) that merely adding the world-to-mind direction of fit does not solve the problem for the Anscombean view only applies to the incidentalist variety.
I owe a great deal of thanks to Jonathan Dancy for many conversations in the early and middle stages of developing this paper. Thanks also to two anonymous referees for their thoughtful comments.