Actions and questions Lilian O’Brien What distinguishes actions which are intentional from those which are not? The answer that I shall suggest is that they are the actions to which a certain sense of the question ‘Why?’ is given application; the sense is of course that in which the answer, if positive, gives a reason for acting. (Anscombe 2000: 9) Anscombe claims that the applicability of such ‘why’ questions – ‘Why are you leaving?’, ‘Why did you do that?’ – is a distinguishing feature of intentional action. It is a key early claim in her highly original, and strikingly social, approach to intentional action. It promises insight into intentional action that stems, not from mental states and causal relations, but from an independently identifiable social practice. Here I will critically discuss a near-identical claim – the Distinguishing Feature Claim (DF Claim)1 – as a widely accepted necessary condition claim that is inspired by Anscombe. For brevity I will refer to the social practice of everyday enquiry into, and explanation of, intentional action in terms of reasons as the Reasons Explanation practice or RE: (DF Claim) If A is an intentional action, then an RE-question is given application to A. My main aim in this paper is to present robust reasons for thinking that DF Claim is false. First, there are intentional actions for which RE-questions are unsound, and so these questions are not, I argue, ‘given application’. Second, when these questions are ‘given application’ this is plausibly explained, not in terms of some necessary link obtaining between intentional action and such questions, but in terms of the fact that agents have exercised certain rational capacities in acting. 1 Schwenkler (2020: 20) interprets Anscombe as offering a biconditional claim. As these questions may elicit reasons for attempts and failures rather than intentional actions, it may not be sufficient for something’s being an intentional action that a why question is ‘given application’ to it. Teichmann (2015) argues that Anscombe’s interest in everyday language games should guide our interpretation of her work, and so he may reject characterizing her claims about questions in terms of an attempt to specify necessary conditions on intentional action. I cannot hope to enter the exegetical fray here. For my limited purposes I will discuss DF Claim as a necessary condition claim that is inspired by Anscombe and widely accepted – it is accepted by those who adopt an Anscombean approach to action but also by philosophers who disagree with Anscombe on many issues. Analysis Vol. XX | Number XX | XX 2022 | 1–9 doi: https://doi.org/10.1093/analys/anac069 © The Author(s) 2023. Published by Oxford University Press on behalf of The Analysis Trust. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. Downloaded from https://academic.oup.com/analysis/advance-article/doi/10.1093/analys/anac069/6970814 by guest on 21 May 2023 1. Introduction 2 | Lilian O’Brien 2. Proposed necessary condition on a question’s being ‘given application’ Tarts Q1. Why did you steal the tarts? A1. I didn’t steal the tarts! Q1 presupposes that the agent to whom Q1 is directed stole the tarts. ‘I didn’t steal the tarts’ is true, it may be epistemically helpful, but it offers a corrective to Q1. It is common to distinguish direct and corrective answers. Corrective answers conflict with a presupposition of the question (e.g. Belnap and Steel 1976: 15). When a presupposition is mistaken, the enquiry expressed by a question is faulty, and the question lacks a true direct answer. Here I take it that a ‘sound’ question is one that defines a set of answers and at least one of the set is true.2 An ‘unsound’ question, by contrast, lacks a true direct answer. The kind of question that we are concerned with seeks, as Anscombe says, the reason or reasons that the agent is acting on.3 We can think of these questions as defining a set of possible answers that present the agent’s reasons for acting. Set of answers defined by Q1: I stole them because they looked so good. I stole them because I wanted to prevent the queen of hearts from eating them. Etc. My suggestion is that it is a necessary condition on a question’s being ‘given application’ that it is sound. Why? As DF Claim concerns successful intentional action and its relationship to a common kind of enquiry into such actions, DF Claim should presumably be concerned with well-formulated instances of this kind of enquiry. It is plausible to suppose that soundness is an element in being well-formulated. Later in the discussion this assumption will be questioned, but for now I accept it: 2 This relies on Hamblin’s idea that a question sets up choice among answers (Hamblin 1973: 48). 3 ‘Reasons’ are interpreted very broadly throughout to include favourers, desires, intentions, policies, social rules, an overall end towards which one aims – something that can be cited by an agent to allow her interlocutor to see the point, appeal or appropriateness of her action. Downloaded from https://academic.oup.com/analysis/advance-article/doi/10.1093/analys/anac069/6970814 by guest on 21 May 2023 In order to assess DF Claim we should better understand what it is for REquestions to be ‘given application’ to an action, to use Anscombe’s intriguing coinage. Consider the following case: Actions and questions | 3 Necessary Condition: A question, Q, that expresses an RE-enquiry is given application to an intentional action only if it is sound. Let us turn to a problem for DF Claim: Now of course a possible answer to the question ‘Why?’ is one like ‘I just thought I would’ or ‘It was an impulse’ or ‘For no particular reason’ or ‘It was an idle action – I was just doodling’. (Anscombe 2000: 25) The challenge is clear: there are intentional actions where RE-questions are not ‘given application’. Anscombe counters: I do not call an answer of this sort a rejection of the question. The question is not refused application because the answer to it says that there is no reason, any more than the question how much money I have in my pocket is refused application by the answer ‘None’. (Anscombe 2000: 25) But it is not clear that the putative analogy – between an RE-question that elicits ‘No reason’ and the question ‘How much money …?’ that elicits ‘None’ – holds. The ‘how much’ question defines a set of answers that present quantities of money. And the true answer gives a quantity: zero. Of course, the questioner may be surprised, as she may have expected the other to have some greater amount of money, but the question ‘how much’ does not obviously presuppose that the quantity is greater than zero. We have reason to think that the answer is direct and that the question is sound. By contrast, I think that we can make a convincing case for the view that the putatively analogous RE-question has a mistaken presupposition and a corrective answer. Let us consider a simple example: No Reason Q2: Why are you raising your hand? A2: No reason. Set of answers defined by Q2: I raised it because I want to stretch my muscles. I raised it because I am trying to catch my friend’s attention. Etc. Assuming that A2 is true, we should ask whether it is included in the set of possible answers defined by Q2. Here I accept the usual Downloaded from https://academic.oup.com/analysis/advance-article/doi/10.1093/analys/anac069/6970814 by guest on 21 May 2023 3. A challenge to DF Claim 4 | Lilian O’Brien 4. An alternative Anscombe readily concedes that the answers giving RE-questions application ‘… are more extensive in range than the answers which give reasons for acting’ (Anscombe 2000: 29). This sounds like she rejects Necessary Condition. Given the difficulties that this condition poses for DF Claim, but making no pretence to exegesis, let us consider a modified version of DF Claim that does not rely on Necessary Condition. Suppose that practical knowledge – secure epistemic access to an intentional action that is exclusive to the agent and related in suitable ways to the practical reasoning on which she acts – is essential to the performance of intentional action. Suppose also that practical knowledge is what RE-questions elicit. We might, then, amend DF Claim as follows: (DF Claim – Practical Knowledge (PK)) If A is an intentional action, then an RE-question is given application to A by eliciting the agent’s practical knowledge of A. Downloaded from https://academic.oup.com/analysis/advance-article/doi/10.1093/analys/anac069/6970814 by guest on 21 May 2023 interpretation of RE-questions found in the literature, namely, that they seek the agent’s reasons for action. I also accept that the epistemic value of knowing the agent’s reasons, broadly understood, is that they can lead the enquirer to see what the agent ‘... held dear, thought dutiful, beneficial, obligatory, or agreeable’ in their action (Davidson 1963: 685). These assumptions shape the suggested answers for the set defined by Q2. But A2 does not look like the answers in the set: it does not give the agent’s reasons and it does not allow an enquirer to see in the action a feature that recommended it to the agent. We can suppose that the case is one where the agent is very idly performing the intentional action without particularly wanting to do what she is doing. In such a scenario an honest answer of ‘No reason’ seems to involve denying that one has something informative to say about why one’s action is ‘... dutiful, beneficial, obligatory, or agreeable’, as Davidson says. And in such a scenario, A2 seems epistemically helpful because it corrects the presupposition that the agent is acting for reasons that, once understood by an interlocutor, would illuminate something positive in the action and thereby allow an interlocutor to understand why the agent is performing it. If this is on the right track, A2 is corrective. If A2 is corrective, then Q2 is unsound. If we accept Necessary Condition – an RE-question must be sound to be given application to an intentional action – then this seems to be a case of intentional action where an RE-question is not given application. Not only should we reject Anscombe’s analogical reasoning because one question (i.e. ‘How much?’) seems sound while the other is not, we should reject DF Claim. Actions and questions | 5 Downloaded from https://academic.oup.com/analysis/advance-article/doi/10.1093/analys/anac069/6970814 by guest on 21 May 2023 In No Reason S answers Q2 on the basis of her practical knowledge of her action: she has practical knowledge of the fact that she is only A-ing, and so that she is not A-ing in order to B or C etc. Although Q2 is unsound, because Q2 seeks the agent’s reason for acting and S denies that there are such reasons, the question is ‘given application’ in virtue of the fact that it elicits S’s practical knowledge. We could also fruitfully compare No Reason with a case in which S is not performing any intentional action at all. In such a case a reason-seeking question makes a different kind of mistake: it mistakes S’s non-intentional behaviour for an intentional action. S’s corrective answer ‘I didn’t realize I was doing that!’ does not need to be based on any practical knowledge of S’s. And we may think that this case is also clearly one in which the RE-question is not ‘given application’. DF Claim – Practical Knowledge (PK for short) may, pending further exploration, provide a persuasive way to demarcate the class of intentional actions by reference to whether or not RE-questions are ‘given application’ to behaviours. I assume that any worthwhile variant on DF Claim will preserve the ambitions of the original: it will offer a necessary condition on intentional action in terms of its intimate relationship to the social practice of RE. If RE-questions are ‘given application’ to intentional actions even when the questions are unsound, these questions are based on mistaken presuppositions, and they seek an aspect of an action that it does not have. How could they be the basis for an illuminating necessary condition on the thing that they so clearly misrepresent? Notably, the thing that unifies the disparate cases of sound and unsound RE-enquiry is practical knowledge. In fact, it seems that if anything here illuminates intentional action, it is practical knowledge. But when we take these implications of PK into account, RE-enquiry looks like a theoretical third wheel. Insofar as PK does not preserve the theoretical ambition of using RE-enquiry as an independent source of insight into intentional action, it is inadequate to fill the role of DF Claim. It should, then, be rejected as a suitable substitute for the original DF Claim. We might reject the assumption that I have been relying on so far concerning RE-questions: their aim is to elicit the agent’s reasons for her action. Instead, we may suppose that their aim is just to elicit practical knowledge. On this approach we could revert to accepting Necessary Condition – only sound RE-questions are ‘given application’ – thereby avoiding the problematic idea that unsound questions provide an illuminating necessary condition on intentional action. Would this save PK? It still faces serious problems. In abandoning the dominant characterization of RE, it seems to abandon reliance on an independently-given social practice as a theoretical foothold from which to understand intentional action. It seems to rely instead on a theory-driven, even gerrymandered, category of enquiry: questions that elicit practical knowledge whether or not the true direct answers feed our curiosity about what the agent saw as desirable or required in her action. What unifies this category 6 | Lilian O’Brien 4 This depends, in part, on how one characterizes practical reasoning. The relationships among RE, practical reasoning, and practical knowledge require a careful discussion that goes well beyond my scope here. Downloaded from https://academic.oup.com/analysis/advance-article/doi/10.1093/analys/anac069/6970814 by guest on 21 May 2023 of enquiry is that it elicits practical knowledge. And so, we are again faced with the problem that if anything illuminates intentional action, it is practical knowledge. The theoretical ambitions of DF Claim are again abandoned. In Intention, Anscombe argues for a neat mapping of RE on to practical knowledge and for a neat mapping of practical knowledge on to practical reasoning. If the foregoing arguments are sound, these views are not quite right. When an agent says ‘no reason’ in a context where she herself sees little appeal in her own action – perhaps she is very idly but intentionally drumming her fingers on the table – she seems to both express her practical knowledge and offer a corrective answer to an REquestion. She seems to thereby deny the applicability of the RE-enquiry to her intentional action. And if she has practical knowledge of the fact that she is acting for no reason, this practical knowledge may also outstrip her practical reasoning.4 It may be thought that the arguments here underestimate the resources of Intention. For one example, Anscombe argues that ‘… the concept of voluntary or intentional action would not exist, if the question “Why?”, with answers that give reasons for acting, did not’, thereby affirming an exceptionally tight connection between intentional action and RE (Anscombe 2000: 34). Could such a claim and the argument it is based on provide the necessary support for DF Claim? It is not clear. First, Anscombe’s argument for this claim remains subject to ongoing exegetical discussion. (e.g. Schwenkler 2020: 56–63, Teichmann 2008: 44–46, Vogler 2008: 205–12, Wiseman 2016: 114–20) There is no straightforward interpretive path from her views to a rejection of the arguments here. Second, Anscombe’s claim seems to be that we only acquire the concept of intentional action by engaging in the social activity of asking for and giving reasons for action. This may be so, and yet it does not follow that intentional actions are actions that are performed for reasons. Anscombe herself seemed to think that some actions that are not performed for reasons can be intentional actions. Given the challenges involved in mining the rich resources of Intention for solutions, I will not attempt to say more. Although there may be a solution available from within an Anscombean framework, it should be borne in mind that the target here is the widely-accepted DF Claim, and an Anscombean solution may rely on claims that some adherents of DF Claim would not accept. I set further enquiry into Anscombe’s views aside, and turn to assessing one last variant on DF Claim. This variant also aims Actions and questions | 7 to preserve the spirit of the original, while avoiding the problems encountered with the original and with PK. We can borrow an analogy from John Schwenkler that is inspired by Anscombe: As a pocket is something about which one can sensibly ask what it contains, and a competitive game something about which one can sensibly ask who won, so an intentional action is something about which one can sensibly ask for a reason-giving explanation. (Schwenkler 2020: 43) Perhaps we can accept that a game without a score is defective qua game, and that an intentional action that does not yield a true direct answer to an RE-question is defective qua intentional action. Put another way, intentional actions that yield true direct answers to RE-questions set the standards by reference to which all intentional actions are to be measured as defective or exemplary. This suggests the following variant on DF Claim: (DF Claim – Paradigmatic) If A is a paradigmatic intentional action, then an RE-question is given application to A. This makes a troubling concession: DF Claim promises that as long as we can understand what it is for an RE-question to be ‘given application’, we will have insight into intentional action. But Paradigmatic complicates this by raising questions about what subset of intentional actions RE will give us insight into and why. We must solve, not just for ‘is given application’, but also for ‘paradigmatic’. To sharpen the worry, some reflection on intentional actions suggests that we can find significant commonalities among members of the category that have no clear connection to successful RE-enquiry. For example, intentional actions typically, if not necessarily, involve (i) an agent’s knowingly tokening a specific act-type; (ii) having a special kind of epistemic access to the action; (iii) employing knowledge-how in acting; (iv) a specific relationship between the action and some motivational or executive state; (v) an exercise of control that results in a relationship of correct match between an action and the plan of action that the agent represents in executive thought. These conditions involve the exercise of rational capacities and most, if not all, of the conditions would seem to be satisfied in a case such as No Reason. Why is such an action not a ‘paradigmatic’ intentional action?5 5 As mentioned earlier, for Anscombe the concept of intentional action is tightly bound up with our social practices of asking for, and giving reasons for, intentional actions. On this approach the idea that the sketched conditions could be sufficient for paradigmatic intentional action is fundamentally mistaken. Downloaded from https://academic.oup.com/analysis/advance-article/doi/10.1093/analys/anac069/6970814 by guest on 21 May 2023 5. A final alternative: paradigmatic intentional actions 8 | Lilian O’Brien Downloaded from https://academic.oup.com/analysis/advance-article/doi/10.1093/analys/anac069/6970814 by guest on 21 May 2023 Paradigmatic may be thought to have a decisive advantage here, namely, the strong intuitive pull of the idea that intentional action for reasons is the paradigmatic case of intentional action. I will suggest an explanation for the appeal of this idea. If this explanation is correct, Paradigmatic may be true, but its truth does not yield a defence of DF Claim. In fact, this explanation gives us, I will suggest, an additional reason to abandon attempts to find a successful variant of DF Claim. Some agents, such as neurotypical adult humans, have sophisticated rational – epistemic, prudential, moral – capacities, and they live in complex social settings. Given their complex capacities, and their sensitivities to rational pressure from their own plans and from moral and other considerations, they typically only act when the action recommends itself to them in one or more ways. Typically, in fact, in doing something as effortful, consequential or revealing as acting intentionally, they weigh pros and cons, or settle on a rationally intelligible overall goal, or conform their actions to social or other rules and so on. And when such agents act intentionally without taking their action to be recommended in one or more ways, these actions are, we might say, non-paradigmatic for such agents. Given their sensitivities and capacities, these actions are ‘nonstandard’ or ‘defective’. But these intentional actions are not defective in themselves. Rather, they are defective qua action-for-such-an-agent. If this is correct, it explains why actions that yield positive answers to REquestions strike us as paradigmatic actions. And although Paradigmatic may be true, rather than illuminate the nature or concept of intentional action, it illuminates the relationship between RE-questions on one hand, and the sensitivities and capacities of sophisticated agents on the other. If so, the truth of Paradigmatic does not preserve the ambition of the original DF Claim. Thus, this final attempt to vindicate DF Claim is unsuccessful. This explanation also promises to make sense of why it is natural to ask a kind of reasons-seeking question that seems similar to RE-questions about such agents’ beliefs, intentions, preferences, commitments to policies and practical roles, evaluative attitudes and even their failed attempts to perform intentional actions. Because of such agents’ capacities and sensitivities, they will usually do these things, or have these plans and evaluative attitudes and so on, only when the agents regard them as required, good or appropriate in some other way. Should we suppose that the reasons-seeking questions concerning these agents’ diverse behaviours will feature in necessary conditions on these types of behaviour, as DF Claim suggests? Hardly – the more economical thought is that it is because these are the behaviours of agents with certain rational capacities and sensitivities that these reasons-seeking questions are typically appropriate to address to these agents concerning their behaviours. Even if there are important differences between RE-questions and these other subspecies of reasons-seeking enquiry, it is not implausible to hypothesize that they are all sub-types of a general type of enquiry. If this Actions and questions | 9 Helsinki Collegium for Advanced Studies University of Helsinki Finland lilian.obrien@helsinki.fi References Anscombe, G.E.M. 2000 [1957]. Intention. Cambridge, MA: Harvard University Press. Belnap, N.D. and T.B. Steel. 1976. The Logic of Questions and Answers. New Haven: Yale University Press. Davidson, D. 1963. Actions, reasons, and causes. Journal of Philosophy 60: 685–700. Hamblin, C. 1973. Questions in Montague English. Foundations of Language 10: 41–53. Schwenkler, J. 2020. Anscombe’s Intention: A Guide. Oxford: Oxford University Press. Teichmann, R. 2015. Why ‘Why?’? Actions, reasons, and language. Philosophical Investigations 38: 115–32. Teichmann, R. 2008. The Philosophy of Elizabeth Anscombe. Oxford: Oxford University Press. Vogler, C. 2008 [2002]. Reasonably Vicious. Cambridge, MA: Harvard University Press. Wiseman, R. 2016. Anscombe’s Intention. Abingdon: Routledge. 6 I would like to thank two referees for this journal whose questions and comments helped to improve the manuscript. Thanks to Antti Kauppinen for helpful comments on an earlier draft. My thanks to audiences at the University of Messina and the University of Hradec Králové for engaging with the ideas presented here. Downloaded from https://academic.oup.com/analysis/advance-article/doi/10.1093/analys/anac069/6970814 by guest on 21 May 2023 is on the right track, the tight connection between intentional action and RE should not be understood in terms of some necessary link between these two phenomena, but in terms of the fact that RE-enquiry is a sub-type of enquiry that we pursue when an agent exercises, or seems to us to exercise, certain of their rational capacities. When we put this last line of thought together with the other sceptical lines of argument in this paper, we have robust reasons to think that DF Claim is false, and to accept that neither of the modified versions that we have considered – DF Claim – Practical Knowledge and DF Claim – Paradigmatic – successfully preserves the ambition of the original. We should, I think, be sceptical about the prospect of mining RE, an independently given social practice, for a necessary condition on intentional action.6