Section 1: what is rationalizing action explanation?

In our everyday practices of explaining intentional actions, we appeal to temperament, character, proclivity, pathology, past experience, social class, educational status, among other things. But we also sometimes appeal to the desires, beliefs, and intentions that the agent acted on. And we sometimes appeal to these specifically to reveal what the agent saw in her action. This is Davidson’s well-known characterization of rationalizing action explanation:

I am going to refer to this sub-species of everyday action explanation as ‘Rationalizing Action Explanation’ or ‘RAE’.

‘Action explanation’ is often used interchangeably with ‘reasons explanation’ and some philosophers take it that action explanation is the explanation of an intentional action where the reason or reasons that favoured the action play a key explanatory role. According to one way of understanding such reasons, they are not ordinarily psychological states.Footnote ⁵ I am not aiming to address such non-psychologistic views in this paper, interesting though they are. It should be noted that they face large obstacles that have been discussed in the literature (e.g. Alvarez Reference Alvarez2010; Mele Reference Mele, Laitinen, Sandis and D’Oro2013). Psychologistic views of RAE remain alive and well and they will be my exclusive target here (e.g. Baker Reference Baker2013; Mele Reference Mele2000, Reference Mele, Laitinen, Sandis and D’Oro2013; Roth Reference Roth1999; Ruben 2003; Stueber Reference Stueber2012).Footnote ⁶

Discussions of RAE often take it that it is distinct from other kinds of explanation because it involves revealing the intentional action as rationally intelligible. There has been a lot of controversy about what rational intelligibility amounts to.Footnote ⁷ Many philosophers converge on the view that the action is explained, not when it is revealed as objectively good or right, but when it is revealed as appropriate given what the agent took into account when she decided to act as she did. If I ask you why you are opening the window and you tell me that you are trying to get some cool air, I see what you saw in your action, even though I think that what you should do is turn on the air conditioner instead. Given the warm room that we find ourselves in, and some things that I take you to believe, such as that the air conditioner is broken – a belief that I know to be false – I see why you thought that opening the window was the appropriate thing to do. Your action is revealed as rationally intelligible and my RAE enquiry comes to a successful conclusion.

But if you tell me that in opening the window you are trying to quench your thirst, I may not come to see what you saw in your action. Opening a window seems deeply irrational, given what, I assume, you believe and value. Of course, I might learn more about your beliefs, and come to see how what you believe about the relationship between window-opening and thirst-quenching led you to think that opening the window is an effective means to quenching thirst. In such a scenario your action may be revealed as rationally intelligible and my RAE enquiry comes to a successful conclusion.

It might be thought that the antidote to such incomprehension should be characterized as additional knowledge about the beliefs, desires, intentions, values, feelings that led to the action – more knowledge about the agent’s practical perspective on her action. But this is a bit quick. It is possible that even if I come to have extensive knowledge of what shaped your decision to act as you did, too many of the things that you relied on as given when you deliberated simply provoke more why-questions for me. For example, your beliefs about the causal relationship between window-opening and thirst-quenching merely prompt me to wonder why anyone would ever believe such things. It seems that in such a scenario I do not come to see what you saw in your action. Rather, I remain pretty mystified about why you did what you did (and believed what you believed). I may even revert to alternative kinds of explanation that don’t have the aim of revealing your action as rationally intelligible, but which give me some other kind of understanding (e.g. I understand your actions as the result of psychopathology P).

What is required to successfully complete an RAE-enquiry if it is not more knowledge of the agent’s practical perspective on her action? A plausible view is that the recipient of answers to an RAE-enquiry must use that knowledge to arrive herself at a judgment that the action was an appropriate response for the agent to make, given what the agent took into account. This would explain why even quite comprehensive knowledge of an agent’s practical perspective can leave an enquirer mystified about why the agent did what she did. It would also clarify what it takes to ‘see what the agent saw in her action’ and what it is for an RAE-enquiry to reveal the action as rationally intelligible: the successful endpoint of an RAE-enquiry is a judgment by the epistemic agent pursuing an RAE-enquiry that the action was an appropriate response by the agent to what she thought her situation was. If this is correct, we can make the following general claim about RAE-enquiries:

Having clarified what is sought in an RAE-enquiry and what its successful endpoint looks like, let’s consider some objections. Let’s suppose for the sake of argument that an agent can perform an intentional action while seeing no good in what she does. To borrow (and somewhat misuse) an example from Anscombe (Reference Anscombe1957/2000), suppose that I see you eating a saucer of mud, and I ask you why you are doing that. You tell me that you are doing it ‘for no reason’. Your response may seem to bring my enquiry to a successful conclusion – after all, I come to know that you have no aim in your action other than ingesting mud. Nevertheless, I don’t judge your action appropriate. As the endpoint does not fit with Positive Evaluation, isn’t Positive Evaluation mistaken?

It is common in theorizing about questions to distinguish between direct and corrective answers. (Belnap and Steel Reference Belnap and Steel1976) A corrective answer denies one or more of the presuppositions of the question, whereas direct answers do not. A corrective answer can be very informative, while revealing that the question does not have any true direct answer. Suppose that I asked my question of the mud-eater presupposing that she wanted some further thing from eating the mud. A further end might reveal the eating as an appropriate thing to do. But if this is so, a presupposition of my enquiry is mistaken, and my enquiry doesn’t have any true answer. Although my enquiry was one that is typical of RAE, the response is a corrective answer, and so, does not seem to be a continuation of that enquiry. A mistaken enquiry can yield an epistemic gain, and yet, the presuppositions it rested on may be mistaken. Consequently, this scenario is not a case of successful RAE and it is not a challenge to Positive Evaluation.

There are lots of other examples that may seem troubling for Positive Evaluation. Someone might kick me, and I may not know whether they have done so intentionally. With this in mind, I may ask why they did that (Smith, Reference Aguilar and Buckareff2010). I leave open whether the answer to my question will reveal why they thought this was an appropriate thing to do or whether it will reveal the causal history of, for example, an involuntary jerk. The explanation that the enquiry might yield could, then, result in full understanding, but no judgment of appropriateness. And yet, isn’t this a case of RAE?

We might think of the question as expressing two different enquiries – one that seeks the rational intelligibility of the action and one that seeks the cause of the action. Where the first shares the aim and presuppositions of the kind of practice I have been interested in – RAE – the second does not. The first qualifies as RAE, but there is little pressure to suppose that the second must also be counted as such. Consequently, I don’t see a serious challenge here to the claim that Positive Evaluation is a defining feature of successful RAEs.

Suppose that you think I’m irredeemably irrational, and you have no interest in trying to understand what I saw in my action, but you need to know whether I had purpose A or B in my action. Suppose that you are eating a saucer of mud and I think that you might be eating it for health reasons, or as an act of protest. You ask me why I did what I did and I tell you that the eating is an act of protest. Isn’t this RAE, and so, isn’t this example another challenge to Positive Evaluation?

In this example it is not presupposed that the agent is acting in her capacity as competent rational animal, nor is the aim to see ‘what she saw in her action’. There are, then, important differences between this case and the paradigmatic cases of RAE. This is a good reason to think that they are not the same kind of explanation, and so, that this is not a counter-example to Positive Evaluation either.

It is worth noting that the enquiries that are not governed by Positive Evaluation do not feed easily into other social practices in the way that RAE does. In RAE, my exclusive concern here, we seek information about the practical perspective that an agent has on her action so that we may understand why she sees it as an appropriate thing to do. This may allow us to learn from the agent, to remonstrate with her, to praise her, or to blame her. For example, we ask the expert why she ran the test at t₁ instead of at t₂. Understanding her action on the basis of an RAE may be our own path to expertise. Or, it allows us to understand what we believe are the mistakes of the putative expert so that we can both fully understand and then correct those mistakes. Praising and blaming usually requires understanding the agent’s action in light of her practical deliberation, because without that, we will not know why she thought it an appropriate thing to do, and we need this if we are to rationally blame or praise her. RAE is a kind of explanatory enquiry that is directed towards understanding agents when they act in their capacity as rational animals, and as such, it plays a distinctive role in social life.

While the phrase ‘action explanation’ may be used to refer to a wide variety of explanatory practices, I am focussing on one important sub-species of action explanation – RAE. RAE is, if the quotation by Davidson is anything to go by, the kind of explanation that has been of concern to philosophers of action under such rubrics as ‘rationalization’, ‘rationalizing action explanation’, and ‘reasons explanation’.

Section 2: (i) motivating causalism by appeal to the putative fact that it involves the explanation of an event; (ii) explanatory and presuppositional roles

There may be explanations in mathematics, philosophy, and elsewhere that are not causal explanations. But it is both a widely accepted assumption and a meticulously defended claim that explanations of particular events are causal explanations (see e.g. Skow Reference Skow2014; but see also Baker Reference Baker2005; Lange Reference Lange2013). If we assume that RAE involves the explanation of a particular event, there is a good reason to think that causalism about RAE is correct. There is, one might say, a ‘default causalism’ about RAE that comes with the assumption, which I accept here, that intentional action is a species of event.

In this section I will present reasons for thinking that this default causalism provides little support for causalism about RAE. I argue that there are at least two kinds of explanation concerning events: causal explanations of events and non-causal event-involving explanations. I argue for this distinction by appeal to examples of explanation that are not RAEs. Focussing on non-RAEs is not just a distraction: one aim of my discussion is to motivate the claim that there are many explanations that are non-causal event-involving explanations. If this is correct, and if RAE is not a species of causal explanation as I will argue in Section 4, then RAE is not just an oddity in being non-causal and event-involving. It is one of a slew of non-causal event-involving explanations. The discussion also provides evidence for a distinction between explanatory and presuppositional roles, which will feed directly into the argument against causalism about RAE in Section 4. Finally, the discussion casts some light on why inferences from Causal Claim to Causalist Conclusion are more problematic than is often appreciated.

What is it for something to be an explanation of a particular event? A reasonable view is that it is an explanation of the occurrence of the event in terms of the event’s causal antecedents – a causal explanation. But there are other kinds of explanation involving events. These are, we might say, ‘event-involving’ explanations that are not causal explanations. Consider the following example:

Q: Why did she lose the game?

A: If you sink the 8-ball before sinking the other balls that have been assigned to you – the solids in this case – you lose the game.

Let’s suppose that Q is asked by someone new to the game of pool. A is a direct (non-corrective) answer to Q – one that grants all of the question’s presuppositions. A is the only true answer to Q. A presents the explanation of the losing, but this doesn’t involve an explanation of the occurrence of the losing in terms of its causal antecedents, alternative possible causal histories, causal laws, or in terms of dispositions to cause things – things we might expect to see in a causal explanation. Rather, it involves the explanation of the losing by appeal to the rules governing the game of pool. Given those rules, the sinking qualifies as losing.

As specified in the example, the person who poses Q is puzzled about the rules of the game – she asks her question because she doesn’t see why the sinking involves losing, as opposed to winning, or gaining an advantage, or incurring a penalty etc. She is not interested in the causal history and we can make this vivid by considering the following alternative answers to Q:

A* and A** describe features of the actual and alternative possible causal histories of the sinking to explain the losing. But for someone who doesn’t know the rules of the game, such answers are unhelpful: they simply presuppose and do not illuminate the nature of the rule-based relationship between the sinking and the losing, and as such, they are inadequate answers to Q. Given the questioner’s epistemic situation, the focus of Q and A seems to be exclusively on this rule-based relation.

If the rules governing the sinking explain the losing, then it is possible to have event-involving explanations that do not amount to explanations of the occurrence of the event in terms of its causal antecedents.

Pool then offers support for the view that even if we assume that an action is an event, and so, that RAE involves an explanation concerning a particular event, it does not follow that action explanation is causal explanation. RAE could be a non-causal event-involving explanation like the one in Pool.

The second point is that Pool nicely illustrates how causal facts may be presupposed to hold by those seeking and offering explanations, but the causal facts may not themselves play a role in closing the gaps in understanding that motivate the questions asked. For example, the answer A, which I’m stipulating is a direct and the only true answer to Q, presents an explanation of the losing that presupposes a certain causal history for the sinking – the causal facts seem to play a role in framing the explanation, rather than in closing the questioner’s gap in understanding.

And this is not an isolated phenomenon – consider this example (inspired by Dretske Reference Dretske1989), which makes even clearer the questioner’s presuppositions about the causal facts:

Let’s suppose that Q1 and Q2 are asked in sequence by someone who doesn’t know what happened between our soprano and conductor. She knows only that the conductor’s heart is broken. She doesn’t speak German. She accepts A1 as the correct answer to Q1 – let’s stipulate that it is the only true answer – before asking Q2.

In A1 the soprano’s utterance of ‘Nein’ plays an explanatory role in virtue of its causal features – the utterance plays a key causal role in bringing about the grief. In A2, the utterance seems to be explanatory in virtue of semantic features alone. It is the meaning of ‘Nein’ alone that seems to close the questioner’s gap in understanding. It seems that the questioner can understand the causal facts while failing to understand the semantic facts. It is noteworthy that the enquirer who poses Q2 already accepts that the utterance of ‘Nein’ fulfilled whatever causal conditions it had to fulfill for the soprano’s utterance to count as an act of refusal and an act that caused grief. What she wants to have explained when she asks Q2 is which features of an utterance of ‘Nein’ – other than these causal features – make the utterance one of refusal. The meaning of ‘Nein’ alone closes this specific gap in her understanding. If this is right, it gives us an excellent reason to think that semantic facts taken alone are explanatory in this context. In other words, Soprano, and specifically the enquiry and explanatory answer captured by Q2-A2, is, like Pool, an event-involving non-causal explanation.

I am assuming that if we look closely at what rational questioners accept as true when they pose a question, at what precisely they are curious about, and at what facts must be described in a direct, complete, and true answer to such a question, we will shed light on which of a phenomenon’s many features are explanatory and which are not. The central question is whether when specific semantic or legal or rational features close gaps in understanding, they also close gaps in causal understanding that the enquirer seeks to close. If they close gaps in semantic or rational understanding without producing such causal understanding, and the rational enquirer’s curiosity is fully satisfied, we have a reason to think of the answers in such contexts as presenting non-causal explanations.

If what was said about Pool and Soprano is on the right track, it raises the question of what status the causal facts have in these scenarios. The causal facts are not irrelevant: it is important to the viability of the enquiry and explanation expressed in Q2-A2 that the enquirer accepts A1 before posing Q2. In accepting A1 she presupposes that the scenario is one in which the soprano’s singing fulfills certain causal conditions relevant to its being an act of refusal. If A1 were false, because, for example, the soprano’s singing wasn’t caused by her hearing the conductor’s proposal and forming an intention to refuse, but was caused merely by an intention to exercise her vocal chords, then presuppositions of Q2 wouldn’t hold, and there would be no correct answer to the question. Clearly, the causal facts have a role to play in the success of the explanation involved in A2.

A plausible explanation is that there is a distinction between two ways in which a fact can play a role in an explanatory context: it can play either an explanatory role or a presuppositional role.Footnote ⁸ I will say a little more about this distinction but a full discussion goes beyond my scope here. Following Hamblin and others, I will assume that a sound question is one that defines an exhaustive set of mutually exclusive possible answers and the truth of one entails the falsity of the others.Footnote ⁹ As mentioned before, a direct answer to a question will be treated as one in which no presuppositions of the question are challenged or denied. I will usually talk about explanatory questions – questions that seek explanations. The question ‘How many chairs are in the room?’ is not an explanatory question, but the question ‘Why are there 10 chairs in the room?’ is. Finally, I will talk about questions as expressing explanatory enquiries and answers as expressing explanations – this is because it seems that one and the same enquiry could be expressed by different questions. For example, the questions ‘Why is Donald doing that?’ and ‘What the hell is Donald doing now?’ may be uttered simultaneously by two spectators of Donald’s bizarre actions. In spite of the differences between the questions, both express the same enquiry. Similarly, different sentences uttered in answer to a question could express one and the same explanation.

We might begin to characterize the difference between explanatory role and presuppositional role in terms of (i) what must or (ii) need not be referred to in the true answer to a sound explanatory question.

Explanatory Role makes use of the problematic notion of the explanation. In this paper, for brevity and simplicity, I will work with a rough and incomplete notion of the explanation as that which is described by the true answer to a sound question.

A necessary condition on a fact that plays a presuppositional role can be understood as follows:

In Pool and Soprano facts that must obtain for presuppositions to hold don’t play an explanatory role as I have described that: they need not be referred to in the true answers (A in Pool, A2 in Soprano) to the explanatory questions (Q2 in Soprano) asked in those scenarios. The viability of the enquiries nevertheless depends upon the presuppositions holding. That certain causal facts obtain is essential to the success of the relevant enquiries in Pool and Soprano.

The distinction between explanatory and presuppositional roles sheds light on why the inference from Causal Claim to Causalist Conclusion (in the Introduction) is more problematic than is sometimes appreciated. Causal Claim does not specify which of the two roles we have identified – explanatory or presuppositional – the causal features of an explanatory intention play, and Davidson’s challenge does not register the possibility that they might play only a presuppositional role. Causalist Conclusion cannot be drawn without clarifying the role of the causal features in the context of RAE. And as long as that is not done, one can rationally accept Causal Claim while doubting Causalist Conclusion.

But I think that there is evidence for a stronger claim. It is plausible to suppose that the causal features of the intention in an RAE context must play an explanatory role and not just a presuppositional role if Causalist Conclusion is to be drawn. I will address this issue in Section 3. In addition, it may be much harder than the causalist thinks to establish that the causal features play an explanatory role. In Section 4, I will argue that, in fact, the causal features of the explanatory intention in an RAE context only play a presuppositional role, and so, that Causalist Conclusion is false.

Section 3: a necessary condition on causal explanation?

To this point, little has been said about what causal explanation is. It is too large an issue for the scope of this paper. Nevertheless, here is a suggested necessary condition:

Necessary Condition is arguably too restrictive. On Skow’s (Reference Skow2014) view, for example, information may provide a partial causal explanation of a particular event if it specifies, not what an event’s occurrence actually causally depended on, but what alternatives to the event would have resulted given variations to its causal history. In addition, many philosophers regard explanation by omission or disposition as causal explanation while denying that omissions or dispositions are causes. (McLaughlin Reference McLaughlin, Laitinen, Sandis and D’Oro2013; 102–104; Lange Reference Lange2013; 493) Should Necessary Condition on Causal Explanation be rejected as too restrictive?

As noted earlier, Davidson and other causalists claim that RAE is a species of causal explanation. On such a view RAE involves specifying ‘… which pair, from among the vast number of belief–desire pairs that were suited to cause the action, actually did cause it.’ (Davidson, 1987, 109) It seems that if RAE is a species of causal explanation, it is explanation whose aim is to isolate the actual cause of the action from among a contrast class of possible causes. Although we may reject Necessary Condition as too restrictive for an adequate characterization of causal explanation in general, I think that it is on pretty firm ground at least when it comes to RAE. In my discussion I will assume that the following modified version – Necessary Condition on Causal Explanation – RAE – is correct:

Section 4: defending non-causalism about rationalizing action explanation

Let us turn, at last, to the argument for a central thesis of this paper, which is that RAE is not a species of causal explanation. Although the preliminary discussion has taken a lot of space, establishing the possibility of event-involving non-causal explanations, and arguing for claims such as Positive Evaluation, Explanatory Role, Presuppositional Role, and Necessary Condition on Causal Explanation – RAE will prove important in the discussion of this section.

I will draw on a simple but representative case of RAE, TicketFootnote ¹²:

I am stipulating that Q3 and A3 are a paradigmatic case of explanatory success. A3 is the only true answer to the sound explanatory question, Q3.

The aim of RAE is spelled out in Positive Evaluation: in posing Q3 Quentin aims to come to see what Vera saw in her action, and a successful conclusion of the enquiry is one where Quentin judges her action an appropriate thing to have done, at least given what Vera took into account in her practical deliberation.

If anything allows Quentin to see what Vera saw in her action, it is the contents of the practical deliberation that rationally supports, or seemed to Vera to support, its performance. More specifically, it is the rational role that the contents of that deliberation play that qualifies them as explanatory: it is in virtue of the rational relations that they stand in to one another in practical thought that they can close the gap in a questioner’s understanding. If this is correct, the following claim, Contents, is true of Ticket:

It will be objected that Contents can’t be the whole story. After all, surely we aren’t just concerned with the rational intelligibility of Vera’s conclusion that she should buy a ticket to Venice in terms of considerations that rationally support it. Rather, we are explaining why Vera bought a ticket to Venice – performed that intentional action – on the basis of an intention and belief that in the course of practical deliberation yielded the conclusion that she should do this. The contents and rational relations mentioned in Contents would have to have been casually efficacious in deliberation and in the production of Vera’s action if they are to be explanatory. According to this view, Contents is false because it gives an incomplete account of what is playing an explanatory role in A3. The following should be accepted instead:

Note that our objector bases her claim that Psychological States is true on the fact that the contents of Vera’s deliberation wouldn’t be explanatory if certain additional causal facts didn’t obtain. While this line of reasoning sounds right, is consistent with Causal Claim, and echoes Davidson’s challenge, it does not show, as I have argued in section 2, that the causal facts play an explanatory role. This must be established if Causal Conclusion is to be drawn

And it is not clear that our objector can go on to establish this. In framing his enquiry, Quentin presupposes that Vera is a competent rational agent performing an intentional action, and that she is acting on what she takes to be reasons that favour her doing what she does. This is typically, if not necessarily, what is presupposed by those engaging in RAE-enquiries. Because Quentin presupposes this, he presupposes that the action fulfilled whatever conditions it had to fulfill to count as an intentional action. If the causalist about intentional action is right, the action had to have had a certain kind of causal history. Given what Quentin is presupposing, he is not curious about what kind of causal history the action had. In addition, Positive Evaluation is true of this kind of explanation, so Quentin expects the answer to his enquiry to reveal what the agent saw in her action. He remains in the dark about the precise contents of the psychological states and these contents are of interest to him because they will, he hopes, reveal the action as rationally intelligible. Insofar as Ann’s answer, A3, is a non-corrective answer to Q3, she also accepts the presuppositions of Quentin’s enquiry. So, in presenting Vera’s deliberation in her answer, she also seems to presuppose that the deliberation satisfies whatever conditions it would have to satisfy for Vera to be acting on the conclusion of her practical deliberation. Insofar as Ann is a co-operative interlocutor who is appropriately sensitive to Quentin’s curiosity, she ignores information concerning the kind of causal history the action had, and gives him information that reveals the rational intelligibility of the action. These considerations offer support for the thesis that while the contents and rational relations among them play the explanatory role, the causal facts are in the background, playing, at most, a presuppositional role.

Of course, if Vera’s explanatory intention must, as causalists maintain, cause her intentional action, then the true answer to Q3 will provide information about the causal history of Vera’ action. Quentin could infer things about the causal history of Vera’s action from Ann’s answer together with causalist claims about what it takes for Vera’s action to be an intentional action. Is this sufficient to make Ticket a case of causal explanation? In general, that some information is conveyed in an answer to an explanatory question is not a reason to define the type of enquiry by reference to that information. We can learn lots of things from an answer that are not part of the explanation proper. I may learn that the player who threw the baseball that broke the window was not abiding by the rules of the game when she threw. But if I seek causal understanding and the information about the rules does not contribute to this, we should be reluctant to class this as an explanation that is concerned with rules. For causal facts to qualify as explanatory, we need a better argument than that the causal information can be inferred from the answer given. For example, we should be able to see that both of the sentences in the following answer to Quentin’s question are playing explanatory roles:

A3* strikes me as confusing. If the first sentence captures something that is explanatory, there is no point in adding the second – the first wouldn’t be explanatory if the second were not true. When we think of A3* as uttered by Ann in answer to Quentin’s question, the second sentence seems to violate Gricean maxims of quantity and relevance. Because there is a kind of ordering between the sentences, where the explanatory value of the first sentence depends on the truth of the second, explicitly adding the second threatens to call the explanatory value of the first into question. It treats the facts that the first depends on as if they need to be confirmed. But, then, it’s not clear that the first could be offered by a co-operative interlocutor as the explanation if that interlocutor is also offering the second as a part of the explanation. If this is correct, it offers further support for the idea that the first sentence refers to features playing an explanatory role and the second refers to features playing a presuppositional role. Consequently, it is not at all clear that A3* can be understood as having two complementary explanatory elements.

We have, I think, robust reasons to think that the causal features referred to in Psychological States are only playing a presuppositional role. We can grant the objector’s point that the causal facts must obtain, but it doesn’t follow that these facts are playing an explanatory role, nor that Contents is incomplete. Furthermore, Ticket seems to be analogous to Pool and Soprano (Q2-A2) in that causal features relevant to the success of the explanation do not play an explanatory role. Finally, if we accept Necessary Condition on Causal Explanation – RAE, it looks like this case of RAE is not a species of causal explanation. We should deny that explanations like Ticket are causal explanations, and we should not subscribe to the general and widely accepted claim that RAE is a species of causal explanation.

But consider the following objection: imagine that Quentin has two hypotheses in mind when he asks why Vera bought the ticket. He knows that she bought the ticket either because she genuinely intends to visit Venice, or because she doesn’t really intend to go to Venice, but she intends by her buying to show her boss how committed she is to her job – there will be a work conference in Venice at the time for which the ticket is valid. Quentin asks the question because he wants to know which of the two hypotheses is correct. Let’s call this scenario Ticket* to distinguish it from the original. Suppose that Ann tells Quentin ‘Vera bought the ticket because she genuinely intends to visit Venice’. Doesn’t Quentin’s question seek causal understanding and doesn’t Ann’s answer facilitate it? And isn’t this a case in which the action is rendered rationally intelligible and in which it is causally explained?Footnote ¹⁴

If this is a case in which the action is rendered rationally intelligible – and it must be if it is to qualify as an RAE – then in accordance with Positive Evaluation, the successful completion of the RAE-enquiry will involve Quentin coming to judge that Vera’s action was appropriate. Let’s grant that Ann’s answer allows Quentin to both form the judgment that Vera’s action was appropriate and that it allows him to come to know which intention was causally efficacious. But Ann’s answer wouldn’t be relevant to the epistemic goal of revealing the action as rationally intelligible if it weren’t accurate about the causal facts. This is the problem of ‘ordering’ that we have seen in the discussion of A3*. If the two kinds of fact – causal and rational – are ordered so that one cannot understand the rational facts without first knowing the causal ones, we still have a reason to think that the causal facts are playing a presuppositional role in an explanatory context where rational intelligibility of the thing to be understood is of paramount importance.

It will be pointed out that Quentin is actively and simultaneously seeking to fill gaps in both his causal and ‘rationalizing’ understanding in asking his question. And we can stipulate that Ann is aware that Quentin has two hypotheses and that Ann intends of her answer that it closes both kinds of gap in Quentin’s causal understanding. This, it should be conceded, looks like a case of RAE in which there is causal explanation.

But there are two reasons to think that this is nevertheless not a victory for causalism about RAE. First, we might think of the deep structure of Ticket* as analogous in key respects to Soprano. It seems that the causal and rationalizing facts are such that if one is curious about both kinds of fact, one must be sure of the causal facts before the rational facts can produce understanding. This suggests, in turn, that there are two kinds of explanation in play in Ticket*, one a causal explanation, and one an explanation concerning only rational intelligibility. The deep explanatory structure could be understood as follows:

We do not need to suppose that Quentin must first ask Q4 and then Q5, it just has to be the case that there is a non-contingent ordering to what Quentin grasps if he is to achieve the understanding that he seeks. But if we concede this, Ticket* seems to be analogous to Soprano. It seems that the recipients of answers to questions in both cases must first understand the causal facts before they can understand the rational ones. And if we grant that there is an instructive parallel between Ticket* and Soprano, and if we think that Soprano (Q2-A2) involves non-causal explanation, then we have a good reason to grant that the second component of Ticket* (Q5-A5) is also non-causal explanation. Although the objector pushes us to acknowledge that Ticket* involves causal explanation, this claim is challenged by the claim that Ticket* consists instead of two distinct explanations, one causal and one non-causal, where the latter is a non-causal explanation that, in its focus on rational intelligibility, looks like a typical RAE explanation.

The second reason to deny that the objection is a serious one is that Ticket* is not a paradigmatic case of RAE. Note that we had to stipulate that a number of special conditions had to be met to get us to the conclusion that Ticket* involved causal explanation in some way. Two competing causal hypotheses are introduced into the epistemic context, the questioner is seeking knowledge about which of these is true, and the interlocutor is aware of this. In paradigmatic cases of RAE, causal hypotheses do not explicitly shape the epistemic context. The questioner (perhaps naively) sees what they take to be an intentional action and they wonder what the agent saw in this intentional action, why the agent thought it desirable, dutiful, and so on. They pose their question supposing that the action has fulfilled whatever conditions it had to fulfill to qualify as an intentional action, and what these conditions are – causal or otherwise – are effectively ignored for the purpose of attaining a rationalizing understanding. The enquirer wants to see what the agent saw in her action, she wants to have the action made rationally intelligible. Although cases like Ticket* may happen, they are, at best, outlying cases.

It might be objected that I am carving the social world of action explanation, not at the joints, but to suit my conclusion. But I don’t think that Ticket* is the paradigmatic case in any meaningful way. It is common in everyday social contexts to ask ‘Why did she do that?’ with no prior hypotheses about why he might have done what he did. The agent may be unknown to us, or we haven’t taken the time to think of some hypotheses, or we have taken the time, but we can’t think of any, and so on. In such contexts our sole epistemic aim seems to be to come to see what the agent saw in her action. And we trust (perhaps naively) our interlocutors to have good enough epistemic access to the actual contents of the agent’s practical thought so that the presuppositions of our interlocutors’ sincere answers hold. It should be noted that in understanding paradigmatic cases in this way, I am not out of step with the philosophical literature. Most philosophers, causalist and non-causalist alike, take RAE to be centrally concerned with what the agent thought justified her action. In the literature RAE is treated as distinctive precisely because it reveals the action as rationally intelligible. Even for causalists, the causal facts seem to play some kind of supporting role.

There are other objections to be explored – causalism has not been the dominant view for so long for no reason. Nevertheless, I hope to have shown that close attention to epistemic agents’ curiosity and presuppositions should lead us away from a traditional causalism about RAE.

Let us turn to a brief review of some of the central claims of the paper. It is implausible to suppose that we can explain the very occurrence (performance) of an intentional action without direct appeal to the causal facts. We have reason, however, to deny that RAE involves explaining the very occurrence of an intentional action. First, like Pool or Soprano (Q2-A2), it seems that RAE could be a non-causal event-involving explanation. In non-causal event-involving explanations, we can draw a distinction between explanatory and presuppositional roles – an event or action is presupposed to have occurred or have been performed, and its semantic or rational or legal features alone create gaps in understanding. The semantic or rational or legal features close these gaps in understanding – they play explanatory roles independently of their causal roles.

Second, when we take Positive Evaluation seriously and we look closely at the presuppositions of a case of RAE such as Ticket with this in mind, it seems that the performance of an intentional action by a pretty competent rational practical agent (and hence, by a particular kind of causal route) is presupposed by those seeking such understanding of actions. Roughly, it is the appeal of the action in the eyes of that practical agent that piques the questioner’s curiosity. If this is right, it seems that the contents of the practical agent’s thought and the rational relations among them will close the gap in the enquirer’s understanding, and so, that they alone, independently of their causal role, play the explanatory role. But as the causal facts remain relevant to a correct explanation, it is plausible to suppose that they play a presuppositional role.

The power of Davidson’s challenge, far from being denied, is, I think, accommodated. It is acknowledged that the contents of Vera’s practical thought could not be explanatory if they were not causally efficacious. As mentioned earlier, this is a key difference between what is defended here and other approaches to defending a non-causal view of RAE. But if we attend closely to questions and answers to try to understand what is doing explanatory work in RAE and what is not, Davidson and other causalists seem to elide the distinction between explanatory and presuppositional roles and they do not pay enough attention to the presuppositions and curiosity of those who pursue RAE-enquiries.

Section 5: davidson’s challenge again

I have been stressing that what I argue for here is compatible with causalism about intentional action. But far from being a selling point, this may be regarded as a liability. It may be objected that at least some causalists about intentional action are not in disagreement with what I argue for here. In later work, for example, Davidson does not repeat the claim that ‘action explanation is a species of causal explanation’, but makes claims that might be interpreted as consistent with what I have argued for. He says,

Although it is natural to read Davidson as defending causalism about RAE, a weaker reading may also be possible. A reason couldn’t explain an action if it didn’t cause it and so its explanatory force derives from its causal features. But a source of ‘explanatory force’ might be any feature that plays a presuppositional or explanatory role. On this weak reading Davidson is not committed to Causalist Conclusion and I am not adding much to what he argued for.

But talk of the ‘explanatory force … com[ing] from …’ the specification of the cause of the intentional action is not informative about what exactly is playing an explanatory role and what is playing a presuppositional role. Without such distinctions, we lack insight into how precisely RAE works. Consequently, we remain unclear about what epistemic agents presuppose and are curious about in RAE contexts. In addition, we are not given much help in addressing puzzles about RAE, such as whether it fits easily within a naturalistic and non-first-personal framework (e.g. see the discussion in Kim [Reference Kim, Bransen and Cuypers1998, Reference Kim2010]). And we will have a harder time getting to grips with the problems of explanatory competition and putative exclusion when the cognitive sciences meet so-called ‘folk’ psychology. Even if causalists would concede that Causalist Conclusion was something of an exaggeration, then, the foregoing discussion helps to clarify this claim, and to forestall hasty acceptance of RAE’s naturalistic credentials.

It may be objected that the discussion has lost sight of one of the clear advantages of causalism over any version of non-causalism about RAE. Discussing George Wilson’s and Scott Sehon’s non-causal views of intentional action, Al Mele presents a case in which an agent, Norm, has an intention to act that causes Martians to intervene in the causal chain between Norm’s intention and movement. They do this, not to prevent or pervert the course of his movements, but to ensure that the relevant movements fulfil Norm’s intention (Mele Reference Mele2003, 45–51). In one scenario, Norm is unaware of the Martian’s intervention and takes himself to be successfully directing his own actions. But it does not seem that he is performing an intentional action, and although Norm himself would take his intention to reveal his action as rationally intelligible, it is plausible to suppose that a successful RAE is not possible in such a context. The causalist about intentional action and RAE is surely right to view the problem with Norm as causal in nature. Given this, shouldn’t we pursue a causal theory of both intentional action and RAE to explain how Norm fails to perform an intentional action and to explain why his action is not amenable to being rationalized?

Although my project in this paper is not to develop a theory of RAE, and so, a full discussion of a range of cases will not be offered, I believe that a non-causal theory would be able to exclude such a case. If Norm is not performing an intentional action, then presuppositions of RAE-questions about Norm’s actions don’t hold. Given this, RAE-questions asked by those unaware of the Martians’ intervention are unsound, and there are no true answers to them. There is, then, no RAE of Norm’s action to be had. Assuming that a successful non-causal theory of RAE should imply that there is no RAE to be had of Norm’s movements (although see Sehon, 169–171]), this is the right result.

But a non-causal theory of RAE will not illuminate the metaphysical differences between Norm and the normal practical agent, and it might be objected that a theory of RAE should say more about the metaphysics of intentional action. But it is not clear why we should accept this contention. If I am right, Davidson’s challenge is problematic because it encourages a swift move from a metaphysical claim about the ‘primary reason’ and its relationship to action to a claim about how RAE works. But how RAE works does not seem to be straightforwardly determined by what may well be the causal nature of the primary reason and intentional action. To get to grips with RAE, we must look closely at the presuppositions and curiosity of interlocutors in everyday contexts, and this data should, I think, drive our theory of RAE.

It may be worried that in trying to remain compatible with causalism, the kind of non-causal view that I favour will simply inherit the problems of both causal and non-causal views of RAE, while failing to have the benefits of either view.Footnote ¹⁶ The view will not, for example, be free of the deviance problem that dogs causalism, but by insisting on non-causalism, it will nevertheless be unable to meet Davidson’s challenge. First, although for the sake of discussion, I have often assumed that some version of causalism about intentional action is correct, I am not committed to causalism about intentional action. I assume only that there will be some important metaphysical differences between intentional actions and other movements, and between the intention acted on and other psychological states not playing a role in the production of action. Consequently, I am not committed to a view that is afflicted by the deviance problem.

I think that the second worry – that a non-causal view of RAE will not be able to meet Davidson’s challenge – is harder to address without a fully developed theory in hand and that is not the project of this paper. But the worry downplays hitherto largely untapped resources in theorizing about RAE. We may develop a more detailed account of the presuppositions of RAE-questions, RAE-answers, and the kind of curiosity that drives RAE-enquiries. And drawing on these resources and the rich literature on questions and explanation, we may find alternative ways to approach the question. The discussion of this paper suggests that it is a problematic assumption that we can only understand what makes the explanation the correct explanation in causal terms, rather than, for example, in terms of true answers to sound questions. There is a case to be made, I think, for re-orienting our theory of RAE away from the metaphysics of action and towards a detailed understanding of the questions, answers, and the curiosity motivating RAE-enquires.

Concluding remarks

Davidson’s challenge casts a long shadow over the literature on RAE. It seems to rule out the viability of non-causal views. But if the arguments of this paper are on track, we may grant that the ‘primary reason’ that rationalizes the action must cause it, while denying that RAE is a species of causal explanation. If what I have said about Ticket is correct, we may go beyond merely questioning the power of Davidson’s challenge: we have compelling evidence that RAE is not a species of causal explanation.