This paper has two main aims. The first aim is to defend a pair of contrasting claims about the de se and the de nunc: the de se is not essential for (intentional) action, and the de nunc is essential for action. This contrast result is a standalone contribution to the debate about perspectival/indexical thought. But it also serves as a case study in support of a broader claim. The broader claim, arguing for which is the second aim of the paper, is that the de se–de nunc contrast is best explained by a picture on which different kinds of perspectival thoughts are seen as islands. On this picture, an account of the conceptual role of any one of them won’t settle much about any of the others. Perspectival thought is standardly framed as a natural cognitive kind. Looking at how the argument for the de se–de nunc contrast works helps us see why this natural kind picture is incorrect, and the islands picture correct. The reason the de se is not needed in agency, while the de nunc is, has to do with the level of metaphysics, or reference. It has to do with the fact that de se thought is about an agent, and de nunc thought is about a moment of time.
I begin with some reminders of the kind of examples that generate this debate about perspectival thoughts, and then discuss some standard takes on those examples.
I, the satisfier of description d or and referent of name n, am well aware that ‘d is in immediate danger’, equally that ‘n is in immediate danger’, but fail to take any appropriate action because I am unaware of ‘I am in danger’. That’s an interesting possibility. But surely it isn’t specific to the first-person, a mental or linguistic item that refers to a person or thinker. A relevantly similar case involves thought about the current moment. The current moment, the satisfier of description dt and referent of name nt, is the time at which a meeting I wish to attend is scheduled, but I fail take any steps towards going to the meeting because although I am aware of ‘The meeting begins at dt’ and of ‘The meeting begins at nt’, I am unaware of ‘The meeting begins now’. This example, involving a time not a person, seems similar. This piece of the jig-saw puzzle, the one that is just a few centimetres in front of me, is piece 456. But I don’t pick it up because although I know ‘Piece 456 is the jigsaw piece needed to finish the hat’ I don’t know ‘This is the piece needed to finish the hat’. This example, involving a jigsaw piece, not a time or a person, again seems similar.1
One can have an impression of similarity across these three cases without having an explicit view about what the similarity consists in. How one attempts to articulate one’s view about that will interact with the next question: is there a fourth case involving two non-perspectival terms: for example, the two proper names ‘Hesperus’ and ‘Phosphorus’, that is as similar to the first three cases as they are to each other? I might want to send a missile to Hesperus, know how to get a missile to arrive at Phosphorus, but not take any appropriate action because I am unaware of ‘Hesperus is Phosphorus’. Does that illustrate the same phenomenon?
If someone answers this question with a ‘yes’, then their view need not embrace perspectival thought as a natural kind. They can see all four cases as illustrating a generic point about thoughts, for example, the point that co-referential expressions cannot be substituted salva veritate in action-explanation contexts.2 Mammals native to Wyoming or France is not a natural kind, because although things in the category have important things in common—warm-blood, giving birth to live young, etc.—those are just the things that mammals have in common. Similarly, perspectival thought is not a natural kind since although perspectival thoughts have important things in common those are just the things that thoughts have in common.
But most people aren’t going to answer ‘yes’ to the last question. The following quotes are more standard takes on the category of perspectival thought:
But there is also the kind of thought we operate with when actively engaged in the world, when the world is directly presented to our consciousness; and this seems shot through with indexicality. . . . It thus seems that what the phenomenologists would call ‘being-in-the-world’ presupposes indexical thought. (McGinn 1983: 91)
[N]o transformation from indexical to non-indexical is possible without affecting the cognitive significance of the utterance and therefore changing the thought it expresses. . . . The irreducibility and indispensability of indexicals is widely acknowledged. (Recanati 2007: 243).
It is widely agreed that agents need information in an egocentric form: they must think of places as “here” and “there”, times as “now” and “then” if they are to be able to act on what they know. (Owens 2011: 267)
A strong contemporary current runs to the effect that the ability of an agent to project knowledge of the world into relevant action in the world depends upon the ability to think indexical thoughts. (Millikan 1990: 723)
. . . practical guidance is, in Perry’s phrase, essentially indexical, in the sense that its function depends not only on which of many propositions it expresses but also on how that proposition is determined by the context—specifically, on its being determined in the same way as the reference of indexical expressions such as “I”, “you”, “here”, and “now”. (Velleman 1999:64)
These quotes vary quite a bit. Some focus explicitly on agency but not all do (e.g., Recanati uses the broader idea of ‘cognitive significance’). Of the ones that mention agency, what they say about it and perspectival thought is not transparently the same. Some focus on perspectival thought as something required for agency, but not all do (e.g., Velleman). Among those that focus on perspectival thought as something required for agency, it’s not clear that what is being required is the same (e.g., the view Millikan mentions seems to be that what’s required for agency is some kind of perspectival thought or other whereas Owens’s quote suggests the stronger view that agency depends on having each of several varieties of perspectival thought). So there’s at least some diversity here and the diversity suggests some significant philosophical difficulty in answering the question ‘What is the role of perspectival thought in agency’? But the important point is that none of these authors would think it appropriate to add proper names, like ‘Hesperus’, to their list of examples. All would accept that the three examples I started with showcase something important that distinguishes perspectival thoughts from non-perspectival ones.
Two statements of opposition to this standard view are Herman Cappelen and Josh Dever in their co-authored work, and Ofra Magidor. Their position is that the interesting features of perspectival thoughts are just the interesting features of thoughts. This is a one-tier picture on which all thoughts are in principle equally intimately related to agency. Suppose you think that this is wrong. Then you might think that the standard two-tier picture on which perspectival and non-perspectival thoughts play two basically different kinds of roles in the explanation of action has to be right. But there is a third picture that deserves consideration besides these familiar two, and that I flesh out in this paper. It’s a picture of islands. The de nunc is one island, and obviously differs in important ways from thoughts of the ‘Hesperus’-kind or of the ‘3pm’-kind. But it isn’t part of a perspectival archipelago that constitutes a natural cognitive kind.
What’s a general reason to doubt that perspectival thought is a natural cognitive kind, and to expect variety within the category of perspectival thoughts? Answer: a ‘metaphysics first’ approach to cognition. Christopher Peacocke, from whom I borrow the phrase, expresses his approach by saying that, in general, an account of the metaphysics of a domain is prior to an account of thought about the things in the domain.3 One way of hearing this is as follows: we need to sort out Metaphysics 101 topics before we sort out Mind 101 topics. Want an account of de se/first-person thought? First, you’d better tell me what a person is, focusing especially on the facets of being a person that hold on your account—for example, ‘a person is something with the persistence conditions of a biological organism’—but not on other peoples’. Want an account of the de nunc? First, you’d better tell me what times are, focusing especially on the facets of being a time that hold on your account but not on other peoples’. This may or may not be the best way of understanding Peacocke, and it may or not be true. But it anyway isn’t what I have in mind here. Instead, I argue that features of persons and times that pretty much any metaphysical theory will accept significantly constrain an account of the de se and the de nunc. Want an account of the de se or de nunc? The picture defended here is that it isn’t possible to give one in a way that abstracts from the most central, uncontentious features of the natures of times and persons. More generally, an account of a particular kind of perspectival thought will depend on the nature of its referent.
One way to break the mould of the natural kind picture of perspectival thought is by working through a case study in which the metaphysics of a domain determines a different thought-kind. That is the plan here. §1 gives a précis of the recent debate about agency and perspectival thought, providing evidence that the natural kind picture is not just accepted but doing important work in determining how the significance of some key cases is framed. §2 introduces the recipe Magidor, and Cappelen and Dever, have provided for action without the de se. §3 considers whether this recipe extends to the de nunc. Its first subsection argues for some general constraints on action explanation. Its second section argues that these are satisfied in Magidor, and Cappelen and Dever’s, de se free cases, but couldn’t be in parallel de nunc free cases. Having given that argument, I provide a comparison between it and the argument of two contributions that might appear to anticipate this paper, Ludlow (2019) and my own earlier paper Morgan (2019).4 The similarity between Ludlow (2019) and this paper turns out to be quite small. Like Ludlow (2019), this paper starts from the idea that the de nunc is underexplored. But Ludlow (2019) accepts the natural kind picture of perspectival thought and thinks that action depends equally on the de nunc and the de se. The similarity with my own paper is much greater. In it, I suggest that it ought to be possible to give a metaphysics-first argument for the de nunc but not the de se being required for intentional action. The current paper actually gives such an argument, and it locates the significance of that argument as a problem for the natural kind picture of perspectival thought, and as a support for the islands picture. §4 returns to the bigger picture, and the significance of my argument that there are no de nunc free cases. It considers an objection from an opponent who agrees with what I say about the de nunc but who is committed to the natural kind picture of perspectival thought and wants to use what I say about why there can’t be action without the de nunc in arguing that there also cannot be action without the de se. In replying to this objection, I bring out the crucial role metaphysics has in explaining the original de se–de nunc contrast.
1. The Debate about Agency and Perspectival Thought: The Natural Kind Picture
Suppose one wants to articulate the distinctive contribution to action explanation of perspectival thought. One wants to only say true things about perspectival thought. And one wants to say things that are only true of perspectival thought. One might start by saying something like this:
Some intentional action explanations involve a de se component.
This seems true. For example, one might point out that, in explaining why I took some appropriate action it was essential that I thought ‘I am in danger’, thinking ‘d is in danger’ wouldn’t have done, even though I am d. The problem is that it doesn’t seem only true of perspectival thoughts. The failing to send a missile example mentioned above suggests that there are action explanations where it matters than an agent is thinking of Hesperus ‘as Hesperus’.
That provides a motivation for focusing not on the existential generalization, but on the universal generalization:
Every intentional action explanation involves a de se component. (Everyde se)
The corresponding claim, concerning the non-perspectival way of thinking of Hesperus associated with the proper name ‘Hesperus’, would be
Every intentional action explanation involves a ‘Hesperus’-component. (EveryHesperus).
This claim surely exaggerates the significance of a kind of thought that, even for philosophers and classicists, is not especially central. So it seems plausible that if Everyde se is true it really captures something distinctive about perspectival thoughts. In the recent debate, skeptics about perspectival thought, like Cappelen and Dever (see 2013:36) and Magidor (see 2015: 258–59), have taken Everyde se as the claim they need to offer counterexamples to. Fans of perspectival have typically sought to show that those counterexamples don’t work (see, e.g., Bermudez 2017).
What’s striking is the significance the dispute about Everyde se has been taken to have, by both sides. Both sides take for granted that the truth of the following claim
Every intentional action explanation involves a de nunc component. (Everyde nunc)
stands or falls with the truth of Everyde se. The more general tendency this illustrates is to treat discussion of one species of perspectival thought as a proxy way of understanding every other (in nearly every case, the de se is the species that is discussed in detail, and the other species are understood by proxy5).
The risks inherent in the understanding-by-proxy approach will perhaps be readily conceded, once it’s explicit what the approach is. Is this approach really common in the literature? With this worry in mind, I run through some examples.
The title of Cappelen and Dever’s book is The Inessential Indexical and the first sentence of their chapter on agency is: ‘This chapter argues that there are no interesting or distinctive explanatory connections between indexicality and agency’ (2013: 31, my italics). But the only cases they discuss are de se cases. Unless it’s a given that, if there are counterexample cases to Everyde se, then those cases, or suitably adjusted ones, will also be counterexamples to Everyde nunc, this is fallacious. Action might depend on the de nunc but not the de se, the possibility I explore in this paper.
José Luiz Bermúdez (2017) offers a reply to Cappelen and Dever (2013). In it, he takes for granted that, to rebut their claim that action does not depend on indexical thought, he’s obliged to show that it does depend on de se thought. Again, this ignores the possibility that action might depend on the de nunc but not the de se.
Magidor’s focus might seem more restricted, since ‘de se’ figures in the title of her paper, ‘The Myth of the De Se’. But, in fact, she defines ‘de se’ states as states ‘that are typically expressed using indexical expressions such as “I” and “now”’ (2015: 249). That is, as she uses ‘de se’, it is a synonym for ‘indexical’ or ‘perspectival’. So, Magidor’s denial of the claim that ‘de se attitudes, as opposed to other kinds of attitudes, are an essential component of any practical deliberation or intentional action’ (2015:58) is a denial of a claim about perspectival attitudes. But, just like Cappelen and Dever, the example she uses to support that denial at most shows how to act without a de se attitude (in the narrow sense), not how to act without a perspectival attitude.
Those are three fairly recent contributions to the essential indexical debate. But they follow a pattern that goes back to the original papers by Lewis6 and Perry that started the debate and runs through the quotes offered above by Velleman, Recanati, Owens, Millikan, etc.
In this paper, I argue that Everyde nunc and Everyde se don’t stand or fall together, and that the understanding-by-proxy approach is a mistake.
2. Action Without the De Se
Here is Magidor’s putative counterexample to Everyde se.7 The agent, let’s say Ofra herself, has the intention that Oxfam receive a donation. Of course, if she is to execute her intention, she will have to do something, selecting some further means. Let’s say what she does is: she fishes in her pocket for a note. But, the idea is, she does not form any further intention directed at this particular means. Her selection of means is subpersonally achieved, just as her selection of means might be subpersonally achieved in a case when she fishes around in her pocket for a note by performing a series of movements with her shoulder blade without having an intention directed at those shoulder blade movements. So the idea is that this is a case of action in which Ofra just forms a non de se ‘Oxfam receives a donation’ intention, and acts on it directly.8
I’ll pick out two ideas that the case depends on. We can think of them as together constituting a recipe for generating counterexamples to Everyde se.
The first idea concerns intentions. It’s the idea that some intentions are not de se. An intention that I donate to Oxfam is clearly a de se intention. An intention to donate to Oxfam is clearly a de se intention (on the face of it, the very same one9). But, the idea is, Ofra’s intention that Oxfam get a donation is not a de se intention. Someone might say that, really, ‘S intends that p’ ascribes to S the intention to bring it about that p, which is de se. The recipe assumes that we should instead take such ascriptions at face value as ascribing a relation, of intending, that holds between S and p.10
The second idea concerns the limited role intention plays in means-selection. As the ‘fishing in pocket by doing something with her shoulder blade’ example brings out, it’s indisputable that selecting the means to the execution of a particular intention doesn’t always involve forming a further intention. Granted that possibility, there’s no reason why it couldn’t in principle happen that means-selection for a given action is subpersonally implemented at such an early stage that, at the point where person-level explanation gives out, the agent themselves is not yet an object of any representation, and, in particular, not of any de se representation. For example, in the example discussed above, by the time the level of person-level explanation gives out, Oxfam is in view, but Ofra herself is not. That’s the second idea the recipe depends on.
So, a summary of the recipe to hold in view, for subsequent comparison with the de nunc case, is:
Agent has a worldly (i.e., not self-directed) intention.
Selection of means to the satisfaction of this worldly intention is subpersonally achieved.11
3. Action Without the De Nunc?
Arguments about what specific action explanations are adequate have to appeal to some general constraints. In the first part of this section, §3.1, I motivate some general constraints. In the second part, §3.2, I argue that while those constraints do not rule out action without the de se (§3.2.1), they do rule out action without the de nunc (§3.2.2).
3.1. Some General Constraints on Action
Anscombe said that intentional actions were ones to which a certain sense of ‘Why?’ can be applied (see 1957). The main focus for this ‘Why?’ she had mind was ‘Why are they φ-ing, as opposed to refraining from φ-ing?’ I’ll call the question with this focus the ‘Why φ-ing?’ question. There are other possible focuses for ‘Why?’, asked about an agent’s action, that are relevant here. One is: ‘Why are they φ-ing, in contrast to other agents who are not?’ I’ll call this the ‘Why them?’ question. The other is: ‘Why are they φ-ing now, in contrast to at other times, when they are not?’ I’ll call this the ‘Why then?’ question.
The three ‘Why?’ questions we’ve distinguished—‘Why φ-ing?’, ‘Why them?’, ‘Why then?’—often won’t have answers that connect up an action-triggering state with a state involving some further conceptual material. For example, I might be juggling, and there might not be more to be said in answer to the ‘Why φ-ing?’ question other than that I wanted to juggle. I need not be juggling with some further end in view, for example, acting on the thought ‘Juggling would be a nice way to entertain the children’. Similarly, it need not be that I am juggling now, as opposed to a few minutes before or after, because I think ‘The time to juggle is when the cake comes in’ and ‘The cake is coming in now’. And it need be that I am juggling, in contrast to others who are not, because I think ‘The person who ought to juggle is the person who’s been hired for that express purpose’ and ‘I am the person who’s been hired for that express purpose’.
Even so, these three ‘Why’ questions answers should get answers, if only answers that are ‘superficial’ in the sense that they don’t bring in extra conceptual material relevant to the agent’s motivation. It’s bizarre to think that that there could be a case in which I do count as juggling, intentionally, but I neither see juggling as a means to some further desirable end, nor even just want to juggle. So the ‘Why juggling?’ question at least gets the superficial answer ‘They wanted to juggle’. Similarly, it doesn’t seem there could be a case of intentional action where either of other two ‘Why?’ questions fails to get even a superficial answer. To anticipate, I am going to argue that a superficial answer to the ‘Why them?’ question need not involve the de se (e.g., in cases of the sort Magidor highlights it won’t) but a superficial answer to the ‘Why then?’ question must involve the de nunc. That’s why even though intentional action without the de se is possible, action without the de nunc isn’t.
3.2.1. ‘Why Them?’ and De Se Thought
Suppose Ofra donates to Oxfam. We’re interested in why she is donating, in contrast to the many others who are not. We’re not assuming that there must be a non-superficial answer to this question—that Ofra thinks something like ‘The person who has the most money ought to donate’ and ‘I have the most money’. But we assume that there is some state that answers the ‘Why them?’ question, if only in a superficial way. The question is whether this state has to be de se—for example, something like a state of Ofra intending ‘to donate to Oxfam’ or ‘that I donate to Oxfam’ or ‘that I bring it about that Oxfam receives a donation’.
On the face of it, it doesn’t. Suppose that Ofra simply has the non de se intention ‘Oxfam receives a donation’. Citing the fact that she is the one whose intention it is does seem to make sense of the fact that she is the one who is donating. After all, in who else’s action if not Ofra’s might the intention instead get expressed?
I think there is a metaphysics-first explanation of why the non de se intention can answer the ‘Why them?’ question. The pertinent metaphysical fact is that who the owner of an intention is fixes in whose action it will get expressed. So, whatever the further specifics of the intention are, and in particular whether or not it is de se, the fact that its owner is Ofra is enough to explain why it is in Ofra’s action that it gets expressed.
I put this as an observation about intentions, which are in focus here because they are the mental states most proximate to action. But the point equally applies to other mental states. Mental states of S’s that get expressed in action get expressed in actions of S. Suppose that at t an intention or desire that Oxfam receive a donation were Ofra’s and at t plus five minutes that same intention was magically transferred to Herman. Then merely citing the fact that it was Ofra’s intention or desire at t wouldn’t fix whether it got expressed in Ofra’s action (at t) or in Herman’s (at least five minutes later). But that isn’t possible. Mental states are non-transferable. There’s only one person in whose agency the states of a given person get expressed: that person themselves. That metaphysical truth is crucial to the possibility of a non-de se intention answering the ‘Why them?’ question.
Since the non-transferability claim is so crucial, we had better be sure that it really is true. One way of stress-testing it is by thinking through a case that might appear to be a counterexample. The following case seems at least conceptually possible: I become wired to a body other than the one that I normally control. For example, your consciousness is temporarily muted, and, while it is, I am somehow able remotely to make use of your body. In this way, you become my avatar. I use your body to execute my intention to shoot the President.
Might this be a case of transfer, that is, a case in which my intention gets expressed in your action?12 I say not. Transfer would have to happen in one of two ways. It would happen if my intention became your intention, and got expressed in your action. But this is not a case in which, while unconscious, you formed an intention to shoot the President. If forming an intention to kill is morally appalling, then I have done something morally appalling, but you haven’t. Alternatively, transfer would happen if my intention, while remaining my intention, got expressed in your action. But that doesn’t happen either. You haven’t performed any action, granted that ‘action’ refers to intentional action. If killing is morally appalling, then I have done something morally appalling but you haven’t (your role, later, may be more like that of a traumatised witness). Of course, it will look to an observer as though you are acting. Moreover, the specific means via which I acted might only be available to me because I am using you as an avatar. For example, a bazooka was my weapon of choice even though my own body is far too weak for such a heavy gun. To that extent, the action bears the distinctive agential signature of your bodily capacities. But that is analogous to the way that, if I am in a Ferrari, how I accelerate may bear the signature of the car’s distinctive capacities. It doesn’t mean that you are really the agent, any more than it means the car is really the agent.
We’ve now discussed the action without the de se cases the literature explicitly focuses on. Assuming you buy that action without the de se cases are possible, should you also accept the possibility of action without the de nunc?
3.2.2. ‘Why Then?’ and De Nunc Thought
If an agent ever does anything, there will be some particular time at which they do that thing. We can ask ‘Why were they doing it then, as opposed to a few minutes before or after?’ There won’t always be a non-superficial answer to this question (e.g., ‘I want to do it to coincide with x. x is happening now’). But there will always be at least a superficial answer to this question. The question is whether the superficial answer could fail to be de nunc. For concreteness, let’s say the action is: doing a certain physio exercise. The time of the action is: 12.03 (on a given day). I, Daniel, am the agent. I’ll first distinguish a few different kinds of possibilities for answering the ‘Why then?’ question that one naturally thinks of for a case like this, and note how they all on examination turn out to be de nunc. I’ll then explain why the ‘action without the de se’ recipe the Oxfam example exemplifies fails to provide a model for constructing a different, non de nunc kind of answer.
In one kind of de nunc-involving case, I form an intention, at some earlier point, to do my physio exercise at some time or other and, at 12.03, this gives way to a de nunc intention, and yields action then. There could be some feature of that moment that makes it seem like the right time to act. But there need not be. Time is a fertile source of Buridan’s ass cases. People regularly act at one time in full awareness that acting a bit or a lot later would be equally fine. I can plump for a time.
In a different kind of de nunc-involving case, I form an intention, earlier, to do my physio exercise at some specific time, 12.03, and, at 12.03, this gives way to a de nunc intention because, at 12.03, I know ‘It’s 12.03’. I execute a previously formed, temporally determinate plan.
Neither of these cases is a counterexample to Everyde nunc since each involves the de nunc. What would a de nunc-free case look like, and how might the ‘Why then?’ question be answered in it? Why shouldn’t the way in which a non-de se state answers the ‘Why them?’ question serve as a useful model here?
A reminder that the relevant state, mentioning which answers the question in the Oxfam case, is:
Ofra’s intention that Oxfam receive a donation.
A non-de se state like this might not be about any individual, but it does have an individual who owns it. I have argued that the fact about who owns it makes it fit to answer the ‘Why them?’ question, because mental states are non-transferable. Can we pull off the same trick in the de nunc case? In the de nunc case, the parallel answer would note when the intention appealed to was had. For example, ‘Why was Author doing their physio at 12.03?’ Candidate non-de nunc states to mention in an answer include:
Author’s 12.03 intention to do their exercise at 12.03.
And
Author’s 12.03 intention to do their exercises at some time or other.
The fact about that the intentions is had at 12.03 would answer the question about why I am doing my physio at 12.03 if intentions were ephemeral, that is, if there were always just one moment at which an intention is had, the moment at which it was first formed, and it was always expressed in action at this same moment. Ephemeralness is a temporal analogue of non-transferability. Then we could say that the fact that the intentions is had at 12.03 explains why I am acting at 12.03. The key rhetorical question (parallel to the rhetorical question mentioned in the de se discussion—‘in who else’s agency might it be expressed?’) would be: ‘at what other moment could it be expressed?’
But, of course, intentions aren’t ephemeral. Intentions can be acted on as soon as formed. But it’s also commonplace for them to get expressed in action only some time after they were first formed (on many views of intention, facilitating coordination across time is their raison d’etre). The rhetorical question gets the answer: ‘at any subsequent moment’.
Since the timing of the intention (when it was had) can’t answer the question about the timing of the action, the slack needs to be taken up by something else about the intention—its temporal content. It’s easy to see how that slack could be taken up by an intention with a de nunc content. ‘At 12.03, Author intended to be doing their physio exercise then’ makes it intelligible that I am doing my physio at 12.03. But we are trying to do without the de nunc.
As between our non de nunc two candidates—Author’s 12.03 intention to do their exercise at 12.03, Author’s 12.03 intention to do their exercise at some time or other—it’s the second that more closely resembles Ofra’s non-de se Oxfam-directed intention. Ofra’s Oxfam-directed intention is not self-directed, and an ‘at some time or other’ intention is not present-directed. But an ‘at some time or other’ intention patently contributes nothing to explaining why I am doing my physio at 12.03, rather than at some other time.
That leaves us with the other candidate. Why is Author doing their physio at 12.03? The answer given is: ‘At 12.03, they intended to do their physio at 12.03’.
This sentence might well be offered as an answer to the question. But someone doing so would probably be assuming that, at 12.03, I knew what time it was. That is, what happened at 12.03 is that I formed a de nunc intention because I was aware that the time I had already pegged for action, 12.03, had come round. This is the second of the two de nunc-involving cases we already looked at, the one in which I execute a previously formed, temporally determinate plan.
If we clarify that this case isn’t that case—that the only present-directed thought I have is an intention to do my physio-exercises ‘at 12.03’—the case becomes mysterious. Why would someone’s wanting, at 12.03, to do her physio exercises at 12.03, simply conceptualized ‘as 12.03’, make them act? The central virtue of the action without the de se recipe Magidor and Cappelen and Dever constructed was that it wasn’t just a case in which the agent was supposed to have a self-directed, non de se attitude—for example, a desire of Ofra’s that Ofra donate to Oxfam—that they acted on. If it was such a case, it would have attracted the reaction, whose potency Lewis and Perry highlighted decades ago: ‘But why should that lead to action, if Ofra doesn’t know that she is Ofra?’ The key to making the action without the de se possibility plausible is to extrude the self-directed aspect, not just the de se aspect. If we keep in the self-directed aspect, we get the ‘but that’s not the action-triggering way of thinking of the self’ worry. Similarly, if we keep in the present-directed element, we get the parallel ‘but that’s not the action-triggering way of thinking of the present’ worry. So far so similar between the de se and the de nunc. The difference is that when we extrude the present-directed element of a state (focusing on my 12.03 intention to do the physio at some time or other), we have nothing left to explain why the action occurs when it does. In particular, the time at which intention is had doesn’t explain that.
In the previous subsection, I argued that the ‘Why them?’ and ‘Why then?’ questions always get answered. In the two parts of this subsection, I’ve argued that this constraint rules out action without the de nunc, even though it doesn’t rule out action without the de se. That completes the section’s ‘Why?’-based argument that there can’t be action without the de nunc, even though there can be without the de se. I note, before moving on, two things that help contextualize the argument.
First, the two ‘Why?’ questions the argument focuses on naturally suggest the perspective of an onlooker, trying to understand another agent’s action. From the agent’s own perspective, the contrast looks a bit different. It isn’t that the agent faces both a ‘Why me?’ and a ‘Why now?’ question, and, from their perspective, the ‘Why me?’ question can be answered by a non-de se state but the ‘Why now?’ question can’t be answered without the de nunc. It’s rather that, from the agent’s perspective, only one of these two questions arises. In acting, I have to select a moment to act at, and it’s hard to see how this selection happens without something like the thought ‘now is the time to do it’. But, in acting, I don’t have to select an agent via whom to act. That choice is already made for me: platitudinously, the agent via whose agency I act is myself. So there’s no parallel need for something like the thought ‘I am the one to do it!’ A metaphysical contrast—agents spans times, but not agents—yields a role for the de nunc in action that the de se simply does not have.
Second, I provide a comparison between my argument and the argument of Ludlow (2019) and Morgan (2019), two contributions that might appear to anticipate this paper. Both of these papers agree that the de nunc is importantly underexplored. In the case of Ludlow (2019), that is the extent of the similarity with this paper. But in the case of Morgan (2019), the similarity goes beyond that. I take the two comparisons separately.
3.2.2.1 Ludlow
It’s helpful to begin by focusing just on the comparison between Ludlow’s take on the Cappelen and Dever/Magidor recipe, and mine. Ludlow (2019) engages with Cappelen and Dever’s example of the recipe, in which an agent, Herman, is supposed to be saving a particular child, Nora, while just having mental states directed at Nora, and none at himself. Ludlow comments as follows:
It is really hard to make sense of Herman’s beliefs and desires apart from some additional perspectival facts. There are billions of children in the world. Why is it that Herman cares about Nora? Why does he desire that she in particular be safe? And why does he think that he, among all the adults in the world, is responsible for taking care for her. Well, one natural explanation is that she is his daughter . . .
The creep of assumed perspectival contents does not end there. It seems that smuggled into Cappelen and Dever’s rationalizations above are some of the key perspectival contents that this book is arguing for—the temporal ones. When we say that Herman’s belief is that Nora is in danger, we are saying his belief is that she is in danger now. When we say that he believes that if the door will be closed she will be safe, we are obviously saying that if the door is closed now, she will be safe now. . . . All of these are temporal perspectival contents. (2019:29)
Ludlow’s position here is that Herman’s action requires the de se and the de nunc. The Nora case is unsuccessful, in his view, because of the creep of ‘assumed perspectival contents’. The assumed perspectival contents he mentions first are de se. But he thinks it would be a mistake to stop there. On examination, it turns out that there are other equally essential assumed perspectival contents, the de nunc contents, which he thinks have been unjustly neglected.
Another similar take is Ludlow’s comment on Velleman’s view about specifically ethical agency. Velleman (1999) says that the Kantian maxim that I should only do things I could will to be universal law is essentially perspectival in that it essentially involves the de se. Ludlow agrees. But he thinks it would be a mistake to stop there. Ludlow argues that, on examination, the maxim also turns out to involve the de nunc (2019: 9) This is consistent with everything that Velleman says, but equally not something Velleman mentions.
The similarity between Ludlow (2019) and this paper, which comes out in the examples I’ve just mentioned, is that Ludlow anticipates something that this paper has made a great deal of—the de nunc is underexplored. The difference between Ludlow (2019) is in what is supposed to happen when, having recognised that fact, we make the de nunc our explicit focus. According to me, what happens is that we discover something that appeared to hold for the de se, and in fact does not, in fact does hold for the de nunc—action requires it. This goes against the natural kind picture. According to Ludlow (2019), what happens is that we see that parallel interesting connections that obtain for the de se also hold for the de nunc. Herman needs the de nunc, and not just the de se, to save Nora. The Kantian promiser needs to be thinking about herself in a perspectival way, but also to be thinking perspectival thoughts about when the promises she is making are to be kept. This is exactly what, for example, Perry (1979) or Lewis (1979) suggested we would find, and entirely in keeping with the natural kind picture of perspectival thought.
To see that there’s nothing in Ludlow’s overall project to disrupt the natural kind picture, it’s additionally helpful to look at his summary of his project. In the abstract to his concluding chapter, he writes that his key ideas are:
that perspectival contents are ubiquitous and run through our accounts of human action, emotion, normativity, computation, and information and furthermore that communication across diverse perspectives is not only possible but routinely accomplished thanks to our ability to dynamically track modes of expression across perspectival positions. (2019:182)
The key point to note is that this summary is pitched at the level of perspectival thought, and does not mention any de se–de nunc contrasts. That’s because no such contrasts are central to Ludlow’s project.
3.2.2.2 Morgan (2019)
The similarity between this paper and Morgan (2019) goes beyond the idea that the de nunc is underexplored. In that paper, I claim that action requires that de nunc but not the de se. Moreover, I sketched a metaphysics-first argument for that claim:
Intuitively, the source of the need for de nunc attitudes in action has something to do with the fact that agents exist over time and have to choose one particular time at which to perform a given act, from all the different times at which they could perform it. If the source of the (claimed) need for de se attitudes were the same, it would have something to do with the fact that agents exist ‘over agents’ and have to choose, from the all the different agents through which they might act, just one agent through which to act. That doesn’t sound right. The kind of thing that chooses, an agent, spans times but does not span agents. (2019:60)
This is exactly the sort of argument I’ve tried to give in this paper! So, of course, I regard it as insightful. It is just an argument sketch though, which needs unpacking. Presented in this form, there are lots of ways for someone hostile to its conclusion to respond. Granted, times and agents are radically different sorts of thing, but why should that yield a dependence of action on the de nunc but not the de se? Maybe the de se is as necessary as the de nunc because in some sense an agent does have to select an agent through which to act (just like they have to choose a time at which to act) even though, of course, agents don’t exist ‘over agents’. Or maybe the suggestion that the source of the apparent need for the de se has to do with ‘selecting an agent through which to act’ is just false. Maybe the need for the de se comes from somewhere else. These are, in my opinion, reasonable worries to have. I’ve tried in this paper to allay them, by giving a fleshed out argument, oriented around the ‘Why them?’ and ‘Why then?’ questions, of the sort that I earlier said it should be possible to give. In the next section, I try to locate the broader significance of that argument.
4. Islands of Perspectival Thought
I’ve argued that action explanation always has a de nunc component. I take this to be a contrast with the de se case, where there are counterexamples to Everyde se to be had. What if someone is committed to the natural kind picture and so endorses the biconditional that there will be counterexamples to Everyde se if and only if there are counterexamples to Everyde nunc? An alternative response they might have is to agree with what I’ve said about the de nunc and see it as helping pinpoint what’s wrong with the original cases Magidor and Cappelen and Dever gave against Everyde se. Saying how I want to reply to this objection gives me an opportunity to say what I think the significance of the de se–de nunc contrast is and the role metaphysics has in explaining it—to say why, if you buy what I’ve already argued, you should reject the idea of perspectival thought as a natural kind and embrace the islands picture.
Here is the objection:
Your argument that action requires the de nunc is plausible. But it doesn’t support any de se -de nunc contrast. On the contrary, we can use your argument about the de nunc to pinpoint where the action without the de se recipe goes wrong. Your argument was that acting always involves selecting one’s moment for action, and there’s no selecting one’s moment without the de nunc. The contrast you claimed was that one does not, in parallel fashion, select an agent when one acts. Delegation falsifies this. In action, one always selects an agent. Most commonly, one selects oneself. But one can also select others (delegation cases). We might even view a phenomenon you mentioned in your argument about the de nunc—forming an intention at an earlier point and acting on it later—as a special case of delegation. When I form an intention, which I will act on later, I am delegating to my future self. In other cases, when I form an intention, and act on it immediately, the reason it triggers immediate action is that it is de nunc. Similarly, in cases where I act on an intention, and act on it directly, the reason it triggers my own action, rather than another’s, will be that it is a de se intention. If, in fact, Ofra selected herself as the agent of her action, then she must have formed the intention ‘I bring it about that Oxfam receive a donation’. Without the de se, nothing about the states Ofra acted on explains why she acted directly, rather than via delegation. That is why the de se is always essential for action.
There must be something wrong with this argument. First, focusing on the final concluding line, this argument couldn’t establish a general connection between agency and the de se (which is required to yield Everyde se) precisely because what it says is that there’s a connection between the de se and a particular subspecies of intentional agency—intentional agency not involving delegation.13 Second, the specific proposal about how the ‘de se’ figures in cases where someone acts not via delegation cannot be right. The ‘I bring it about that p’ formulation is maximally coy about means selected. So, in particular, it’s compatible with the means one selects involving delegation.
Lurking beneath these two objections is a simpler point, which captures a sense in which ‘temporal delegation’ (doing something one formed an intention to do at some earlier point) is different from person-to-person delegation. Suppose I form an intention at 9am to do my physio at 11am, which I then execute at 11am. It’s only at 11am that any action is performed. I don’t perform an action at 9am. The intention I form at 9am simply persists over two hours, updating at 11am into a de nunc intention and, at that point, yielding action. By contrast, if Ofra’s intention leads her to delegate to Herman, then Ofra still acts. In particular, she acts by: delegating to Herman. Perhaps she utters the sentence ‘Please Herman can you make a donation to Oxfam?’ Herman now has his own intention that Oxfam receive a donation, and he has that intention because Ofra has delegated to him. It’s not chillingly atomistic to think what happens in person-to-person delegation is that one person’s goal becomes the other’s, through the other person forming their own intention, as a result of something the first person has done (request, threaten, instruct etc.).
In delegation, each person, the delegator and the delegate, acts on an intention that they have. There is no action of S’s without S doing something, even if what S does is: delegate. So, the claim that the de se is involved in selecting oneself as the agent of one’s action could be the claim that the de se is involved in selecting oneself ‘qua ultimate source’ of one’s action. Or it could be involved in selecting oneself ‘qua means of execution’.
The first claim should be rejected. As a metaphysical fact, there is only one person who is the ultimate source of an action yielded by states of Ofra: Ofra herself. When Ofra acts, Ofra acts. So there is never any need for her to select who will play this ‘ultimate source’ role. The ‘Why them?’ question, seen as asking why someone is the ultimate source of a given action, can be answered simply by noting who owns a given intention. Hence, there is no need to answer that question by reference to the de se.
The ‘Why them?’ question could be understood in the second way. It could be understood as asking about the means via which someone is acting. Why is it that this action, whose ultimate source is Ofra, is being executed by Ofra herself, rather than by Herman on Ofra’s behalf? And the answer could be: this was a case in which Ofra’s intention that Oxfam get a donation led her to form the follow up intention ‘I fish in my pocket for a note’, not the follow up intention ‘I persuade Herman to fish in his pocket for a note’.
What’s hard to see is why we should expect the ‘Why them?’ question, disambiguated as a version of the ‘Why this means?’ question, always to be answered at the person level. We know that the ‘Why this means?’ question is very often not answered at the person level. We accept that it’s a brute fact about how Ofra is hardwired that the intention ‘Move my arm’ gets executed by her doing one thing rather than another with her shoulder muscles even though she has no shoulder-muscle directed intentions. Why shouldn’t it be a brute fact about how Ofra is hardwired that her intention that Oxfam receives a donation leads next to her body (her arms) moving in such a way that she fishes in her pocket in her note rather than leading to her body (her lips) moving in such a way that she utters the words ‘Herman, please give something to Oxfam’. There’s no obvious reason why there couldn’t be an agent whose states are exclusively directed outward, at the things in the world on which they are acting, and where the changing relations between themselves and those things, and the consequent changes in what specific movements are required to produce what outcomes, are monitored subpersonally. And, to recall, since Everyde se is a claim about any possible instance of intentional agency, the possibility of such an agent is all that is needed for a counterexample.
Having seen what my reply is to the hypothetical objection, let’s remind ourselves of the larger project my case-study involving the de nunc and the de se is intended to support. It’s to argue that different sub-types of perspectival thought ought to be regarded as islands, where the default expectation is there won’t be similarity of conceptual role between one subtype and another. How exactly does our case-study support this claim?
The topic of perspectival thought is broader than our discussion in two ways. First, there are kinds of perspectival thought other than the de se and the de nunc that would have been used for a different case study—for example, thought about a place ‘as here’, or a world ‘as actual’, or about an object ‘as that’. Second, someone might be interested in the connection between perspectival thought and perception, or perspectival thought and memory, as opposed to perspectival thought and agency. Agency is just one important aspect of the conceptual role of thought. So there are many interesting questions about perspectival thought that have not been in focus. Still, I think the evidence we’ve assembled is strong enough to support a view about the broader category of perspectival thought. At its most basic, the evidence suggests that the diversity of the category is larger than it’s usually seen to be. The frame of a natural kind offers a compelling way of understanding the course the philosophical discussion of perspectival thought has taken. A scientist looks at one sample of water and records their finding, expecting what they find will hold good for other samples of water, because water is a natural kind. The analogous practice of looking at one sample of perspectival thought (nearly always: the de se), recording one’s finding and expecting what one finds will hold good for other samples of perspectival thought is precisely what we find in the literature, as discussed above. And it’s hard to see what could justify that practice other than the assumption that perspectival thought is a natural kind. But that assumption, on the evidence of the previous two sections of this paper, distorts the various kinds of perspectival thought we’re trying to understand.
Here is a sketch of an alternative way of thinking about perspectival thought. There’s a pull towards similarity in the conceptual role of perspectival thoughts, associated with the hard to pin down but natural idea that in all cases of perspectival thought one thinks about an entity through exploiting a perspective one has on that entity, in contrast to other cases of thought—for example, thought that would be expressed with names—where one does not. The less highlighted pull, which this paper has focused on, is the pull towards difference. The de se and the de nunc refer to radically different kinds of thing. When we look at the conceptual roles of the de se and the de nunc in detail it turns out that an absolutely central factor is the kind of thing each is about. Put another way, you can’t describe the conceptual role of a particular kind of perspectival thought in abstraction from the nature of the thing it involves perspective on. We can think of a conceptual role of a particular kind of thought as a function of at least two different arguments. One argument is: what the thought is about. Another is: whether the thought has a perspectival or a non-perspectival reference-fixing story. There are going to be similarities between ‘NN’-thoughts and (co-referential) de se thoughts, associated with their similarity in relation to the first argument. There are going to be similarities between de se thoughts and de nunc thoughts, associated with their similarity in relation to the second kind of argument. There are going to be differences between de se thoughts and de nunc thoughts, associated with their difference in relation to the second kind of argument. And there are going to be differences between ‘NN’-thoughts and de se thoughts associated with their difference in the first kind of argument. I think the cross-perspectival similarities (e.g., similarities between de nunc and de se thoughts), and the cross-referential differences (e.g., differences between de se and ‘NN’-thoughts), are more attention grabbing than the cross-referential similarities or the cross perspectival differences. Papers by Lewis and Perry, early adopters of perspectival thought as a cognitive frame, did a brilliant job of emphasizing the former. But they are, on my way of thinking, just half of the full story. And the reason they’re attention grabbing, and that it took until the half way through the 20th century to come fully into focus, is that the importance of the other source of differences/similarities is so much more obvious. Samuel Johnson famously observed, about a dog walking on its hind legs: ‘It is not done well; but you are surprised to find it done at all’. This could turn out to be a good way of thinking about de se–de nunc analogies in particular and about the category of perspectival thought in general.
5. Conclusion
The de nunc is usually discussed only by proxy: someone argues for a thesis about the de se and banks on some analogy carrying over to the de nunc. My strategy in this paper has instead been to make de nunc thought my direct focus and to foreground the fact that de nunc thought is a kind of thought about time. I’ve argued that agents need to select a time for their action, but do not make a parallel selection concerning the agent of their action. This difference is a consequence of the metaphysical fact that agents span times but do not span agents. And it in turn has an impact on the kind of explanatory work there is for the de nunc to do in agency, and the kind of work there is for the de se to do. It means that action depends on the de nunc even if not on the de se. This argument about the distinctiveness of the de nunc is standalone. But I’ve made the case that we can put it to broader service, in undermining the idea of perspectival thought as a cognitive natural kind, and the associated practice of using discussion of the de se as a proxy way of understanding other species of perspectival thought, and replacing that with the picture of perspectival thoughts as islands. We can’t understand perspectival thoughts in abstraction from the metaphysical natures of the things on which we have perspectives, so we should expect the former to be as diverse as the latter.
Notes
- Classic discussions of such cases include Castañeda (1967), Perry (1979), and Lewis (1979). For a recent defence, in this journal, see Shaw (2019.) ⮭
- Cappelen and Dever (2013) and Magidor (2015) both develop this suggestion. ⮭
- See Peacocke (2019: chapter 1). ⮭
- Thanks to an anonymous reviewer for the suggestion that I do this. ⮭
- A crude but revealing statistic is that googling ‘de se + philosophy’ brings up 278 million results whereas googling ‘de nunc + philosophy’ brings up 1 million. ⮭
- A revealing thing to note here is that the practice of using ‘de se’ as a synonym for ‘indexical’ or ‘perspectival’, which I mentioned in discussing Magidor, comes from Lewis (1979). Obviously, this terminological decision risks promoting a ‘seen one kind of perspectival thought, seen them all’ approach. ⮭
- See Magidor (2015:259). ⮭
- For Cappelen and Dever’s similar case, see (2013: 43). Their case involves Herman saving someone else, Nora, while thinking exclusively about her, and not himself and his body’s movements. To keep things simple, I mostly focus on Magidor’s case. ⮭
- Not everyone agrees with this. E.g., Recanati (2007) and Perry (1986) say that one is ‘implicitly’ and the other is ‘explicitly’ de se. ⮭
- See Morgan (2018) for defence of the face-value view and Babb (2016) for criticism. ⮭
- No doubt not everyone will be convinced that this recipe yields genuine counterexamples to Everyde se. A main aim of this paper is to correct the literature’s de se fixation by focusing on the de nunc, so I’m not going to consider every possible move that might be made in this debate about a claim about the de se. I note though that there’s a risk of dismissing the recipe overhastily if two nearby questions are allowed to blur. One question is whether there are possible action explanations in which a de se intention does not figure. The other is whether mature human agents often act without having de se intentions. There are two relevant gaps, which explain why the answer could be ‘yes’ to the first (and so Everyde se false), even if the answer is ‘no’ to the second. First, it seems extremely easy for mature human agents to gain de se knowledge of what intentions they have. So it’s difficult to think of a case where a mature human agent, S, clearly has a ‘p’ intention but doesn’t know ‘I intend that p’. Additionally, if S knows that they have a ‘p’ intention, they will be aware that the only way this intention is going to get executed is if they do something. So, if S has de se awareness of the non de se ‘p’ intention, that makes it natural for them to form, and to be attributed, the de se intention, ‘I bring it about that p’. Together, these two facts mean that it’s difficult to think of a case where S clearly has a ‘p’ intention but clearly doesn’t also have an ‘I bring it about that p’ intention. But that’s compatible with the intentions being separable in principle—and they would in fact come apart for thinkers who have less de se self-knowledge than mature humans do (or for those mature humans who are especially good at repressing their intentions). Second, even if it were the case that anyone who ever acts has a de se intention, it doesn’t follow that, whenever there is action, a de se intention is contributing something to the explanation of that action. ‘Present for’ does not imply ‘explanatory of’. The de se belief ‘I am I’ is present for every action I perform, but doesn’t seem explanatory of any of them (other than, perhaps, the verbal action of uttering the sentence: ‘I am I’). An argument that the recipe for de se free action does not work should highlight some explanatory question that needs to be answered in any case of action, and that a non de se intention could not answer. My discussion in section §3, especially §3.2.1, supports the view that there isn’t any explanatory question that only a de se intention can answer. It thus provides a new argument that the recipe works, although, as noted already, defending the recipe isn’t a central aim of the paper. ⮭
- Thanks to an anonymous referee for raising this challenge. ⮭
- This, of course, might be regarded as an interesting claim in its own right. There may indeed be several claims about the de se that are more restricted than Everyde se and still interesting. E.g., a different claim, that results from a restriction to a different subtype of action than the one mentioned in the text, is Stokke’s (2022) claim that navigation always involves the de se. One might also propose a claim that is restricted in the type of agent it applies to. E.g., García-Carpintero (2017) argues that, in beings that are like us, action always involves the de se. Engaging with any of these interesting claims would take me too far from the current topic. For my purposes it’s enough that, even if one or several of them are true, Everyde se itself is not true. By contrast, I will argue, Everyde nunc is unrestrictedly true. That is an important de se de nunc contrast, and I think it has a metaphysical explanation. ⮭
Acknowledgements
Thanks for help with this paper to audiences at Nantes, Barcelona, York, Umeå, Oxford and to the Mindwork Group.
References
Anscombe, Gertrude E. M. (1957). Intention. Harvard University Press.
Babb, Matthew (2016). The Essential Indexicality of Intentional Action. Philosophical Quarterly, 66(264), 439–57.
Bermúdez, José L. (2017). Yes, Indexicals Really Are Essential. Analysis, 77(4), 690–94.
Cappelen, Herman and Joshua Dever (2013). The Inessential Indexical: On the Philosophical Insignificance of Perspective and the First Person. Oxford University Press.
Castañeda, Héctor-Neri (1967). Omniscience and Indexical Reference. Journal of Philosophy, 64(7), 2013–210
García-Carpintero, Manuel (2017). The Philosophical Significance of the De Se. Inquiry: An Interdisciplinary Journal of Philosophy, 60(3), 253–76.
Lewis, David (1979). Attitudes De Dicto and De Se. The Philosophical Review, 88, 513–43.
Ludlow, Peter (2019). Interspectival Content. Oxford University Press.
Magidor, Ofra (2015). The Myth of the De Se. Philosophical Perspectives, 29(1), 249–83.
McGinn, Colin (1983). The Subjective View: Secondary Qualities and Indexical Thought. Clarendon Press.
Millikan, Ruth Garrett (1990). The Myth of the Essential Indexical. Noûs 24, 723–34.
Morgan, Daniel (2018). Impersonal Intentions. Philosophical Quartlerly, 68(271), 376–85.
Morgan, Daniel (2019). Temporal Indexicals Are Essential. Analysis, 79(3), 452–61.
Owens, David (2011). Deliberation and the First Person. In Anthony Hatzimoysis (Ed.), Self-Knowledge (261–77). Oxford University Press.
Peacocke, Christopher (2019). The Primacy of Metaphysics. Oxford University Press.
Perry, John (1979). The Problem of the Essential Indexical. Nous, 13(1), 3–21. (Reprinted in The Problem of the Essential Indexical and Other Essays, 3–52, by John Perry, 1993, Oxford University Press)
Perry, John and Simon Blackburn (1986). Thought without Representation. Proceedings of the Aristotelian Society, Supplementary Volumes, 60, 137–51.
Recanati, F. (2007). Perspectival Thought: A Plea for Moderate Relativism. Oxford University Press.
Shaw, James (2019). Frege’s Puzzle and De Se Exceptionalism. Ergo: An Open Access Journal of Philosophy, 6, 1057–86.
Stokke, Andreas (2022). Navigation and Indexical Thought. Erkenntnis, 87(4), 1659–81.
Velleman, J. David (1999). The Voice of Conscience. Proceedings of the Aristotelian Society, 99(1), 57–76. (Reprinted in Self to Self, 110–28, by J. David Velleman, 2006, Cambridge University Press)