What Makes Requests Normative? The Epistemic Account Defended

  • Daniel Weltman orcid logo (Ashoka University)


This paper defends the epistemic account of the normativity of requests. The epistemic account says that a request does not create any reasons and thus does not have any special normative power. Rather, a request gives reasons by revealing information which is normatively relevant. I argue that compared to competing accounts of request normativity, especially those of David Enoch and James H.P. Lewis, the epistemic account gives better answers to cases of insincere requests, is simpler, and does a better job incorporating the importance of relationships. I also defend the epistemic account from three objections: that requests are important because they constitute consent, that requests and wishes must be normatively different, and that some requests do not provide any new information.

How to Cite:

Weltman, D., (2023) “What Makes Requests Normative? The Epistemic Account Defended”, Ergo an Open Access Journal of Philosophy 9: 64. doi:



Published on
18 Jul 2023
Peer Reviewed

I request that you hand me the salt, so you hand me the salt. Handing me the salt is rational for you to do, and it is rational in light of the relevant reasons. My request is tied to the relevant reasons somehow. How? This is the question of the normativity of requests. Why does my requesting that you pass the salt ever give you any sort of reason to pass the salt? One shared view amongst many who have addressed this topic is that there is a wrong answer, which I will call the epistemic account of requests. The epistemic account holds that a request gives reasons merely by providing information. In this article I defend the epistemic account of requests.

1. The Epistemic Account of the Normativity of Requests

The epistemic account says that my request that you X gives you a reason to X only by revealing information to you. It does not alter your objective normative domain, although it does alter your subjective normative domain.1 Your objective normative domain contains the objective normative reasons that are relevant to your potential actions. Objective normative reasons are reasons that apply to you regardless of your knowledge of them, insofar as such reasons exist. So, perhaps you have an objective normative reason not to drink a glass of petrol, even though you are unaware of this reason’s relevance, because you think the petrol is gin.2 Your subjective normative domain contains the subjective normative reasons that are relevant to your potential actions. Subjective normative reasons are the objective normative reasons that you know about, the normative reasons that you mistakenly take to be objective normative reasons, and the normative reasons that exist only insofar as you know about them.3 So for instance you could have a subjective normative reason to drink any nearby gin and even a subjective normative reason to drink the petrol, because you justifiably believe it is gin.4 From here on out, I will talk just of objective and subjective reasons, but I will be discussing normative reasons, which are the reasons which describe what it is rational for us to do, rather than motivating reasons (the reasons on which we act) or any other sorts of reasons (Whiting 2014).

According to the epistemic account of the normativity of requests (henceforth the epistemic account of requests), a request alters your reasons only by revealing information to you, information which alters your subjective normative domain, typically by adding a new reason to that domain. So, for instance, when I request that you pass the salt, this gives you a reason to pass the salt because it informs you that I want the salt. This might make it the case that your objective reason to give me the salt I want also becomes a subjective reason to give me the salt I want (if this objective reason in fact exists). The objective reason came into existence before I requested that you pass the salt: as soon as I wanted you to pass the salt, it existed. Someone who thinks that no objective reasons exist would instead describe this case as one in which a subjective reason comes into existence when you learn that I want the salt. This reason is not based on the fact that I requested the salt, but on the fact that you now know I want the salt. You happened to learn about my desire from my request, rather than via some other means, but the reason obtains in virtue of your knowledge of my desire, not in virtue of my request. If you had found out about my desire some other way, you would still have acquired this new subjective reason.

The epistemic account of reasons thus does a good job of accounting for the fact that we can use locutions like “I would like some salt” or “you are standing on my foot” or “I’d love to hear how you attract butterflies to your garden” to issue requests, even though a literal reading of these phrases would suggest that they are merely reports about one’s preferences or the state of the world, rather than requests.5

Almost nobody likes the epistemic account. “We should reject the view that a request does no more than draw attention to a reason to act which the requestee already has,” says Geoffrey Cupit, because “that would leave unclear what the point of requesting is, as against merely stating our wishes” (Cupit 1994: 449). “It would be wrong to regard requests as mere communication of information” about the requestor’s desires, says Joseph Raz (Raz 1988: 36). Peter Schaber claims that “making a request is not simply informing another person that one has a desire that the other person does x” (Schaber 2021: 433). Mark Lance and Quill Kukla (writing as Rebecca Kukla) even argue that “acting as though the request was just evidence of the speaker’s psychological desires” is a way of defying a request (Lance & Kukla 2013: 470). Ezequiel Monti claims that the epistemic view, among other options, is “to be adopted only if no other theory of robust reason-giving is forthcoming” (Monti 2021: fn. 3; see also Gläser 2019: 30). The only defenders of the epistemic account other than myself are N. G. Laskowski and Kenneth Silver, who briefly suggest that the account, which they label the Epistemic-Evidential view, might be defensible because it is supported by some intuitions and because it can be defended from one chief objection (Laskowski & Silver 2021).6

Everyone is happy to admit that “requests certainly have an epistemic dimension to them,” as David Enoch puts it (Enoch 2011: 4). In light of this, everyone can also be happy to admit that requests, insofar as they are normative, sometimes instantiate what Enoch calls “epistemic reason-giving,” which is what the epistemic account says that requests do (Enoch 2011: 4). What people are not happy to accept is that this exhausts the normative role played by requests; hence the existence of other accounts.

Before discussing some of these other accounts, it will help to distinguish two questions: the question of what makes a request normative (when it is normative), and the question of what makes a request a request, as opposed to some other kind of speech act or as opposed to not being a speech act at all. A full accounting of everything relating to requests and reasons requires answering both. However, our focus is on the first question, and so we will largely move past the second.

1.1. Two Questions7

The first question, about what makes a request normative, is a narrower question, and it admits of multiple overlapping answers. As noted above, everyone accepts the epistemic account as a partial answer to this question. Everyone agrees that a request can reveal information, the revealing of which is normatively relevant. This is not a unique feature of requests: practically anything can reveal normatively important information. The epistemic account has been supplemented by many other accounts of ways in which requests can be normative.8

The second question is more fundamental than the normativity question. It is about what makes something a request, as opposed to, say, a command, an order, a question, a warning, a wish, or any other kind of speech act, or no speech act at all. There are varying accounts of this, too, and the epistemic account is largely ecumenical with respect to the options on offer. Some of the proposals are normatively neutral, and thus it is easy to see how the epistemic account can be ecumenical with respect to these proposals. Searle, for instance, defines requests as illocutionary acts regarding a future act of the hearer which they are able to do, which the speaker believes they are able to do, which they will not obviously do of their own accord in the normal course of events, which the speaker wants the hearer to do, and so on (Searle 1969: 66). We can ask questions like whether one speech act (say, a question like “could you pass the salt?”) can also be another speech act (a request that you pass the salt) (see, e.g., Bertolet 2017). But the normativity of requests is a separate issue.

Other accounts of requests are potentially not compatible with the epistemic account. One potential disagreement between Enoch and the epistemic account turns on Enoch’s claim that some things that sound like requests are not “genuine” requests, “precisely because the reason-giving involved is not robust” but is rather epistemic (Enoch 2011: 15). Similarly, James H. P. Lewis claims that epistemic reasons are “prior,” such that epistemic reason-giving doesn’t create new reasons (Lewis 2018: 3).9 He also says that although we might call something a request when it doesn’t attempt to create reasons, he reserves the term ‘request’ only for “the real deal, so to speak,” such that his view is based on intuitions about “only this central subset of what the term ‘requests’ often includes” (Lewis 2018: 4). Thus Enoch and Lewis both use the term ‘request’ in a way exclusive of the epistemic account, although Lewis explicitly sees this as a stipulation, while Enoch does not say this is the case (Lewis 2018: 4). Whether we should stipulate the definition of ‘reason’ in the way Lewis suggests of course depends on whether, as he claims, the epistemic account is false (see Lewis 2018: 5–6).

If this is merely a question of stipulation that turns on the correct theory of request reason-giving, as Lewis claims, then the epistemic account does not need to worry about competing views of what requests are. These competing views are simply entailments of competing views of the reason-giving power of requests. But if Enoch is right to say that to count as a request at all is to give reasons in a non-epistemic fashion, and if this is not merely a stipulation but something supported by independent considerations, then the epistemic account seems to face an objection. It attempts to explain the reason-giving force of requests by explaining away the possibility of precisely what it is that makes something a request: its special reason-giving force.

The epistemic account has three replies to this sort of worry. First, it is at best somewhat overhasty and at worst question begging to claim that something counts as a request only when it robustly gives a reason in the way Enoch describes. There are many competing accounts of request normativity: even if they are all right in thinking that requests have more robust normative powers than those described by the epistemic account, it goes too far to say that it is this precise normative power that makes a speech act a request at all. Despite there being considerable uncertainty and disagreement in the realm of what, exactly, constitutes the normative force of requests, there is little disagreement about which speech acts are requests. Surely we can solve this latter question without making it hinge on the former. So Enoch should not claim that to be a request at all is (in part) to fit his preferred theory of request normativity.

Second, there is a position nearby Enoch’s which works just as well for what he wishes to argue and which is much more ecumenical than Enoch’s claim. Enoch can simply say that requests which lack the normative force he describes are actually requests, but deficient in some way. This is perhaps what he in fact means: he claims a request which fails to robustly give a reason is not a “genuine request” (Enoch 2011: 15). Maybe we can accept that ersatz requests are still requests. Perhaps all requests are genuine requests and “genuine request” is just a circuitous way of saying “request.” If not, however, then this is once again just a question about how we label things: some requests are genuine, and others aren’t, but they are all requests. Thus perhaps Enoch’s view, like Lewis’s, is ultimately one about how we stipulate definitions, rather than about something deeper. (Or, to the extent it’s about something deeper, it is just about whether the epistemic account is correct, in which case to accept Enoch’s view at the outset would be to beg the question against the epistemic account.) As a proponent of the epistemic account, I think it is more charitable to read Enoch as saying that these are still requests, albeit not genuine ones, because if the epistemic account is right, then this is the reasonable approach. Enoch’s view of request normativity is compatible with saying that non-genuine requests are not actually requests at all, though, so he may not agree that this is a charitable reading.

Third, assume for the moment that a speech act is not a request unless it does in fact have the robust reason-giving normative power Enoch describes. Nevertheless, there are speech acts the utterances of which are indistinguishable from requests, and which can be used to accomplish many of the things requests are used to accomplish, and which socially serve the role requests serve, and which we colloquially refer to as requests. One might reasonably wonder whether these speech acts have normative force, and if so, what normative force they have, and how their normative force differs (if at all) from the normative force of requests. Once we put the question this way, it is clear how Enoch’s view does in fact beg the question against the epistemic account: Enoch’s view requires that a speech act have the special force he posits requests have in order to count as a request. But if at the outset we accept that there are a variety of speech acts, some of which may or may not count as requests as Enoch defines them but which seem to be requests in every other sense, and if we want to see whether we can explain how these speech acts work in the same way we explain how the speech acts treated as requests by Enoch work, then we will not begin by accepting Enoch’s definition of requests, nor any other definition which rules out the epistemic account. At least, we will not accept such a definition without further reason, and I am not aware of any arguments to think that requests must have some sort of particular normative force in order to count as requests.10

The above strategies work not just against Enoch and Lewis but against any definition of requests that is exclusive of the epistemic account.11 If we respond like this to the question of what counts as a request, the substantive disagreement to resolve is over the way in which requests are normative (when they are normative). Alternative accounts of requests are either mere stipulative disagreements, or they beg the question against the epistemic account. What we should do instead is evaluate the epistemic account’s plausibility before endorsing a view of requests that requires the epistemic account to be false. Eventually we will need complete theories not just of the normativity of requests but also of what constitutes a request, but the goal here is to focus mainly on the first question, since our answer to the first question helps explain why we shouldn’t beg the question against the epistemic account when answering the second question. I will thus now defend the epistemic account against competing accounts. I will focus on Lewis’s account because it is one of the most recent and because it shares commonalities with many earlier accounts.12 For arguments against other accounts see Weltman (2022a; 2022b; forthcoming).

2. Lewis’s Account

Lewis take’s Enoch’s account and adds a condition to it, so to summarize Lewis’s account I will first summarize some relevant aspects of Enoch’s. Enoch sees request reason-giving as one kind of “robust reason-giving,” which is a type of reason-giving beyond mere epistemic reason-giving (Enoch 2011: 2). For X to robustly give Y a reason via requesting, among other things, X must intend to give Y a reason, communicate this intention to Y, intend Y to recognize this intention, and intend Y’s reason to depend on Y’s recognition of X’s intention (Enoch 2011: 15). Only if X accomplishes this can X’s request robustly give Y a reason. When X robustly gives a reason, this happens via rendering true the antecedent of a conditional along the lines of “if X asks Y to Z, Y has a reason to Z.” These conditionals exist only in certain circumstances: for instance, “if Val asks Adrien to commit murder, Adrien has a reason to commit murder” is not a conditional the antecedent of which can be rendered true via request, because we cannot robustly give people reasons to murder. A full account of the reasons we can robustly give depends on a more extensive account of reasons generally. Because the relevant cases here are the ones in which requests can at least in principle give reasons, we will restrict our inquiry to cases where there is a conditional the antecedent of which can be rendered true via request, like “if Val asks Adrien to pass the salt, Adrien has a reason to pass the salt.” In other words, the most relevant parts of Enoch’s account are the one described above, according to which for X to robustly give Y a reason via requesting, among other things, X must intend to give Y a reason, communicate this intention to Y, intend Y to recognize this intention, and intend Y’s reason to depend on Y’s recognition of X’s intention. The other details we can ignore.

Lewis adds another condition and claims that only if this condition is fulfilled can requests create reasons in the way Enoch says they do. The condition is that the requestee places discretionary value in the requestor. When this condition is satisfied, the account is identical to Enoch’s account. When it is not, the request cannot create a reason (except perhaps incidentally via epistemic reason-giving).

Lewis adds this condition because he thinks that Enoch’s account fails to explain why acceding to requests is discretionary. According to Enoch (says Lewis), the conditional which is made true by the request (e.g., “if X asks you to Z, you have a reason to Z”) leaves no room for Y reasonably refusing to Z once X asks them to, and thus the account fails to capture the discretionary normativity of requests (Lewis 2018: 11). Lewis’s account, meanwhile, can easily explain this discretionary feature, because of course we place discretionary value in the requestor only at our discretion.

There are good reasons to reject Lewis’s theory (and other competing theories) independently of what we say about the epistemic account. There are also things the epistemic account can say in reply to the objections these theories offer, and ways in which the epistemic account can capture what is central to these theories.13 Here though I will just respond to one objection Lewis’s theory can offer, an objection it shares with at least Cupit’s and Enoch’s theories, which is that the epistemic account is committed to the normativity of certain insincere requests, whereas these other accounts are not.

3. The Problem of Insincere Requests

Lewis’s theory (like Enoch’s) implies that some insincere requests, because of their insincerity, are not normative. By “insincere requests” I mean requests made not for the sake of getting the requestee to undertake the requested action but for some other reason.14 So, for instance, if you offer me $5 to request of Val that Val hand me a book, and I accept your offer and make the request only in order to get the $5, this is an insincere request. Insincere requests include requests that the requestor doesn’t care about and also requests that the requestor actively does not want fulfilled, like Lance and Kukla’s example of my request that you “give me a fifty-page draft of a dissertation chapter for me to comment on over the weekend” (Lance & Kukla 2013: 462).

Among these insincere requests, some (according to Enoch and thus Lewis) are not normative, because the requestors lack the intentions that (according to Enoch and thus Lewis) are required for a request to be normative. Recall Enoch’s claim that “it seems like a necessary condition for something to qualify as a request that the person making the request intend to thereby give a reason to the addressee”—no intention, no normativity (Enoch 2011: 14). Moreover, for requests to be normative, further intentions must be present—X must intend Y to recognize X’s intention to give Y a reason, and intend Y’s reason to depend on Y’s recognition of X’s intention. If any of these intentions are lacking, the request is not normative. One way to lack any or all of these intentions is to be insincere: if I request that Val hand me the book just for the money, I can do this without any of the intentions Enoch and Lewis discuss. I can be entirely indifferent as to whether (for instance) Val’s reason to hand me the book depends on Val’s recognition of my intention to give Val a reason. I might not even intend to give Val a reason in the first place: I may intend simply to get $5.

One might think that Lewis’s account is more compelling because it says these insincere requests are not normative. That may seem like the correct answer. Instead, I will argue that these insincere requests can give us reasons to comply in the same way sincere requests do, and thus it is in fact a weakness of an account if it claims that these insincere requests can’t generate reasons.1516

3.1. Requestee’s Knowledge or Lack Thereof about the Insincerity of the Request

First, consider a case in which the requestee knows that the request is insincere. Val, for instance, is standing right next to us when you make your offer. In this instance, the epistemic account can happily agree with the other accounts that the request is not normative, because it doesn’t give Val a reason to accede to the request. This is because Val knows that I don’t care whether Val accedes, and Val knows that I know that Val knows this, and so on. The request therefore does not give Val the impression that I have some desire that would be fulfilled if Val accedes, or anything like that. So, there are no considerations in favor of doing as I request and handing me the book. It would not be impolite to fail to hand me the book, just as it is not impolite to refuse to accept an apology if the apology is clearly in jest. For example, if you jokingly apologize to me for the weather because it is cold and drab when I visit your city, it is not impolite of me if I fail to forgive you. Val knows that handing me the book would not make me better off in any way; that I don’t actually want the book; and so on. So, my request does not offer Val any information about my desires or anything else which would alter Val’s subjective normative domain.

Next, consider a case in which the requestee does not know that the request is insincere. In this case, Val walks up after you have made your offer, and then I request that Val hand me the book. In this case, it makes sense to say that my request is normative: it alters Val’s subjective normative domain, and Val acts reasonably by acceding to my request and handing me the book. On the basis of my request, Val justifiably forms a new belief, the belief that I desire the book near Val. This, in conjunction with Val’s standing reasons to give me what I want so long as it comes at little to no cost to Val, and so long as there’s nothing immoral about granting my wishes, and so on, combines to give Val a reason to hand me the book. If I reply, upon being handed the book, “listen, Val, there was no reason for you to do that,” Val could fairly reply that there was in fact a reason: I asked for the book! There is a sense in which Val is perhaps mistaken: Val’s objective normative domain was not altered by my request. So, if Val says “you asked me for the book, and so objectively speaking I had a reason to pass it to you,” Val would be wrong.17 But we act on the basis of our subjective normative domain18 and there is no question that Val is perfectly justified in altering that on the basis of my request. Val doesn’t know my request is insincere. It would be wrong for me to think Val is unreasonable or irrational for acceding to my request. So, we shouldn’t balk at the claim that insincere requests can be normative in virtue of altering the requestee’s subjective normative domain. Since this is the only sort of alteration the epistemic theory compasses, it is fine that the story only works for the subjective domain and not also the objective domain. This allows us to explain the normativity of these insincere requests the same way we explain the normativity of all requests, rather than either implausibly denying there is any normativity to these requests, or coming up with an ad hoc explanation of their normativity.

An opponent of the epistemic view might reply that insincere requests are not really normative: they are just apparently normative, and people mistakenly take them to be normative. But the epistemic account can account for the idea that people are mistaken, because an insincere request is only normative (or apparently normative) in virtue of the information it appears to convey. That information is misleading in the case of an insincere request. So for instance when I insincerely request the salt, you mistakenly believe I want the salt. So, the epistemic account can capture the sense in which insincere requests leave people mistaken about the normative fact of the matter: they are mistaken about the information that the request putatively revealed. The request is not “really” normative if by “really” normative we mean that the request tells you about your objective normative domain, or that the request is normative in the way requests are usually normative, which is by revealing true information (rather than false information). It is apparently normative, though, and apparent normativity is the only normativity we can act on.19

3.2. Insincere Requests and Insincere Promises

A second reason for thinking that insincere requests provide reasons is that insincere promises can also provide reasons. If I insincerely promise you something, I still undertake an obligation (Searle 1969: 62; Owens 2008). I cannot get out of a promise by proving to you that I didn’t mean it. If anything, that only makes me look worse. It certainly does not wipe out my obligation. It’s reasonable to think promises and requests are similar in this way, since they are similar in many other ways (Cupit 1994). So, if we can accept normative insincere promises, we should accept normative insincere requests.20

This argument is not entirely decisive, because it is open to the opponent of the epistemic theory to offer different accounts of sincere and insincere promises in order to explain how, although both give reasons, they give reasons differently, and then to use these two accounts to explain why only the sincere requests give reasons. Reasons of parsimony tell against this, as do cases where someone is conflicted about their promise and thus their promise is neither entirely sincere nor entirely insincere. In such cases we would not want to excuse them from having reason to fulfill the promise. If we need an answer for in-between cases, such an answer ought to suffice for both clear cases: those of sincere promising and those of insincere promising. This is a larger issue than can be fully resolved here. But, insofar as we have any reasons to treat promising and requests as similar in this way, and any reasons to treat sincere and insincere promises as normatively on a par, this tells against refusing to treat some insincere requests as normative.21

3.3. Enoch’s Example

Enoch describes a case which, he claims, demonstrates that some insincere requests do not provide reasons. Enoch utters the words “would you mind having a look at my draft?” to you while the chair of your department walks by. You have a reason to keep your chair happy, and your chair thinks that you should do what Enoch asks you to do, so now you have a reason to read Enoch’s draft. Enoch only uttered these words because he knew that if the chair overheard him, this would give you a reason to read his draft. “It seems to me,” says Enoch, “that this case—in which I intend to give you a reason, and succeed in doing so by uttering words that sound like the making of a request—is not a genuine request at all” (Enoch 2011: 15). This is because Enoch did not intend for your reason to read his paper to depend on your acknowledging his intention that you read his paper. All it depends on is your desire to keep your chair happy.22

Is it true, though, that Enoch has not made a normative request, and that your reason to read the paper thus does not depend on Enoch’s request? It’s true that it’s not merely Enoch’s request that provides your reason for reading the paper. Your desire to keep your chair happy plays a large role. But Enoch’s request also plays a crucial role in giving you a reason to read his paper. Enoch could have given you this reason some other way, say by offering to pay you or threatening you. But instead he did it by requesting that you do so while your chair walked by. If you tell someone why you ended up reading Enoch’s paper, you could felicitously say that he requested that you read it. The whole story is more complicated, and so, for instance, if a good friend asks you why, you’ll say not just that Enoch asked you to but that he did so within earshot of your chair. But, if you are being diplomatic, you can truthfully cite Enoch’s request as the genesis of your reason to read your paper, leaving out the stuff about the chair, and at most this constitutes an omission of certain other relevant facts about Enoch’s Machiavellian machinations, rather than an outright lie. And anyways, the whole story is always more complicated, even with normal requests. If you request that I help you lift something, and this gives me a reason to help you lift something, that’s not the whole story. It must also be the case that there’s nothing immoral about lifting the thing; I don’t have other duties to attend to; the object is not too heavy for me to safely assist with the lifting; and so on. There are always many other things beyond the bare request which explain why the request is or isn’t normative, no matter which account of request normativity we adopt.

The case Enoch gives here is rather convoluted and abnormal. It should not surprise us that most requests don’t give reasons like this, and that we are thus inclined to see requests as giving reasons in more traditional ways that are tied to the concerns which are central to Enoch and Lewis: interpersonal relationships. But the fact that requests typically work in light of a certain fact should not mislead us into thinking that requests only work in light of this fact. Our intuitions about requests and their normativity depend on how requests typically function, epistemically speaking. Requests, like other sorts of speech acts, “depend upon and make use of existing normative contexts and roles for their production” (Lance & Kukla 2013: 457). Conventions around what sorts of requests are acceptable to make, who it is acceptable to make these requests to, how to acceptably phrase these requests, and all sorts of other things all go into determining the epistemic content of even a very simple request like “pass the salt,” let alone much more normatively laden requests, like my request that you help me move a body.23 Thus although the epistemic account might on its face seem to be not just simple but in fact overly simple, and thus unable to capture the important aspect of relationships present in Enoch’s account, nothing could be further from the truth.

For example, I learn very different things about you when you request that I pick your child up from school than I learn about my spouse when they ask of me the same thing. My spouse’s request typically just communicates that they are busy or tired. Your request communicates (for instance) that something has come up, the importance of which requires that you delegate a task normally reserved for family members and family friends. This may make my reason to pick up your child very strong: presumably you are dealing with some sort of emergency, or some extremely crucial deadline, or something like this. Or this may reveal that our relationship is closer than I might have previously thought, and thus I have reasons to be helping you out that I was not aware of. Or it may communicate something else. And it never just communicates one thing: lots of information goes along with each request we make, especially once we take into consideration the context and the way the request is made. This is less true of Enoch’s strange case, which mostly just communicates the fact that he is a master manipulator (and that he thinks you must be trapped into reading his paper, and thus that your relationship with him is not strong enough to give you a reason to read the paper already, and that the paper is likely crummy, and so on).24

The point here is that the epistemic account is quite nuanced, because the social conventions around requests are quite nuanced, and requests therefore come with lots of information attached. This is how the epistemic case responds to Enoch’s simpler chairless argument that “if you come to believe my request was not in fact sincere . . . you will come to the conclusion that you don’t have any reason to read my draft” (Enoch 2011: 15; see also Schaber 2021: 434). When this is true, it is in virtue of realizing that my request did not reveal the information you thought it revealed. You thought it revealed that I wanted you to read my draft, but in fact it did not, because my request was insincere.

So, if our intuition in Enoch’s chair case is that the request is not normative, probably this is just picking up on the fact that the request doesn’t give us the sort of information we usually get from a request. Normally when Enoch requests that you read his paper, this tells you about how important he thinks it is to get feedback, how much he values your feedback, and so on. These are all sources of reasons to read it. Since these are absent in the chair case, it is tempting to say the request gives no reason at all, which is what Enoch wants to claim. But that would be false. The request still gives you a reason to read the paper, because it informs you that if you don’t read the paper, your chair will be mad. It’s true the request does not give you a reason the way requests normally give you reasons. It gives you a reason via a strange route. But it still gives you a reason.

Enoch might reply that this would allow crazy causal sequences involving requests to count as request reason-giving (see also Monti 2021: 3740–44). So for instance Enoch requests that you read his draft and the chair overhears the request. The chair communicates this information to their spouse. Their spouse tells it to a friend. The friend knows Enoch’s draft ostensibly has a funny joke and the friend asks you whether this is true. You have yet to read the draft, but now you are intrigued because you like jokes, so you have a reason to read the draft. Did Enoch give you a reason to read the draft by asking you to? The epistemic account can give one of two responses.

First, it can say no. Your reason to read the draft is that you think it has a funny joke, not that Enoch asked you to read it. Your reason to think it has a funny joke is not based on information revealed when Enoch asked you to read it, but on information revealed when your friend mentioned this. So, there is no problem here. The relevant epistemic information, which is that the draft might have a funny joke, was not directly revealed by the request, so the request did not give you a reason. This is in contrast to the chair case, in which the relevant epistemic information (your chair is going to be mad if you don’t read the draft) is directly revealed by the request.25 So, the epistemic account could hold that a request must directly provide the information in order to be normative. This is a plausible reply because we do not need to commit to the view that a request gives us reasons merely because a request occurs somewhere in the vicinity of the reason acquisition. Indeed, one of the epistemic account’s key features is that requests are not normatively special, and so there is no need for us to locate the normativity of any particular case of reason-giving in a request, as opposed to something else that may seem more apt in the particular circumstance. When presented with an oddly circuitous causal chain resulting in reason-giving, if our intuition tells us that the request part of the chain was not involved in the reason-giving, the epistemic account will typically have an easy way to accept this. The causal chain is long and odd and there will be something else somewhere in the chain that we can point to in order to explain the reason-giving.

Second, the epistemic account might instead bite the bullet and say yes. Enoch’s request gave you a reason to read his draft, albeit via a circuitous route. This is an acceptable bullet to bite (although we need not bite it if we accept the first response). These strange cases are outliers, and our inclination to reject them as examples of request reason-giving is plausibly as much a result of their strangeness as it is any commitment to any sort of principle. We are inclined to believe that requests only give reasons the normal way (e.g., via revealing information about the requestor’s priorities, desires, and relationship with you, if the epistemic account is correct) and when requests give reasons some other way this unusualness suggests to us that we must not be dealing with the same thing. But, we are. A request has given us a reason.26

4. The Objection from Paternalism and Autonomy

There is another sort of case that the epistemic account seems to have trouble dealing with, and that is a case where the mere fact that I have made a request seems to be normatively significant in ways that override any epistemic information. These are cases where it would violate someone’s autonomy to act on relevant epistemic information absent a request from that person that you act on that information. One example is paternalism: you know that I want to eat healthily, and that I am planning on asking you to interfere with me if I try to order something unhealthy at a restaurant when we are dining together. So, without asking me, and without my having asked, you interrupt me when I try to order a bucket of french fries at dinner. This seems objectionably paternalistic. The same goes for other autonomy violations. To take perhaps the most extreme example, we might think it is entirely irrelevant that I know Val wants to have sex with me: unless Val requests that we have sex, it would be not just objectionable but monstrous for me to have sex with Val. Nobody is ever “asking for it” except people who literally ask for it. The epistemic account of requests seems ill-suited to dealing with these cases. Surely one must wait to be asked before interrupting someone’s food order, having sex with them, or otherwise circumventing their will, no matter how much someone might want to be overridden, even in fact if they wish that someone would act in these ways without having been asked. That is, even if you know I’d rather you act paternalistically and stop my french fry order without my having to ask, chances are it is still wrong for you to do this.

These cases point to the importance of consent. There are some things we ought not to do to people unless they consent, and merely knowing that they would consent, or want to consent, or want us to ignore their lack of consent, is not enough to justify bypassing consent.27 Or, to put it in terms amenable to the competing accounts of requests, sometimes it seems like the normatively important thing is that I ask, not that I would ask or will ask or want to ask. I have to actually make the request before it’s even permissible for you to act on it, let alone before you have a good reason to act on it.

How can the epistemic account respond to these cases? The key is consent. We can take requests to constitute evidence of consent because issuing a request can itself constitute consent. Thus if I request that you X, this constitutes evidence that I consent to you Xing. It constitutes evidence that I consent to you Xing because my requesting that you X constitutes consent to you Xing, and things are evidence of themselves.28 Thus in cases where the permissibility of your Xing turns on whether I have consented, my request is relevant in precisely the way the epistemic account says it is relevant: it provides you evidence about the normative situation.

Now, it’s true that my request doesn’t just provide evidence about the normative situation. It also alters the normative situation, because my request constitutes consent, and consent modifies the normative situation. So, haven’t I conceded that the epistemic account of requests is false? In this case, doesn’t the request do more than merely reveal information about reasons? Doesn’t the request create a reason via non-epistemic means?

It’s true that the request creates a reason. But, it does not create this reason in virtue of constituting a request. It creates this reason in virtue of constituting consent (or an expression of consent). The fact that it constitutes a request has nothing to do with it constituting consent.29 This is because instead of making the request, I could have consented in some other way. So for instance when I request you pass the salt, this also constitutes my consent that you pass the salt (and evidence of my consent that you pass the salt). But I could consent that you pass the salt without issuing the request. I could say “I consent to your passing the salt, if you should happen to desire to do so.” So, it is irrelevant whether I consent in the form of issuing a request. When I do, my consent is concurrent with, and constituted by, my request. But my consent does not constitute consent in virtue of being a request. It constitutes consent in virtue of instantiating whatever an expression needs to instantiate in order to constitute consent. Requests can instantiate these properties, but so can other expressions and actions.30

To put it another way, something can be evidence of X and also constitutive of X without it being one of these things in virtue of being the other thing. My dipping my paintbrush into the can of blue paint rather than the can of white paint can be evidence of my deciding to paint the wall blue and also constitutive of my deciding to paint the wall blue, but it is not evidence of my decision because it constitutes my decision. It could be evidence without constituting the decision (maybe I decided yesterday) and it could constitute my decision without being evidence of it (maybe I’ve been dipping the paintbrush for days as a way of expressing my indecision, but in this case it is a decisive act, although someone observing me wouldn’t be able to tell the difference). In this case it just so happens to be both.

The fact that I am making a request does not do any normative work, except incidentally, because my request also constitutes consent (and thus the request is evidence of consent in the same way anything is evidence of its own existence). Monti rightly describes the epistemic account as claiming that requests are only indirectly explanatory: what is directly explanatory is some other normative fact (Monti 2021: 3740). In this case that fact is consent. There’s nothing inherently normative about my request that you override me when I order unhealthy foods or Val’s request that we have sex. Rather, what’s normative is consent to these things (and also the fact that the other party is now aware of one’s consent, and also aware of one’s desires that these things occur, and the other relevant information). This is clear to see if we imagine I first consent and then issue the request. I might say “I’m not asking you to do this for me, because I know you might not have this impulse in the first place, but I consent to your overriding me whenever I try to order unhealthy foods in the future, just in case you ever want to do so.” I might later say “remember that food thing I consented to earlier? I am now requesting that you do this. It would be a big help.” In this case there is nothing wrong with the epistemic account of this request’s normativity. In issuing the request I merely reveal to you that I’d like you to do the food thing. If you had somehow caught wind of this information prior to my request but subsequent to my consent (say, you heard from a mutual friend that I want all my friends to do this), there would be nothing objectionable to your overriding me prior to my making the request.31 I have already consented, so you are good to go, so long as you have good reason to think this is what I want.

So, here is the general template for responding to worries about consent. Take the case in which a request is supposedly normative and separate the consent from the request. Imagine that consent is expressed absent any request, and then later the request is made with the understanding that consent has already been granted (and that consent has not been rescinded). Ask what is doing the main normative work: the consent, or the request. If the answer is the consent, then the epistemic account stands. If the answer is the request, the epistemic account is in trouble. But, my claim is that the answer will always be that consent, rather than the request, is doing the main normative work. The request still does some work: it informs the requestee that the requestor has not merely consented to the thing in question, but that they also actively desire it, or whatever else is revealed via request. And that is important too. But, the original worry was that no amount of epistemic information could ever make it okay to undertake certain actions absent a request. We have defused that worry by showing that this is true because some actions require consent, which can be divorced from the request, even though in many actual cases, the request also constitutes consent.32

5. Further Objections to the Epistemic Account

5.1. Requests Aren’t Wishes

We’ve already seen the notion that requests must be normatively different from wishes. After all that has been said, one might still cling to the intuition that something must be going on when we request that is normatively distinct from making a wish. And it is difficult to see how we can draw this distinction with the epistemic account, since a wish seems like it can offer just as much information as a request. To help partially defuse this intuition, it is important to remember that, as highlighted in Section 1.1, the epistemic account is an account of the normativity of requests, not of what constitutes a request. The epistemic account doesn’t claim that requests are wishes or that wishes are requests. We can hold these two to be separate speech acts and still endorse the epistemic account. The epistemic account only claims that insofar as requests are normative, they are normative like wishes (insofar as wishes are normative). That requests and wishes are normatively on a par hardly means that they are the same in every other way. For example, following Searle, we might say that an essential feature of a request is that it counts as an attempt to get someone to do something, whether sincerely or insincerely (Searle 1969: 66). Both sincere and insincere attempts to get someone to do something entail taking responsibility for attempting to get someone to do something (Searle 1969: 62). So, when I request something, and someone acts on my request, I am to some extent responsible for this, whereas when someone acts on a wish, I am not.33 This is a big difference.34 So there is no worry that if we equate the normativity of requests and wishes, we thereby remove any distinction between requests and wishes.

The main reply to this worry, though, is that the conventions around requests, wishes, and all other speech acts are such that often the sort of information that one conveys with a request differs from the sort of information one conveys with a wish. I cannot replace a wish with an equivalent request or vice versa without changing what is implied. This explains why requests and wishes differ normatively in a way that comports with the epistemic account. So for instance if I request that you X rather than wish that you X, this typically implies that I care more that you X than would have been implied if I had chosen to wish instead. This is not a logical entailment of requesting rather than wishing: I can request that you X no matter how I feel about it, or wish that you X no matter how I feel about it. But one main reason I would choose to request rather than wish in some context is that I want to increase the chances that you will X compared to if I were to wish that you would X (albeit at the cost of my being responsible if you act on my request, and so on). So, I issue a request, the conventional import of which is such that you justifiably believe I want you to X more than you would believe if I had merely wished you would X. Thus for all practical intents and purposes, there is a normative distinction between a request and a wish, because in any given context, a request pragmatically communicates different information than a wish.

Because requesting, wishing, and other practices are constituted by and inseparable from social contexts, it is hard to come up with a case in which the request and a wish are normatively on a par because they provide the same information (at least in American English) without the wish sounding like a request.35 Compare “would you wear something nice tonight?” with “I wish you would wear something nice tonight”—if anything the latter sounds almost like a command or at least a request rather than a wish. Social conventions in contemporary American society (which I use as my example because I am most familiar with it) dictate that if one merely wishes that someone else do something, one typically ought to keep this to oneself. Thus if one expresses a “wish,” it is taken to constitute a request, albeit one expressed via the locution of a wish.

But, we can try to generate a case in which a request and a wish have the same epistemic value so that we can see if they also have the same normative value, which is what the epistemic account predicts. Imagine I give the following speech at a party I am hosting: “Welcome to my party. I am going to do something strange for the sake of vindicating an argument. Tonight, I want you all to assume that every request I make and every wish I make are on a par when it comes to how I feel about the content of my request or my wish, and when it comes to my opinion of the person I’m talking to, and so on. For instance, if I request that you turn on the porch light, this implies that I desire the light to be turned on to some degree, and that I desire you do it, and so forth. If I wish that you turn on the porch light, this implies the exact same things. The same goes for all requests and wishes. This does not mean a request is a wish or vice versa: I’ll be clear about whether I’m requesting or wishing.” Having given this speech, I later request that you hand me a carrot on your left and wish that you would hand me a carrot on your right. In a moment, partygoers are going to eat whichever carrots you haven’t handed me, and you only have time to hand me one, so you must choose. If my request gives you just as much reason to hand me the carrot on your left as my wish gives you to hand me the carrot on your right, you have most reason to hand me the carrot on your right, since you are right-handed and this will be easier and less likely to result in failure (like dropping the carrot or knocking over the vegetable tray). If my request gives you more reason to hand me the carrot on your left, you should do that, because it is hardly onerous to use your left hand rather than your right. What should you do?

You should hand me the carrot on your right. This supports the epistemic account. Now that my speech has eliminated any reason to think my request implies things my wish doesn’t, the request and the wish are normatively on a par. Because they are on a par, the ease of grabbing the right side carrot wins out. I do not have an argument for this beyond the fact that it strikes me as compelling. Hopefully that will suffice. My claim is that, if anyone does not find this answer compelling, it is only because they have not fully internalized the speech I gave. That at least is the answer the epistemic account gives.

5.2. The “No New Information” Objection

One might object that we can imagine requests that give us reasons but which do not give us any new information, which suggests the epistemic account must be wrong (see, e.g., Lewis 2018: 5–6).36 In these cases, the epistemic account must say that these requests do not give any reasons, but such cases are extremely rare to the point of being almost impossible. There is practically always some information conveyed by a request, even if it’s just the information that the requestor’s desire that the requestee carry out the request is large enough to make a request socially acceptable in the present context. Say for instance that it is common knowledge that our cat’s litter box needs to be cleaned, that it is your turn to clean it, and that you ought to have cleaned it yesterday. I then request that you clean the litter box in order to prompt you to do it.37 How can this request have any new information? Doesn’t this example defeat the epistemic account?

In normal circumstances, my request will in fact supply a fair amount of information. It will imply that I’m annoyed with your tardiness, or I’m worried that maybe you’re dealing with some issue and I’d like to give you an opportunity to explain to me what’s going on in case you’d like some support, or that I’m not sure that it is in fact common knowledge that the litter box needs cleaning or that it’s your turn, or any of a dozen other things. Any one of these pieces of information can give you a reason to clean the litter box. Learning that I’m annoyed will reveal that there are higher costs to letting the litter box remain dirty than you previously thought. Learning that I’m worried will reveal that your litter cleaning habits have caused me to form some speculative beliefs about what is going on in your life. Learning that I’m not sure you realize the litter box needs cleaning or that it’s your turn tells you that maybe you’ve put off the cleaning longer than it’s reasonable to do so. Learning any of these things can reveal that you’ve put off the cleaning long enough to cause me to take notice in a way that might damage our relationship going forward. And these are just a few of the many possibilities.

Even if we stipulate all this away, my request will still tell you that I care enough about your tardiness that I think it is acceptable to make this request. Requests are not appropriate in every context. If you are not late in cleaning the litter, and you never have been, and yet I still request that you clean it, I’m a jerk. Assuming (for the sake of the argument) that I’m not a jerk, my request implies that I think your dereliction of duty is grave enough to warrant a comment. It is hard to see how you could know that unless I in fact comment. You could find it out if you are a mind reader, and thus the epistemic account has to say that if you read my mind prior to my request, then the request itself will not be normative.38 For those who lack psychic powers, though, my request will let you know how I feel. My feelings are certainly normatively relevant, and my feelings exist prior to the request. I could’ve expressed the feelings some other way, but I happen to have done it via request. You have more reason to clean the litter box if I feel strongly about it than you have if I don’t. So, the request’s normative value here consists merely in its evidence about the way I feel. Ultimately your reason to clean the litter box comes from my feelings, and my request merely gives you information about my feelings. Had you found out about my feelings before I requested, you would already have had a reason to clean the litter box.

We can go even further and say that requests with no new information can still be normatively relevant in virtue of the evidential role they play. Laskowski and Silver describe how. Briefly, their argument is that even if we grant that everyone involved has knowledge of the relevant reasons, this does not preclude the request from providing evidence about the relevant reasons. You can know X and yet still receive evidence for X (Laskowski & Silver 2021). Thus there is nothing untoward about suggesting that requests provide evidence even when the epistemic situation is one in which nobody learns anything new. The evidence, they suggest, can cause the requestee to attend to the reasons which they already know (Laskowski & Silver 2021). For instance, you may know the litter box needs cleaning, but my request may cause you to attend to this knowledge in a way you were not previously attending: perhaps you weren’t attending to it at all, or you were, but not as much as you are now that I have issued the request. The request doesn’t change what reasons you have, or even what reasons you know you have. It just changes the degree to which you are attending to your reasons to clean the litter box.39

One might object to Laskowski and Silver that requests are normatively irrelevant if their only effect is on our attention to already existing reasons. According to this objection, if there is anything normatively relevant here, it is the existing reasons and our reasons to attend to those existing reasons. The request gets you to attend, and thus to comply with your extant reason to attend, but the request doesn’t give you the reason to attend. If you don’t already have a reason to attend, a request cannot give it to you, we might think. For instance, if the litter box does not require cleaning, or it’s my turn to clean it, and I request that you clean it, then there isn’t even a reason to clean it for you to attend to, let alone a reason for you to attend to the reason to clean it. So, the request has no power here. Only the reasons (including reasons to attend to other reasons) have power.

The epistemic account bites the bullet and says that in cases where requests cause us to attend to something, there is nothing normatively special about the request. The only normative strength is in the reasons that are attended to (which already exist) and the reasons to attend to those reasons (which also exist prior to the request). The request has a normatively valuable result—getting you to attend to something you have reason to attend to—but no normative power itself.40 But, after all, this is just the bullet the epistemic account bites more broadly. In every case, not just attention cases, requests have no special normative force. They are only evidence of things that have normative force. Thus it is no extra cost for the epistemic account to accept that cases like those described by Laskowski and Silver are ones in which the request is merely evidence of something else that is normatively relevant.

6. Conclusion

The epistemic account is a serious challenger to other theories of the normativity of requests. Unlike other theories, which invest requests with a special normative force, the epistemic account locates the normative force of requests in the information that they reveal. This means that each particular request is normative insofar as it communicates information that is normatively relevant. There is no general theory of what information is normatively relevant because this is quite situational. That it is Tuesday is normatively relevant when you’ve promised to meet someone on Tuesday, but not if you haven’t, and so on. This article’s title contains the question “what makes requests normative?” The simple answer is that the information they reveal makes them normative. Or, more accurately, the normative import of the information they reveal makes them normative. Or, more broadly, the normative import of the information they reveal or invite us to attend to or otherwise involve makes requests normative. Or, finally and most accurately: requests aren’t normative. They are a red herring, normatively speaking. What are normative are the reasons which exist either prior to the request, or which are created not in virtue of the request qua request but in virtue of what the request is evidence of.41 Requests are, as Parfit puts it, normative in the “reason-involving” sense: their normativity is derivative, since the reasons they give us come from the information they reveal, not directly from the requests (Parfit 2011: 39).42

The epistemic account, even after all of the arguments I have given, may seem like a rather lackluster theory. Perhaps it still seems wrong on its face.43 Why might we ever be tempted to endorse such a theory?

6.1. Simplicity

First, the epistemic account is simpler in one way. We need not posit a special sort of normativity for requests. If we can account for the normative features of the world without complicating our picture of normativity, this is nice. One might reply that we already posit normative powers in other realms, like consenting and promising, and it is not more complicated to take one of those powers and apply it to requesting. However, attributing a special normative power to requests still counts as an increase in complexity, because the more phenomena we bring under the umbrella of, say, discretionary normativity, the more complicated we make that realm. Even if we accept the possibility of accounting for requests with some existing normative power, every account of requests must also say that requests can be normative in the way the epistemic account says they are normative, because certainly requests sometimes give reasons merely by revealing information. So, since everyone already admits the epistemic account is operating, it is simpler if we can explain requests entirely by adverting to that account.

6.2. Sociality

Second, the epistemic account helps us fully appreciate how the normativity of requests (and the normativity of related behaviors, like wishing) takes its strength and timbre from the social contexts that also constitute these practices in the first place. Requests, like other complex social behaviors, are quite nuanced. Consider for instance the choices we have amongst the “hierarchy of request types” described by Raymond W. Gibbs Jr. (Gibbs Jr. 1985: 108). Gibbs sets out 14 different locutions one can use to make a request, and notes considerable variation among which locutions are conventional in which situations (Gibbs Jr. 1985: 101–4). Making a request at all, and making it in one way rather than another, can indicate many different things, some of which vary a lot depending on the context.44

We need to make use of that nuance to explain all the sorts of reasons that requests can give us, according to the epistemic account. Other accounts, which posit a special sort of capacity to create reasons in virtue of relationships, have a more straightforward way of explaining how a request gives reasons. But if this straightforward way does all the work, we miss out on the true complexity of requests.

Take Cupit’s puzzlement over why someone would request rather than wish if both are merely epistemic sources of reasons (Cupit 1994: 449). Elda Weizman suggests that requesting via the locution of a wish can be useful because it allows someone to plausibly deny that they made a request, and it also allows the requestee to plausibly deny that the requestor has made a request of them (Wiezman 1993: 125). Thus when one unambiguously uses the locution of a request to make a request, one does not avail oneself of this plausible deniability. The reasons for and against engendering plausible deniability are multifarious, but the general point is that quite a bit goes into making a request versus a wish versus something that is ambiguous between the two, and the epistemic account properly accords all of this context pride of place when it comes to determining normativity.

Other accounts do not necessarily have to deny that all this information is contained in one’s choice to request versus to wish versus doing something that could be either, but they do have to say it is all normatively secondary or perhaps even inert because it is swamped by the other special normativity that attaches to requests. We should not say this. Instead we should realize that all of the information encoded into the context of any request is complex enough to do all the normative work we need. So, somewhat paradoxically, the epistemic account is good not just because it’s simpler (as described above), but also because it allows for more complexity.45


For their comments, I am thankful to David Enoch, David Killoren, Max Lewis, Sushruth Ravish, Preston Werner, the attendees of the 2019 Southampton-Humboldt Normativity Workshop, and reviewers for Philosophers’ Imprint, Pacific Philosophical Quarterly, and Ergo.


  1. More precisely, a request that gives you a reason, i.e., a normative request, alters your subjective normative domain. The epistemic account, like the other accounts described below, is an account of the normativity of requests, not of requests more generally, so cases in which requests are not normative are outside the ambit of this theory. See Section 1.1 for further discussion.
  2. The gin and petrol example is from Bernard Williams’s discussion of internal and external reasons (Williams 1982: 102).
  3. One could think that the only normative reasons one can justifiably take to be objective normative reasons are actual objective normative reasons: in other words, there are no justifiable mistakes. I say nothing here about what it means to justifiably take something to be a reason, as this is a large separate topic. Any plausible account will suffice. Debates like those between perspectivism and objectivism about what one ought to do are thus separate from this discussion (Way & Whiting 2017).
  4. The distinction between objective and subjective reasons is one that is commonly drawn, and my argument does not depend on any particular specification of either sort of reason or even on the view that objective normative reasons exist (Schroeder 2008; Vogelstein 2012; Wodak 2017; 2019).
  5. I do not mean for this consideration to be at all dispositive in favor of the epistemic account. One might for instance think that these are merely ways of making a request more polite by way of obscuring it, rather than evidence that requests take the form of information proffered. I offer it only as a way of helping to make the epistemic account clear. Cf. Searle (1969: 68), Raz (1988: 36), Cupit (1994: 448), Schaber (2021: 433).
  6. Their main argument defends a view about the wrongness of certain kinds of requests and their central point is formulated such that it does not require the truth of the Epistemic-Evidential view. I am happy to accept their claim that the epistemic view enjoys some intuitive support, although in my experience people tend to find the view quite unintuitive. Their main argument can be fruitfully compared with Cupit’s account of requests, since both revolve around the importance of trust and relationships. For extensive discussion of Cupit’s account and mine see Weltman (2022a). Because Laskowski and Silver, unlike Cupit, use trust to explain the wrongness of certain requests rather than the normative force of those requests, their account supports the view I defend here (while Cupit’s account opposes my view).
  7. I thank a reviewer for this journal for suggesting I address the distinction between these questions.
  8. For discussion of accounts beyond those examined below see Weltman (2022a; forthcoming).
  9. Whether reasons given by epistemic reason-giving are “prior” depends on a variety of things, most notably whether there are objective reasons (which could easily exist prior to a request), or subjective reasons that we don’t know about, or both. These are controversial topics which are orthogonal to the discussion here and thus I will not offer any particular view. Lewis’s claim that epistemic request reason-giving gives reasons that existed prior to the request relies on answering these questions in some way as to make this possible. The epistemic account, meanwhile, could in principle say the reasons are new, but they are created merely in an epistemic sense. (Thus the epistemic account is more ecumenical.) Some cases where this occurs are discussed below in Section 4.
  10. Enoch’s only argument is that it “seems to him” that requests that merely have epistemic reason-giving force are not genuine requests (Enoch 2011: 15). Two can play at that game: these seem to me to be requests (and I do not know or care if they are genuine, so long as they are requests). Linguists would typically classify them as such: a typical linguistic definition of a request is a “communication that indicates the speaker’s desire for the hearer to bring about some desired state or event which would not have occurred otherwise” (Kim & Bresnahan 1994: 319). This definition has nothing to say about reason-giving.
  11. See, e.g., Enoch’s discussion of Cupit (Enoch 2011: 15 fn. 43).
  12. Lewis’s account is a direct continuation of Enoch’s earlier account (Enoch 2011). It also incorporates aspects of Lance and Kukla’s account (Lewis 2018: 4 fn. 7, 13 fn. 26) and shares some commonalities with Cupit’s (Lewis 2018: 4 fn. 6). Another recent account is Schaber’s, which is extremely similar to Monti’s (Schaber 2021; Monti 2021: 3751–53). Gläser classes Enoch, Lewis, and Lance and Kukla together (along with some other views) because they all give what he calls a “mere addition account” of requests (Gläser 2019: 32). I do not have space to respond directly to Schaber and Gläser, but many of their objections to opposing accounts can be accepted by the epistemic account, and the epistemic account has responses to features of their accounts which it cannot accept (see Weltman forthcoming). Moreover, because Gläser thinks requests do not give reasons, his account is compatible with mine, although I cannot show this here (Gläser 2019: 47).
  13. I argue for these points in Weltman (2022a).
  14. Most alternative definitions of insincere requests are compatible with my argument below. Neri Marsili, for instance, defines insincere requests as requests that falsely express a desire that the requestee do what the requestor is requesting (Marsili 2016: 280–81). This definition, and most other alternative definitions of insincerity, can be accommodated by my argument with little to no modification. For references to other definitions of insincerity see Marsili (2016: 280).
  15. I believe Schaber agrees with me that these insincere requests are normative, although he only explicitly says that they are requests, rather than that they are normative (Schaber 2021: 434).
  16. A similar point applies to requests that are allegedly not normative because we do not think the requestee is responsive to reasons of the sort generated by a request. Enoch argues that “in order to robustly give someone a reason . . . I must believe that she can respond to a reason thus given” (Enoch 2011: 7). Their ability is a “normative felicity condition” of request reason-giving, “without the belief in which the request is insincere” (Enoch 2011: 7). (Thus this is a special case of insincere requests which lack normativity, according to Enoch.) This implausibly suggests that if I email you asking you to comment on a draft, thinking that you are not reasons-responsive, but it turns out you are reasons-responsive, my request has given you no reason to comment on the draft. Enoch’s view implies that when you email me back with comments on my draft, I could felicitously claim not to have given you a reason to do that, since I thought that (for instance) you were too cold-hearted to entertain reasons to help me.
  17. Val would be wrong due to thinking that this implied any desire on my part for the book (if Val believes the epistemic theory), or wrong due to thinking that my request added something to the objective normative domain (if Val subscribes to an account of request normativity like Lewis’s).
  18. Or in other words, we act on apparent reasons (Sylvan 2014).
  19. One might think that apparent normativity exhausts normativity, and so in one sense insincere requests must be “really” normative because they are normative in the only way anything is normative. The epistemic account can accept this conclusion.
  20. This argument could be generalized to combat a view like Enoch’s, which holds that there is something special about request normativity above and beyond epistemic reason-giving. If there were some such component, then this component would be present in promising too (Enoch 2011: 1–2). But this would allow me to escape my promise by getting rid of this component somehow. And it is implausible to think we can escape our obligations merely by carrying out this sort of procedure (Sinnott-Armstrong 1984: 253). So this additional component must not be necessary to the normativity of things like promises and requests. This is a larger topic than I can address here.
  21. I thank a reviewer for this journal for suggesting I say more on this point.
  22. Enoch’s view faces its own objection in the form of a similar strange causal chain, and Monti argues that Enoch’s view fails because of this (Monti 2021: 3741–44).
  23. Hence the joke “a friend will help you move, while a good friend will help you move a body.”
  24. One of Monti’s objections to Enoch is that Enoch has not given any way to distinguish between deviant causal chains and typical causal chains in the relevant cases (Monti 2021: 3742). In most cases, the epistemic account can explain this by adverting to the different kinds of information communicated by different sorts of requests. See also the argument below at the end of this section.
  25. The request doesn’t just reveal this information. It also helps make it the case that the chair will be angry if you fail to read the draft. If one worries that requests cannot play a dual role like this, see Section 4 below.
  26. See also my fuller defense of this response in Weltman (2022a).
  27. I am granting this for the sake of the argument, but there may be reasons to deny it. For instance, David Estlund argues that in certain circumstances we can ignore one’s lack of consent if it is morally wrong for one to withhold consent (Estlund 2008: chap. 7). An arch-utilitarian could think consent is relevant only as evidence of what is liable to increase or decrease utility. Since these sorts of suggestions either help my argument or are orthogonal to it, I ignore them.
  28. This assumes that consent is a speech act. There is much to say about what consent is, and whether one’s expression of consent is consent or merely evidence of an internal attitude that constitutes consent, and so on. See for instance Kimberly Ferzan’s endorsement of the “mental act” view of consent (Ferzan 2016). For surveys of the topic see Gruber (2016), Dougherty (2018: 92 fn. 5). I put these questions aside here. If one thinks consent is necessarily an internal mental act, then when I suggest that a request can constitute consent, one ought to read me as saying that a request constitutes very good evidence of consent. Anyone inclined to take this view will be that much more amenable to the epistemic account of requests, because if consent happens entirely within one’s head, and consent alters the normative situation, then the cases I discuss below where requests and consent come apart are always the case. Requests are speech acts and consent (if it is not a speech act) precedes requests. The consent is normative and the request merely gives us evidence of the normative thing, which is consent. See also Enoch (2017).
  29. Gläser describes a similar case where a request constitutes a reason, but not in virtue of being a request (Gläser 2019: 47 fn. 15).
  30. It’s true that we might think requests instantiate these properties in virtue of the request being normative, just like other expressions and actions instantiate these properties when these other expressions and actions are normative. But if requests get all of their normative force from constituting consent, then we will have one theory of the normativity of requests for these cases, and another for cases where requests do not constitute consent. That would be implausible. A theory of request normativity should apply to all cases, not just to cases where requests constitute consent.
  31. Practically speaking it actually probably would be objectionable, because my desire is so strange that you should probably check to make sure I have not rescinded my consent. Here is an example that does not have this issue: I consent to your picking lemons from my lemon tree if you need them. If you later learn I want the lemons picked, you would not have to check with me first before picking the lemons.
  32. A further way to help tease apart requests and consent is to notice that there are other request-adjacent speech acts which are distinct from, but which sometimes imply, consent. See for instance Kukla’s discussion of invitations (Kukla 2018: 81–84). Kukla does not say whether invitations can imply consent, but it is typically infelicitous to invite someone to do something to which you do not consent.
  33. In practice this is complicated by the fact that we can express requests with the locution of wishes, commands, questions, and so on. Thus something like “will no one rid me of this turbulent priest?” can be understood in a number of ways. Someone who hears it as a wish may be inclined to treat it as an idle musing. Someone who hears it as a request may be inclined to treat it as a statement which implies that, if someone were to deal with the priest, they would have the speaker’s blessing. Someone who hears it as a command will be inclined to treat it as threatening punishment if it is ignored. If the speaker hears it as a wish but the listeners hear it as a command, complications may arise.
  34. It is also a normatively important difference, but not in the sense of the normativity we are interested in when we ask why requests are normative. Similarly, if I accidentally reveal what cards I am holding in a game of bridge, I am less culpable than if I intentionally do so, but in each case the cards I reveal give my opponents identical reasons to play certain cards in order to win.
  35. It is doubly hard because conventions about requesting, wishing, and similar speech acts vary from place to place, language to language, context to context, and so on. So, even to claim that X constitutes a request or a wish, let alone that the normative content of X is Y, is to wheel in a lot of baggage. See, e.g., Lee-Wong (1989), García (1993), Wilson (1994), Hassall (1999), Kasanga (2006), Moriizumi (2009), Yu (2011).
  36. This is similar to Enoch’s objection based on epistemically transparent cases (Enoch 2011: 4). Enoch’s objection imagines cases where we know everything and yet the request gives us a reason. The present objection doesn’t depend on epistemic transparency, which is important because, as I argue elsewhere, adverting to epistemic transparency obscures the relevant issue and potentially also begs the question (Weltman 2022a). Laskowski and Silver note the similarity between these two objections (Laskowski & Silver 2021).
  37. We might be tempted to say that I make my request in order to gently remind you, but if I’m actually just reminding you, this vindicates the epistemic account, since I’m merely giving you relevant information about the litter. If it’s truly common knowledge that the litter needs cleaning, then a reminder is otiose.
  38. Psychic powers would turn the case into Enoch’s epistemically transparent situation, as noted above. For discussion see Laskowski and Silver (2021), Weltman (2022a).
  39. My arguments thus complement and strengthen Laskowski and Silver’s: between their reply to this objection, my replies here, and my arguments in Weltman (2022a), it should be clear that the epistemic account has nothing to fear from this objection.
  40. It may technically have normative power insofar as you have more reason to attend to a reason brought to your attention, because (for instance) it would be rude to ignore someone’s having brought it to your attention. But this normative power has nothing to do with the request being a request rather than any other action.
  41. This includes cases where requests are evidence of themselves, in which case the requests are normative but qua something else, rather than qua request. The chief case of this is consent, as discussed above.
  42. I thank a reviewer for this journal for suggesting Parfit’s notion of reason-involvement here.
  43. Pace Laskowski and Silver’s claim that the view is intuitive (Laskowski & Silver 2021).
  44. Examples of distinctions like this are multifarious. See, e.g., Elizabeth de Kadt’s discussion of 9 kinds of locutions, from “hedged performatives” like “I would like to ask you . . .” to “strong hints” like “Why is the door closed?” (de Kadt 1992: 103).
  45. For related reasons to endorse the epistemic theory, see Weltman (2022a).


1 Bertolet, Rod (2017). On the Arguments for Indirect Speech Acts. Philosophia, 45(2), 533–40.

2 Cupit, Geoffrey (1994). How Requests (and Promises) Create Obligations. The Philosophical Quarterly, 44(177), 439–55.

3 de Kadt, Elizabeth (1992). Requests as Speech Acts in Zulu. South African Journal of African Languages, 12(3), 101–6.

4 Dougherty, Tom (2018). Affirmative Consent and Due Diligence. Philosophy & Public Affairs, 46(1), 90–112.

5 Enoch, David (2011). Giving Practical Reasons. Philosophers’ Imprint, 11(4), 1–22.

6 Enoch, David (2017). Hypothetical Consent and the Value(s) of Autonomy. Ethics, 128(1), 6–36.

7 Estlund, David (2008). Democratic Authority. Princeton University Press.

8 Ferzan, Kimberly Kessler (2016). Consent, Culpability, and the Law of Rape. Ohio State Journal of Criminal Law, 13(2), 397–439.

9 García, Carmen (1993). Making a Request and Responding to It: A Case Study of Peruvian Spanish Speakers. Journal of Pragmatics, 19(2), 127–52.

10 Gibbs, Raymond W. Jr. (1985). Situational Conventions and Requests. In Joseph P. Forgas (Ed.), Language and Social Situations, 97–110. Springer-Verlag.

11 Gläser, Micha (2019). The Normative Structure of Request. In Mark Timmons (Ed.), Oxford Studies in Normative Ethics (Vol. 9, 29–49). Oxford University Press.

12 Gruber, Aya (2016). Consent Confusion. Cardozo Law Review, 38, 415–59.

13 Hassall, Timothy (1999). Request Strategies in Indonesian. Pragmatics, 9(4), 585–606.

14 Kasanga, Luanga A. (2006). Requests in a South African Variety of English. World Englishes, 25(1), 65–89.

15 Kim, Min-Sun and Mary Bresnahan (1994). A Process Model of Request Tactic Evaluation. Discourse Processes, 18(3), 317–44.

16 Kukla, Rebecca (2018). That’s What She Said: The Language of Sexual Negotiation. Ethics, 129(1), 1–39.

17 Lance, Mark and Rebecca Kukla (2013). Leave the Gun; Take the Cannoli! The Pragmatic Topography of Second-Person Calls. Ethics, 123(3), 456–78.

18 Laskowski, N. G. and Kenneth Silver (2021). Wronging by Requesting. In Mark Timmons (Ed.), Oxford Studies in Normative Ethics (Vol. 11, 49–69). Oxford University Press.

19 Lee-Wong, Song Mei (1989). Imperative in Requests: Direct or Impolite - Observations From Chinese. Pragmatics, 4(4), 491–515.

20 Lewis, James H. P. (2018). The Discretionary Normativity of Requests. Philosophers’ Imprint, 18(20), 1–16.

21 Marsili, Neri (2016). Lying by Promising: A Study on Insincere Illocutionary Acts. International Review of Pragmatics, 8, 271–313.

22 Monti, Ezequiel H. (2021). Against Triggering Accounts of Robust Reason-Giving. Philosophical Studies, 178(11), 3731–53.

23 Moriizumi, Satoshi (2009). Face Concerns and Requests in Japan: Exploring the Effects of Relational Closeness and Social Status. Journal of Intercultural Communication Research, 38(3), 149–74.

24 Owens, David (2008). Promising Without Intending. The Journal of Philosophy, 105(12), 737–55.

25 Parfit, Derek (2011). On What Matters: Volume One. Oxford University Press

26 Raz, Joseph (1988). The Morality of Freedom. Oxford University Press.

28 Schaber, Peter (2021). The Reason-Giving Force of Requests. Ethical Theory and Moral Practice, 24(2), 431–42.

29 Schroeder, Mark (2008). Having Reasons. Philosophical Studies, 139(1), 57–71.

30 Searle, John (1969). Speech Acts: An Essay in the Philosophy of Language. Cambridge University Press.

31 Sinnott-Armstrong, Walter (1984). ‘Ought’ Conversationally Implies ‘Can.’ The Philosophical Review, 93(2), 249–61.

32 Sylvan, Kurt (2014). What Apparent Reasons Appear to Be. Philosophical Studies, 172(3), 587–606.

33 Vogelstein, Eric (2012). Subjective Reasons. Ethical Theory and Moral Practice, 15(2), 239–57.

34 Way, Jonathan and Daniel Whiting (2017). Perspectivism and the Argument from Guidance. Ethical Theory and Moral Practice, 20(2), 361–74.

35 Weltman, Daniel (2022a). The Normativity of Requests: The Epistemic Theory versus Relation-Centric Theories. Unpublished manuscript.

44 Weltman, Daniel (2022b). The Paper Chase Case and Epistemic Accounts of Request Normativity. Unpublished manuscript.

36 Weltman, Daniel (forthcoming). How Requests Give Reasons: The Epistemic Account versus Schaber’s Value Account. Ethical Theory and Moral Practice.

37 Whiting, Daniel (2014). Keep Things in Perspective: Reasons, Rationality and the A Priori. Journal of Ethics & Social Philosophy, 8(1), 1–22.

38 Wiezman, Elda (1993). Interlanguage Requestive Hints. In Gabriele Kasper and Shoshana Blum-Kulka (Eds.), Interlanguage Pragmatics (123–37). Oxford University Press.

39 Williams, Bernard (1982). Internal and External Reasons. In Moral Luck: Philosophical Papers 1973–1980 (101–13). Cambridge University Press.

40 Wilson, Steven R. (1994). A Cross-Cultural Comparison of Implicit Theories of Requesting. Communication Monographs, 61(3), 210–35.

41 Wodak, Daniel (2017). Can Objectivists Account for Subjective Reasons? Journal of Ethics and Social Philosophy, 12(3), 259–79.

42 Wodak, Daniel (2019). An Objectivist’s Guide to Subjective Reasons. Res Philosophica, 96(2), 229–44.

43 Yu, Kyong Ae (2011). Culture-Specific Concepts of Politeness: Indirectness and Politeness in English, Hebrew, and Korean Requests. Intercultural Pragmatics, 8(3), 385–409.