Options
Article

Advice for Analytic Naturalists

Author
  • Jesse Hambly (Australian National University)

Abstract

In this paper I argue against Analytic Normative Naturalism by suggesting that the view cannot capture the way that normative concepts figure in advice. To establish this conclusion, I identify several links between normative concepts and advice and argue that, if Analytic Normative Naturalism were true, these links would not obtain.

How to Cite:

Hambly, J., (2023) “Advice for Analytic Naturalists”, Ergo an Open Access Journal of Philosophy 9: 59. doi: https://doi.org/10.3998/ergo.3584

Downloads:
Download XML
Download PDF
View PDF

604 Views

334 Downloads

Published on
2023-07-18

Peer Reviewed

License

Creative Commons Attribution-NonCommercial-NoDerivs 4.0

Introduction

In this paper I argue that Analytic Normative Naturalism cannot capture the way that normative concepts figure in advice. I offer several arguments for this conclusion. These arguments identify links between normative concepts and advice and suggest that, if Analytic Normative Naturalism were true, these links would not obtain.

The structure of the paper is as follows: in Section 1 I clarify Analytic Normative Naturalism (henceforth ‘Analytic Naturalism’). In Section 2 I identify two connections between normative concepts and advice or recommendation (I’ll use these terms interchangeably) through an examination of an account of the nature of recommendation offered in an important recent defence of Analytic Naturalism by Stephen Finlay. In Section 3 I use these connections to develop two arguments against Analytic Naturalism and defend these arguments against objections. I suggest that the most promising reply to these arguments available to the Analytic Naturalist involves adopting a particular analysis of the concept advice. In Section 4 I show that adopting this analysis has the consequence that Analytic Naturalists are unable to capture another important link between normative concepts and advice. Section 5 concludes.

1. Clarifying Analytic Naturalism

I understand Analytic Naturalism as comprising two core claims:

(AN1) Normative properties are nothing over and above or reducible to natural properties; and

(AN2) We can establish (AN1) by giving an analysis of normative concepts in non-normative terms.1

In this section, I’ll clarify (AN1) and (AN2).

Focusing first on (AN1), I understand the normative in terms of reasons and oughts. The relevant reasons and oughts are ‘all things considered’ or simpliciter reasons and oughts. (There are interesting challenges to the idea that we can make sense of claims about all things considered or simpliciter reasons and oughts. However, I believe that the arguments I give in this paper can be used to help defend the coherence of claims about such reasons and oughts by helping to identify the conceptual role of the ought simpliciter.)2 I will interpret the claim that normative properties are nothing over and above or reducible to natural properties as suggesting that for all normative properties F, for all x, any fact of the form Fx is either identical to or fully grounded in a natural fact (i.e., in the instantiation of a natural property).3 For the purposes of this paper, I hold that it’s a sufficient condition for a property to be natural that we can characterize the property in non-normative terms.4

Turning to (AN2), an analysis of a concept tells us what it takes to fall under a concept C; it provides metaphysically necessary and/or sufficient conditions for falling under C. In this paper my focus will be on ‘definitional’ analyses which state necessary and sufficient conditions for falling under a concept.5 (Also, the type of analyses that I’m concerned with here are descriptive as opposed to ameliorative or revisionary analyses.)6 To take as uncontroversial an example as possible, an analysis of the concept aunt tells us that x falls under the concept aunt just in case x is the sister-of or wife-of-a-sibling-of a parent. Applying this discussion to normative concepts, an analysis of, say, ought, tells us that something falls under the concept ought just in case it has natural property N. Analytic Naturalists need to hold that N is either identical to the property of being what one ought to do or that N fully grounds the property of being what one ought to do. Without this assumption, the analysis won’t help to establish that the normative is reducible to the natural.

My argument in this paper is that non-normative analyses of normative concepts can’t capture the conceptual role of such concepts, specifically their role in advice. (By the ‘conceptual role’ of a concept I mean “its role in thinking, problem solving, deliberating and the like—and, in general, in mediating between sensory inputs and behavioural outputs.”)7 Given that I’ve characterized conceptual analysis as telling us about what it takes to fall under a concept, it may be a little unclear why an analysis of a normative concept would need to capture the role of the concept in advice. I’ll return to this issue in Section 3.3.

There is a final point concerning conceptual analysis that I need to discuss: proponents of Analytic Naturalism commonly hold that agents who possess normative concepts typically lack explicit knowledge of the correct analyses of these concepts. For example, Stephen Finlay claims that “ordinary competence [with a concept] is a form of tacit-knowledge or know-how: a disposition or ability to follow a rule R that functions unconsciously and typically doesn’t rise to the level of explicit knowledge or theory” (2014: 7).8 Similarly, Michael Smith asserts that “Even though someone who has mastery of some concept C must have certain inferential and judgemental dispositions, it may not be transparent to her what these inferential and judgmental dispositions are” (1994: 38).9 The significance of this point—call it ‘the unobvious analysis response’—is that it provides Analytic Naturalists with a way of responding to challenges to the view based on the fact that an agent who possesses a normative concept can doubt or dispute the truth of any proposed analysis of the concept. (The classic challenge of this kind is the Open Question Argument.)10 The Analytic Naturalist can accept that this is true but maintain that, given that the correct analysis of a normative concept might be unobvious to someone who possesses the concept, this doesn’t show that the proposed analysis is false. A central aim of this paper is to develop a line of argument against Analytic Naturalism which isn’t threatened by the unobvious analysis response.

2. Finlay on Advice

In this section I’ll examine an account of the nature of recommendation offered in an important recent defence of Analytic Naturalism by Stephen Finlay. I have two reasons for discussing Finlay’s views on advice: Firstly, I’ll use a problem that I identify with Finlay’s account to motivate two connections between normative concepts and advice. These connections will form the basis of the arguments I go on to give against Analytic Naturalism in Section 3. Secondly, Finlay’s account illustrates some common but, I think, mistaken ideas about advice in contemporary (meta)ethics. Given that my arguments in this paper will rely on claims about the nature of advice, addressing these ideas will prove useful for what follows.

Finlay suggests that “to recommend to s that she φ is to draw s’s attention to the possibility of φ-ing ostensibly with the intention to motivate s toward φ-ing in pursuit of her own [s’s own] desired ends” (2014: 128 and 181).11 Finlay’s account of recommendation entails that: (R1) if A recommends to s that she φ, then A is drawing s’s attention to the possibility of φ-ing ostensibly with the intention to motivate s toward φ-ing in pursuit of s’s own desired ends; and, (R2) if A draws s’s attention to the possibility of φ-ing ostensibly with the intention to motivate s toward φ-ing in pursuit of s’s own desired ends, then A is recommending to s that she φ.

I think that (R2) is false and that we can learn some important lessons from its failure. However, before I discuss (R2), I’ll explain why (R1) is also questionable. Why think it is true that when A recommends to s that she φ, A must ostensibly intend to motivate s toward φ-ing in pursuit of s’s own desired ends? I might recommend s take some medicine which will ensure she lives many more years of healthy, happy life, even having acknowledged that s does not have any desires that will be promoted by taking the medicine.12

This counterexample to (R1) might be disputed on the grounds that advice is defective if the advisee cannot be motivated to perform the action in question13 and that a necessary condition for being motivated to φ is having a desire that will be promoted by φ-ing.14 The first thing to note about this response is that it’s not clear it supports (R1) rather than (R1*): If A non-defectively recommends to s that she φ, then A is drawing s’s attention to the possibility of φ-ing ostensibly with the intention to motivate s toward φ-ing in pursuit of s’s own desired ends. Putting aside this issue, the second conjunct of the response is false. Why does one need to have a desire that will be promoted by φ-ing in order to be motivated to φ? Notice that this is not the position defended by proponents of the Humean Theory of Motivation (HTM). According to the HTM, one must only have a desire (the object of which) one believes will be promoted by φ-ing in order to be motivated to φ.15

Suppose that, considering this point, the proponent of the response we are examining slightly modifies the second conjunct so that it claims that ‘a necessary condition for being motivated to φ is having a desire one believes will be promoted by φ-ing’ and they stipulate that s doesn’t have a desire that she believes would be promoted by taking the medicine. This still doesn’t establish I can’t recommend s take the medicine. The fact that s currently doesn’t have any desires she believes will be promoted by taking the medicine doesn’t mean s couldn’t acquire a new desire she believes would be promoted by taking the medicine after the advice is given. s could then be motivated to take the medicine by this desire. For example, in virtue of having received the advice s might acquire a desire to do what she has been advised to do.16

To see why (R2) is problematic, imagine that I stand to gain from s’s φ-ing. I might draw s’s attention to the possibility of φ-ing ostensibly with the intention to motivate her toward φ-ing in pursuit of her desired ends even though I believe it would be a terrible mistake for her to φ. In such a case, I have merely pretended to advise s to φ.

To deal with this case, Finlay might drop the ‘ostensibly’ from the account of advice and suggest that this revision allows for a straightforward response to the counterexample: in the case that I’ve described, I do not intend to motivate s toward φ-ing in pursuit of her desired ends. Rather, I intend to motivate s to do what I want her to do. However, imagine that the proverbial eccentric billionaire of philosophy examples has promised to give me a billion dollars if I can motivate s to φ in pursuit of her desired ends. In this case, I have the specific intention to motivate s to φ in pursuit of her desired ends, not just the more general intention to motivate s to do what I want her to do. I am clearly concerned with the way in which s comes to φ, not merely that she φ’s. I would not be content, for example, to use some kind of mind control device on s to ensure that she φ’s.

A better response to the counterexample is to suggest that I have in fact advised s to φ, but that I’ve done so insincerely as opposed to merely pretended to advise s to φ.17 For the purpose of the arguments against Analytic Naturalism that I want to develop in this paper, I’m happy to concede that my counterexample only shows that (R2) can’t serve as a sufficient condition for sincere advice.18

What lessons can we draw from the failure of (R2) as a sufficient condition for sincere advice? Firstly, (R2) conflicts with the fact that in order to sincerely recommend that s φ one must do so on the basis of certain facts one thinks help justify or support s’s φ-ing. Or, to put it another way, if one sincerely advises s to φ, one must do so on the basis of facts one thinks of as reasons for s to φ. Call this the Advice Base Link. Secondly, (R2) conflicts with the fact that one cannot sincerely recommend that s φ if one believes that s ought not to φ. Call this the Ought Not/Advise Not Link.

3. Advice and Analytic Naturalism: Two Arguments

3.1. The Advice Base Argument

I’ll now offer two arguments against Analytic Naturalism built around the links between normative concepts and sincere advice I identified in Section 2. I’ll start with the Advice Base Link. I formulated this link as claiming that if one sincerely advises s to φ, one does so on the basis of the belief that certain facts are reasons for s to φ. (Note, from this point on, by ‘recommendation’ or ‘advice’, I’ll exclusively mean sincere recommendation or advice unless expressly stated otherwise.) However, it’s not entirely clear that this is correct as it stands. Perhaps we sometimes advise agents on the basis of merely taking a fact to be a reason. After all, we often don’t have explicit thoughts about reasons when we give advice.19 In light of this point, my revised statement of the Advice Base Link is the following: if one advises s to φ one does so on the basis of either taking or believing some fact r to be a reason for s to φ.20 I’m now in a position to formulate what I’ll call the Advice Base Argument.

AB-1) If one advises s to φ one does so on the basis of taking or believing some fact to be a reason for s to φ.

AB-2) It’s not the case that if one advises s to φ one does so on the basis of taking or believing some fact to be a N* fact (where N* represents an analysis of the concept reason in non-normative terms).

AB-3) If (AB-1) and (AB-2) then Analytic Naturalism is false.

Therefore,

AB-4) Analytic Naturalism is false.

I hope that my discussion of the problems with (R2) in Section 2 has made (AB-1) plausible. To evaluate (AB-2), I want to focus on what is probably the best developed Analytic Naturalist theory in the literature, Stephen Finlay’s ‘end-relational’ view. According to Finlay, to believe that s ought to φ is, roughly, to believe that some contextually specified (non-normative) end is most likely if s φ’s and to believe a fact is a reason for s to φ is to believe that this fact explains why some end would be more likely if s φ-ed (2014: chs. 3 and 4).21 An ‘end’ here is a “proposition conceived as a potential outcome” (2014: 32).22 Finlay’s account of normative concepts interacts with his account of advice that I discussed in Section 2: what makes a particular utterance of a normative sentence an instance of advice, according to Finlay, is that the contextually relevant end is one which is desired by the audience (2014: 139). The question we are faced with, then, is the following: is it true that if one advises s to φ, one does so on the basis of believing or taking some fact to explain why some end (you believe) s desires would be more likely if s φ’s? I think we can clearly conceive of someone advising another agent to φ without doing so on this basis. For example, I might advise s to φ on the grounds that φ-ing would bring them pleasure or be morally best. I even think that one might advise s to φ on these grounds if one explicitly believes that these facts don’t explain why φ-ing would promote one of s’s desires, but this isn’t necessary for the success of my argument.23

My argument can, I think, be extended to target any proposed analysis of reason in non-normative terms; for any analysis of reason in non-normative terms, we can conceive of an agent advising s to φ without doing so on the basis of the fact picked out by the analysis in question.

3.2. The Ought Not/Advise Not Argument

The other link between normative concepts and advice I posited in my discussion of Finlay’s account of advice, the Ought Not/Advise Not Link, suggests that if you believe s ought not to φ then you cannot advise s to φ.24 The argument built around this link, the Ought Not/Advise Not Argument, claims:

ON/AN-1) If you believe s ought not to φ then you cannot advise s to φ.

ON/AN-2) It’s not the case that if you believe N- (where N- represents the Analytic Naturalist’s analysis of ought not), then you cannot advise s to φ.

ON/AN-3) If (ON/AN-1) and (ON/AN-2), then Analytic Naturalism is false.

Therefore,

ON/AN-4) Analytic Naturalism is false.

(ON/AN-1) is just a statement of the Ought Not/Advise Not Link. My discussion in Section 2 has hopefully already made the link plausible, but I have more to say in its defence. I’ll provide two arguments for the link. The first of these arguments draws on connections between the concept ought and advisability:

  1. If you believe s ought not to φ, then you believe or take s’s φ-ing not to be advisable.

  2. If you believe or take s’s φ-ing not to be advisable you cannot advise s to φ.

  3. Therefore, if you believe s ought not to φ, you cannot advise s to φ.

Premise (1) is very plausible. If someone were to say that they believe s ought not to φ but then go on to maintain it’s advisable for s to φ we would wonder about their grip on the relevant concepts; for some action or attitude to be advisable is just for that action or attitude to be advice-worthy and it’s hard to see how one could think what someone ought not to do (in the all things considered sense) is advice-worthy. Premise (2) might be disputed on the grounds that if one is not attending to the fact one believes or takes s’s φ-ing not to be advisable then it’s not the case that it’s impossible to recommend s φ. If this really is an issue—and I’m not convinced it is—I could qualify premise (2) so it applies only to cases in which your attitudes are fully transparent to you.25

My other argument for the Ought Not/Advise Not Link involves thinking about our interest in soliciting advice. Our primary aim in soliciting advice is to help us determine what it is we ought to do.26 It’s unclear whether advice could play the role of helping us determine an answer to the question of what we ought to do if we could advise agents to do what we believe they ought not to do; being advised to φ would (at least qua advice) leave you no better placed to answer this question than you were before the advice was given.

Having argued for (ON/AN-1), I can move on to defending (ON/AN-2). (ON/AN-2) claims that it’s not the case that if you believe N- (where N- represents the Analytic Naturalist’s analysis of ought not), then you cannot advise s to φ. Suppose the Analytic Naturalist claims that we can analyse ought in terms of what would best promote some contextually specified non-normative end and, consequently, ought not as what would do the least to promote the relevant end.27 Suppose first that the contextually specified end is an end desired by the advisee. It doesn’t seem true that one can’t advise s to φ if one thinks φ-ing is the action available to s that would do the least to promote one of s’s desired ends. This is true no matter what the desired end may be. For instance, I could advise s to φ despite believing that φ-ing would do the least to promote their desires overall or do the least to bring them pleasure, perhaps because I think that φ-ing would be morally valuable. Importantly, the issue here is not whether my advice is bad advice—for all I’m concerned to establish this might be true—only whether I count as advising s at all.

Perhaps the lesson of my discussion in the previous paragraph is that a proponent of the Analytic Naturalist view which I’m using as a test case—which is roughly Stephen Finlay’s view explained in Section 3.1—should say that the relevant end is an end desired by the advisor rather than the advisee. Is it the case that you can’t advise s to φ if you think φ-ing is the action available to s that would do the least to promote one of your own desired ends? I think it’s clear that the answer is ‘No’. For example, I might advise my friend to apply for a job she is highly likely to receive, despite the fact that I desperately want the job in question.

On a related note, Finlay (amongst others) has drawn attention to the limitations of views which try to explain the way normative concepts are used to give advice in terms of these concepts being used to express an agent’s own attitudes:

Expressivists characteristically explain the interpersonal influence of normative speech as a kind of psychological contagion, manipulation, or “osmosis”, appealing to a contingent human disposition to be motivationally influenced by others’ evident motivations. To advise you about what you “ought” to do is to try to manipulate you psychologically into having the motivations I want you to have; to ask for normative advice about what I ought to do is just to invite psychological manipulation . . . as a general account of normative discourse and recommendatory force this verges on the scandalous. (2014: 143)28

Something similar to the point Finlay is making here equally applies to views that suggest that normative concepts are used to describe facts about one’s own desired ends; why would telling someone about what would promote some end you desire constitute advice?

At this point the Analytic Naturalist might suggest that instead of focusing on ends desired by the advisor, we need to focus on ends endorsed by the advisor. Is it true that you can’t advise s to φ if you think φ-ing is the action (available to) s that would do the least to promote some end which you endorse? The answer to this question depends on whether endorsement is a normative state—that is, a mental state which is explicated using normative concepts. If endorsement is a non-normative state, the answer is ‘No’. As I’ve argued, I can advise someone to perform some action which would do the least to promote some end which I desire, and I take it that a non-normative account of the state of endorsement will hold that endorsement is a state related to desiring. However, on the other hand, if endorsement is a normative state, then perhaps it is impossible to advise someone to do something that you think would do the least to promote some end you endorse. However, this is of no help to the Analytic Naturalist. An Analytic Naturalist can’t provide an analysis of ought which employs concepts which are themselves explicated using normative concepts on pain of the resulting view being circular or failing to establish normative facts are nothing over and above natural facts.29 (This point also explains why an Analytic Naturalist can’t adopt something like Finlay’s end-relational approach to analysing normative concepts but hold that the relevant ends are normatively specified ends like ‘doing what one ought to do’.)

The points that I’ve been developing in the preceding paragraphs to defend (ON/AN-2) can, I think, be employed against any Analytic Naturalist theory: for any analysis of ought in non-normative terms, N, an agent could believe N- yet advise another agent to φ.

3.3. Objections

I’ll now consider several objections to the two arguments that I’ve offered. The first objection targets (AB-3) and (ON/AN-3). These premises tell us that if non-normative analyses of normative concepts can’t capture the connections between normative concepts and advice I’ve posited (the Advice Base Link and the Ought Not/Advise Not Link), then Analytic Naturalism is false. So far, I’ve taken these premises for granted. However, Analytic Naturalists might dispute that their analyses need to capture such connections.

If a successful analysis must capture a concept’s conceptual role, then (AB-3) and (ON/AN-3) are unproblematic. That this is a requirement on a successful (definitional) analysis is assumed in several arguments against Analytic Naturalism. For example, Ralph Wedgwood develops an objection to Analytic Naturalism based around the claim that necessarily, if one is rational, then, if one judges that one ought to φ, one intends to φ. (Wedgwood calls this thesis ‘normative judgement internalism or ‘NJI’.) Wedgwood contends that Analytic Naturalists can’t explain the truth of NJI because for “every possible non-normative content, it is possible for there to be a perfectly rational agent who judges that content but is not motivated to form the corresponding intention to act. . . . So it seems, no [non-normative] conceptual analysis [of ought] can possibly explain why NJI is true” (2007: 71). Similarly, Paul Boghossian argues that the problem with any non-normative, definitional analysis of the concept good is that

1. It is necessary that, if S judges that x is good, S has certain positive dispositions towards x [i.e., dispositions to behave in certain ways and feel certain emotions]. 2. It is not necessary that if S judges that x is D, where D is any descriptive property, S has any positive dispositions towards x. Therefore, 3. For any D, ‘good’ cannot mean the same as D. (2021: 373–74)30

However, suppose an Analytic Naturalists denies that a successful analysis needs to capture a concept’s conceptual role on the grounds that all that is required of a successful analysis is that it captures metaphysically necessary and sufficient conditions for falling under the analysandum concept (i.e., the concept being analysed).31 This cannot, however, be sufficient for a successful analysis; necessary a posteriori identities state metaphysically necessary and sufficient conditions for falling under a concept but are not analyses.32 A key feature which differentiates an analysis from a necessary a posteriori identity claim is that one’s justification for believing the analysis rests on one’s understanding of the analysandum concept.33 I’ll now argue that this feature can be used to support the claim that Analytic Naturalists need to capture the connections between normative concepts and advice I’ve posited.

Focus on the Ought/Not Advise Not Argument. I’ve argued that your understanding of ought tells you that you can’t advise someone to φ if you believe that they ought not to φ. (Note that the fact that some would deny this putative conceptual truth can’t be a fatal challenge to its status as a conceptual truth—at least if Analytic Naturalism is to have any hope of success.)34 I’ve also argued that it’s not the case that you can’t advise someone to φ if you believe that φ-ing is N-. If I’m right, this suggests that to think someone ought not to φ isn’t a matter of classifying φ-ing as N-. But this makes it hard to see how reflection on the concept ought could reveal that something is what you ought to do just in case it is N.35

Moving on to another objection to my arguments, an Analytic Naturalist might deny (ON/AN-1) and perhaps also (AB-1)—that is, the Ought Not/Advise Not Link and the Advice Base Link—by appealing to cases like the following: suppose I know you desire to φ. I might provide you with information about how to φ or how best to φ. For example, suppose I am a financial advisor and you tell me that you want to put your money in cryptocurrency. If I tell you which cryptocurrency would be best for you to invest in, it might seem natural to describe me as having advised you to put your money in cryptocurrency. However, I could provide you with this information even if I believe you ought not to put your money in cryptocurrency. Perhaps I even believe you have no reason to do so. Similarly, suppose you tell me that you are trying to decide whether to go law school or do a PhD in philosophy. If I tell you to go to law school, it again might seem natural to claim that I have advised you to go to law school. However, I could tell you to go law school in this case even if I believe that you ought not go to law school (because there is some option available to you other than these two which I think that you should take).36

My response to these cases is built around the fact that I could, apparently without contradiction, tell you to put your money in cryptocurrency A (rather than B or C) or tell you to go to law school, yet add that I’m not advising you to put your money in cryptocurrency or advising you to go to law school. To see this, focus on the case of telling you to go to law school. It seems perfectly coherent for me to tell you to go to law school rather than do a PhD in philosophy but add that I would advise you not to go. What this suggests is that either these cases don’t involve advice at all or, more plausibly, that they involve a kind of advice that is relativized to an end similar to the way instrumental ought claims like ‘If you are going to be a famous mass murderer, you ought to kill as many people as you can’ are relativized to an end.37 (Notice that in both of the cases of putative advice we are considering it would be natural to use an instrumental ought claim. For instance, I might say ‘If you are going to go to law school or do a PhD in philosophy, then you ought to do a PhD in philosophy.’) If the first disjunct is true, such cases obviously don’t threaten the links between advice and normative concepts that I have posited. On the other hand, if the second disjunct is true, I can respond that (ON/AN-1) and (AB-1) don’t involve the kind of instrumental or qualified advice which features in these cases. Rather, they feature what we might call ‘categorical’ or ‘all things considered advice’. Why think that we can give advice of this kind? Derek Parfit’s example of advising an agent to take some medicine which will ensure she lives many more years of healthy, happy life, even knowing that the agent does not have desires that will be promoted by taking the medicine, illustrates the possibility of such advice (2011: 281). However, there are less extreme (and less controversial) examples: compare being asked by a student about whether to apply to philosophy graduate school with being asked by the student about how to apply to graduate school. In the first case, unlike the second, you don’t assume some end and then provide information about the (best) way to achieve it. Of course, you may well consider facts about your student’s desired ends when advising them about whether to go to graduate school, but this hardly shows that such advice is relativized to such ends, only that we often treat people’s desires as relevant when considering what it is advisable for them to do.38

I think that some philosophers are likely to insist that all advice is relativized to some (non-normative) end that the advisor believes is desired by the advisee. To put the point another way, they will insist that all advice is advice about how to φ, given you believe that the advisee wants to φ. (Note that the kinds of cases that I’ve been considering in the preceding paragraphs which might seem to help motivate this position do not support the claim that advice is relativized to the advisee’s desires taken collectively or to their deepest desires. When I tell someone about which cryptocurrency to invest in—given that they want to invest in cryptocurrency—my claim isn’t somehow inapt or inapplicable if I believe that investing in cryptocurrency is in tension with many of the agent’s other, possibly deeper, desires.)39 I have several comments to make about this position: firstly, the position isn’t the view that you can’t advise an agent to φ if you think that it’s impossible that they could be motivated to φ. Perhaps this is true. However, this is perfectly compatible with it being the case that not all advice is relativized to some desired end of the advisee. Secondly, note that the view that proponents of the position that all advice is implicitly relativized to the advisee’s desired ends are committed to regarding a case like Parfit’s medicine case isn’t simply that it is a mistake to advise this person to take the medicine—something which is perfectly compatible with everything I have had to say about advice—but the much stronger position that this is conceptually impossible. I worry this view might gain what plausibility it has from an illicit slide from conclusions about when advice is inappropriate (because inefficacious, imprudent, or just plain mistaken) to conclusions about whether someone is engaging in the activity of advising at all. What I take to be central to the activity of advising is that you are telling an agent about what they have reason to do40 and I can think that the agent has reason to take the medicine.41 Thirdly, advising is clearly not something that is constrained by the existing descriptive beliefs of the advisee; when advising people about what to do you don’t tailor your advice to their existing descriptive beliefs about the world. Why then is advice constrained by an agent’s existing desires? Is there some principled reason for this difference other than the thought that advice must be capable of motivating—a thought which doesn’t require that advice be constrained in this way?42 Finally, even if it is true that advice is constrained by an agent’s existing desired ends (so that you can’t advise an agent to do something that you believe wouldn’t promote their existing desires), this doesn’t entail that advice is relativized to such ends in the sense that all advice is advice about what to do in order to promote such ends.43 More needs to be said to establish this conclusion.

The final response to the Advice Base Argument and the Ought Not/Advise Not Argument I’ll consider suggests these arguments are problematic because we can analyse the concept advice in terms of conveying to an agent that they ought to φ. An Analytic Naturalist might suggest the following analysis of advice: ‘you advise an agent, s, to φ just in case you convey to s that they ought to φ’. (I am using ‘convey’ here in a stipulative fashion to incorporate both asserting that s ought to φ and implicating that s ought to φ.)44 This analysis plausibly explains why the two links between advice and normative concepts that I’ve posited hold: the Ought Not/Advise Not Link holds because you can’t convey to an agent that they ought to φ if you believe they ought not to φ. (I’m assuming here that conveying that p entails believing that p.)45 Similarly, the Advice Base Link holds because if you convey to an agent that they ought to φ you do so on the basis of taking or believing some fact to be a reason for the agent to φ.

What is useful about this analysis of advice for the Analytic Naturalist is that it implies that intuitions about the concept advice are not independent of intuitions about the concept ought. An Analytic Naturalist can claim that the intuition that one can advise an agent to φ even if one believes N- (i.e., (ON/AN-2)) is explained by the intuition that one can convey an agent ought to φ even if one believes N-. And, similarly, the Analytic Naturalist can claim that the intuition one can advise an agent to φ without doing so on the basis of believing or taking some fact to be a N* fact (i.e., (AB-2)) is explained by the intuition that one can convey that an agent ought to φ without doing so on the basis of believing or taking some fact to be a N* fact. In both cases, the second intuition looks vulnerable to the unobvious analysis response; the Analytic Naturalist can contend that these intuitions can be explained away by the fact that the correct analyses of ought and reason might be unobvious to an agent who possesses these concepts. So, with the analysis of advice we’ve been considering in hand, the Analytic Naturalist can suggest that my two arguments are no improvement on the Open Question Argument.

Evaluating the plausibility of this response to my arguments involves evaluating the plausibility of the analysis of advice on which it rests. However, I’m just going to assume for the sake of argument that this analysis succeeds. I’ll now argue that embracing this analysis leaves the Analytic Naturalist open to another advice-based argument against the view, which I call the Advice Argument. An interesting and important feature of the Advice Argument is that it is not vulnerable to the unobvious analysis response.

4. The Advice Argument

According to the Advice Argument:

AA-1) Conveying to s that they ought to φ is sufficient to advise s to φ.

AA-2) Conveying to s that some non-normative fact, N, obtains is not sufficient to advise s to φ.

AA-3) If (AA-1) and (AA-2), then Analytic Naturalism is false.

Therefore,

AA-4) Analytic Naturalism is false.

If an Analytic Naturalist commits themselves to the analysis of advice according to which one advises s to φ just in case one conveys to s that they ought to φ, then they are committed to (AA-1); (AA-1) is just the right-hand conditional of the analysis.

Turning to (AA-2), suppose the Analytic Naturalist suggests ought can be analysed in terms of what an agent would most strongly desire themselves to do if they had informed and coherent desires. Suppose I convey to s that φ-ing is what they would most strongly desire themselves to do if they had informed and coherent desires. Have I recommended to s that they φ? I don’t deny that there are circumstances in which conveying this information to s could constitute advice. However, it seems clear that there are also circumstances in which I could convey this to s yet not be advising s to φ. For example, suppose that I believe that Jones would most strongly desire to kill some person who has insulted him if he possessed an informed and coherent set of desires. I also believe that if he kills the person he will be arrested and punished with hard labour for the rest of his life. I might convey to Jones that he would most strongly desire to kill this person if he had informed and coherent desires yet not be advising him to kill his enemy.46,47

This point, I believe, can be extended to any analysis of ought in non-normative terms. For any such analysis, there will be cases in which one can convey to s that N obtains, yet not advise s to φ. To be clear, by saying this I’m not denying that you can advise a person to φ by conveying to them that some non-normative fact obtains. For example, if I say to you ‘Buying that house leaves you exposed to serious financial risk’, I might be advising you not to buy the house. All I’m suggesting is that, for any non-normative fact, N, conveying to s that N obtains doesn’t entail that you have advised s to φ.48

Analytic Naturalists might be tempted to reply to (AA-2) by offering the following response: consider an agent who lacks explicit knowledge of the correct analysis of ought. Imagine this agent conveys to s that N. This agent has in fact advised s to φ, but is not aware that they have done so because they don’t realize that by conveying that N they have conveyed to s that s ought to φ. My concern about this response is that it involves suggesting an agent can inadvertently advise another agent to φ. I think we should be suspicious of this claim. The idea of inadvertently giving advice is of dubious coherence. Giving advice is like commanding or requesting; whether we do it depends on whether we intend to do it. The lesson of my discussion in this paragraph is that the nature of advice creates problems for employing the unobvious analysis response against the Advice Argument.

Another response to the Advice Argument suggests that, given that I’ve granted the analysis of advice in terms of conveying to an agent that they ought to φ, it may be that we can substitute ‘conveying to s that they ought to φ’ for ‘advice’ in the premises of the Advice Argument. Call this version of the Advice Argument the Advice Argument’:

AA1') Conveying to s that they ought to φ is sufficient to convey to s that they ought to φ.

AA2') Conveying to s that some non-normative fact, N, obtains is not sufficient to convey to s that they ought to φ.

AA3') If (AA1') and (AA2') then Analytic Naturalism is false.

Therefore,

AA4') Analytic Naturalism is false.

Suggesting that the Advice Argument can be restated this way is not in and of itself an objection to the argument. It doesn’t show that any of the premises of the argument are false. An Analytic Naturalist might claim that once we see that (AA-2) can be restated as (AA-2') denying (AA-2) looks more plausible. However, I don’t see why this is true. The example I gave to motivate (AA-2) works just as well in the case of (AA-2'). What if the Analytic Naturalist suggests that the way of explaining away the plausibility of (AA-2) I considered above appears to work better in the case of (AA-2')? That strategy was to suggest that (AA-2) merely seems true because an agent might convey to s that N obtains, but not realize that they have advised s to φ. So, in the case of (AA-2'), the suggestion would be that while (AA-2') is false, its plausibility is explained by the fact that an agent might convey to s that N, but not realize they have conveyed to s that s ought to φ. I claimed the strategy we are considering failed in the case of (AA-2) because we cannot inadvertently give advice. But, if this claim is true, then, if the analysis of advice we are considering has any chance of success, it must similarly be true that we can’t inadvertently convey to an agent that they ought to φ.

I’ve defended the Advice Argument on the assumption that (AA-1) is true—an assumption which follows from a particular analysis of advice which I granted for the sake of argument. However, I’m not confident that this analysis succeeds. Nor am I confident about the prospects of giving an independent defence of (AA-1). This isn’t to say that I’m convinced that (AA-1) is false. I think (AA-1) may well be true,49 it’s just that I don’t have an argument for this conclusion. Also, there are some challenges to (AA-1).50 Given these issues, I only want to employ the Advice Argument against Analytic Naturalists who accept the analysis of advice in terms of conveying to an agent that they ought to φ. Of course, Analytic Naturalists are free to reject this analysis. However, if they do so, they leave themselves open to the Ought Not/Advise Not and Advice Base Arguments.

5. Conclusion

My aim in this paper has been to show that when we attend to various important connections between normative concepts and advice, we see that Analytic Naturalists are unable to vindicate these connections. I’ve argued for this conclusion by developing two links between advice and normative concepts and showing that either Analytic Naturalists can’t capture these links or, if they can (through adopting a particular analysis of advice), they are left unable to capture another such connection.

I want to conclude on a methodological note. I’ve relied heavily on appeals to intuitions concerning advice. Analytic Naturalists are hardly in a position to take issue with my use of intuitions; whatever else is involved in conceptual analysis it centrally involves drawing on intuitions.51 Drawing on intuitions concerning advice and its connections to normative concepts can hopefully advance the debate between Analytic Naturalists and their opponents. Such intuitions can serve as data to guide or constrain our theorizing about normative concepts.

Notes

  1. Proponents of Analytic Normative Naturalism include Finlay (2014), Smith (1994), and arguably Williams (1981; 1995). Jackson and Pettit defend Analytic Moral Naturalism, the view that we can establish that moral properties are natural properties by giving analyses of moral concepts in non-moral terms. See Jackson and Pettit (1995), Jackson (1998: 113–62), and Pettit (2001). For a similar view see Lewis (1989). ⮭
  2. For worries about the coherence of the ought simpliciter see Baker (2018). For defences of the ought simpliciter see Wodak (2019), Dorsey (2014: ch. 1), and Case (2016). ⮭
  3. My discussion here draws on Rosen (2018). Cf. Leary (2021). ⮭
  4. What is it to ‘characterize’ a normative property in non-normative terms? For different ways of cashing out this idea (any of which would serve for my purposes) see Wedgwood (2007: 145), Rosen (2018: 154–55), and Streumer (2017: 5). ⮭
  5. For discussion of non-definitional analyses see Ludwig (2013: 322) and Chalmers and Jackson (2001: 322). Can Analytic Naturalists claim that they only need to give non-definitional analyses of normative concepts which provide naturalistic sufficient conditions for falling under such concepts? For worries about non-definitional, non-normative analyses of normative concepts see Hattiangadi (2018: 595–98) and Boghossian (2021: 374–79). ⮭
  6. For a discussion which bears on the prospects of giving revisionary analyses of central normative concepts like ought see Eklund (2020). See also Boghossian (2021: 379–82). ⮭
  7. Block (1987: 160). Sometimes ‘conceptual role’ is used more narrowly than the way that I’m using the expression here. For relevant discussion see Greenberg and Harman (2008: 296–97). ⮭
  8. For relevant discussion see Laskowski and Finlay (2018: 546–49). ⮭
  9. For more on how correct analyses can be opaque to agents see Jackson (2010: 179–82). ⮭
  10. For a sophisticated development of the Open Question Argument see Greco (2015: 509–14). ⮭
  11. Finlay describes this as a “definition” of recommendation (2014: 181). ⮭
  12. This example is Derek Parfit’s (2011: 281), although he uses it in a different context. ⮭
  13. This response is suggested by some of Bernard Williams’s remarks about advice (1995: 40). See also Wiland (2021: 122). ⮭
  14. One might support the first conjunct of this response by suggesting that the constitutive aim of advice is to get an agent to perform an action by heeding the advice. For relevant discussion see Southwood (2016: 24). ⮭
  15. Proponents of the HTM claim that a “desire is necessary for action, and no mental states other than a desire and a means-end belief are necessary for action” (Sinhababu 2009: 465). Notice that the way Sinhababu formulates the HTM leaves it open that an agent could be motivated to φ without having a means-end belief. He does this because “one may still be a Humean whether or not one holds that desires to engage in immediate bodily movements—for example, a desire to move my hand right now—can cause action without the assistance of a means-end belief” (2009: 465, footnote 2). I’m going to ignore the complication that this raises for my discussion because I don’t think that it threatens the point I’m making here. ⮭
  16. Cf. Wiland (2003: 303–4). ⮭
  17. Eric Wiland (2021: 171–72) gives some cases of insincere advice. Wiland suggests that a key difference between advice and command is that “(felicitous) advice typically aims to direct its recipient to do what they already have reason to do. Command need not, and often does not, have this aim” (2021: 126). I’m not sure how Wiland thinks of the relationship between felicitous advice and sincere advice. ⮭
  18. There is another response to my challenge to (R2) built around the claim that we can sincerely advise an agent about how to φ—given that they have adopted φ-ing as an end—even if we believe that φ-ing would be a terrible mistake. I will discuss roughly this claim in detail in Section 3.3 below. For the moment, however, I just want to note that it seems unlikely to me that the truth of this claim will establish that there are no cases where (1) I draw s’s attention to the possibility of φ-ing with the intention to motivate her toward φ-ing in pursuit of her desired ends, even though I believe it would be a terrible mistake for her to φ and (2) I have not sincerely advised s to φ. All I require is one such case (one case where (1) and (2) are true) to establish that (R2) fails. ⮭
  19. For a discussion of what is involved in taking something to be a reason see Schlosser (2012). I think that the idea of taking something to be a reason is one that we need for all sorts of purposes including explaining what is involved in acting for a normative reason and the process of inference. For the former point see Singh (2019). (Singh talks of ‘representing considerations as reasons’ rather than ‘taking them as reasons’, but his use of ‘represent as a reason’ tracks the way other philosophers use ‘take as a reason’.) For the latter point see Boghossian (2019). ⮭
  20. It’s arguable that believing that r is a reason entails taking r to be a reason, making the second conjunct redundant. This may be true if taking something if taking something to be a reason can be understood in terms of implicitly believing it to be a reason. For relevant discussion see Enoch (2011: ch. 9.1). ⮭
  21. Finlay is committed to the view that a conceptual analysis tells us about what the concept contributes to the content of beliefs which feature the concept. See (2014: 142 and 209). This view—call it ‘the content claim’—is common amongst proponents of Analytic Naturalism. See, for example, Smith (1994: 153 and 177), Jackson and Pettit (1995: 21 and 30), and Jackson (1998: 155). ⮭
  22. Importantly, according to Finlay, such ends needn’t be desired or intended by s (2014: 178). ⮭
  23. I’ll have more to say about the relationship between advice and an agent’s desires in Section 3.3 below. ⮭
  24. Does the truth of the Ought Not/Advise Not link depend on whether the ‘ought’ in the link is interpreted as a subjective (belief or evidence-relative) ought or an objective ought? It might appear that if the ‘ought’ is interpreted as a subjective ought then the link is false. This is because there are cases where an advisee’s evidence or beliefs support φ-ing, but you (the advisor) have information which suggests that this is not the best choice. In such a case, it appears that the following is true: the advisee ought subjectively to φ (and we can suppose that you believe this) but you nevertheless sincerely advise them not to φ. (Frank Jackson’s classic three medicines case can be used to illustrate the sort of case that I’m thinking of here. See 1991: 462–63.) Perhaps the lesson of such cases is that I must say that the Ought Not/Advise Not Link features an objective ought; you can’t advise an agent to do what you believe that they ought objectively not to do. This commitment seems very natural given that what these cases appear to establish is that we don’t limit our advice based on the beliefs or evidence of the advisee. However, some philosophers have argued that an advisor’s statement that an advisee ought to φ in the sorts of cases we are considering can be true, even if the ‘ought’ in the statement is read as a subjective ought. See Kiesewetter (2011). Very roughly, Kiesewetter’s idea is that the advisor’s statement at t1 can change the evidential situation of the advisee, making it true that they ought subjectively to φ at t2 (i.e., the time of action). If something like this proposal can be made to work, then the Ought Not/Advise Not Link may be true even if the ‘ought’ which it features is read as a subjective ought. For worries about Kiesewetter’s argument see Graham (2021: 43–45). ⮭
  25. By ‘fully transparent’ I mean that one has explicit knowledge that one has the attitudes in question. Cf. Worsnip (2018: 188). ⮭
  26. It’s important to see that our aim in soliciting advice is not just to determine what to do. To determine what to do one could simply flip a coin. See Finlay (2014: 142–43). ⮭
  27. I’m using ‘end’ here in the way that Finlay uses the term, which I explained in Section 3.1 above. ⮭
  28. See also Scanlon (2014: 58–59). ⮭
  29. Given the close connection between normative concepts and evaluative concepts like good or valuable, the same problems that I’m drawing attention to here will arise if endorsement is understood in evaluative terms. For helpful discussion of the connections between normative and evaluative concepts see Wedgwood (2009). ⮭
  30. By ‘meaning’ here what Boghossian has in mind is roughly something like the possession conditions for a concept—i.e., the conditions which tell us what is involved in possessing the concept. See Peacocke (1992: ch. 1) for a classic discussion of concepts and possession conditions. Wedgwood similarly appears to think that an analysis needs to capture a concepts’ meaning (2007: 58–61 and 66). ⮭
  31. What is the relationship between the analysandum concept and analysans concept (the concept expressed by an analysis)? For discussion of this issue see Horvath (2018) and Earl (2007). Both Horvath and Earl develop a view according to which these concepts are not identical but the analysans concept tells us about the nature of the analysandum concept. ⮭
  32. E.g., x is a gold just in case x has 79 protons in its nucleus. There are also non-scientific examples: ‘x is a reason just in case x is Derek Parfit’s actual favourite property’ states a metaphysically necessary and sufficient condition for falling under the concept reason. For relevant discussion see Horvath (2018). ⮭
  33. To put the point another way, I’m claiming that Analytic Naturalists are committed to the view that (the proposition expressed by) an analysis like ‘φ-ing is what one ought to do just in case φ-ing is N’ is epistemically analytic. For discussion of epistemic analyticity see Paul Boghossian’s contributions to Boghossian and Williamson (2020). Boghossian discusses whether normative propositions are epistemically analytic in his (2021). ⮭
  34. For discussion of the possibility that p can be a conceptual truth (i.e., a truth such that one can be justified believing it in virtue of understanding it) even if p can be rejected by someone who fully understands it see Horvath (2020). ⮭
  35. Importantly, I don’t think that anything I’ve said here is inconsistent with it being the case that that one ought to φ just in case φ-ing is N. My claim is just that this isn’t something you can learn through conceptual reflection, because your understanding of ought suggests that the concept doesn’t somehow encode the information that φ-ing is what you ought to do just in case φ-ing is N. ⮭
  36. These cases are based on cases discussed in Andreou (2006: 59–61). ⮭
  37. This example is Stephen Finlay’s (2014: 50). Finlay uses the example to illustrate the way that instrumental conditionals have non-detachable consequents (i.e., instrumental conditionals don’t license inference by modus ponens). ⮭
  38. Eric Wiland similarly draws a distinction between what he calls ‘categorical’ and ‘instrumental’ advice. He frames the distinction as a matter of the difference between “advice about what to do” and “advice about how to do something” (2021: 128). ⮭
  39. Cf. Andreou (2006: 60–61). ⮭
  40. Cf. footnote 17 above. ⮭
  41. Note that this might be the case even if they don’t have such a reason because some form of (existence) internalism which rules this out is true—at least if the relevant form of internalism isn’t understood as a conceptual truth. For discussion of the internalism/externalism debate see Paakkunainen (2018) and Brunero (2017). I admit that Analytic Naturalists who think that a relevant form of internalism is a conceptual truth may be unconvinced by the claim that I can think that this agent has a reason to take the medicine. ⮭
  42. My discussion here is inspired by Asarnow (2019). My discussion in Section 2 and footnote 24 are also relevant. ⮭
  43. For a defence of this claim see Wiland (2021: 128–29). ⮭
  44. For discussion of the relationship between assertion and implicature see Pagin and Marsili (2021). ⮭
  45. I’m ignoring the complication raised by insincere assertion, among other issues. ⮭
  46. This case is Derek Parfit’s. He uses it to defend something roughly like the Advice Argument (2011: 284). Parfit’s (2011: 278–88) suggestive but elusive discussion of the relationship between Analytic Naturalism and advice is the inspiration for this paper. ⮭
  47. Why would I tell Jones this information if I didn’t mean to advise him to φ? Perhaps he knows I have this information and has threatened to kill me if I do not reveal it. ⮭
  48. Might an Analytic Naturalist argue that: (1) by conveying to s that they ought to φ one pragmatically expresses one’s pro-attitudes towards s’s φ-ing and that this explains why conveying that an agent ought to φ has the connection to advice described in (AA1); and (2), the truth of (1) can be used to explain why (AA2) doesn’t present a problem for their theory? One issue for this strategy is that the putative explanation of the connection between conveying that an agent ought to φ and advice in (1) is problematic for the same sorts of reasons that Finlay holds that expressivist accounts of advice are problematic. (See Section 3.2 above.) ⮭
  49. (AA-1) may need to be slightly modified: there is some plausibility to the idea that to advise s to φ I must see myself as having appropriate standing vis-à-vis s. Suppose I have seriously wronged you in the past. It seems I could convey to you that φ-ing is what you ought to do but deny I am advising you to φ, claiming that it is not my place to give you advice. (There is an interesting parallel here with blame. It seems I can regard a person as blameworthy but believe it is not my place to blame them. For example, perhaps I believe it would be hypocritical for me to blame them given I have committed the same wrong.). It’s easy enough to modify (AA-1) to take into account this point: if you convey to s that they ought to φ and you believe that you have appropriate standing vis-à-vis s, then you advise s to φ. ⮭
  50. Stephen Finlay argues against the claim that conveying to s that they ought to φ entails advising s to φ, contending that “Many people find nothing incoherent about somebody saying, ‘You ought to φ, of course. But don’t do that! Come ψ with me instead—it’s much more fun!’, and such an utterance wouldn’t naturally be interpreted as incorporating a recommendation to φ” (2014: 133). I think we can largely explain away the apparent plausibility of this counterexample by noticing that it’s natural to interpret the ‘ought’ in this sentence as being a moral ought or some other qualified ought. However, suppose we insist on giving the ‘ought’ in this utterance an all things considered reading. I agree with Finlay we don’t naturally interpret the utterance as incorporating a recommendation to φ. This is because we interpret it as incorporating a recommendation to ψ instead. In fact, if we don’t interpret the utterance as incorporating a recommendation to ψ it no longer seems to be a counterexample to (AA-1). However, if this utterance incorporates a recommendation to ψ there is a problem: I seem to be recommending you do something I am committed to thinking you ought not to do. Why am I committed to thinking that you ought not to ψ? By stipulation, I believe you ought to φ. If I believe you ought to φ, I am committed to the position that you ought not to ψ (on the assumption that φ-ing and ψ-ing are mutually exclusive actions and I don’t think you ought to (φ or ψ)). Therefore, in Finlay’s case, I am advising you to do something I am committed to believing you ought not to do. However, I argued above (in Section 3.2) that one can’t advise an agent to do something one believes that they ought not to do. ⮭
  51. Arguably, conceptual analysis also involves abductive reasoning. For discussion of this point see Balcerak-Jackson (2013: 3066–72). ⮭

Acknowledgements

I’m grateful to Nicholas Southwood, Philip Pettit, John Broome, Thomas Schmidt, Josef Holden, Benjamin Kiesewetter, two anonymous referees for Ergo, and the Ergo area editor for their helpful comments.

References

1 Andreou, Chrisoula (2006). Standards, Advice, and Practical Reason. Journal of Moral Philosophy, 3(1), 57–67.

2 Asarnow, Samuel (2019). Internal Reasons and the Boy Who Cried Wolf. Ethics, 130(1), 32–58.

3 Baker, Derek (2018). Skepticism about Ought Simpliciter. In Russ Shafer-Landau (Ed.), Oxford Studies in Metaethics (Vol. 13, 230–52). Oxford University Press.

4 Balcerak Jackson, Magdalena (2013). Conceptual Analysis and Epistemic Progress. Synthese, 190(15), 3053–74.

5 Block, Ned (1987). Functional Role and Truth Conditions. Proceedings of the Aristotelian Society, 61, 157–83.

6 Boghossian, Paul (2019). Inference, Agency and Responsibility. In Magdalena Balcerak Jackson and Brendan Balcerak Jackson (Eds.), Reasoning: New Essays on Theoretical and Practical Thinking (101–24). Oxford University Press.

7 Boghossian, Paul (2021). Normative Principles are Synthetic A Priori. Episteme, 18(3), 367–83.

8 Boghossian, Paul and Timothy Williamson (2020). Debating the A Priori. Oxford University Press.

9 Brunero, John (2017). Recent Work on Internal and External Reasons. American Philosophical Quarterly, 54(2), 99–117.

10 Case, Spencer (2016). Normative Pluralism Worthy of the Name is False. Journal of Ethics and Social Philosophy, 11(1), 1–20.

11 Chalmers, David and Frank Jackson (2001). Conceptual Analysis and Reductive Explanation. The Philosophical Review, 110(3), 315–60.

12 Dorsey, Dale (2014). The Limits of Moral Authority. Oxford University Press.

13 Earl, Dennis (2007). A Semantic Resolution of the Paradox of Analysis. Acta Analytica, 22(3), 189–205.

14 Eklund, Matti (2020). Variance Thesis in Ontology and Metaethics. In Alexis Burgess, Herman Cappelen, and David Plunkett (Eds.), Conceptual Ethics and Conceptual Engineering (187–204). Oxford University Press.

15 Enoch, David (2011). Taking Morality Seriously. Oxford University Press.

16 Finlay, Stephen (2014). Confusion of Tongues: A Theory of Normative Language. Oxford University Press.

17 Graham. Peter A. (2021). Subjective Versus Objective Moral Wrongness. Cambridge University Press.

18 Greco, Daniel (2015). Epistemological Open Questions. Australasian Journal of Philosophy, 93(3), 509–23.

19 Greenberg, Mark and Gilbert Harman (2008). Conceptual Role Semantics. In Ernest Lepore and Barry C. Smith (Eds.), The Oxford Handbook of the Philosophy of Language (296–322). Oxford University Press.

20 Hattiangadi, Anandi (2018). Moral Supervenience. Canadian Journal of Philosophy, 48(3–4), 592–615.

21 Horvath, Joachim (2018). Philosophical Analysis: The Concept Grounding View. Philosophy and Phenomenological Research, 97(3), 724–50.

22 Horvath, Joachim (2020). Understanding as a Source of Justification. Mind, 129(514), 509–34.

23 Jackson, Frank (1991). Decision-Theoretic Consequentialism and the Nearest and Dearest Objection. Ethics, 101(3), 461–82.

24 Jackson, Frank (1998). From Metaphysics to Ethics. Oxford University Press.

25 Jackson, Frank (2010). Conceptual Analysis for Representationalists. Grazer Philosophische Studien, 81(1), 173–88.

26 Jackson, Frank and Philip Pettit (1995). Moral Functionalism and Moral Motivation. The Philosophical Quarterly, 45(178), 20–40.

27 Kiesewitter, Benjamin (2011). “Ought” and the Perspective of the Agent. Journal of Ethics and Social Philosophy, 5(3), 1–25.

28 Laskowski, Nicholas and Stephen Finlay (2018). Conceptual Analysis in Metaethics. In Tristram McPherson and David Plunkett (Eds.), The Routledge Handbook of Metaethics (536–51). Routledge.

29 Leary, Stephanie (2021). Normativity. In Michael J. Raven (Ed.), The Routledge Handbook of Metaphysical Grounding (472–83). Routledge.

30 Lewis, David (1989). Dispositional Theories of Value. Proceedings of the Aristotelian Society, 63, 113–37.

31 Ludwig, Kirk (2013). Methods in Analytic Epistemology. In Matthew C. Haug (Ed.), Philosophical Methodology: The Armchair or the Laboratory (217–39). Routledge.

32 Paakkunainen, Hille (2018). Internalism and Externalism about Reasons. In Daniel Star (Ed.), The Oxford Handbook of Reasons and Normativity (143–70). Oxford University Press.

33 Pagin, Peter and Neri Marsili (2021). Assertion. In E. N. Zalta (Ed.), The Stanford Encyclopedia of Philosophy (Winter 2021 ed.). Retrieved from https://plato.stanford.edu/archives/win2021/entries/assertion/

34 Parfit, Derek (2011). On What Matters (Vol. 2). Oxford University Press.

35 Peacocke, Christopher (1992). A Study of Concepts. Oxford University Press.

36 Pettit, Philip (2001). Embracing Objectivity in Ethics. In Brian Leiter (Ed.), Objectivity in Law and Morals (234–85). Cambridge University Press.

37 Rosen, Gideon (2018). Metaphysical Relations in Metaethics. In Tristram McPherson and David Plunkett (Eds.), The Routledge Handbook of Metaethics (151–69). Routledge.

39 Scanlon, T. M. (2014). Being Realistic About Reasons. Oxford University Press.

40 Schlosser, Markus E. (2012). Taking Something as a Reason for Action. Philosophical Papers, 41(2), 267–304.

41 Singh, Keshav (2019). Acting and Believing Under the Guise of Normative Reasons. Philosophy and Phenomenological Research, 99(2), 409–30

42 Sinhababu, Neil (2009). The Humean Theory of Motivation Reformulated and Defended. The Philosophical Review, 118(4), 465–500.

43 Smith, Michael (1994). The Moral Problem. Blackwell.

44 Southwood, Nicholas (2016). Does “Ought” Imply “Feasible”? Philosophy & Public Affairs, 44(1), 7–45.

45 Streumer, Bart (2017). Unbelievable Errors: An Error Theory About All Normative Judgements. Oxford University Press.

46 Wedgwood, Ralph (2007). The Nature of Normativity. Oxford University Press.

47 Wedgwood, Ralph (2009). The “Good” and the “Right” Revisited. Philosophical Perspectives, 23, 499–519.

48 Wiland, Eric (2003). Some Advice for Moral Psychologists. Pacific Philosophical Quarterly, 84(3), 299–310.

49 Wiland, Eric (2021). Guided by Voices: Moral Testimony, Advice, and Forging a ‘We’. Oxford University Press.

50 Williams, Bernard (1981). Internal and External Reasons. In Bernard Williams (Ed.), Moral Luck (101–13). Cambridge University Press.

51 Williams, Bernard (1995). Internal Reasons and the Obscurity of Blame. In Bernard Williams (Ed.), Making Sense of Humanity (35–45). Cambridge University Press.

52 Wodak, Daniel (2019). Mere Formalities: Fictional Normativity and Normative Authority. Canadian Journal of Philosophy, 49(6), 828–50.

53 Worsnip, Alex (2018). What is (In)coherence. In Russ Shafer-Landau (Ed.), Oxford Studies in Metaethics (Vol. 13, 184–206). Oxford University Press.