1. Introduction
In metaethics, it is often said that some standards – call them ‘substantively normative standards’ – are normatively authoritative: paradigm examples include morality, prudence, and epistemic justification.Footnote 1 Other standards – call them ‘merely formally normative’, or ‘mere formalities’ for short – are said to lack this feature: paradigm examples include etiquette, grammar, and codes of honor, and law. As we will see, this distinction also appears in everyday folk discourse when we talk about what we really, truly, or genuinely ought or have reason to do.
Many doubt that, say, morality is authoritative, but accept that some standards are (e.g. prudence) and some are not (e.g. etiquette). For our purposes, that’s fine. This is not a paper about the authority of morality. It’s a paper about the very possibility of normative authority. That very possibility has been called into doubt by skeptics about authority, who treat claims like ‘etiquette is not genuinely normative’ as meaningless: they are either empty rhetoric, or a symptom of conceptual confusion.
Skepticism about authority is an unattractive thesis. In this paper, I aim to undermine the case for skepticism via an unexplored route: instead of focusing on what makes standards like morality special, I focus on what makes standards like etiquette ‘not really normative’. I defend a form of fictionalism about mere formalities on which etiquette is ‘not really normative’ in roughly the same way that Sherlock is ‘not really a detective’. And I show how adopting fictionalism about mere formalities undermines the case for skepticism about the possibility of authority.
Here’s the plan. First, I argue that we should want an account of how (say) morality can be authoritative and etiquette can be ‘not really normative’. Second, I outline the case for skepticism, and show why it seems formidable on traditional approaches that focus on what makes standards like morality special. Third, I outline a different route wherein we focus instead on what makes standards like etiquette ‘not really normative’, and show why fictionalism is an attractive position once we take that route. Fourth, I argue that fictionalism undermines the case for skepticism about authority. Finally, I consider an important objection to fictionalism.
2. Authority and ‘non-genuine normativity’
In this section I argue that we should want an account of how it is possible for some standards to be authoritative and others to be ‘not really normative’ to provide a charitable interpretation of folk and philosophical discourse, and to explain a distinct kind of normative mistake. Before doing so, let’s clarify the lay of the land. What is meant by claims about (1) standards being ‘authoritative’, (2) standards being ‘mere formal’, and (3) standards being ‘not genuinely normative’?
Claims that some standards are authoritative have been a familiar feature of metaethics for centuries. A common Kantian gloss of such claims was that morality issues categorical requirements, whereas etiquette et al. do not. This Kantian gloss was famously questioned by Philippa Foot, who argued that morality and etiquette are both categorical. ‘Both are inescapable’, she argued, in the intuitive sense ‘that behavior does not cease to offend against either morality or etiquette because the agent is indifferent to their purposes and to the disapproval he will incur by flouting them’ (Foot Reference Foot1972, 311). It has since become common to gloss claims about authority in terms of claims about ‘necessary reason-giving force’.
Claims that some standards are ‘mere formalities’ are newer, and there is no convergence on terminology here. There has been a trend, following Foot, to discuss contrast putatively authoritative and non-authoritative standards, like morality and etiquette. The latter is offered as an instance of a relevant class of standards. Derek Parfit, for instance, contrasts ‘reason-involving normativity’ with ‘rule-involving normativity’, which:
involves requirements, or rules, that distinguish between what is correct and incorrect, or what is allowed and disallowed. Certain acts are required, for example, by the law, or by the code of honour, or by etiquette, or by certain linguistic rules. It is illegal not to pay our taxes, dishonourable not to pay our gambling debts, and incorrect to eat peas with a spoon, to spell ‘committee’ with only one ‘t’, and to use ‘refute’ to mean ‘deny’ (Parfit Reference Parfit2011, vol. 1, 144–145, emphasis his).
Tristram McPherson offers a similar gloss on this class of standards (McPherson Reference McPherson2011).Footnote 2 I am using his terminology, ‘merely formal normativity’, because it has a nicer nominalization, and because Parfit’s terminology is misleading and inapt.Footnote 3
What about claims that some standards are ‘not genuinely normative’? These claims arise in debates about particular standards. For instance, some claim that structural rationality is not truly normative.Footnote 4 Such claims also arise in debates about particular metaethical theories. For instance, metaethicists claim that all standards would turn out to be not really normative if naturalism were true (see e.g. Parfit Reference Parfit2011, vol. 2, ch. 25, and cf. Rosati Reference Rosati1995, and references therein), or if constitutivism were true (see Silverstein (Reference Silverstein and Shafer-Landau2016), and references therein), or if non-naturalism were true (see Dasgupta (Reference Dasgupta2017), and references therein). Such claims are prompted by the contention that the relevant standard(s) do or would lack authority.
Skeptics often contend that claims about ‘normative authority’ and/or ‘non-genuine normativity’ are either compatible with their deflationary views about normativity, or, if they aren’t, such claims can be dismissed on the ground that they are vague and confused. While there are significant differences between his view and the views of the skeptics I discuss later, Stephen Finlay provides a nice initial case in point.Footnote 5 Finlay argues that his view explains how our talk about morality is (like etiquette) categorical in roughly the sense(s) described by Foot.Footnote 6 But he then denies that there is good evidence for some further ‘special categoricity of moral language’ to be found in our uses of such language:
Is there evidence of any other kind of categoricity in normative language not yet considered? […] The only other evidence offered [for that claim] is the intuitions and protestations of some people that they mean something different by their moral claims, which Richard Joyce vaguely calls ‘practical oomph’. But they decline to analyze or explain what this is: primitivists like Derek Parfit maintain it is ineffable, while error theorists like Joyce and Wittgenstein hold the concept to be inchoate and nonsensical. Such a ‘fugitive thought’, which makes no testable predictions about moral practice, poses no genuine challenge (Finlay Reference Finlay2014, 196)Footnote 7.
If claims about ‘normative authority’ and/or ‘non-genuine normativity’ were all just vague protestations of philosophers with metaethical axes to grind, Finlay might be on firm ground. But I do not think that this is the case. I take it that important features of folk discourse get at the same idea. Consider how we mark claims about ‘non-genuine normativity’ by talking about what we really, truly, or genuinely ought to do. This point was recently made by Michael Ridge, in discussing the ‘deep discontinuity between the ‘ought’ of etiquette, on the one hand, and the ‘ought’ of morality and the all things considered ‘ought’, on the other’, and between ‘reasons of etiquette’ and ‘standards of practical reasoning which we think deserve to be taken seriously’ (Ridge Reference Ridge2014, 93, 16, 40). In contrasting these, Ridge noted something important about folk normative discourse:
there are certain distinctive ways in which we mark genuinely normative claims. The contexts in which we make such conversational moves are familiar enough. We use modifiers like ‘really’ [and] ‘genuine’... to modify words like ‘ought’, ‘must’, ‘reason’ and the like. […] [S]omeone skeptical about etiquette might remark, ‘I know that etiquette demands that I put the fork here, but do I really have to put the fork here? I mean, in the grand scheme of things, why should anyone care about etiquette anyway?’Footnote 8
In other words, contra Finlay, we do have a ‘testable prediction’ about our discourse: when one takes some standard to be not genuinely normative, one will be disposed in certain conversational contexts to use modifiers like ‘really’ and ‘genuine’ before normative terms like ‘ought’ and ‘reason’. And one does not always do so with a metaethical axe to grind.
It is a cost to a theory if it cannot offer a charitable interpretation of this feature of folk normative discourse. We should not, like the skeptic about authority, dismiss such claims as empty or confused.Footnote 9 Some may disagree; revisionary metaethicists abound. But the skeptic is in no position to disagree, as she motivates her view by saying that she is taking the features of our ordinary normative discourse at face value: we talk of ‘legal reasons’, so we must accept that law is also necessarily reason-giving in some sense. If ordinary people who are skeptical that law is ‘genuinely normative’ mark this by using such modifiers in the way described above, the skeptic should not dismiss their claims out of hand.
A similar point applies to philosophical discourse about particular standards. Of the putative mere formalities mentioned above, the one discussed most is legal normativity.Footnote 10 The philosophy of etiquette is not a large field. But the philosophy of law is. And a wide variety of philosophers of law claim that it is ‘not genuinely normative’. The leading natural law theorist Finnis noted that in deeply unjust legal systems we can create ‘legal requirements’, but ‘cannot create what any self-respecting person would count as a genuine obligation’ (Finnis Reference Finnis2002, 22–23). Similar language is used by legal positivists. David Enoch agrees that legal requirements necessarily give us legal reasons, but he denies that these are genuine reasons: ‘the word ‘legal’ in ‘legal reason’ functions like the word ‘imaginary’ in ‘imaginary friend’’, in that the fact that there is a legal reason to φ does not entail that there is a reason to φ, just as the fact that one has an imaginary friend does not entail that one has a friend (Enoch Reference Enoch, Green and Leiter2011, fn. 28). Enoch notes that this view dates back at least to H. L. A. Hart Essays on Bentham (Reference Hart1982, 267). Like Finlay, some find this view to be puzzling. ‘When you say that legal reasons are not real reasons, ’ Scott Hershovitz wrote, ‘you owe us an account of what they are, since the claim is that they are not what they purport to be’ (Hershovitz Reference Hershovitz2015, fn. 55). Fair enough: we are owed an account of what claims about non-real obligations and reasons amount to, and none has been provided. But it is still true that these features of philosophical discourse (and folk discourse) give us a strong reason to want an account of how it is possible for some standards to be authoritative and others to be ‘not really normative’.
Another reason to want such an account is that we should want to be able to explain a distinct kind of normative mistake. Consider an actual incident described by Kevin, a pimpFootnote 11. Kevin spoke of the ‘the rules of the game’, which were ‘adhered like laws’, even though some are immoral. When a prostitute called Lois – for whom Kevin had much respect and affection – was disobedient, he ‘had to do something’, ‘had to act’, so he punished her violently: he slapped her repeatedly, then ‘whacked her across her back with this pool stick just about as hard as [he] could’, breaking ‘the pool stick in half’. Kevin knew this was immoral, but he did it because it was what the pimp norms required (see Leslie Reference Leslie2015).
On skeptical views, the requirements of morality and pimp norms are on a par; according to morality, punishing Lois was a mistake; according to the pimp norms, letting the matter slide would be a mistake too. The norms conflict, but neither has ‘authority’ over the other. This is deeply unattractive. We should want an account of how it is possible for some standards (like morality) to be authoritative, so that we can make sense of how Kevin could make a normative mistake by prioritizing the wrong standard (pimp norms) in practical deliberation about what to do.
3. Of skeptics and hamburgers
So we should want an account of how it is possible for some standards to be authoritative and others to be ‘not really normative’. The skeptic, however, contends that there can be no such account. And the case for skepticism about authority looks formidable, if we take for granted a common assumption about how we should think about mere formalities.
Call this assumption the ‘Hamburger Model’. On this view, the relation between mere formalities and substantive normativity is like the relation between hamburgers and hamburgers with cheese: the former provides minimal conditions for something being a normative standard; the latter consists of those conditions plus some additional feature (‘cheese’).Footnote 12
This assumption is implicit in much of the (admittedly limited) literature on mere formalities. McPherson wrote that in ‘explaining the difference between [substantive] and merely formal normativity’, the first criterion would be to locate ‘a feature (call it F) that explained [substantive] normativity’.Footnote 13 We start with a notion of mere formalities, then just add F. David Copp’s view is similar. He calls mere formalities ‘generic normativity’ (Copp Reference Copp and Schaber2004). We start with a generic sense of normativity (a hamburger), then add differentia (cheese) to get substantive normativity; mere formalities are the genus without that feature. (McPherson recently adopted this terminology, ‘generic normativity’, for mere formalities (see McPherson Reference McPhersonforthcoming). On Copp’s view, a generically normative standard just is a system of requirements, recommendations, and permissions that can be correctly expressed using deontic modals.Footnote 14 There’s nothing further to say about mere formalities. The only further question is then: what’s the cheese?
The skeptic about authority accepts that standards like morality and etiquette are normative in this minimal sense. But she denies that any standard like morality could be authoritative, Footnote 15 and that any standard like etiquette can be ‘not genuinely normative’. Talk of ‘non-genuine normativity’, on her view, is either conceptually confused or purely rhetorical. The skeptic’s case here turns on the difficulty of identifying the cheese. ‘The relevant notion of authority, ’ the skeptic contends, ‘is difficult to characterize non-circularly’ (Tiffany Reference Tiffany2007, 248), or is inexplicable (see Copp Reference Copp1997), or is incoherent (see Baker Reference Bakerforthcoming).
Many think it is easy to characterize authority: they identify it with some normative feature, like ‘necessary reason-giving force’. So how does the skeptic present a formidable case to the contrary? By presenting two important demands for explanation. The first concerns the normative notions that appear in accounts of authority, like ought simpliciter or reasons simpliciter. Foot wrote: ‘This ‘ought’ – in the sentence ‘One ought to be moral’ – is supposed to be free floating and unsubscripted, and I have never met anyone who could explain the use of the word in such a context’ (Foot Reference Foot1977, 169). Foot’s thought is that we start by taking at face value folk normative discourse, which relativizes normative notions to a plurality of standards. She formalized this by subscripting ‘ought’ et al.: ‘oughtM’ and ‘oughtE’ (the moral ought and the ought of etiquette), ‘reasonsM’ and ‘reasonsL’ (moral and legal reasons), and so on. Then she ask, rhetorically: how could we explain ‘ought’ or ‘reason’ in terms of these starting points, the subscripted notions like ‘oughtM’ or ‘reasonsL’? So the skeptic declares that the concept of authority is empty, inexplicable.
The skeptic’s second demand is meant to show that this concept, even if explicable, is defective. Say we can explain a notion of ought simpliciter or reason simpliciter. How can it play the role that authority is meant to play? Consider the conflict between morality and the pimp norms. As Derek Baker puts the point, ‘the initial conflict resulted from too many oughts, not too few. So how does adding another ought help? Why doesn’t it simply breed more conflicts?’ The thought is that we need to appeal to a normative notion like ought simpliciter to explain how (say) morality is authoritative, in the sense that we ‘ought’ to resolve conflicts between, say, oughtM and oughtE in favor of the former. But then ‘the initial conflict will reappear at the level of which ought we ought to heed’, as ought simpliciter tells us to resolve the conflict in favor of morality while oughtE tells us to resolve it in favor of etiquette. Why go with ought simpliciter? If we say ‘because it is authoritative’ then our attempt to explain authority ‘non-metaphorically makes tacit appeal to the original metaphor’ (see Baker Reference Bakerforthcoming).
What makes the skeptic’s case formidable is that it works equally well for any normative notion that one might appeal to as the feature that we add to ‘generic normativity’ to get our account of substantive normativity. That is, it works equally well in terms of ‘ought’ or ‘reason’ or ‘value’ or ‘fittingness’ or whatever: for any normative notion, we seem to have merely formal and substantively normative instances thereof; so we cannot, it seems, appeal to any such notion as a feature that we add to mere formalities to get our account of substantive normativity.
Some might deny this. You might, for instance, grant that we have reasonsM and reasonsL, but deny that they both involve the same normative notion of reasons. The similarity between the two subscripted notions is superficial, like the similarity between financial banks and river banks: they are utterly different entities. So the presence of reasonsL is no threat to an account of authority in terms of ‘reason-giving force’.
But this move is both ill-motivated and ultimately unhelpful. It is ill-motivated because reasonsM and reasonsL seem very similar: they are considerations that count in favor of, morally or legally speaking; we appeal to them to explain and justify actions and reactive attitudes (see Hart 1961/1994, 56–57, 89–91, 114–116). And it is unhelpful because to take reasonsM and reasonsL to be utterly distinct entities does not explain how they can conflict in the first place: weighing them against each other would be like adding apples and oranges.Footnote 16 How can reasonsM override reasonsL if they cannot be meaningfully compared?
It should be clear why skepticism leaves no room for an explanation of how Kevin could make a normative mistake by prioritizing the wrong standard in practical deliberation about what to do. It may not yet be obvious why the skeptic’s view leaves us without a charitable interpretation of folk and philosophical discourse about ‘non-genuine normativity’. Baker considers claims like ‘etiquette is not really or genuinely normative’, and says that their ‘problem … derives from [their] obscurity’, as they rely on a special sense of really ‘that is only conveyed through italics, scare quotes, or table-thumping’, or on some ‘resort to metaphor (‘grip, ’ ‘normative force’) which remains unpacked’ (Baker Reference Baker, McPherson and Plunkett2016, 577). Baker considers one literal interpretation, but it is confused: ‘In the literal sense, the prudential ought really – that is, just as genuinely and accurately – tells you what to do as any other ought’, like the ought of etiquette (Baker Reference Bakerforthcoming).
Putting this point in terms of the Hamburger Model, we can see why such claims seem puzzling. On one hand, people could be saying that a hamburger is ‘not a real hamburger’: an x is not really an x. This is what Baker meant by the ‘literal’ or, better, the existential sense of ‘real’. And as he notes, in this sense it is clear that oughtM and oughtE both tell you what to do: both are normative in the minimal sense. They are systems of deontic requirements, recommendations, and permissions. On the other hand, people could be saying a hamburger is ‘not a real hamburger with cheese’: an x is not really an x + F. Contra Baker, this does not invoke a special sense of ‘really’; it invokes the familiar sense of ‘really’ as an intensifier or a bar-raiser.Footnote 17 The trouble here is the same one from before: to make sense of how the bar is being raised from some minimal notion of normativity (x) to that notion plus some further feature (x + F).
For instance, consider the view that we are raising the bar from being normative to being normative and reason-involving. This might make sense of claims like ‘legal obligations are not genuine obligations’. But what about claims like ‘legal reasons are not real reasons’? Perhaps one could insist that ‘real’ is just raising the bar from weak to weighty reasons. Now we have a different view: the bar is raised from being normative to being normative and weighty-reason-involving. It’s not clear if this view is much better. In context, it does not offer a charitable interpretation of Enoch:
from there being a legal reason to φ it does not follow that there is a reason to φ. It does follow, of course from the fact that I have a strong reason to φ that I have a reason to φ, because a strong reason is a reason, but a legal reason is not necessarily a reason (Enoch Reference Enoch, Green and Leiter2011, footnote 28).
Enoch is clearly not just asserting that ‘from there being a legal reason to φ is it does not follow that there is a weighty reason to φ’. And more importantly, if we make this move we can no longer explain authority in terms of reasons, as we have granted that reasonsL and reasonsE are instances of reasons simpliciter. We must make sense of what seemed to be a difference in kind (substantive and merely formal normativity) in terms of a difference in degree (weighty and weak reasons); all hamburgers turn out to have cheese, but some are cheesier than others.
This is not a reductio. But it carries some unpalatable commitments. Do pimp norms necessarily give Kevin a (weak) reason to assault women? What about the racist requirements of the law (or etiquette, for that matter) during the antebellum era and Jim Crow? Or the the ruthless rules of omertà? This is an implausible defense of the possibility of authority.Footnote 18
So, once we start with the Hamburger Model, it is difficult to locate an additional normative feature (like reasons) that we could add to mere formalities to generate substantively normative standards. This is ultimately because for any such normative feature, we seem to have merely formal and substantively normative instances thereof. There may well be more to say for proponents of the Hamburger Model. But this is enough to motivate exploring a different route. The key will be to stop focusing on what makes standards like morality special, and focus instead on what makes standards like etiquette ‘not really normative’.
4. Fake ducks and fictionalism
A good place to start is with J. L. Austin’s remarks about a sense of ‘real’ which is typically synonymous with ‘genuine’: ‘a definite sense attaches to the assertion that something is real, a real such-and-such, only in the light of a specific way in which it might be, or might have been, not real’. To illustrate his point, Austin wrote ‘I don’t know just how to take the assertion that it’s a real duck unless I know just what, on that particular occasion, the speaker has it in mind to exclude’ (Austin Reference Austin1962, 70). What would the speaker have in mind to exclude? Fake ducks. This is not because fake ducks don’t exist (so this is not the existential sense of ‘real’). Nor is it because we raised the bar for counting as a duck from looking a certain way to having a certain biology. It would be foolhardy to think that the relation between fake ducks and ducks is like the relation between a hamburger and a hamburger with cheese. We do not start with an account of what it is to be a fake duck then add something to get an account of what it is to be a duck. Rather, we start with an account of what it is to be a duck then understand fake ducks as a non-real version thereof.
This suggests an alternative to the Hamburger Model, ‘the Fake Ducks Model’. On this view, substantive normativity and mere formalities are like ducks and fake ducks. That is, this alternative reverses the order of explanation: we start with the genuine article (substantive normativity), and understand mere formalities to be a fake version thereof.
The fact that the Fake Ducks Model is an attractive alternative to the Hamburger Model is enough to show that the latter should not be assumed without argument. Especially because the Fake Ducks Model makes better sense of our discourse. Think of Enoch’s analogy to friends and imaginary friends. ‘Imaginary friends are not real friends’ is clearly meaningful; indeed, it is clearly true. It does not mean that an x is not an x, or that an x is not an x + F. When ‘imaginary’ modifies friends, it does not carve out a sub-category of friends. Some terminology is helpful here. We can interpret ‘legal reasons’ adjectivally, where ‘legal’ picks out a sub-category of reasons, or we can understand it perspectivally, where the claim means ‘reasons from a legal point of view’ (see especially Shapiro Reference Shapiro2011, 184–188). If ‘legal reasons’ can be understood perspectivally, the commitments of natural law theorists like Finnis and positivists like Enoch and Hart need not be dismissed: ‘legal reasons are not real reasons’ and ‘legal obligations are not genuine obligations’ can be given a more charitable interpretation.
The question is how we can understand such claims perspectivally. A natural option is to understand such claims in terms of the perspective of a fiction, via a fictionalist theory of mere formalities.
The starting point for a fictionalist theory is to note that our uses of ‘real’, ‘really’ and ‘genuinely’ to modify words like ‘obligation’ and ‘reason’ look a lot like our uses of such modifiers in claims about fictions:
Sherlock Holmes is a detective, but he’s not a real detective.
Holmes is a detective, but he’s not genuinely a detective.
Holmes is a better detective than my local sheriff, but he’s not really a better detective [than my local sheriff].
Given their surface similarity, it is worth considering how we explain (1)-(3), and whether it can apply to claims about ‘non-genuine normativity’.
There is much disagreement about the semantics and metaphysics of fictions: philosophers disagree about whether (1)–(3) are true, are implicitly dominated by operators (‘According to the fiction, p’), and whether fictional characters like Sherlock Holmes exist. But there is a core of agreement in their explanations for (1)–(3): speakers accept but do not believe or reflectively avow, say, that ‘Sherlock Holmes is smart detective’, which is why they seem to assert ‘p, but not really p’ (see, inter alia, Sainsbury Reference Sainsbury and Lihoreau2011; Thomasson Reference Thomasson, Kim, Sosa and Rosenkrantz2009). That one asserts but does not reflectively avow p is often taken to be a marker that one accepts a fiction that p without actually believing p.Footnote 19 By extension, perhaps when we accept but do not believe claims about ‘legal reasons’, we can meaningfully say things like ‘legal reasons aren’t real reasons’.
This motivates a simple, schematic fictionalist theory of mere formalities. For our purposes, all we need is an existentially quantified thesis:
Strong Fictionalism
For some standards like etiquette, to take that standard to be a mere formality is to take a fictive attitude towards that standard.
Fictionalism is first and foremost a claim about our attitudes. A full theory would tack on further semantic and metaphysical commitments. There are different options here, modelled on different views about fictions. For simplicity’s sake, I assume that (1)–(3) are typically false and not implicitly dominated by operators. But I want to flag upfront that this there can be non-error-theoretic ways of developing Fictionalism.Footnote 20 In any case, while this fuller version of the theory will do some work for us below, the core claim about our attitudes will do most of the heavy lifting. So it is this core claim about our attitudes that we should focus on.
What does this core claim about our attitudes amount to? Let’s start with an illustrative example. Imagine that a group of children walking in the woods see tree stumps and start to act and speak as if they are bears (see Walton Reference Walton1990). Upon seeing a stump they might point, scream ‘Bear!’, run away in fear, and so on. Intuitively, they created a fiction in which <Those things are bears>, by accepting such propositions. As this illustrates, engaging with fictions is sometimes more creative and active than, say, reading a book.
The key idea for Fictionalism is that one can take etiquette to be a mere formality by taking the same kind of attitude towards the relevant propositions that the children take towards <Those things are bears>. What are relevant propositions? There are, again, different options here.Footnote 21 A simple view would be that they are specifications of the standard’s salient requirements, recommendations, or permissions: <I must put the fork here>, <I should not rest my elbows on the dinner table>, and so on. What is it to take fictive attitudes towards those propositions? Most obviously, it is to accept those propositions; there may well be other fictive attitudes, but we can focus on acceptance to keep things simple.
What is it to accept propositions like <I must put the fork here>? And how is this different from believing such propositions? Much has been said about this. I’ll stick to three ecumenical points that will prove helpful. First, there are significant similarities between accepting that p and believing that p, including that they share very similar motivational profiles.Footnote 22 Believing <That’s a bear> can motivate you to run away, and so can accepting this proposition. Likewise, believing <I must put the fork here> can motivate you to act, so can accepting this proposition. This helpfully explains something noted above: that we appeal to moral and legal reasons in explaining and justifying similar patterns of behavior.
Second, there are significant differences in our linguistic practices when we believe p rather than accept but do not believe p: when we believe <That’s a bear> we are disposed to assert and reflectively avow that this is so, but when we accept but do not believe this proposition we are disposed to make claims like (1)-(3) in certain contexts. This helpfully explains our claims like ‘I must put the fork here, but I don’t really have to do so’.
Third, believing subjects you to stricter norms than accepting. A belief that <That’s a bear> is subject to stricter norms about evidence, coherence, and context-independence; accepting this proposition is not (see, among others, Stalnaker Reference Stalnaker1984, 80–81; Cohen Reference Cohen1995, 13; Bratman Reference Bratman1992). The children can accept that <Those things are bears> on Monday but not Tuesday, do so without evidence that it is true, and do so without revising their beliefs about biology. This is not to say that acceptance floats free of any norms whatsoever: it is sensitive to what’s true in the fiction, but those propositions can be false, or incoherent.Footnote 23 This will be helpful in §VI in showing how fictionalists can respond to the objection that we do not take our own attitudes towards etiquette et al. to be fictive.
So far we’ve unpacked the key idea that one can take etiquette to be a mere formality by taking the same kind of attitude towards the relevant propositions that the children take towards <Those things are bears> . Now let’s return to that illustrative example and see how it can motivate Fictionalism. Imagine the children’s practices evolving in various ways. They sneak up on stumps and tie ropes around them, declaring that this ‘captures’ bears. They act and speak as if they ‘must’ capture bears. They act and speak as if one ‘can’ also untie and thereby release bears. They say they ‘should’ avoid getting close to dangerous untied bears.
The children first created a descriptive fiction in which <Those things are bears> by accepting that this is so. Then they created a normative fiction in which it <We must capture bears>, <We can release bears>, <We ought to avoid untied bears>, by accepting these propositions. In other words, the children created a game with a distinct standard – a system of requirements, permissions, and recommendations. Plausibly, the children take this standard to be a mere formality. After all, in certain contexts they will say things like ‘We have to capture bears, but we don’t really have to do so.’ Moreover, their attitudes towards this standard have all the hallmarks of acceptance: their attitudes towards the normative propositions like <We must capture bears> seem of the exact same kind as their attitudes towards the descriptive propositions like <That is a bear>.
A natural concern here is that the children only have fictive attitudes towards <We must capture bears> because of the underlying descriptive fiction about bears. But on reflection, this concern is ill-founded. If we told the same story without the original descriptive fiction – if it were just a game in which the children must capture tree stumps, and so on – the children’s attitudes would still seem to be of the exact same kind. It is possible to accept purely normative fictions: that is, to accept exclusively normative propositions about what we ought, must, and can do. This suggests that at least one instance of taking a standard to be a mere formality is best understood in terms of taking fictive attitudes towards it. For the children to take the requirements of the Tree Stump game to be mere formalities is for them to take fictive attitudes towards that game.
This get us to our existentially quantified view, Fictionalism. If we accept it, there is some pressure for reasons of parsimony to universalize it:
Strong Fictionalism
For all standards like etiquette, to take that standard to be a mere formality is to take a fictive attitude towards that standard.
But I will ignore that move, as it raises extensional questions that are best left to one side here for the sake of clarity and brevity. To defeat skepticism about the possibility of authority, we will only need Fictionalism to be a viable view about at least some mere formalities: that will suffice to show that some standards can be authoritative over others.
5. Skepticism about authority
Grant that mere formalities are like fake ducks and that taking some standards like etiquette to be mere formalities involves adopting fictive attitudes. Does this undermine the case for skepticism about authority?
Recall the skeptic’s two demands for explanation. How could we explain ‘ought’ or ‘reason’ in terms of ‘oughtM’ or ‘reasonsL’ as starting points? And, how could any ‘ought’ or ‘reason’ explain why we should resolve conflicts between oughtM and oughtE in favor of standards like morality?
I argue that the skeptic’s reasoning goes awry because (1) she inherits the burden of explaining our unsubscripted normative notions, and (2) Fictionalism allows us to easily discharge the second explanatory debt.
Let’s take these points in order. As a complaint against her opponent, the first demand for explanation makes sense if we’re assuming the Hamburger Model. But it makes no sense if we assume the Fake Ducks Model. Unless the skeptic can deny that there are mere formalities, she owes us an account of what it is to be a mere formality. (Moreover, and more pertinently, Footnote 24 unless the skeptic can deny that anyone takes any standards to be mere formalities, she owes us an account of what it is to take a standard to be a mere formality.) And since the Fake Ducks Model reverses the order of explanation, that means that she owes us an account of what it is to be a duck. She doesn’t need to say that ducks are instantiated. But she needs to grant that we have a concept of ducks.
Fictionalism allows us to make this point more precisely by contrasting two types of fictions. Mirroring fictions are where we take one thing to be another according to a fiction: according to a fiction, inanimate pieces of yellow rubber are ducks, like those animate feathered creatures. Original fictions are where we say that according to a fiction there is some entity that is utterly unlike anything else: supernatural fictions about ghosts and faeries might be like this. With mirroring fictions, we say something is a fake x: a fake duck, an imaginary friend, not a genuine detective. With original fictions, we say that the category is fake: Caspar is not a fake ghost; rather, ghosts are fake. If my view of mere formalities is correct, taking a standard to be a mere formality can involve a mirroring fiction, a la fake ducks. So constructing this fiction involves a normative notion we already have. If that’s right, the skeptic also owes us an account of the free-floating notion of ‘ought’ or ‘reason’ that appear in such fictions.
The point above is limited in several respects. For one, it leaves open whether we should understand authority in terms of reasons or values or whatever.Footnote 25 This is a virtue, not a vice: the view is intended to be ecumenical with respect to different metaethical views on this issue. What the view does is remove a path towards a skeptical view about the normative notions that appear in such accounts of authority. And it does so by showing that we do not start with a plurality of subscripted notions like reasonsE, to which we add something to get reasons simpliciter. We start with an unsubscripted notion, from which we derive fake versions thereof.
For another, the point is intended to silence many of the skeptical charges about such notions. If we have a concept of reasons simpliciter and use it in constructing fictions, that concept is not empty or inexplicable. This leaves open the skeptical charge that the concept of reasons simpliciter is defective, like ‘square circles’ (which is not empty or inexplicable, but is self-contradictory). And the skeptic’s second demand for explanation is intended to show that even if there is a non-empty concept like reasons simpliciter, it is defective because it cannot do its job of non-arbitrarily resolving conflicts between reasonsE and reasonsM. So to fully answer the case for skepticism, we need to address this second demand too.
This demand is hard to satisfy by taking reasonsL to be (perhaps weak) instances of reasons simpliciter; that route involves unpalatable commitments about pimp norms, Jim Crow laws, and omertà rules. But it also looks hard to satisfy if we take reasonsL and reasonsM/reasons simpliciter to be utterly different entities, as in that case comparing the two would be adding apples and oranges. So how can we meet the challenge?
Fictionalism gives us an easy solution. It allows us to make meaningful comparisons between reasonsM and reasonsL without taking the unpalatable view that both are classes of reasons simpliciter. Why? While we cannot add apples and oranges, we can draw meaningful comparisons between relevantly similar real and fictional entities. We do so all the time: recall example (3), ‘Sherlock is a better detective than my local sheriff.’
To make this point more precise, consider an example involving weights. If you imagined or drew a one ton turkey, you could intelligibly ask ‘Is this turkey heavier than any real turkey?’, and thereby compare the weights of real and imagined turkeys. Similarly, we can compare the weights of real and fictional reasons. In doing so, we compare reality to a representation (i.e. to a perspective).
There are many ways of answering questions that compare reality to fiction. One involves importing real things into fictions: you import my local sheriff into Conan Doyle’s stories and note that Holmes is cleverer. Another way is privileged in certain contexts. If you were asked to bring a heavy turkey for Thanksgiving and brought your drawing of a one ton turkey, you’d have made a mistake. Any real turkey was heavier than your imagined turkey because only real turkeys have weights. In contexts involving practical deliberation about what to do, you should not import real things into the fiction; you should only compare actual weights. Similarly, while in some contexts we might aptly import reasonsM into a fiction to work out whether they’re stronger than reasonsE, this is the wrong way to compare the two when engaged in practical deliberation about what to do. When the fiction is false, real reasons non-arbitrarily beat fictional reasons by default: only real reasons have weights.Footnote 26
This isn’t an explanation for why morality in particular is authoritative and etiquette is not: I have said nothing about whether moral reasons are real and reasons of etiquette are fictional. Instead, it is an explanation of how it is possible for the notion of authority to be non-empty, explicable, and non-defective. If moral reasons are real and reasons of etiquette are fictional, we can meaningfully compare the two even though moral reasons would non-arbitrarily win by default: so the claim that morality is authoritative in a way that etiquette is not is coherent, and possibly true.
Nor is this an explanation for the authority of any standard y over any standard x. The explanation appeals to Fictionalism, and I have not argued that this view is true of all mere formalities (all the xs). But even if one denies Strong Fictionalism, it may be that a similar pattern of explanation could be told in terms of non-fictional perspectives, Footnote 27 in keeping with the broader commitments of the Fake Ducks Model. And moreover, the test was not whether all claims about authority can be true and explicable; the skeptic’s case was that authority is impossible. I have shown that this case fails because (a) the skeptic also owes us an account of the free-floating normative notions that appear in accounts of authority, and (b) such notions, contra the skeptic, can be used to explain how, say, reasonsE should be overridden when they conflict with reasonsM.
6. ‘But isn’t fictionalism a non-starter’?
There are multiple ways for the skeptic to push back. She might, for instance, argue that Fictionalism entails an error theory about ‘legal reasons’ and ‘legal obligations’, and hence entails that obviously true claims like ‘The U.S. legislature creates legal obligations’ are false.Footnote 28 As I alluded to before, I do not think this is the best place to object to Fictionalism, as it is compatible with the wide variety of positions on the semantics and metaphysics of fictions. For instance, many argue that similar claims like ‘Doyle created Sherlock Holmes’ are true; these claims motivate various realist (non-error-theoretic) views about fiction.Footnote 29
The better place for the skeptic to push back, then, will be against the core of Fictionalism, which is a claim about our attitudes: that what goes on when someone takes, say, etiquette to be a mere formality is the same kind of thing that goes on when a child plays a game of make-believe in which tree stumps are bears. And there is an obvious objection here. The fictionalist’s hypothesis about our attitudes is surprising. We do not typically think of our attitudes towards etiquette in the same way that we think of our attitudes towards Harry Potter. When we read Harry Potter we have a kind of metacognitive recognition that our attitudes are fictive. Why don’t we have that same kind of metacognitive recognition about our attitudes towards standards like etiquette? Without a good answer to that question, Fictionalism may seem to be a non-starter against the skeptic.
This is a specific instance of a general objection to (hermeneutic) fictionalisms. For instance, here’s Jason Stanley making a similar point:
If the hermeneutic fictionalist is correct, then x can bear the propositional attitude of pretense toward a proposition, without it being in principle accessible to x that x bears the propositional attitude of pretense towards that proposition. But this introduces a novel and quite drastic form of failure of first-person authority over one’s own mental states (Stanley Reference Stanley, French and Wettstein2001, 46).Footnote 30
Stanley’s version of the objection may seem more forceful, but is less plausible. There are arguments for fictionalism, so it is in principle accessible to us that we have fictive attitudes (acceptance, pretense). We can follow these arguments and justifiably infer that our attitudes are fictive. By contrast, the objection I sketched does not allege that metacognitive recognition is impossible. It asks fictionalists to explain why metacognitive recognition is immediate with literary fictions, but so difficult with our attitudes towards mere formalities. This comparison is what makes the objection to (hermeneutic) fictionalism seem damning.
There are a few things that the fictionalist can say to soften the blow here. She can say that her view does not require everyone to take fictive attitudes towards etiquette; Santa Claus is fictional even if some children believe he exists, because enough people accept the relevant fiction.Footnote 31 This still requires the fictionalist to explain why those of us who do accept the relevant fiction fail to recognize this. Here, she can note that failures of metacognitive recognition are not rare. Metacognitive knowledge is not immediate and infallible; the nature and content of our own mental states is often opaque (see, e.g. Williamson Reference Williamson2000). And knowing whether our attitudes are fictive is often an achievement, requiring rare theoretical reflection (see Joyce Reference Joyce2001, 292). This is especially true of accepting that p and believing that p, since they share so many important similarities, including in their motivational profiles.
Moreover, among the philosophers of law who have engaged in rare theoretical reflection about the attitudes expressed in statements about what people ‘legally ought to do’, many have been independently attracted to the view that (many) such statements express fictive attitudes, and/or are best understood in terms of implicit fictional operators.Footnote 32 No putative paradigm case of a mere formality has been subjected to the same degree of scrutiny as legal normativity. So the fact that many philosophers of law take some form of fictionalism seriously for independent reasons suggests that Fictionalism should not be dismissed because others who have not engaged in such reflection don’t find it immediately obvious.
Beyond softening the blow, fictionalists can and should do more to answer the objection by explaining the comparison upon which it rests. Why do we readily recognize that our attitudes towards most fictions are fictive, but not that our attitudes towards mere formalities like etiquette are fictive? This is where two final responses to the objection are helpful.
First, metacognitive recognition that our attitudes are fictive is harder to come by when fictive attitudes are interspersed with garden variety beliefs about the exact same propositions. Consider historical fictions. I believe the proposition:<According to Saramago’s The History of the Siege of Lisbon <The Portuguese laid siege to the Moorish city of Lisbon>>: it is true, because the embedded proposition is true according to the fiction. But the unembedded proposition <The Portuguese laid siege to the Moorish city of Lisbon> is also true. So part of the content of the fiction is true. And when I read Saramago’s book, there are many propositions that I believe and know to be true, yet I accept them all the same.Footnote 33
Why would the normative fictions we are concerned with be like historical fictions in this respect? Say that we take etiquette to be a mere formality, and take reasons of etiquette to correspond to reasons to do ‘what’s done’ according to a fiction. We might think that heeding local norms of etiquette is often important in order to, among other things, communicate respect (see Buss Reference Buss1999; Calhoun Reference Calhoun2000). So there may often be good (moral) reasons to do what etiquette requires. When etiquette requires us to RSVP in the third person, we might accept a fiction according to which <There is reason to RSVP in the third person>, while also believing that <There is reason to RSVP in the third person>. And this belief might be true, partly because of the fiction.
Crucially, if our fictive attitudes are interspersed with beliefs about the same propositions, this will make it harder to recognize that our own attitudes are fictive. For one thing, if we were to engage in uncommon reflection about our attitudes towards etiquette, we would not come to believe that all of those attitudes are fictive because they are not all fictive. Some are garden variety beliefs. For another, if we were to engage in such reflection, it will be harder to detect that any of our attitudes are fictive because the markers of fictive attitudes will be masked. When I believe and accept that <The Portuguese laid siege to the Moorish city of Lisbon>, I will assert and reflectively avow that this is so. The same holds when I believe and accept that <There is reason to RSVP in the third person>. And it holds for each marker that distinguishes belief from acceptance.
Of course, if mere formalities are like historical fictions, it is still easier to recognize that our attitudes towards the latter are fictive. Why is that? This takes us to the second response: failures of metacognitive recognition are especially likely for some standards that we take to be mere formalities because we rarely enter or exit their jurisdiction. Plays and films begin when the curtains raise and end when the lights come back on. This matters because a key marker of fictive attitudes is their context-dependence. When the fiction has clear entry and exit points, fictive attitudes do too; the distinctiveness of the attitudes is made stark and salient when the curtain drops and we return to reality. Plausibly, this is part of how we recognize that our attitudes to, say, Saramago’s novel are fictive. After we close the book our beliefs stay put, while fictive acceptance shortly dissipates.
Some standards that we take to be mere formalities are like plays. Even the most creative children enter and exit games of make-believe, taking respite in reality in between. But others are not. Imagine someone who takes the local etiquette norms to be a mere formality but never left her community. She never entered or exited etiquette’s jurisdiction, so the context-dependence of her fictive attitudes would not be made salient. A similar point applies to a jet-setter who constantly jumps between different etiquette norms. If she takes etiquette to be a mere formality, no change in context ever prompts the sudden absence of her fictive attitudes to etiquette-related propositions. She would be like a child constantly jumping between games of make-believe: the contents of her fictive attitudes change, but the attitudes themselves never abruptly drop away.
Putting these two responses together, we can predict that failures of metacognitive recognition will be more common and explicable to the extent that (1) we accept and believe propositions like <There is reason to RSVP in the third person>, because we take the mere formality to bear important connections to substantive normativity; and (2) our fictive attitudes rarely drop away, because we rarely exit the jurisdiction of the relevant mere formality. If this is right, and I think it is, we can explain how Fictionalism about mere formalities can be plausible about standards like etiquette despite resting upon a surprising hypothesis about the nature of our attitudes. More generally, these responses illustrate how fictionalists can appeal to resources drawn from thinking about fictions to respond to seemingly damning objections.Footnote 34
7. Conclusion
Since Foot’s ‘Morality as a System of Hypothetical Imperatives’, philosophers have argued about authority while largely taking mere formalities for granted: once we know we are not dealing with the genuine article, it is cast aside, ignored. This was a significant mistake. Mere formalities are interesting in their own right: they are not just minimally normative standards that lack some feature (authority); rather, they have distinctive features that warrant explanation (like being in some sense fake). And focusing on that feature helps keep the barbarians at the gate by explaining how authority is possible and explicable.
Moreover, by focusing on that feature we can explain common claims in metaethics, philosophy of law, and folk normative discourse. Metaethicists say that etiquette isn’t genuinely normative; philosophers of law say that legal reasons aren’t real reasons; the folk say that they have to do something, but don’t really have to do it. There’s a meaningful interpretation of this discourse in terms of perspectives, and, in particular, fictional perspectives. That interpretation rests on a hypothesis about our attitudes that is surprising. But it is a defensible hypothesis all the same.
