1. Introduction
The point of this paper is to draw attention to an important feature of neo-pragmatism as a general philosophical method, and to appeal to this feature in solving a problem that challenges both expressivist and cognitivist accounts of moral thought and talk. The important feature of neo-pragmatism is its focus on language as a social practice, in contrast to the derivative issue of what a given individual can manage to do with a token assertion. The problem is explaining how people with different moral views can be understood to be talking to each other, rather than past each other. In the context of expressivism, this is what Tristram McPherson and Gunnar Björnsson call ‘The Specification Problem’: the problem of isolating a specific attitude that is expressed by the use of a given normative term (Björnsson and McPherson Reference Björnsson and McPherson2014). In the context of cognitivism, it is the problem of isolating a specific meaning that underwrites genuine disagreement while at the same time meeting what Laura Schroeter calls ‘The Flexibility Constraint’: the need to accommodate the fact that different speakers will have very different criteria for the use of the relevant normative term (Schroeter Reference Schroeter2012; See also Schroeter and Schroeter (Reference Schroeter and Schroeter2014, 19)).
Björnsson and McPherson offer an expressivist-friendly solution to the problem, while Laura and François Schroeter offer a more representationalist-friendly solution. I will agree with many elements of both solutions: for example, the increased emphasis on the role of the social. Moreover, I will agree with Björnsson and McPherson (as against the Schroeters) that no substantial notion of reference is required in order to make sense of moral discourse. And I will agree with the Schroeters (as against Björnsson and McPherson) that we do not have to know virtually anything about the attitudes of two speakers in order to be justified in taking them to be involved in a genuine moral dispute. The resulting neo-pragmatist co-option of what is correct in both Björnsson and McPherson's view and in the view defended by the Schroeters yields something both more general and more attractive. But before getting to that, let me explain what I take neo-pragmatism to be.
2. Neo-pragmatism
Neo-pragmatism is an approach to philosophical questions that have traditionally seemed to require us to do some metaphysics. We start out wondering what, for example, numbers or values might be, and the metaphysician calls out to us with various proposals. Really good metaphysicians manage to come up with proposals that save most of the first-order phenomena. And a subset of those metaphysicians offer proposals that identify the target of our worries with something that we are not similarly worried about. Worried about numbers? They’re just sets, no different really than sets of chairs or spoons. Worried about normative reasons? They’re just our desires. Worried about possibilities? They’re just stories. What all these proposals share is a presupposition about what the best strategy for addressing our worries should look like. It should look like an identification of some entity or property that lies at the non-linguistic end of a reference relation. This is a very significant presupposition, and one of the neo-pragmatist's defining features is the acceptance of a picture that denies it.
What is the neo-pragmatist's contrasting picture? It is one that highlights the overt, observable aspects of our practices with sounds, marks, and signs, analogizing them with games that make use of physical pieces. One useful aspect of this analogy is that there is little temptation to think of pieces in games in terms of reference relations to anything. The neo-pragmatist picture also highlights the ways in which our linguistic practices are continuous with similar practices that other animals engage in. For example, consider the dances of bees. It is not hard to imagine a story that explains how this sort of complex information-transmitting behavior evolved, without having to appeal, at any level, to reference relations or to any other semantic notions, or even to the idea that the dances express a certain mental state. Of course, analogies with bees are unlikely even to seem relevant to someone with antecedent representationalist views about how human language works. So let me now develop an example that involves animals more closely related to human beings: non-linguistic primates.
Consider the transmission of disgust reactions in groups of non-linguistic primates.Footnote 1 Disgust is a contagious response. If one animal exhibits disgust at a certain substance for what we might call ‘primary reasons’ – reasons that have to do with actual physical contact with a toxic or infectious substance – a second animal who witness this reaction will acquire a disposition to be disgusted by that same kind of thing, even without the same sort of physical contact. And later on, when that second animal manifests disgust at that same kind of item, a third animal will acquire a similar disposition. Nothing in this explanation requires appeal to semantic notions, or even an appeal to the notion of belief. Moreover, it is easy to see the usefulness of the mechanism here, since it allows groups of primates to acquire the disposition to avoid certain dangerous substances, without each of them having to have a dangerous form of contact with it.
As it happens, disgust is hard to unlearn. But it would in some cases be useful for a community to unlearn it. Suppose, for example, that the initial expression of disgust at a certain item had nothing to do with the properties of that item; rather, suppose that a causally independent illness just happened to reach its nauseating peak soon after the animal put a bit of genuinely healthy fruit in its mouth. If the resulting disgust propagates through the community with nothing to check it, then this sort of food will be needlessly avoided. So let me now indulge in some primatological fiction. Suppose that there is a distinct response to expressions of disgust – distinct, that is, from the response of acquiring a disposition to be disgusted by the same sorts of items. This distinct response is manifested at expressions of disgust when they are directed at items with which the witnessing animal has had direct contact, and which have proved to be tasty and energy-providing. Call this second-order emotion ‘progust.’ According to my fictional account, an animal whose disgust mechanism has been calibrated to be triggered by a certain substance can have it reset by exposure to an animal that exhibits progust in response to observed disgust. The paired disgust and progust reactions together serve to get a community of primates to converge on a more useful uniformity of response-dispositions. As will emerge later, the usefulness of this sort of uniformity plays a crucial role, for the neo-pragmatist, in explaining the emergence of semantic terms such as ‘true’ and ‘false’. Indeed, it may be useful to think of expressions of disgust as corresponding to a claim – ‘That's not to be eaten’ – and to think of expressions of progust as corresponding to an assessment of that first claim as false.
According to the neo-pragmatist, we should think of human linguistic practices on analogy with the dances of bees and expressions of disgust and progust. Of course human language is much richer than either of these examples. But we humans are nevertheless animals; our behavioral dispositions, including our linguistic capacities, are amenable to biological-evolutionary and cultural-evolutionary explanations. If we can sketch such explanations of the linguistic practices in which we make the assertions that raised the mystifying questions about the nature of numbers, possibilities, and values, then we should cease to be mystified about the objects of these sorts of assertions. That is, we should abandon the metaphysics of mathematics, possibility, and value, replacing them with explanations of the ways in which we use number language, modal language, and evaluative language.Footnote 2
Against the neo-pragmatist program I’ve just described, those who are partial to more traditional metaphysical methods might initially make the following protest: ‘It's all well and good to explain how we came to say such things as “There are three positive even numbers less than six”, or “It is simply not possible that I could have been born as a pine tree”. But, however it happens that we utter these sentences, some turn out to be true, and others to be false, and we need to do some metaphysics in order to understand what the difference is.’ But this protest simply assumes what the neo-pragmatist denies: that truth and falsity are robust properties, and that they to be explained in terms of relations – including reference relations – between assertions and the entities and properties that those assertions are about. If this sort of account of truth and falsity were correct, then metaphysics might well be required to explain difference between true and false propositions. But the neo-pragmatist will favor deflationary accounts of truth and falsity.Footnote 3 That is, the neo-pragmatist will explain our use of the words ‘true’ and ‘false’ in a way that does not depend on appeal to a property of truth or falsity.
The account of truth-talk that I favor is one that has been developed and defended by Huw Price (1988, 2003). His idea is that for some behavioral dispositional states it is useful if a community converges in certain ways. For example, it is useful if everyone is disposed to count in the same way, since then the words used in counting can serve all sorts of useful functions. As a result, it is also useful to have terms such as ‘false’ or ‘wrong’ that function to correct people when their verbal behavior deviates from the useful counting pattern. And those same words – ‘false’, ‘wrong’, and so on – can be used in response to an open-ended class of other linguistic performances: ones we might call ‘assertions about possibility’, ‘assertions about the moral status of an act’, and ‘assertions about the locations of objects the physical environment’. But according to neo-pragmatism, the common assertoric form – shared whether one is talking about numbers, values, possibilities, or moral status – does not owe its existence to the fact that it allows for a truth-constituting correspondence between parts of sentences – or the propositions they express – and parts of reality. Rather, the explanation of the shared assertoric form is that (a) it invites the use of words like ‘true’ and ‘false’, and that (b) one common feature of number-talk, possibility-talk, and moral-talk is that it is useful if a community of speakers converge in certain ways with respect to the assertions its members are prepared to make in those arenas. Of course, as these examples makes clear, mere convergence is not typically the useful thing: rather, certain foci of convergence will be especially useful.
One important kind of assertion is a description of the physical environment. Truth-assessments of these assertions can, it is true, be seen as responsive to a kind of correspondence between, on the one hand, words for things and properties and, on the other hand, those very things and properties. But it would be a mistake to take this one case – one in which a kind of causal/historical account of reference is plausibly extensionally adequate – as a model for all reference or representation. Rather, the neo-pragmatist offers a completely general explanation of our use of ‘true’ and ‘false’ (and other semantic terms) that focuses on establishing certain sorts of harmony in the states that underlie our counting things, our assessing possibilities, our evaluations of things, and our descriptions of the physical environment. In the latter case it turns out that the useful harmony is one in which the psychological state involves some causally mediated relations to the physical world. But even in that case the causal relations do not constitute the relation of reference or representation, since there is no such relation. One way of putting this point is that the causal/historical theory provides a good account of the reference of certain sorts of terms (names for physical objects and properties), but not of the concept of reference in general. By way of analogy, ‘ants and grubs’ is a good description of the diet of certain sorts of animals, but does nothing to clarify the concept ‘diet.’
Against the view just described, a common idea among representationalists is that we should take truth-assessments of descriptions of the physical world to be making use of a notion of genuine truth, and that while truth-talk has a home in domains such as the normative or the modal, its function there is a derivative one. On this view, there is a radical difference between genuine truth-assessments and what we might call assessments of quasi-truth. The former paradigmatically apply to descriptions of the physical environment, while the latter apply to linguistic performances that have the outward form of assertion, but that actually serve to express some kind of non-cognitive attitude. But that is not the right way to think of truth-assessments, according to neo-pragmatism. Rather, according to neo-pragmatism, all truth-assessments are in a very important sense in the same business: helping to push a community of speakers into a useful harmony.
Given the above explanation of truth-talk, the similarity between assertions in virtue of which they all count as assertions is a very superficial one, and tells us nothing about their function. Following Wittgenstein, Price likens the similarity in the superficial grammatical form shared by all assertions to the similarity in the form of the various different handles one might find in the control room of a train.Footnote 4 What makes the handles look so similar is that they are designed to be grasped by a human hand. But what that hand does with them – pumping, putting pressure, switching from an on to an off position, rotating – can vary, as of course can the whole mechanism to which the handle is connected, and (therefore) the function of the handle in the overall operation of the train. The fact that a given handle is hand-sized and conveniently shaped tells us absolutely nothing about what the handle does or how to operate it. Similarly, Price argues that the assertoric form in human languages allows for the kind of pressures that truth-assessments provide: to put it in terms of the analogy with handles, the assertoric form allows assessments of truth and falsity to ‘get a grip’ on the attitudes of speakers. Because of this grip, they help push a community of speakers into a useful uniformity of attitude. But those attitudes are as various as the mechanisms to which the handles in a control room might be connected.
In explaining the relevant sense of harmony we should think of it as related to success in social action. If we all count in the same way, for example, this allows number words to function in their characteristic ways in requests, orders, descriptions, and so on. And if we are in overwhelming agreement in terms of what sorts of behavior lay one open to various sort of sanction, this allows us to go about our daily lives without having to take so many costly precautions to protect us from harm at the hands of other people. Also, of course, agreement in the states that find expression in verbal descriptions of the locations of things in physical space is hugely helpful. But the notions of truth and falsity that apply to assertions in all three domains – the arithmetical, the moral, and the physical – are the same. Roughly put, calling an assertion that p ‘true’ only when one could also sincerely assert that p, and calling it ‘false’ only when one could also sincerely assert that not p helps bring about the relevant harmony in the behaviors connected with number-talk, moral-talk, physical object-talk, and so on.
According to the neo-pragmatism I am defending, beliefs are functional states in a vast psychological architecture that includes all sorts of sensory input, and that has behavior – including verbal behavior – as output. But that characterization doesn’t differentiate beliefs from sensations, desires, and so on. What beliefs share is the following: they are states that have ‘characteristic assertions’ that are subject to our practice of truth-assessment because of the usefulness of those practices in the domains of discourse to which the characteristic assertions belong. Moreover, the usefulness of truth-assessments depends on the usefulness of there being a convergence on beliefs within a linguistic community, which means that a given assertion cannot be both true and false. This characterization helps explain why belief is what Mark Schroeder calls an ‘inconsistency-transmitting’ attitude (Schroeder Reference Schroeder2008). That is, it explains why it is problematic to hold beliefs that have inconsistent contents, in contrast to the perfectly reasonable case of wondering whether p while also wondering whether not-p. And the neo-pragmatist account of belief does this despite the fact that belief is not, in an important sense, a homogeneous category. Rather, an attitude counts as a belief if it can find expression in assertion. And what marks off assertions from questions, commands, greetings, and so on, is that they are offered up in a form that invites truth-assessments. So the neo-pragmatist does not explain why belief is inconsistency-transmitting by reference to a supposedly unified metaphysical nature of belief. Rather, what does the work is the practical usefulness of truth-assessments: assessments that presupposes that – of a pair of conflicting assertions – only one can be called ‘true’.
It should be stressed that the usefulness of our practice of assertion and truth-assessment is not to be found in token cases. Making a correct assertion might well do nothing but harm; so too might informing someone else that he has spoken falsely. Rather, the benefits of having the assertions and truth-assessments can be seen when we consider whole domains of discourse, and when our view is long-term and social.
3. The specification problem
Let me now explain how the sort of neo-pragmatism I’ve just presented sidesteps a problem that arises both for expressivists and for more traditional normative realists. Expressivists take normative claims to express non-cognitive attitudes, rather than beliefs. Moreover, for each normative predicate, expressivists have tended to assume that there is some specific non-cognitive attitude it expresses, distinct from the attitude expressed by other normative predicates. Gunnar Björnsson and Tristram McPherson note this tendency and a certain pressure towards it (Björnsson and McPherson, Reference Björnsson and McPherson2014). The pressure stems from the need to explain what a specifically moral disagreement is, and why there need not be a similar disagreement between someone who thinks a certain policy is not morally wrong and someone who thinks it is stupid. This assumption of univocality in expressed attitude gives rise to a significant problem for expressivists, which takes the form of a dilemma. To the degree that the attitude we propose as moral disapproval is specific – for example, a certain type of anger based in specific aspects of its target, combined with a distinct form of approval for punishment – it looks like it will not be held by some people who make sincere and competent moral claims that a certain action is immoral. But to the degree that we make it more general – making it more like plain old disapproval – we lose the capacity to distinguish claims about moral wrongness from claims about stupidity or bad taste.Footnote 5
The dilemma I’ve just described for expressivists might seem to press for a cognitivist account of moral judgment. But, as Björnsson and McPherson stress, it turns out that cognitivist accounts face quite a similar dilemma, at least if they rely on a substantial notion of reference. This dilemma is central to the well-known Moral Twin Earth arguments developed by Terry Horgan and Mark Timmons (Reference Horgan, Timmons and Heil1991). Such arguments ask the cognitivist for her theory of reference and then cook up a version of Moral Twin Earth according to which the seemingly moral language of the Twin Earthlings ends up referring to a property distinct from the property of moral wrongness to which we refer here on Actual Earth. As a result, Earthlings and Twin Earthlings will – according to the cognitivist – be talking past each other when they get into what looks like a moral dispute. But many theorists have the strong sense that this is not right: that Earthlings and Twin Earthlings are capable of having genuine moral disagreements. So it looks like cognitivists and expressivists alike will have difficulties specifying exactly what it is in the semantics of moral vocabulary that explains when we have a genuine moral disagreement, and when we don’t.
If the specification problem afflicts both cognitivists and expressivists, what are we to do? In this section I will examine two attempts to deal with the issue. The first is developed by Björnsson and McPherson themselves, and is a version of non-cognitivism. The second is developed by Laura and François Schroeter, and is a version of cognitivism. Both solutions shift the focus of theorizing from individuals to communities. From the neo-pragmatist perspective this is, of course, a promising move. But I will argue that both Björnsson and McPherson on the one hand, and the Schroeters on the other, do not distance themselves sufficiently from the initial commitments of the more individualistic versions of their views. In particular, Björnsson and McPherson give a central role to a very specific attitude: paradigmatic moral judgment. And the Schroeters continue to rely on a substantive notion of reference. As I will argue, these commitments are unnecessary and constraining. When they are jettisoned, the result is a neo-pragmatic solution to the specification problem that has all the virtues of their solutions, without their liabilities.
3.1. Björnsson and McPherson's non-cognitivism
Although Björnsson and McPherson offer a solution to the specification problem on behalf of the non-cognitivist, they are certainly not traditional non-cognitivists. One of their central departures from traditional views comes in the form of a denial ‘that every party to a genuine moral disagreement must share the same type of noncognitive attitude’ (Björnsson and McPherson, Reference Björnsson and McPherson2014, 3). What makes them count as non-cognitivists, despite this disavowal, is that a certain kind of noncognitive attitude nevertheless plays a central role in their account. They call this attitude a ‘paradigmatic wrongness judgment, ’ and they think that ‘nearly all of our experience with what we think of as wrongness-judgements’ are constituted by encounters with this specific attitude (Björnsson and McPherson, Reference Björnsson and McPherson2014, 13). Paradigmatic wrongness judgments are characterized by five features. The first three are shared by other sorts of negative normative judgments, such as aesthetic ones, or judgments of irrationality. These include (1) aversion to the objects of the judgments, (2) person-level acceptance (this sort of acceptance distinguishes them from mere appearances, which one might regard as false or misleading), and (3) a disposition to engage with others when there are conflicts in such attitudes. There are, however, two distinctive features of paradigmatic moral judgments. First, they are grounded in a perception or belief that the target has intentionally harmed someone or violated some boundaries crucial to social cooperation. Second, they include a syndrome called ‘social hostility’. This includes hostility not only towards those whom we judge to have acted wrongly, but also towards those who give expression to moral judgments that conflict with our own. Björnsson and McPherson argue that the five features of paradigmatic wrongness judgments explain why those who make such judgments will be quite accommodating in interpreting other speakers as appropriate targets for engagement in moral discussion. This allows their account of moral disagreement to accommodate quite a divergence in the attitudes expressed by two disputants (Björnsson and McPherson, Reference Björnsson and McPherson2014, 17).
What I want to argue in this section is that Björnsson and McPherson make an assumption that they need not make. And when we reject that assumption, what we get is not best viewed as a ‘non-cognitivist solution to the specification problem’, as they describe their own proposal (Björnsson and McPherson, Reference Björnsson and McPherson2014, 21). Rather, we get a neo-pragmatist solution. The assumption with which I am concerned is contained in the very name of the problem: the specification problem. This label suggests that what we need to look for is some specific thing – either an attitude or a property – that both (a) plays a special role in explain the meaning of a moral term and (b) at least commonly plays that role in individual uses of that term. This way of viewing the difficulty shares too much with approaches to specification problem that view it as a challenge to find the right semantic values for moral terms.
Admittedly, one might try to solve the specification problem by reference to a specific attitude without taking that attitude to play the role of a semantic value. Björnsson and McPherson might well describe their own project in this way. Nevertheless, their focus on a specific attitude that can form part of the psychology of a single person does not give sufficient weight to the fact that language – including, of course, moral language – is a social practice. As such a practice, it can serve its purpose even if most of the individuals who take part in it do not manifest the same attitude by uttering – sincerely and competently – the same sentence. Of course, in a trivial sense, they do express the same attitude: belief in the proposition expressed by the sentence. But for the neo-pragmatist, the notions of belief and proposition are to be understood in a deflationary way, and cannot be deployed in explanations of the meanings of utterances.
How will the neo-pragmatist think of moral talk, in order to avoid Björnsson and McPherson's idea that there is a specific non-cognitive attitude that is the psychological state expressed by paradigmatic moral judgment? One way is to think of moral discourse on analogy with a multi-player game. Typically the point of making a move in a game is not to express a mental state. Similarly, on the neo-pragmatist view, the point of uttering a sentence is not to express a mental state; it is to make an overt move in a social practice.
What, then, is the neo-pragmatist's solution to the specification problem? It is that we should count two people as having conflicting moral beliefs if they are in a position to make (roughly: if they are disposed to make) conflicting sincere moral assertions. If they actually address such assertions to each other then they not only have conflicting views, but are having a moral disagreement. By ‘conflicting moral assertions’ here is meant something quite superficial; not precisely grammatical, but almost.Footnote 6 I say ‘almost’, because seeing that a pair of assertions conflict sometimes requires one to employ the linguistic skills involved in following anaphoric chains, in disambiguating homophonic lexical items, in using context to fill in ellipses, and so on. Still, if the relevant claims are made by people speaking the same language, we can typically see that claims conflict just in virtue of their form.
Of course, two people need not speak the same language in order to have conflicting moral beliefs, and one might worry that the specification problem re-emerges for the neo-pragmatist at the level of linguistic practices taken as wholes. How, that is, do we know that two distinct linguistic communities have the same practice – for example, one that we might call, in our own case, ‘moral discourse’ – so that when they meet their members can be said to disagree (or agree) about a common subject matter? Indeed, once this inter-community question is raised, even the intra-community issue might seem problematic. How do we know that two speakers are making assertions that are both part of the same linguistic practice, and not simply using homonyms? There are a number of answers to these questions.
The first answer begins by noting that the things the neo-pragmatist is comparing – linguistic practices – are quite different from what expressivists and robust-referential-realists are comparing. The standard expressivist (not Björnsson and McPherson) needs to assure herself that two disputants, in a given token discussion, are using the moral term at the center of the dispute to express the very same non-cognitive attitude. How on Earth – or Twin Earth, for that matter – can one have any reason to believe this? Attitudes of the complex sort that expressivists typically assign to moral assertions are not visible.Footnote 7 Similarly, the robust-referential-realist needs to assure herself that somehow or other the moral term, as it emerges from the mouth of each of the disputants, is linked by the reference relation to the very same property. This is even trickier than the problem faced by the expressivist, since the reference relation is not only invisible, but mysterious and complicated as well.Footnote 8 Indeed, its very existence is a matter of dispute.
The neo-pragmatist, in contrast with the standard expressivist and realist, is comparing easily observable public practices. The question as to whether two people, raised in the same linguistic community, are engaged in the same linguistic practice when they are arguing over the moral status of some kind of act is not typically more difficult to answer than the question of whether each of the twenty-two players currently running around on the soccer field is playing the same game. It is not relevant here to protest that the only reason we find this latter question so easy is that we are familiar with soccer and similar games, and that an alien ‘plunked down’ into the soccer stadium would perhaps think that each player was simply doing her own thing. That is correct, but irrelevant; the neo-pragmatist is allowed to appeal to our having been inducted into various practices – both recreational and linguistic – since all he is attempting to explain is why we treat certain assertions as conflicting with others.
Even when we consider two completely isolated linguistic communities – as in Moral Twin Earth thought experiments – it is not hard to see how we might come to the justified conclusion that they each have their own version of what is basically the same linguistic practice. Indeed, the possibility of identifying a linguistic practice as one that concerns morality in particular is presupposed by the thought experiments that are used in order to get the specification problem going for both expressivists and realists in the first place. This is true even of Björnsson and McPherson. The recipe they use in order to generate counterexamples to the standard versions of expressivism and realism ask us to imagine communities in which the
word ‘wrong’ functions much as our word ‘wrong’ does: as a term of criticism that plays a distinctive role in deliberation and social coordination. The term is used to characterize a very similar range of actions, and there are very similar controversies about its proper extension. (Björnsson and McPherson, Reference Björnsson and McPherson2014, 8)
Their appeal to ‘a distinctive role’ here reveals that Björnsson and McPherson are not worried about identifying the role that moral vocabulary plays in our social practices. They would not, after all, have been happy talking about ‘a distinctive attitude’ expressed by our moral judgments, or ‘a distinctive property of actions’ referred to by our moral terms.
Björnsson and McPherson's use of the phrase ‘a distinctive role’ suggests that they are assuming that we know that role when we see it. And I think this assumption is not particularly problematic, since we can see it, unlike invisible attitudes, reference relations, or abstract referents. This capacity to note the role a term plays need not be deployed to recognize genuine disputes within a linguistic community. The reason for this is that the default assumption – the default way of playing the game – is that two speakers who make superficially contradictory moral claims in the same language are disagreeing; that they are using a term that has a shared role and a univocal meaning in the language.Footnote 9 The capacity to note the role of a term in a language comes into play primarily when we consider translation, or thought experiments involving distinct linguistic communities: whether English speakers and Spanish speakers, or English speakers and twin-English speakers.
The neo-pragmatist focus on linguistic practices as social phenomena explains the truth of Björnsson and McPherson's claim that
intuitions about agreement and disagreement in wrongness-judgements seemingly require neither specific substantive assumptions about cognitive content, nor specific substantive assumptions about non-cognitive attitudes. (Björnsson and McPherson, Reference Björnsson and McPherson2014, 12)
But in an important contrast with Björnsson and McPherson, the neo-pragmatist emphasizes the role of language. Björnsson and McPherson often write as if the important things, in explaining particular cases of agreement and disagreement, are the psychological states of a pair of judgers, rather than the claims they do or would make. Consider their suggestion that we should think of moral judgments as functional kinds, on the model of hearts or automobiles. On this view moral judgments will include, in the first instance, paradigmatic moral judgments and, also, various sorts of deviations from that paradigmatic case. In offering this suggestion they claim to be explaining our ‘rough tendencies to treat various attitudes as wrongness-judgements’ even when they lack certain paradigmatic features – just as we classify dead hearts as hearts and wheel-free cars as cars (Björnsson and McPherson, Reference Björnsson and McPherson2014, 28). There is no mention of language here, and one gets the very strong impression that Björnsson and McPherson understand moral assertions derivatively, as expressions of moral judgments, which are doing the real explanatory work.
One way of seeing the plausibility of the neo-pragmatist's linguistic re-casting and co-option of Björnsson and McPherson's account is to note certain suspicious elements in their own explanation of the nature of paradigmatic moral judgments. Is it really true that someone who makes a moral judgment is even typically disposed to engage with the moral judgments of others, as they claim? Moreover, the distinctive grounds of paradigmatic wrongness judgments simply don’t seem essential to a judgment being a moral one – though many of us think they are essential to their being true. Recall, the distinctive grounds are, according to Björnsson and McPherson,
intentionally harming or risking harm to others or things that they care about, and failing to respect certain boundaries that play a central role in sustaining social cooperation. (Björnsson and McPherson, Reference Björnsson and McPherson2014, 15)
But there are loads of moral judgments about sexual matters that do not fit this pattern, and they need not be regarded as inviting divine smiting either, which might – depending on the form that smiting took – put an end to social cooperation. Some people seem simply to think that masturbation, pre-marital sex, and homosexuality are wrong. What explains Björnsson and McPherson's controversial claims about the nature of paradigmatic moral judgment? It is that when one looks for the shared features of even paradigmatic moral judgments understood as attitudes, one is bound to distort the phenomena. What really does the unifying is to be found at the level of linguistic practice.
One response to worries about the specificity of the attitude Björnsson and McPherson describe might be a reminder that what they are concerned with is not what is typical, but what is paradigmatic. That is, they could claim that moral judgments based in purely sexual matters are non-paradigmatic. But this is to ignore their claim that paradigmatic moral judgments constitute ‘nearly all of our experience with what we think of as wrongness-judgements’? (Björnsson and McPherson, Reference Björnsson and McPherson2014, 13). Moreover, even overlooking this fact, the worry would then be that they are using ‘paradigmatic’ in a technical way that robs it of explanatory power. Of course some people make judgments with the five features they list. But why is that meant to be important for understanding what moral judgment is supposed actually to be, if loads of people make non-paradigmatic ones? This will be especially pressing if we want to hold that two people, each of whom makes quite differently-constituted non-paradigmatic moral judgments, can nevertheless be rightly regarded as having a genuine dispute.
What Björnsson and McPherson do seem to me to show is that it is no surprise that many people who make moral judgments will be quite liberal in interpreting other people as making conflicting judgments. I agree that morality is ‘a practice with a purpose, ’ and that the purpose is to sustain certain patterns of social interaction that minimize harm (or something like that). As a result, it's no surprise that when someone who makes paradigmatic moral judgments encounters someone who makes judgments that resemble paradigmatic ones sufficiently closely, the paradigmatic judger will be willing to engage in moral discussion. Such willingness plausibly serves the purpose of the practice.
Where Björnsson and McPherson seem to me to go wrong is in eliding the purpose of the practice as a whole with the purpose of each token moral judgment or assertion. This elision emerges in their appeal to the analogy with functional kinds such as hearts and cars. They analogize these functional kinds with moral judgments, and not with the practice of moral judgment. That is how they explain why a non-motivating moral judgment nevertheless counts as a moral judgment. But the function of moral discourse is best viewed at the level of the whole practice. Indeed, it simply cannot be achieved in any other way.
Because the purpose of the practice of moral discourse is served by the whole practice, and not by individual utterances, there is no call to identify necessary and sufficient conditions for paradigmatic individual moral judgments, as there is a need to explain what a paradigmatic heart or car is. What we can say instead is that there are moral assertions because, given what human beings are like, the language of morality – thought of as a social practice – serves a purpose, and can be taught from one generation to the next. The neo-pragmatist easily sidesteps the specification problem because the focus of attention is not on the semantics of particular moral assertions. Rather, it is on the language game of morality as a whole. And one can make a move in that practice for a host of quite distinct reasons, and with a host of quite distinct attitudes.
Before moving on, let me mention an interpretation of Björnsson and McPherson's paper that I have not discussed. I have been writing as if their goal was the relatively ambitious one of providing a non-cognitive account moral judgment that solves the specification problem. In contrast to this reading, one might take their paper as intended only to explain why it is no surprise that those who make paradigmatic moral judgments often engage in moral discussion with those who make other sorts of judgments – judgments that might or might not be moral judgments. The functional account they offer might be taken not as an account of moral judgments in general, but only of paradigmatic moral judgments. And Björnsson and McPherson might simply be taken as silent as to what counts as a moral judgment that is not paradigmatic. Call this the ‘modest reading’ of their paper.
Against the modest reading, of course, is Björnsson and McPherson's explicit definition of paradigmatic moral judgments as possessing the five features described above, and their claim that moral judgments form a functional kind (Björnsson and McPherson, Reference Björnsson and McPherson2014, 13–14). Moreover, there is admittedly something odd about taking deviant cases – like the moral judgments of young children, depressed people, and psychopaths – as paradigmatic, which we would have to do if the functional account they offer is interpreted as an account of only paradigmatic moral judgments. On the other hand, there are places in which Björnsson and McPherson do describe paradigmatic moral judgments as constituting a functional kind – including in the abstract of their paper (Björnsson and McPherson, Reference Björnsson and McPherson2014, 1, 26). But regardless of whether one takes the modest reading to reflect their intentions or not, it remains a view that one might defend. The question is: how does it fare in comparison with the neo-pragmatist view?
The modest reading can fit, virtually unmodified, within a more general neo-pragmatist view. Moreover, it would benefit from being so located. Keep in mind that Björnsson and McPherson are not offering an account of moral discourse at all. Rather, they are offering an account of a certain functionally-defined attitude-type, and of the interactions between those who hold such an attitude and those who evince attitudes that are sufficiently similar. These interactions can be characterized as efforts to produce a greater uniformity in attitude. The neo-pragmatist can supply Björnsson and McPherson with a ready-made account of the two primary mechanisms used in such efforts: (1) assertions and (2) assessments of truth and falsity.
Neo-pragmatism can also soothe some of the dissatisfaction one might feel at the modest reading's explicit silence as to the nature of moral judgments. Partly this is because the neo-pragmatist explicitly rejects two impulses that generate the dissatisfaction. The first is the impulse to want metaphysical answers to questions such as ‘What is the nature of a moral judgment’. And the second is the impulse to take language as merely the clothing of thought, rather than as the primary phenomenon. Moreover, the neo-pragmatist actually has something like an answer to the question of when it is someone is making a moral judgment. According to the neo-pragmatist, if a person makes a sincere and competent moral assertion, that person has expressed a moral judgment. What Björnsson and McPherson contribute to this answer is partial explanation of existence of moral discourse in the first place, as well as an explanation of the tolerance that certain participants in that discourse will exhibit in interpreting others as worth engaging with in moral discussion.
3.2. Laura and François Schroeter's community-referential account
Laura and François Schroeter solve the specification problem for cognitivists by solving a more general problem: explaining how people with very different criteria for the use of a referring term can nevertheless be taken to be talking about the same thing. In offering their explanation, the Schroeters explicitly reject the central assumption of what they call ‘the template model’ of concept identity (Schroeter and Schroeter (Reference Schroeter and Schroeter2014, 4–5); See also Schroeter (Reference Schroeter2012, 186–187)). The template model holds that concept identity, and therefore the possibility of genuine agreement and disagreement, is type-identity of something internal to a pair of speakers: for example, it might be identity of reference-fixing criteria, or of the non-cognitive attitude expressed by the relevant term, or of the pattern of inferences accepted as conceptually justified. Against this, the Schroeters hold that the basic unit to which we ought to be assigning a semantic value is a socially and temporally distributed linguistic tradition.
The basic conceptual tool the Schroeters employ in constructing linguistic traditions is what they call ‘the appearance of de jure sameness’: the subjective appearance of obvious, incontrovertible and epistemically basic sameness of subject matter (Schroeter Reference Schroeter2012, 181). Appearances of this sort characterize what is going on in a case in which you and I are talking with each other about the Provost by using her given name. They also characterize what is going on in such a conversation when I use the word ‘she’ in such a way that the anaphoric link is to the Provost's name. They do not characterize what is going on when I use the phrase ‘the positive square root of sixteen’ and you say ‘that is, four’. So an appearance of de jure sameness is something that represents the referents of two phrases more closely even than an assertion of necessary identity. It involves, as the Schroeters put it, something more like the total absence of what Frege would call ‘a cognitive difference’.
In our conversations with each other – and in reading books, watching television, and otherwise producing and consuming language – appearances of de jure sameness of topic link together a vast multitude of token uses of terms into what the Schroeters call a ‘representational tradition’ that extends, often, far into the past.Footnote 10 Representational traditions are messy things, since the token uses of the relevant terms are produced by people with quite widely divergent criteria for the use of the term, and some of the utterances may well be assertions that are blankly false and wildly confused. Still, representational traditions, taken as wholes, allow for charitable interpretations: interpretations that seek to identify ‘the most important practical and theoretical interests that have been subserved by the categorizing practices’, and that yield, in favorable cases, univocal referents (Schroeter and Schroeter Reference Schroeter and Schroeter2014, 14 n. 22). These interpretations also end up classifying some appearances of de jure sameness of topic as mere appearances: as when I misinterpret your claim about your cat, Genghis Khan, as a claim about the leader of the Mongol empire. When we exclude these mere appearances, what remains are genuine relations of de jure sameness of topic. The equivalence classes of token uses linked by this ‘cleaned up’ relation all express the same concept, allowing for agreement and disagreement about a common topic.
On reflection a number of important points of similarity between the neo-pragmatist and the Schroeters should be apparent. The neo-pragmatist, like the Schroeters, questions the usefulness of an individualistic metasemantics, and focuses instead on linguistic practices and their purposes. Moreover, the Schroeters’ use of the notion of apparent de jure sameness in order to explain when two people are talking about the same thing results in a view that is extremely neo-pragmatist friendly. That is because appearances of de jure sameness of topic depend essentially on the sort of overt features of language on which a neo-pragmatist will also focus when explaining language as a social practice. Here, for example, are two of the ways the Schroeters describe such appearances:
[T]he psychological mechanisms that generate the initial appearance of de jure sameness in conversation are not directly sensitive to sameness of core topic-fixing criteria – they’re sensitive to the syntactic and pragmatic features that allow us to parse sentences.
[I]n parsing another person's speech, you automatically hear their use of a familiar English expression as pertaining de jure to the very same topic you yourself associate with that expression.Footnote 11
The Schroeters’ strategy for solving the specification problem depends crucially on the two theoretical commitments they share with the neo-pragmatist: a focus on the linguistic community as a whole, and a focus on its overt, public aspects. It is by focusing on overt linguistic phenomena – the ‘public practice’ of language – that they avoid thinking of appearances of de jure sameness as resulting from some sort of comparison of the subject-specific topic-fixing criteria two different people are using. In this way these sameness-appearances are very different from, say, appearances of sameness of shape. Things appear the same in shape because we compare the two shapes. In contrast, the Schroeters think we regard ourselves as talking about the same topic as someone else because of much more superficial – largely syntactic – features of that person's speech. And the essential appeal to the community as a whole comes when the Schroeters select, as the unit of semantic interpretation, the whole body of token uses of particular phrases that are linked together across time and people by the appearance of de jure sameness.
The point of contention between the Schroeters and the neo-pragmatist has to do with the notion of reference. The neo-pragmatist understands reference-talk in a deflationary way, and denies that the reference relation – if we want to call it a relation at all – can bear any theoretical weight. The Schroeters, on the other hand, seem to rely on it. True, they allow the formal possibility that the best interpretation of moral discourse might not involve reference. But they think this possibility is unlikely to be realized, for the following reason:
rationalizing interpretation [of a linguistic tradition] is biased in favor of a traditional, context-invariant realist interpretation of normative concepts because [certain] realist assumptions are central to our normative thought and talk. (Schroeter and Schroeter Reference Schroeter and Schroeter2014, 22)
This is a crucial claim, since these realist assumptions include ‘that a single property is picked out by all competent subjects independently of their idiosyncratic interests or circumstances’. The Schroeters take this univocal ‘picking out’ (i.e. reference) to explain how it is possible that ‘we correct, refine, and precisify prior categorizations in light of this reflective theorizing’. This goes against standard versions of expressivism and speaker-relativism. And I think that the Schroeters think that it also goes against neo-pragmatism. But if neo-pragmatism is on the table, no appeal to these sorts of ‘realist assumptions’ can push it off, even if they make use of the notion of ‘picking out’. The neo-pragmatist can endorse any realist presuppositions the Schroeters offer, given her endorsement of deflationary accounts of truth, reference, belief, and related notions. Indeed, it is notoriously difficult to find a criterion of realism that the neo-pragmatist cannot accept. In any case, the Schroeters have certainly not provided one.
Why do the Schroeters take moral talk to be representational? They seem to think it is part of the best explanation of the fact that normative discourse looks like it does – involving genuine disagreement, argument, and so on:
if you assume that we are all co-referring on normative questions, there will be direct epistemic pressure to take others’ testimony at face value, to give credence to their normative reasoning, and to work toward resolving interpersonal disagreements through distinctions, justifications, and rationales that are intersubjectively acceptable. Without such a presumption of stable co-reference, it's not obvious that these virtuous epistemic practices would be warranted. (Schroeter and Schroeter Reference Schroeter and Schroeter2014, 22–23)
It is clear here that the Schroeters are placing genuine explanatory weight on the notion of reference. But the neo-pragmatist can explain the warrant for our epistemic practices without such an appeal. According to the neo-pragmatist assertoric discourses exist precisely because by opening us up to correction by others they serve to get a community into a useful uniformity of psychological states. That is, we have such things as assertions precisely because of the utility of those practices in which we exercise the dispositions the Schroeters identify as the epistemic virtues. Referential talk is the result of these virtues being virtues in the domains of arithmetic, probability, descriptions of the environment, morality, and so on.
If the Schroeters gave up the idea that truth-aptitude and objectivity require finding a reference, their view would be both more general and more powerful. Vindicating truth-aptitude and objectivity would be a matter of showing that the epistemic practices we employ as a community in resolving disputes about a topic can be expected to yield convergence in all but borderline cases, at least under favorable circumstances. This would allow empirical argument to be one of those epistemic practices in some cases – for example, when we are talking about natural kinds, or about the location of a certain restaurant – but it would also allow mathematical proof to count as such a practice, as well as the distinctive sorts of arguments we use in normative disputes.
Here is one final reason for the Schroeters to abandon their reliance on reference and endorse neo-pragmatism. As they themselves admit, their view makes it impossible for causally isolated linguistic communities to share any concepts. The reason for this is that their model ‘makes real causal-historical connections a necessary condition for concept identity’ (Schroeter and Schroeter Reference Schroeter and Schroeter2014, 18 n. 28). As a result, no alien race could even possibly have the concept of the integers, or of space, or of time, let alone the concept of moral wrongness. And this would be true even if they had concepts that functioned exactly like ours, with the same referents. So the Schroeters reject the presupposition of moral Twin Earth thought experiments. Earthlings and Twin Earthlings are, on their view, incapable of having genuine moral disagreements, at least until their concepts change.
The Schroeters have some responses to the above worry that may serve to blunt its force, but it would be better not to have to face it at all. And the neo-pragmatist does not face it. For the neo-pragmatist, as for the Schroeters, the question of concept identity has to do with how we use the word ‘concept’. But because the Schroeters take the concept of ‘concept’ to be referential, they are compelled to find a referent to validate it. But the neo-pragmatist is not so compelled. As a result, it is possible for the neo-pragmatist to avoid the present worry while agreeing with the virtually all of what Schroeters say in explaining our judgments of sameness of topic within a linguistic community – including the way in which appearances of de jure sameness of topic fit into the picture. Indeed, their view supplements Price's own story about the ways in which truth-assessments produce useful convergence in attitudes within a linguistic community. But the neo-pragmatist can also point to the characteristic usefulness – and therefore the aptness – of truth-assessments of the accuracy of translations. Thus they can say that the English word ‘wrong’ perfectly translates the Twin English word ‘wrong’, so that they mean the same thing and are associated with the same concept. True, such assessments require us to make judgments of identity at the level of linguistic practices taken as wholes – such as the practice of counting, or of assessing probability, or of making moral judgments. But as I have already argued, in section 3.1, this is not especially problematic.
4. Conclusion
Neo-pragmatism avoids the worries that each of the two views I’ve discussed in this paper might raise about the other. Björnsson and McPherson might worry that in relying on primarily syntactic features of sentences to generate the appearance of de jure sameness of topic, the Schroeters fail to do enough to indicate how we should go about the rationalizing interpretation that will exclude some of these appearances as mere appearances. And the Schroeters might worry that in giving paradigmatic moral judgments such a central role in their account, Björnsson and McPherson fail to give competent speakers the kind of latitude they really have with respect to the content and affect those speakers associate with moral terms. Neo-pragmatism avoids both of these problems by shedding the unnecessary commitments that lead to them. It avoids relying on a substantive notion of reference that threatens to limit the sorts of explanations the Schroeters can give of the purposes of moral discourse. And it avoids relying on the idea that there is some special attitude to which we need to appeal in explaining what counts as a genuine moral judgment.
The criticisms the neo-pragmatist levels at both the non-cognitivism of Björnsson and McPherson and the cognitivism of the Schroeters are by no means devastating. Quite to the contrary, I think that it should not be too difficult for them to abandon the assumptions I have been criticizing: (1) the idea that there is a privileged non-cognitive attitude to which we should appeal in explaining moral judgment, and (2) the idea that we need to appeal to a substantive notion of reference to explain the efficacy of the epistemic virtues in the moral domain. What remains, in the case of Björnsson and McPherson, is an explanation for the extent of our willingness to engage with others on moral matters, even when it is clear to us that their moral classifications are quite different from our own. And what remains, in the case of the Schroeters, is an explanation of the cognitive architecture that underwrites our assertoric practices. The neo-pragmatist can embrace both of these explanations. And the hope is, as with other embraces, that it will be returned.
Disclosure statement
No potential conflict of interest was reported by the author.
Acknowledgement
For helpful discussion and written comments on earlier versions of this paper, many thanks to Matt Bedke, Jamie Dreier, Diana Heney, Tristram McPherson, François Schroeter, Laura Schroeter, and Stefan Sciaraffa.
