Quasi-Realism and Ethical Appearances

Quasi-Realism and Ethical Appearances

Edward Harcourt

Quasi-realism in meta-ethics, as advocated by Simon Blackburn, belongs in the expressivist tradition of ethical anti-realism. Insofar as it is expressivist, it is intended to contrast with at least one other type of ethical anti-realism, J.L. Mackie's error theory, according to which the meaning of first-order ethical sentences renders them apt to state facts - if only there were any such facts to state. Within the expressivist tradition, quasi-realism's claim to distinctiveness is that it "saves the ethical appearances": centrally, and in contrast to its emotivist prototype, that it smoothly explains the fact that sentences of first-order ethical discourse are, for example, capable of negation, conditionalization and embedding in propositional attitude contexts.
In a well-known challenge to Blackburn, Crispin Wright has argued that quasi-realism makes such a good job of saving the ethical appearances - that is, allows ethical discourse so many of the features that also belong to discourses which express propositions (or have truth-conditions, or do whatever it is the expressivist must say ethical discourse doesn't do) - that ethical discourse is shown to express propositions after all. It looks as if the quasi-realist has just two responses at his disposal: either to agree that ethical discourse appears to express propositions (have truth-conditions, etc.) and argue that it doesn't really express them, or to deny that there is any such thing as appearing to express propositions. In this paper I try to show that neither response is adequate.
I begin in §1 with an outline of the quasi-realist position. In §2 I identify the principle behind Wright's challenge. The two ways of falsifying the principle correspond to the two different quasi-realist responses. In §3 I show that the first response - on which ethical discourse appears to express propositions but doesn't - turns out to be an untenable error theory of ethical discourse. In §4 I raise some objections to the second response - according to which there is no such thing as appearing to express propositions - but though these hit home against an early formulation of quasi-realism, a more recent ("inclusive") formulation can escape them. I then argue in §5 that even granted the "inclusive" formulation, the idea that there is no such thing as appearing to express propositions has unacceptable consequences, so Wright's principle is confirmed. Finally, I show in §6 how this conclusion converges with the conclusion of a different argument against "inclusive" quasi-realism, to the effect that there is no distinction between ethical discourse's possessing the various surface features which the quasi-realist agrees it has, and its expressing propositions.

1. First, anti-realism. Blackburn's basic understanding of what it is to be a realist or anti-realist about a given region of discourse is underpinned by a distinction between two types of mental state. (I discuss the qualifications to this which Blackburn's more recent work may call for in §4, but it will be helpful to keep matters simple to start with.) On the one hand there are states which represent the world as being a certain way and for which the question of truth and falsity arises (in Blackburn's view, paradigmatically beliefs). On the other hand there are states which do not represent the world at all: these states just are what they are, and the question of truth and falsity cannot arise for them (in Blackburn's view, paradigmatically desires). The quasi-realist, like the emotivist, holds that the states which ethical discourse fundamentally expresses are of the second, non-representational sort. There can therefore be no question on either doctrine of an ethical reality which ethical discourse represents, faithfully or otherwise.
Next, discourses themselves. We can distinguish between discourses which consist (in part, at least) of sentences capable of negation, conditionalization, embedding in propositional attitude contexts, in the context "it is true that ...", and so on, and those which do not. A favoured term of Blackburn's for this family of features is the "propositional surface" of a discourse. But since it matters in what follows to have a label for these features which is neutral as to whether a discourse, in virtue of possessing them, appears to express propositions (have truth-conditions etc.) or not, let us call them simply the T-features. Where the emotivist was forced into retreat by the observation that ethical discourse has the T-features, the quasi-realist takes the observation in his stride:

[C]ommitments seem pretty homogeneous - ... moral ... commitments seem just like other beliefs ... and capable of sustaining propositional attitudes and logical embeddings that might seem to be the exclusive preserve of sentences that express objects of belief.
The difference between a discourse with the T-features where the states fundamentally expressed are representational and one where they are non-representational is that in the latter case we have the effect of projection, "the mechanism whereby what starts life as a non-descriptive psychological state ends up expressed, thought about, and considered in propositional form". However, I shall not say anything about projection as a mechanism and, following Blackburn's usage, shall treat projectivism (in ethics) simply as the conjunction of expressivism with the thesis that ethical discourse has the T-features, whether the conjunction is regarded as having come about as the result of a projective mechanism or not.
So far my presentation of the quasi-realist position has (I hope) followed Blackburn's own. From this point, however, Blackburn's presentation takes an odd turn. He distinguishes between those variants of projectivism which are conservative with respect to the discourse they apply to - i.e. which are such that the first-order discourse is consistent with the anti-realist metaphysics - and those which are revisionary with respect to it. He then identifies projectivism in its revisionary variant with Mackie's error theory. Blackburn's projectivism, by contrast, "is consistent with, and indeed explains, the important surface phenomena of ethics", and "quasi-realism" is another word for conservative projectivism.
But - and the significance of the point will come out in §3 - there are two reasons why quasi-realism can't be a sub-variety of any more general doctrine of projectivism. First, a defining feature of Blackburn's projectivism is its distinctively expressivist anti-realism. But Mackie is no expressivist. He says that his theory, unlike emotivism, is "an ontological thesis, not a conceptual or linguistic one": "the ordinary user of moral language means to say something about whatever it is that he characterizes morally ... and not about, or even simply expressive of, his own or anyone else's attitude or relation to it". And Mackie's error theory earns its name because, according to it, the states fundamentally expressed by first-order ethical discourse are false: the discourse lays claim to the existence of an ethical reality and there isn't one. But ethical discourse couldn't make a claim of this kind if an expressivist construal of it were correct. So although Mackie may use the word "projection" in connection with his own theory, there can be no single doctrine called "projectivism" common to Mackie and Blackburn, and so nothing common to Mackie and Blackburn for quasi-realism to be a sub-variety of.
Secondly, as to Blackburn's own brand of projectivism, to say that it calls for no revision of first-order ethical discourse is just to say that expressivism is consistent with the thesis that ethical discourse has the T-features. But since these two theses each ascribe a property to the same thing - ethical discourse - we would expect them to be consistent anyway, so there's no conceivable revisionary sub-variety of Blackburn's projectivism for the "conservative" label to mark quasi-realism off from. Since ethical discourse's possessing the T-features has sometimes been seen as a decisive objection to expressivism about it, it's not surprising that argument is needed to show that they are consistent, and if one had to find room for a distinction between quasi-realism and Blackburn's projectivism - henceforth simply "projectivism" - quasi-realism could be thought of as the attempt to show that they are consistent. But it is not any distinct thesis which would be established were the attempt to succeed.

2. Now let us look at Wright's challenge to quasi-realism in more detail. Quasi-realism says that the states fundamentally expressed by ethical discourse are not representational. So, since the paradigm of a representational state is belief, the quasi-realist seems committed to the view that ethical sentences do not express beliefs. And - assuming that if one makes a sincere assertion, one has a belief whose content is expressed by the words one uses to make it - he also seems committed to the view that ethical sentences are not assertoric. On the contrary, ethical sentences according to the quasi-realist must be merely quasi-truth-apt - that is, share the T-features with truth-apt sentences without being truth-apt themselves - express quasi-beliefs, and so on. But, Wright objects,

once it has been explained [by the quasi-realist] how the language in some region of discourse may take on all the features of paradigm statement-making, we might as well allow that we therein deal with genuine statements.
That is, the more closely quasi-assertions, quasi-beliefs and the rest can be shown to mimic the real thing, the less reason there is for claiming that they are merely quasi: for surely what looks and behaves as if it's propositional - where "propositional" is to be taken as shorthand for "expresses representational mental states" - is propositional. But anything quasi by definition shares the T-features with the real thing, so it is the real thing.
To assess the challenge, let's begin by distinguishing some varieties of concept. Let us say that "F" expresses an appearance concept if there's such a thing as something's appearing to be F. And within the class of appearance concepts, let us distinguish between a fallible appearance concept and a phenomenal concept. "F" expresses a fallible appearance concept if something's appearing to be F is consistent with its either being or not being F (an example would be "x is fur"); "F" expresses a phenomenal concept if it's the case that if something appears to be F then it is F (an example would be "x is pain").
Now the principle Wright appeals to, that what looks and behaves as if it's propositional is propositional, is true only if "propositional" is a phenomenal concept. But it doesn't go without saying that this is so, and there are two ways in which the principle might be falsified: first, if "propositional" is a fallible appearance concept, i.e. if something can appear to be propositional and fail to be, and secondly, if "propositional" is not an appearance-concept at all, i.e. if there is no such thing as an appearance of propositionality. Corresponding to these two ways of falsifying the principle, there are two different ways in which the quasi-realist's distinctive combination of theses - expressivism, plus the thesis that ethical discourse has the T-features - might be packaged. (I hope this makes it clear why I chose the colourless terminology of "T-features".) On the first - corresponding to the idea that propositionality is a fallible appearance concept - ethical discourse appears, thanks to the T-features, to express representational mental states but doesn't really; on the second - corresponding to the idea that propositionality is not an appearance-concept at all - ethical discourse has the T-features in common with discourses which express representational states and does not itself express such states, but there's no particular kind of state such that, thanks to the T-features, it appears to express states of that kind. Either package would serve equally well as a reply to Wright if it were cogent, though it is not clear that Blackburn has single-mindedly pursued one rather than the other in his writing. Let us look at the two packages in turn, beginning with the first.

3. There is a certain amount of evidence that Blackburn thinks of quasi-realism in terms of the first package. To take "a 'non-descriptive' or nonrepresentational view of our [ethical] commitments" is, he says, to disavow "a realist ... story that says that such commitments do what they seem to do: describe what we take to be the ethical facts". He contrasts an "'emotivist'" or "overtly expressive" language, in which "expressions of attitude wear this function on their faces", with the by implication covertly expressive ethical language we in fact have. He says, with apparent approval, that "according to Wittgenstein, whole areas of language that look as if they are dedicated to describing how things are must be understood in other terms", and terminology such as "realist-seeming" is a recurrent feature of his writing.
However, there is one obvious objection to the first quasi-realist package. On this package, ethical discourse appears - thanks to the T-features - to express representational mental states which - thanks to expressivism - it does not. (It also fails to appear to express states of the kind which, by expressivism, it does express, since for a discourse to appear to express non-representational states would - according to Blackburn - be for it to be an "emotivist" discourse the sentential components of which would not even be in the indicative mood.) But for a first-order discourse to express states which are of one sort while appearing to express states of a different and incompatible sort is for it to give rise to a form of illusion or error. So there is at least one ethical appearance - the appearance of expressing representational states - which quasi-realism fails to save. But quasi-realism is true only if it saves all the ethical appearances, so quasi-realism is false.
How is the quasi-realist to reply? Blackburn is keen that there should be a naturalistic (e.g. evolutionary) explanation of how the (alleged) phenomenon of projection can come about; indeed he says that if we failed to explain naturalistically how it can come about, then we would have to concede that an error theory, rather than quasi-realism, would be the correct one for ethical discourse:

Redeploy the ingredients [of the naturalistic tool-kit] to find [a] natural 'placing' for ethics, or agree with the error theorist that something like a Boo-Hooray language is all that there ought to be.
But this does not seem like much of a reply. It may be possible to explain naturalistically how an illusion arises, but its explicability does not make it any less of an illusion.
An alternative line of response would be to concede that expressing representational mental states is the content of a (false) appearance which first-order ethical discourse presents. However (the response goes), that first-order ethical discourse presents such an appearance is not articulated in any proposition that belongs to first-order discourse, and what Blackburn meant by an "error theory" was a (supposedly true) metaphysical theory whose claims are inconsistent with one or more of the propositions of the first-order discourse to which it applies.
One worry about this response is that it misrepresents what saving the ethical appearances amounts to even in those cases where the quasi-realist succeeds in saving them. For (the worry goes) if the mental states ethical discourse expresses are non-representational, there can be no question of consistency between the meta-ethical sentences - which Blackburn gives us no reason to think of otherwise than as expressing representational states - and the first-order ethical ones, since the non-representational cannot stand in any logical relation to the representational. Successfully saving the appearances can amount to no more than showing absence of inconsistency. Is that enough? It doesn't seem to be: contrasting ethical discourse with religious discourse (about which Blackburn is an error theorist), Blackburn says that whether a projectivist theory is conservative or not will depend on the particular content of the first-order discourse in question. But absence of inconsistency follows straight from expressivism, no matter what the content of first-order ethical discourse. So quasi-realism comes in as conservative merely by default. If on the other hand absence of inconsistency is enough, there is every reason not to appeal to the paradigm of consistency between first- and second-order discourses in order to exclude the appearance that first-order ethical discourse expresses representational states from the class of appearances quasi-realism needs to save.
This particular worry may be ill-grounded, however. For although it is a big problem for quasi-realism if absence of inconsistency isn't enough, it is not a problem that arises here only. As part of its wider programme of saving the ethical appearances, quasi-realism must explain - and I will have more to say about this later - the validity of inferences involving a combination of representational and non-representational contents, and to do this it must have at its disposal some non-standard notion of consistency which applies to contents of both kinds. If quasi-realism can make good such a notion, it can solve the present worry, and if it can't, it won't need this worry to make the quasi-realist project come unstuck. However, even if the quasi-realist can explain what it would be to save the ethical appearances where that is a matter of showing consistency between first- and second-order claims, it doesn't follow that that is all saving the ethical appearances ever amounts to. What really makes the quasi-realist's response out of place is that his own notion of the "ethical appearances" is - rightly - a broad one, encompassing not only the various T-features but also, for example, whatever particular first-order commitments we may have held prior to becoming convinced of projectivism, and the phenomenology of ethical experience, in particular the fact that moral obligations present themselves to us as external and categorical. So the fact that the appearance of expressing representational mental states is not articulated in any proposition of first-order ethical discourse is no reason to think the quasi-realist has no need to save it.
Finally, the quasi-realist might reply that, though ethical discourse does indeed appear to express states of a kind it does not in fact express, it is only philosophers (who have an interest in such questions) who are liable to be taken in by the illusion. But for a philosopher to be taken in by it would not (the reply continues) lead to his doing or saying anything of a first-order sort that he would not otherwise have done or said, but only to his forming a (by quasi-realist lights) false philosophical theory - for example, concluding that ethical realism is true. But quasi-realism is not under any obligation to "save" the theories of its philosophical rivals. But this reply too seems ineffective. It might well be true that only a philosopher would be so far 'taken in' by appearances as to draw a philosophical conclusion from them (such as that ethical realism is true). But the relation of philosophical theory (ethical realism, say) to ethical appearances in this case is (roughly) that of belief to appearance: a philosopher's belief and his being appeared to in a certain way by the surface of ethical discourse are not the same thing, and what the reply needs to show is not that only philosophers would form certain beliefs on the basis of the appearances presented by ethical discourse, but that it is only of philosophers that one can say that the appearances presented by ethical discourse are misleading. Now arguably a forgery, however brilliant, could not be said to present a misleading appearance to a child who cannot tell the difference between real banknotes and Monopoly money, because the discriminations which the forgery would prevent normal adults from reliably making lie beyond the child in any case. But non-philosophical speakers do not stand to philosophers as the child in this example stands to normal adults. For if there could be such a thing as an "emotivist" ethical discourse, i.e. a discourse which both expressed non-representational mental states and appeared to do so, the differences between it and ethical discourse as we now have it would be gross: one would not need to be a philosopher to notice which one was being spoken. So insofar as we don't notice any difference between ethical discourse and discourses expressing representational states, there's a false appearance not just philosophers but everyone is misled by.
On the first package then, quasi-realism turns out to be an error theory. Does it turn out to be "the" error theory, that is, J.L. Mackie's theory of ethical discourse? No: since Blackburn and Mackie mean something different by "projectivism" (see §1), showing that Blackburn's projectivism is not conservative does not yield Mackie's theory. But this is no help to the quasi-realist: objectionable though Mackie's theory may be in its own way, as we saw in §1 it does not take much to show that there is no logical room for a revisionary variety of Blackburn's projectivism. In Mackie, the alleged error consists in a mismatch between a feature of first-order ethical discourse and an ethical anti-realism independently established elsewhere. In Blackburn, by contrast, the error consists in a mismatch between what, on the first quasi-realist package, is a feature of first-order ethical discourse (the fact that it appears to express representational states) and the theory of first-order ethical discourse (that it expresses non-representational states, from which theory ethical anti-realism, of a different sort from Mackie's, is then meant to follow). But to call a theory of ethical discourse an error theory because it's inconsistent with a feature of ethical discourse is like calling a theory of planetary motion an error theory because it's inconsistent with the motion of the planets.
In conclusion, though it would answer Wright's challenge if "propositional" were a fallible appearance concept - that is, if ethical discourse's possessing the T-features were a matter of its presenting fallible appearances of propositionality - this way of packaging the quasi-realist's distinctive combination of theses turns out to be a dead end. Let us turn, then, to the second quasi-realist package, where propositionality is not an appearance-concept at all, and see if this provides the quasi-realist with a better line of reply to Wright.

4. To see the importance in Blackburn's thinking of the idea that "propositional" is not an appearance-concept at all, consider again the quasi-realist idea that ethical sentences are merely quasi-truth-apt. This just means that the behaviour of the word "true" in ethical discourse is indistinguishable from its behaviour in a discourse that expresses representational states. So whether the word "true" in either discourse expresses the concept of truth cannot be discerned from any feature of the surface of the discourse, so propositionality is not an appearance-concept.
The same point is evident from Blackburn's proposed programme of logical reconstruction. As regards syntax, the programme aims to show how a formal language whose components are non-truth-apt and expressive of attitudes could model the patterns of inference of ethical discourse in ordinary English. As regards validity, the programme aims to develop a notion of consistency applicable to both truth-apt and non-truth-apt contents, and thereby to explain the validity of inferences involving (but not necessarily consisting exclusively of) ethical contents, where these are thought of as non-truth-apt. What matters for now is not whether the programme succeeds - and there are undoubtedly technical obstacles in its way - but what the programme is supposed to show. To show that the inference patterns of ethical discourse in ordinary English can be modelled just as well by a logic for expressive, non-truth-apt contents as it is by classical logic (supposing this to be the correct logic for those bits of English which uncontroversially express representational states) would be to show that the logic for expressive contents could be the correct one for ethical discourse, and thus that quasi-realism could be true. (To show that it is the correct one would take further argument.) But it would also be to show that another one of ethical discourse's T-features, its patterns of inference, would be the same whichever logic - classical logic or the expressivist construction - were the correct one for it. So for any given discourse, one could not tell from that feature of its surface whether classical logic was the correct one, or therefore whether the discourse was propositional. So if the reconstruction programme succeeded, it would show once again that "propositional" was not an appearance-concept.
Evidently if there are no appearances of propositionality - no such thing as a discourse's seeming to express representational states - there are a fortiori no false appearances of it, so the charge that quasi-realism turns out to be an error theory lapses. But the idea that there are no appearances of propositionality itself has a good deal wrong with it. Consider the logical reconstruction programme again. The ambitions of the programme as regards logical syntax imply that there could be two logics - a sentential logic with "if ... then ..." as truth-functional arrow, and an alternative "if ... then ..." which operates on non-sentential components, for example - which provide incompatible analyses of (e.g.) "if charitable giving is good then tax breaks for charitable giving are good", "charitable giving is good" and "tax breaks for charitable giving are good" each of which captures the fact that from the first and the second one must infer the third. But could there be two such logics? If there could be, ex hypothesi nothing about the inference patterns of English could decide which of the two connectives the words "if ... then ..." expressed. Blackburn might reply that it was always to be expected that the decision as to which type of states ethical discourse in fact expresses, and therefore which logic gets "if ... then ..." in ethical contexts right, was to come from elsewhere - for example, from Humean considerations about motivation - not from consideration of the discourse's surface features. But this reply is out of keeping with Blackburn's (welcome) interpretationist sympathies: he regards logic as "our way of codifying and keeping track of intelligible combinations of commitment", where "intelligible" signals the combinations of commitment which would be ascribed to an agent by the best possible interpretation of his behaviour, including his verbal behaviour. But the data for such an interpretation - the agent's or his community's inferential practices themselves - would according to the quasi-realist be the same whichever of the two supposed connectives were involved. So either interpretationism is to be rejected and which connective "if ... then ..." expresses is an evidence-transcendent property of a discourse, or else there is no room for the possibility that "if ... then ..." expresses more than one connective. Exactly the same kind of objection applies to the other consequence we noted of the idea that there are no appearances of propositionality: if the word "true" is used in the same way in two different discourses - that is, in two different areas of English - what could warrant interpreting it as the truth-predicate in one but not in the other? The idea that propositionality is not an appearance-concept seems to be in trouble.

5. But we must beware of any over-hasty conclusions, for there has been a development in Blackburn's more recent statements of quasi-realism which takes care of the above two objections. Retreating from the idea that ethical sentences are merely quasi-truth-apt, express attitudes rather than beliefs and so on, Blackburn now agrees with Wright that "attitudes (and the rest) properly sustain the appearance of being beliefs, so they are beliefs". Though quasi-realism "starts by depending on a contrast - that between attitudes on the one hand, and beliefs on the other", an "inclusive" formulation of the position will do just as well: the states expressed by ethical discourse "belong to the subset of beliefs that are also attitudes (dispositions, inference tickets, or whatever)". But expressivism's distinctive claim about ethical discourse is that the states it fundamentally expresses are non-representational and the paradigm of a representational state was belief. So in conceding that ethical discourse expresses beliefs, hasn't Blackburn renounced expressivism, and thus quasi-realism? The answer is no, and any appearance to the contrary is the effect of a verbal coincidence. On the inclusive formulation, "belief" is to be used in such a way that both representational and non-representational states count as beliefs. So Blackburn can concede that ethical discourse expresses beliefs while continuing to be an expressivist about it.
The warrant for the inclusive terminology is best sought in relation to truth rather than belief. On the contrastive formulation, ethical discourse - as we saw - contains a word that behaves just like the truth-predicate, but isn't the truth-predicate: expressivism stayed in business at the cost of driving the property of being the truth-predicate underground. On the inclusive formulation, the property of being the truth-predicate is where interpretationist considerations say it should be, on the surface. Now Blackburn defines minimalism (or "disciplined syntacticism") as the doctrine that any discourse which has the discipline and syntax necessary to sustain a truth-predicate has truth-conditions. "Having truth-conditions" here lines up with expressing representational mental states, so given minimalism, to concede that ethical discourse is genuinely rather than merely quasi-truth-apt would be to make the fatal concession Wright presses for. But Blackburn distinguishes between Wright-style minimalism (henceforth simply minimalism) and what he calls "minimalist or deflationist theories of truth" (henceforth simply deflationism) according to which "'p' is true" is equivalent in meaning to "p". As Blackburn (in "inclusive" mode) sees it, there is a distinction to be drawn between a discourse's sustaining a truth-predicate and its having truth-conditions, and while by minimalism the former implies the latter, by deflationism it does not. The idea is that just because the truth-predicate (according to the deflationist) adds nothing to the sentences of whose designation it is predicated, its presence in a range of discourses, including ethical discourse, is compatible with those discourses' sustaining the distinction between expressing representational and non-representational states which the expressivist needs. Similarly inclusive understandings of assertion, belief and so on then follow smoothly.
By switching from contrastive to inclusive quasi-realism, then, Blackburn isn't giving up on expressivism. Does the inclusive formulation represent a "third way", that is, a reply to Wright's challenge that consists neither in claiming that "propositional" is not a phenomenal but a fallible appearance concept nor in claiming that it is not an appearance-concept at all? I think not: despite Blackburn's "[ethical] attitudes ... properly sustain the appearance of being beliefs, so they are beliefs", it is more helpful to see the inclusive formulation as preserving the view that there are no appearances of propositionality by redrawing the boundaries of the propositional. For by deflationism, there can be two discourses, each of which contains the truth-predicate, but only one of which expresses representational states. The propositionality or otherwise of the discourse resides, as before, in the type of state expressed, but truth-aptness is now classified - like the presence of the mere word "true" on the first quasi-realist package - as a neutral T-feature. This is also the light in which to see Blackburn's concession, relative to the aims of the logical reconstruction programme, that being modus ponens is a matter of "syntactical form", because being modus ponens cannot - presumably for interpretationist reasons - be an undetectable property of inferences. But Blackburn still holds that "the 'deep' semantics of a surface example of modus ponens is to be explained in a particular, perhaps initially surprising, way". Expressivism is still supposedly in play because being modus ponens has moved from the class of features that identify a discourse as propositional to the class of T-features.
Switching from the contrastive to the inclusive formulation thus silences the two particular objections to the second quasi-realist package entertained at the end of §4: that the package implies that being the truth-predicate and being modus ponens are undetectable features of a discourse. However, other objections to the package are close at hand which are addressed to the contrastive and inclusive formulations alike. Though the membership of the class of T-features varies from one formulation of quasi-realism to the other, there is, on either formulation, no particular kind of state which ethical discourse, thanks to its possession of the T-features, appears to express: since this is just another way of putting the point that "propositional" is neither a phenomenal nor a fallible appearance concept, the quasi-realist should have no quarrel with it. The T-features are thus no more characteristic of a discourse that expresses representational states than they are of a discourse that expresses non-representational ones; alternatively, the surface of a discourse, to the extent that this surface is characterized by the T-features, is no more naturally suited to the expression of one of these kinds of state than it is to the other. We can't read off from the fact that the surface of a discourse has the T-features that the states it expresses are non-representational, but we can't read off from this same fact that the states it expresses are representational either; the T-features no more reveal or conceal the fact that a discourse which has them expresses representational states than it reveals or conceals the fact that it expresses non-representational ones. On the second quasi-realist package representational and non-representational states are, as I shall put it, equidistant from the surface of any discourse with the T-features. But this implication (henceforth 'the equidistance claim') is problematic for several reasons.
The equidistance claim raises the question as to why a non-representational state's coming to express itself in a discourse with the T-features should be said to involve projection, while a representational state's expressing itself in that way does not, and indeed calls for no special explanation. Conversely, why should we not say that projection is involved even when a representational state expresses itself in a discourse with those features? But it is absurd to say that projection is involved in both cases: the whole idea of projection rests on the thought that there's an asymmetry between representational and non-representational states in respect of their relation to the T-features. So quasi-realism, on the second package, must give up on the notion of projection. I say this not to announce a disaster for quasi-realism so much as to point out how unfamiliar the outlines of the position become on the second package. Giving up on projection will certainly force some changes of phraseology on the quasi-realist: most notably, "projectivism" as a term for the quasi-realist's distinctive combination of expressivism plus the thesis that ethical discourse has T-features. But the idea of projection as a mechanism plays a remarkably slight role in quasi-realism anyway, so dropping the term should not make much difference. The most important changes required would (I suspect) relate to the alleged advantages of expressing ethical states in a discourse with T-features - to the idea, for example, that a discourse expressing non-representational mental states can, as it were, go further when camouflaged in T-features than it could in its natural state. But the idea of T-features as a handy camouflage is one that the quasi-realist must give up on as soon as he opts for the second way of packaging his distinctive combination of theses over the first.
More substantively, it's a consequence of the equidistance claim that argument will be needed to show not only that ethical discourse expresses non-representational states, but also that discourses which we might assume express representational states do indeed express them. (Interpretationism implies only that the very same behavioural and linguistic data can't be interpreted in terms of two incompatible connectives, or in terms of the truth-predicate and a dummy: it doesn't tell us which of the two rival interpretations is the right one.) However, we have no idea what kind of argument would be needed to show that (say) the description in a guidebook of Salisbury cathedral expressed representational states apart from one which adverted to its T-features - which, by the equidistance claim, cannot settle the issue. Secondly, the very idea that there is such a thing as a discourse that is paradigmatic of the expression of representational states vanishes with the equidistance claim, since it must be a hidden feature of any discourse what kinds of state it expresses. But by doing away with the idea that there are paradigms of the expression of representational states, the equidistance claim renders it a mystery what exactly is significant about the T-features as a class. Of course we do know why it is more interesting that ethical discourse shares the T-features with botanical discourse than it is that these two discourses share some other feature which they may happen to share: it is because the T-features create the presumption that ethical discourse expresses representational states. But the equidistance claim debars us from acknowledging this. Finally, consider Blackburn's Wittgensteinian idea that we are misled by surface similarities between discourses into thinking that ethical (or modal or mathematical) discourse expresses representational states when it doesn't. One does not have to agree with this to see that it is a better idea than its converse: no philosopher (as far as I know) has ever said that we are misled by surface similarities into thinking that representational discourses are non-representational. But if the equidistance claim holds, it is unintelligible why the error should run in one direction only. The implication of all three objections is surely that the equidistance claim cannot be right. But if the equidistance claim cannot be right, the second quasi-realist line of reply to Wright fails too: "propositional" must at least be an appearance-concept. And, since there's no coherent revisionary or error-theoretic version of quasi-realism, "propositional" must be a phenomenal concept.
In the next and last section, I want to show how the above objections to the two quasi-realist packages - which argue against them by showing they have unacceptable consequences - join hands with a set of objections from a different direction, which are designed to show that a set of distinctions which the quasi-realist needs (however his view is packaged) cannot be, or at least have not yet been, successfully drawn.

6. The basic challenge for the inclusive formulation of quasi-realism, however packaged, is that it needs to make good a distinction between a discourse's being truth-apt, or apt to make assertions, or to express beliefs - each of which has become an inclusive notion applicable to ethical and (say) botanical discourse alike - and its having truth-conditions, or expressing representational mental states, which botanical discourse distinctively does and ethical discourse distinctively does not. This distinction isn't needed on the "contrastive" formulation of quasi-realism since, however implausible it is to say that a discourse with the T-features isn't really truth-apt - the implausibility which made us shift to the inclusive formulation as the quasi-realist's best option - at least there's no need for a further distinction between actually being truth-apt and having truth-conditions. But granted the "inclusive" formulation, the distinction is needed just as urgently on the first quasi-realist package as on the second. For on the inclusive version of the first package, about which we've said nothing so far, the quasi-realist must be able to distinguish being truth-apt from expressing representational states in order to make room for the idea that being truth-apt is a potentially false appearance of propositionality.
Truth-apt discourses which have truth-conditions are to be distinguished from those which don't, according to Blackburn, by their "'deep' semantics" - only the former have a truth-conditional semantics; and by their functional role - only the former have a "descriptive" function. If these grounds for distinction can be made good, then the quasi-realist has the distinction between being truth-apt and having truth-conditions, or expressing representational states, that he needs. But both grounds for distinction are themselves problematic, and if they are irreparably so, the surface of a discourse will as it were turn out to be all there is to it: the supposedly deep distinguishing marks of botanical discourse (or whatever) will turn out to be possessed by all discourses with the T-features simply as such, as the principle that what appears propositional is propositional implies.
Let us first consider the possibility - which we set on one side in §5 - that quasi-realism might appeal to differences of "'deep' semantics" to distinguish between discourses which have the T-features in common. The idea, presumably, is that if expressivism is true, a truth-conditional semantics is correct for (say) botanical discourse while ethical discourse requires an alternative semantic account, perhaps along the lines explored by Blackburn and Hale, perhaps involving Gibbard's 'hyperdecided states'. As modus ponens is now to be thought of as a T-feature, ethical discourse must contain instances of it. But if "If charitable giving is good then tax breaks for charitable giving are good, charitable giving is good, therefore tax breaks for charitable giving are good" is an instance of modus ponens, "if ... then ..." must at least be a sentential connective: modus ponens just is a pattern of inference exemplified by declarative sentences. But this then rules it out that the components are merely expressive. So once it's admitted that at the syntactic level the English argument quoted is an instance of modus ponens, there's no room for a semantic account of it on which the components do not have a truth-value. Patterns of inference which call for a non-truth-conditional semantics in terms of hyperdecided states (or whatever) are to be found, if at all, only in the "'emotivist'" ethical language we do not have. And if we are not happy with the idea that modus ponens by definition can be exemplified only by declarative sentences, we could argue more cautiously that on pain of denying that modus ponens is a term for a pattern of inference rather than for some verbal pattern of no logical importance, the components of any two instances of modus ponens must be of the same kind, whether the components are declarative sentences or not. This would still be sufficient to show that two instances of modus ponens cannot differ in their 'deep' semantics.
More complex problems attend Blackburn's second way of distinguishing those truth-apt discourses which have truth-conditions from those which do not, which relies on the supposed descriptive functional role (or absence of it) of the states they express. A potential problem, though it is hard to make precise, is that there is a tension between deflationism, the keystone of the inclusive approach, and the quasi-realist claim that we should try to understand "many areas of discourse in terms other than those of 'representing the facts'". Blackburn's deflationist approach to the truth-predicate extends to "represents facts", "is really true", and (apparently) any other variant on the "representation" locution right the way up to "is really in accord with the eternal harmonies and verities that govern the universe": "just because of [deflationism] about truth and representation, there is no objection to tossing ... [these locutions] in for free, at the end". It seems as if the more successfully deflationism absorbs representational locutions into metaphysically non-committal first-order discourse, the less vocabulary there is left in which to state what it is that ethical discourse distinctively doesn't do.
But how exactly is this objection - which Blackburn dismisses as "confiscating the terminology" - to be sharpened up? One thought is as follows. "Ethical discourse does not represent facts" entails indefinitely many statements of the form "'p' does not represent facts", for ethical 'p'. But deflationism tells us, by a swift step, that "'p' does not represent facts" means the same as "not p". So, the objection runs, quasi-realism entails the negation of any number of ethical propositions which we might otherwise want to assert - scarcely saving the ethical appearances. This objection is fallacious, however, because "represents facts" as predicated of sentences is ambiguous in the same way as "shreds documents" as predicated of office machines: it can mean both "represents a fact" and "has the function of representing facts". The former meaning is the one needed to generate the absurd consequence for the quasi-realist, but the latter is the meaning the quasi-realist intends when talking about ethical discourse.
It would provide the quasi-realist with a convenient reply to the attempt to confiscate his terminology if it were just an accident that we use the same words to mean both these things. But it isn't: we use the same words because the two meanings of "represents facts", like the two meanings of "shreds documents", are related. So deflationism ought to have something to say about "represents facts" in both its meanings. Unfortunately for the objection, the relation between the two meanings is not straightforward, so it's not clear exactly what deflationism ought to say. One tentative thought is this. If, by deflationism, "'p' represents facts" in the first meaning means the same as 'p', then "'p' represents facts" in the second meaning means the same as "'p' is truth-apt", and the same equivalences will hold for the negations of each of these pairs of sentences. "It's not the case that 'p' represents facts" (in the latter meaning) is, for any ethical 'p', something the quasi-realist needs to be able to say in order to mark off ethical discourse from discourses which express representational mental states. But deflationism debars him from saying it, since its equivalent by deflationism - "it's not the case that 'p' is truth-apt" - is something the inclusive approach requires him to deny.
However, let us set this objection on one side and try to undermine the quasi-realist's needed distinctions in a different way, focusing to begin with on the distinction between "belief" (the inclusive notion) and "state with a descriptive function" (which only a sub-class of beliefs are said to have; call these "descriptive states" for short). Blackburn has little to say about what the descriptive and contrasting non-descriptive functions are. He tends to characterize the function of ethical beliefs negatively (they "are first to be understood in other terms than that they represent anything"), taking the nature of the describing or representing function for granted. But to rely solely on a negative characterization is to take a crucial dialectical step without argument. It is a matter of logic that no state can be both descriptive and non-descriptive, where this is just shorthand for "having the descriptive function" and "not having the descriptive function". But that the function which ethical beliefs characteristically have, whatever it may be, is such that they cannot also have the descriptive function is not a matter of logic, since there is nothing in the notion of a function itself to say that two functions which are themselves well-distinguished cannot be possessed by the same thing. (A pair of scales can also make an excellent doorstop.) If one and the same state cannot both have the function characteristic of ethical beliefs and have the descriptive function, that will follow (if it follows at all) from the nature of those two functions, not from the mere fact that they are distinct.
So does Blackburn say enough about the distinctive function of ethical beliefs to show that they cannot also be descriptive, or about the descriptive function to show that ethical beliefs cannot have it? A complication here is that Blackburn's conception of the descriptive function seems to roll two ideas into one, and these need to be disentangled. "Descriptive" is on the one hand intended to demarcate a specialized use of language (and therefore I suppose a special kind of mental state, though the term is more happily restricted to language) on a par with (e.g.) "performative" in Austin's work. To characterize this use in detail would be a major task, but I think we have a rough idea of what's meant: the guidebook entry under "Salisbury Cathedral" counts as descriptive in this sense not simply in virtue of consisting in strings of truth-apt sentences but because it is designed to be carried round the cathedral and compared with its features for accuracy, for example, or because if I wasn't in the cathedral and wanted to know what if anything was inscribed on the font I could find out by reading the guidebook entry. Let us call uses of language that are descriptive in this sense 'descriptions proper'.
Now descriptions proper contrast with various other uses of language, for example, axioms of geometry, statements of intention (not descriptive, because if there's a mismatch with what's done, it's what's done and not the statement that has to change), Austinian performatives, or the mimicry in fiction of descriptions proper. Blackburn's idea is that expressing attitudes is another such specialized use, and that ethical discourse distinctively has it. If that is right, then it seems we have enough of a grip on the notions of description proper and of expression - just as we do on the notions of description proper and of performance - to see that no discourse (or at least no sentence of a discourse) can have both functions at once. So ethical language isn't descriptive proper. Moreover this notion of the descriptive is genuinely distinct from, because narrower than, that of the truth-apt. Performatives or fictional mimicries of descriptions proper aren't descriptions proper but they are truth-apt, and if Blackburn is right about the expressive (and thus non-descriptive-proper) function of ethical discourse then ethical discourse would be another example of the same combination of features. But distinguishing truth-apt states from descriptions proper goes no way towards proving the vital distinction between truth-aptness and having truth-conditions. For, for all that has so far been said, states with the (highly specialized) descriptive-proper function and states with the (distinct, and equally specialized) expressive function could both be both truth-apt and possessed of truth-conditions. So there is no route to the needed distinction via reading the 'descriptive function' in the sense of functioning as a description proper.
However, Blackburn also understands 'descriptive' in terms of direction of fit, as at least co-extensive with having that direction of fit where the state is supposed to fit the way the world is. Let's call states that are descriptive in this sense 'world-fitting' states. Now to show that ethical beliefs have the converse (for want of a better word, 'state-fitting') direction of fit to world-fitting states would show that they were not world-fitting, granted only that no state can have both directions of fit. But would showing that an ethical belief isn't world-fitting be enough to show that it does not have truth-conditions? If we focus on the standard example of a state-fitting state – desire – it can seem as if this is so. I want to argue, however, that it is an accidental feature of desires that they are both state-fitting (and so not world-fitting) and not possessed of truth-conditions. Let us think again about statements of intention, in particular statements of what one is currently intentionally doing such as 'I am pinning the tail on the donkey' as said when blindfold. Statements of intention are, I take it, state-fitting not world-fitting: if I am pinning the tail on the donkey's nose, for instance, then I have to fit the relevant bit of the world (my action) to what I have said, not the other way about. However, there must be something about the statement of intention in virtue of which what one is doing can be said either to be or not to be what one intends, or else there would be no standard to guide one in trying to change one's action to fit the statement. And once again, nothing has so far been said that would warrant classifying this something as truth-aptness but not as the possession of truth-conditions. Nor is it, as in the case of desires, simply that a content-specifying component – the Satzradikal – is true or false of the way the world is, a point which could be made of absolutely any attitude with a propositional content and which would therefore show nothing about the attitude's possession or non-possession of truth-conditions. On the contrary it is the whole statement of intention that is in question: the world doesn't match what the statement of intention says. So although 'state-fitting' may imply 'not world-fitting', from the fact that a use of language (or, by a presumably legitimate extension, a state) is not world-fitting we cannot read off that it does not have truth-conditions. Reading 'descriptive' as 'world-fitting', therefore, ethical beliefs might well fail to be descriptive and yet have truth-conditions nonetheless.
This conclusion has a bearing on Blackburn's last candidate for a genuinely functional characterization of ethical beliefs, namely that ethical beliefs are motivating - as well they may be. That ethical beliefs are motivating shows that they are non-truth-conditional only granted two assumptions: first, that only state-fitting states can motivate and, secondly, that state-fitting states are non-truth-conditional. Someone who went along with the second assumption might challenge the first assumption, and thence Blackburn's conclusion, by citing the fact that ethical sentences are truth-apt, or express beliefs, claiming this as evidence that ethical beliefs, though motivating, are both truth-conditional and world-fitting. The foregoing argument, by challenging the second assumption, opens up the possibility of reaching the same point while leaving the first assumption in place. The fact that ethical sentences are truth-apt, or express beliefs, is indeed prima facie evidence that ethical beliefs have truth-conditions, and the fact that they are also motivating, even if it is true that only state-fitting states can motivate, is no reason to deny this.
Neither the 'deep semantic' nor the 'functional' ground for distinguishing those truth-apt discourses which express representational states from those which do not seems to work. The obvious explanation for this, though it has not exactly been proved, is that the distinction is chimerical: the only notion of possessing truth-conditions going coincides with that of truth-aptness. But that is what the principle behind Wright's challenge - that what looks propositional is propositional - predicts, and the only two ways of denying the principle have been ruled out by the conclusions of §3 and §5 respectively. This pessimistic conclusion still leaves room for one view of the relation between non-representational states and ethical discourse, suggested by some of Blackburn's remarks, which is that although the states expressed by ethical discourse both appear to be and are representational, they "started life" as (i.e. evolved from) non-representational states. This developmental picture gives the notion of projection a more substantive role to play than it has had on any of the quasi-realist options considered so far. But - whether it is understood ontogenetically or phylogenetically - it is not enough for expressivism: expressivism requires the states fundamentally expressed by a mature person's ethical discourse to be non-representational as we find them, now. Blackburn's claim to have developed a form of expressivism that saves the ethical appearances has not therefore been made out.

References

Anscombe, G.E.M., Intention, Cambridge MA: Harvard University Press, 2000
Blackburn, Simon, "Attitudes and Contents", in Essays in Quasi-Realism, pp 182-97

--, "Errors and the Phenomenology of Value", in Blackburn, Essays in Quasi-Realism, pp 149-65

--, Essays in Quasi-Realism, New York/Oxford, Oxford University Press: 1993

--, "How to be an Ethical Anti-Realist", in Blackburn, Essays in Quasi-Realism, pp 166-181

--, "Morals and Modals", in Blackburn, Essays in Quasi-Realism, pp 111-30

--, "Realism, Quasi, or Queasy?", in J. Haldane and C. Wright (eds.), Reality, Representation and Projection, pp 365-84

--, Ruling Passions, Oxford: Oxford University Press, 1998

--, Spreading the Word, Oxford: Oxford University Press, 1984

--, "Wittgenstein, Wright, Rorty and Minimalism", Mind 107 (Jan. 1998), pp 157-181

--, commentary on Ronald Dworkin, "Objectivity and Truth: You'd Better Believe It", e-mail Symposium on BEARS-L, 1997

Haldane, John and Crispin Wright (eds.), Reality, Representation and Projection, New York/Oxford, Oxford University Press: 1993

Hale, Bob, "Can there be a logic of attitudes?", in J. Haldane and C. Wright (eds.), Reality, Representation and Projection, pp 337-63

Hooker, Brad (ed.), Truth in Ethics, Oxford: Blackwell, 1996

Mackie, John, Ethics: Inventing Right and Wrong, Harmondsworth: Penguin, 1977

Wright, Crispin, Truth and Objectivity, Cambridge, MA/London: Harvard University Press, 1992

--, "Truth in Ethics", in Brad Hooker (ed.), Truth in Ethics, pp 1-18

--, review of Simon Blackburn, Spreading the Word, Mind 94 (1985), pp 310-19



I use the term 'anti-realism', as Dummett originally intended it, as the most neutral possible way of flagging opposition to realism. I do not mean to suggest that expressivism and the error theory exhaust the options for the ethical anti-realist.
Perhaps it is misleading to present emotivism and quasi-realism as distinct doctrines: after all, it was never part of emotivism to claim that the ethical appearances didn't need to be saved. Properly speaking, quasi-realism is an evolved version of emotivism which claims to have blunted a well-known objection to it.
"The [quasi-realist] account ... starts with a theory of the mental state expressed by commitments in the area in question" (Blackburn, "Morals and Modals", in Essays in Quasi-Realism (New York/Oxford, Oxford University Press: 1993), p 55).
"On the one hand there is theorizing about some area of our commitments by seeing them in terms of beliefs, answering to facts or possessed of truth-conditions. ... On the other hand there is theorizing about the same commitments by seeing them as something different: expressions of attitude, or of other dispositions" (Blackburn, "Realism, Quasi, or Queasy?", in John Haldane and Crispin Wright (eds.), Reality, Representation and Projection (New York/Oxford, Oxford University Press: 1993), p 365). See also Blackburn, Essays in Quasi-Realism, Introduction, p 3: "a 'non-descriptive' or non-representational view of our commitments ... is easily contrasted with a realist, or descriptive or representational story".
For the idea of "fundamental expression", see Blackburn, "Attitudes and Contents", in Essays in Quasi-Realism, p 192.
I follow Blackburn in using the catch-all term "discourse" here, though it is intended to capture not only ethical talk, but also ethical beliefs, practices, reactions, the sense of obligation, and more besides.
"Realism, Quasi, or Queasy?", p 365.
I say the T-features because it's important for quasi-realism that, whatever T-features a proposition-expressing discourse has, ethical discourse has all of them. However, the presence of the definite article is not meant to rule it out that different features might be reckoned among the T-features on different formulations of quasi-realism: see below, §5.
"Realism, Quasi, or Queasy?", p 365.
Essays in Quasi-Realism, Introduction, p 5.
See "Realism, Quasi, or Queasy?", p 365; also "Errors and the Phenomenology of Value", in Essays in Quasi-Realism, p 152, where the error theory is described as "revisionist projectivism". Even if Blackburn were otherwise right about this identification, a qualification would be needed here: Mackie indeed maintains that first-order ethical discourse has consequences which are contradicted by his anti-realist metaphysics, but - surprisingly, as Blackburn notes - does not say that we should revise the discourse. I keep the label "revisionary" simply to label the fact that there is such a logical collision, but perhaps "potentially revisionary" would be more accurate.
Blackburn, "How to be an Ethical Anti-Realist", in Essays in Quasi-Realism, p 167.
Mackie, Ethics: Inventing Right and Wrong, (Harmondsworth: Penguin, 1977), pp 18 and 33.
Mackie, op. cit., p 35.
As e.g. at Mackie, op. cit., p 42.
Blackburn, "Errors and the Phenomenology of Value", contains a formulation of projectivism which could be said to be common to both Blackburn and Mackie: "Let us call the Humean picture of the nature of morality, and of the metaphysics of the issue, projectivism. On this view we have sentiments and other reactions caused by natural features of things, and we 'gild or stain' the world by describing it as if it contained features answering to these sentiments, in the way that the niceness of an ice cream answers to the pleasure it gives us" (Essays in Quasi-Realism, p 152). But the commonality comes at the price of unhelpful generality, since the formulation masks the fact that projectivism of Blackburn's type is in part a semantic thesis whereas Mackie's is not. In Mackie, "projectivism" simply labels an explanatory thesis about how the error-laden discourse came into being.
Or not at any rate any distinct object-level thesis. This point is obscured, I think, by e.g. Essays in Quasi-Realism, Introduction, p 9, where Blackburn speaks of a "successful projective plus quasi-realist story".
Review of Simon Blackburn, Spreading the Word, Mind 94 (1985), p 318.
Of course as applied to propositionality, the notion of an appearance need not be thought of in perceptual terms. The notion I am trying to capture is rather that of an intrinsic evidential link (fallible or infallible) or, alternatively, the absence of any such link, between propositionality and certain features of a discourse.
Essays in Quasi-Realism, Introduction, p 3, italics mine.
"Attitudes and Contents", in Essays in Quasi-Realism, pp 187-8. Another implication is that the function of linguistic expressions is something that they can 'wear on their faces'.
Ruling Passions (Oxford: Oxford University Press, 1998), p 77, my italics.
E.g. the "realist-seeming grammar" of ethical discourse, "Errors and the Phenomenology of Value", in Essays in Quasi-Realism, p 152. The implications of "realist-seeming" of course go beyond those of the other locutions mentioned: ethical discourse's appearing to be propositional is not the same as its appearing to be such that a realist account is true of it, since the latter "appearance" would require the assumption that the only touchstone of realism is whether or not the states expressed by a discourse are representational or not (conversely, that expressivism is the only alternative to realism in ethics).
See "Attitudes and Contents", in Essays in Quasi-Realism, p 187, for some (doubtfully grammatical) examples of what the sentences of such a discourse would look like.
'Realism, Quasi, or Queasy?, p 379.
See "Morals and Modals", in Essays in Quasi-Realism, p 58.
Incidentally, I take it that Wright's other challenge to quasi-realism - that inferential error for the quasi-realist can amount at most to a moral not a logical defect: see Wright, "Truth in Ethics", in B. Hooker (ed.), Truth in Ethics (Oxford: Blackwell, 1996), p 4 - would be met if such a notion of consistency could be made out. The logic of ethical discourse, according to Blackburn's reconstruction programme, is one of the few things about it that isn't "quasi".
See e.g. "How to be an Ethical Anti-Realist", in Essays in Quasi-Realism, p 172: quasi-realism must show that expressivism does not imply the truth of e.g. "If we had had different sensibilities, cruelty would not have been wrong".
See e.g. "Errors and the Phenomenology of Value", in Essays in Quasi-Realism, p 153: "the objective 'feel' that a properly working morality has".
It's therefore unsurprising that Blackburn sometimes warns us off understanding him in this way: "I have called [my theory] 'projectivism', but that can sound misleading. It can make it sound as if projecting attitudes involves some kind of mistake ... . This is emphatically not what is intended" (Ruling Passions, p 77). See also ibid., p 319.
See Blackburn, Spreading the Word (Oxford: Oxford University Press, 1984), ch. 6, and "Attitudes and Contents", in Essays in Quasi-Realism, and also Bob Hale, "Can there be a logic of attitudes?", in Haldane and Wright, op. cit..
Hale (op. cit., p 342) says the programme aims to show how a "patently expressive language" could do the same as ordinary (ethical) English, but in fact his Eex is a formal representation of the structure of an (actual or possible) language, not a rival natural language.
I stress that the two analyses must be incompatible, since there can obviously be different compatible analyses of the same English argument - revealing different amounts of logical structure - none of which may have an overriding claim to represent the logical form of the argument.
Ruling Passions, p 72.
For Blackburn's version of interpretationism, see Ruling Passions, ch. 3 §2.
It might be objected at this point that the data for interpretation are being too narrowly construed, and that the fact (for example) that ethical attitudes are motivating ought to feature among them too. But though there is nothing wrong with counting that fact among the data, doing so isn't going to provide the quasi-realist with any extra help. For the fact that ethical attitudes are motivating will help the quasi-realist only in conjunction with the assumption that genuinely propositional attitudes are not motivating, and whether ethical attitudes are genuinely propositional or not is, on the quasi-realist's admission, something the data for interpretation leave open.
Cp. Wright, Truth and Objectivity, p 36: "There is no well-conceived deeper notion of a genuine conditional nor, correspondingly, any well-conceived deeper notion of assertoric content which goes along with the capacity to feature as the antecedent of a conditional".
"Realism, Quasi, or Queasy?", p 366. He also accepts Wright's conditional claim that "we cannot say 'such-and-such commitments are not beliefs; they are other states that properly sustain the appearance of being beliefs', if properly sustaining that appearance is all there is to being a belief", but it is his acceptance of the antecedent that is crucial.
From Blackburn, "Realism, Quasi or Queasy?", p 367, as are the terms "contrastive" and "inclusive".
See, for example, Blackburn, "Wittgenstein, Wright, Rorty and Minimalism", Mind 107 (Jan. 1998), p 174, "genuineness of truth-aptness is not in question"; and Ruling Passions, p 79, "since we can handle the ethical proposition exactly like any other, it is not mistaken to say that we voice belief in it, when we do".
See Blackburn, "Wittgenstein, Wright, Rorty and Minimalism", p 159.
And with states which have a descriptive as opposed to a non-descriptive functional role, but more of that later (see below, §6).
Ruling Passions, p 75. Wright of course distinguishes them too: see his Truth and Objectivity (Cambridge, MA/London: Harvard University press, 1992), chs. 1-2.
"Attitudes and Contents", in Essays in Quasi-Realism, p 182.
Ibid., p 183: if modus ponens is not "a diagnosis of ... syntactical form", "then we have no effective procedure for telling when anything is".
Ibid..
Ruling Passions, p 78. My hesitation in attributing the point to Wittgenstein himself is that, though Wittgenstein says something which sounds very similar, I suspect that Blackburn and Wittgenstein mean something different by "descriptive".
Blackburn commends this idea of Gibbard's at Ruling Passions, p 73.
Blackburn seems prepared to acknowledge the conclusion of at least this more cautious argument: as he says, "there is a level of [semantic] analysis at which modus ponens and the rest are no different when their components are evaluative from when they are not" ("Attitudes and Contents", in Essays in Quasi-Realism, p 186). But then it becomes still harder to find something which will distinguish discourses that express representational states from ethical discourse which supposedly does not.
Ruling Passions, p 77.
Ibid., p 79. I'm assuming here that Blackburn is speaking for himself rather than just for Wittgenstein or Ramsey.
Ibid., p 80.
See Blackburn, commentary on Ronald Dworkin, "Objectivity and Truth: You'd Better Believe It", e-mail Symposium on BEARS-L, 1997.
Ruling Passions, p 77. For the "non-descriptive functional location" of ethical commitments, see also "Realism, Quasi, or Queasy?", p 378.
I'm assuming it will be agreed that performatives aren't descriptions proper. But the only reason I can think of for denying this is the desire to oppose the (Wittgensteinianly inspired) idea that performatives don't have a truth-value. This idea is certainly worth opposing, but once we have got hold of the distinction between truth-apt uses of language (a broad category) and descriptions proper (a specialized category) there is no need to oppose it in this way.
This is of course controversial, but it is not the place to enter into the controversy here.
Or as Anscombe put it, 'the facts are so to speak impugned for not being in accord with the words rather than vice versa', G.E.M. Anscombe, Intention (Cambridge MA: Harvard University Press, 2000), p 5.
It might be objected that commands are both state-fitting and such that what's done can be said either to be or not to be what's commanded, but clearly do not have truth-conditions; so the same combination of features in statements of intention implies that statements of intention have no more than what commands are agreed to have, namely fulfilment-conditions. But why assume that having truth-conditions and having fulfilment-conditions are properties of the same level and which exclude one another rather than that the latter is the more inclusive of the two, unless by assuming the point at issue, viz. that state-fitting utterances (etc.) cannot have truth-conditions? If the assumption is not made, we can concede that statements of intention have fulfilment-conditions but, thinking of this as the more inclusive property, maintain that that in virtue of which statements of intention have fulfilment-conditions is precisely their possession of truth-conditions.
See e.g. "How to be an ethical anti-realist", in Essays in Quasi-Realism p 168: "the nonrepresentative, conative function [of ethical attitudes]".
There is of course a third assumption, already mentioned, that states are state-fitting if and only if they are not world-fitting, but I set this aside as before.
See e.g. Essays in Quasi-Realism, Introduction p 9, "[T]he Humean is still in business if [the] irreducibility and particularity [of valuing something] is visibly the upshot of a natural process of voicing and projecting the non-representational states she starts with"; also ibid., p 5 and p 208. This genetic claim also fits something Blackburn says about his "overtly expressive" ethical language: the more closely the overtly expressive language matches actual ethical discourse in expressive power, the less is involved in "jumping ship - changing from an expressivist language to our normal forms" ("Atitudes and Contents", in Essays in Quasi-Realism, p 197). But there is a tension here with Blackburn's claim that ethical discourse complete with declarative sentences capable of negation, conditionalization etc. "is designed or invented or emerges naturally as the focus for our practical transactions" (Ruling Passions, p 70), i.e. as a way of organizing the relations among our ethical attitudes, since the less there is involved in jumping, the harder it is to see why we bothered to jump.
Thanks to Roger Crisp and Alan Thomas for helpful comments.



23