This is a chapter from Nathan Nobis’s Ph.D. dissertation Truth in Ethics and Epistemology: A Defense of Normative Realism.


CHAPTER 3:  Ayer and Stevenson’s Ethical and Epistemological Emotivisms

3.1. Introduction.

A.J. Ayer and C.L. Stevenson advocated ethical emotivisms, non-cognitivist understandings of the meanings of moral terms and functions of moral judgments.  According to ethical emotivism, to make a moral judgment is to express one’s emotions and invite others to share them: it is not to state a moral proposition, be in a representational mental state like belief, or attempt to assert moral facts, attribute moral properties or describe moral reality. 

I argue that the reasons Ayer and Stevenson gave in defense of ethical emotivisms suggest analogous epistemological emotivisms. According to this kind of view, epistemic judgments are neither true nor false either: to judge that some belief is epistemically reasonable, justified, or known; or that some proposition epistemically ought or should be believed; or to that some reasoning is good reasoning is also to express one’s emotions and invite others to share some feelings about a belief.

The common view, however, is that epistemic judgments are propositional attitudes: they are epistemic beliefs, attempts to attribute epistemic properties or state epistemic facts. Even most skeptics agree: they typically argue only that it is false that beliefs are justified or known, not that epistemic judgments are neither-true-nor-false.

Epistemic emotivism is thus at odds with standard assumptions. For many, it might seem obviously implausible and worthy of immediate dismissal: it’s just obvious that epistemic judgments are not emotive, and so it’s obvious that epistemic emotivism is false and the arguments for it unsound. Other might respond that, although it’s not obvious that epistemic emotivism is false, reflection on what epistemic evaluations seem to be like reveals evidence that epistemic emotivism is false or, at least, more doubtful than not: critical reflection reveals more reasons to reject it than accept it.

My response to epistemic emotivism will be more along the second lines.[1] I find no reason why an epistemic emotivism must be false or why it must be unreasonable or unjustified for every person who considers it. Few, if any, views are essentially such that they ought to be rejected, and epistemic emotivism does not seem to be a contender for that possible class of views. The meanings of epistemic terms might be emotive or expressive: there seems to be little or no reason why they must be descriptive or why standard assumptions about the semantics and metaphysics of epistemic judgments must be correct. Nevertheless, I will argue that there are better reasons to reject that view than accept it, especially when we see its far-reaching implications for epistemic evaluations, evaluative judgments involved in reasoning and evaluative presumptions that motivate rational inquiry in general.

I acknowledge that these implications quite literally can be resisted, i.e., someone could sincerely affirm all that follows, as a matter of logical consequence, from epistemic emotivism for the nature of epistemic, and other intellectual, evaluations. I will argue that this, for most people, is an unreasonable position. It strains belief because it is contrary to so much else that, for most people, seems true about epistemic evaluations. Thus, most people should not accept epistemic emotivism and the arguments that might be presented in favor, at least those that are parallel to the arguments Ayer and Stevenson give in favor of their ethical emotivisms. Since these arguments for epistemological emotivism are parallel to the arguments given in favor of ethical emotivism, and the former arguments ought to be rejected as unsound, this provides strong reason to reject the arguments for ethical emotivism. Thus, I undercut the arguments in its favor that Ayer and Stevenson provide.

I will argue also that epistemic emotivism has an immediately important implication for meta-ethics in that it seems to undercut any epistemic support Ayer and Stevenson offered for their ethical emotivisms. This is because if epistemic emotivism is true then all epistemic judgments are neither true nor false, so it is neither true nor false that anyone should accept ethical emotivism or is justified in believing it (or any other view, including epistemic emotivism, for that matter). But if it it’s not true that anyone, from an intellectual point of view, should accept emotivism, then no one should.

Some epistemic emotivists might accept this consequence happily, agreeing that there is no truth to their judgment that they, or anyone, ought to accept epistemic or ethical emotivism. They might affirm this as their position as the basic nature of epistemic evaluations and respond that objections to this position are based only in contrary epistemic expressions which are equally untrue also. That might be the truth about such matters, but, again, I will give reasons to think that this position is, for most people, unreasonable.

I argue that while Ayer and Stevenson might have had strong feelings for their views, those feelings, if they are expressions of epistemic approval, do not provide good reasons to accept ethical emotivism. If that is so then, truthfully, both ethical and epistemological emotivisms should be rejected, and that’s not just how we, or anyone, might merely feel about it.

3.2. Ayer on Ethical Naturalisms and Non-Naturalisms.

Like other logical positivists, Ayer was an ethical emotivist. He also thought emotivism was “valid on its own account” apart from positivistic inspirations.[2] I will argue that the arguments he offers for ethical emotivism suggest analogous arguments for epistemological emotivism as well. I then evaluate this result.

Ayer considered the famous concluding paragraph from Hume’s Enquiry Concerning Human Understanding an “excellent statement of the positivist’s position”[3]:

When we run over [to] libraries, persuaded of these [empiricist] principles, what havoc must we make? If we take in our hand any volume of divinity or school metaphysics, for instance, let us ask, Does it contain any abstract reasoning concerning quantity or number? No.  Does it contain any experimental reasoning concerning matter of fact and existence? No.  Commit it then to the flames, for it can contain nothing but sophistry and illusion.

Positivists accepted Hume’s sentiment: they claimed that a sentence is either true or false, i.e., it expresses a proposition and so might be believed, if, and only if, it is either analytic or empirically verifiable.[4]

A sentence was considered empirically verifiable, roughly, if it was an “observation statement” or entailed by observation statements in conjunction with analytic propositions.[5] Analytic propositions were said to “enlighten us by illustrating the way in which we use certain symbols.”[6] Too little was said about the meaning of “meaning,” but a sentence was considered analytic “if it is true solely in virtue of the meaning of its constituent symbols, and cannot therefore be either confirmed or refuted by any fact of experience”: tautologies and formal propositions from logic and mathematics were considered analytic.[7] Sentences that were either analytic or empirically verifiable were said to be “cognitively meaningful” or have “cognitive meaning.” Only then might they be true or false; only then might they be true.

Ayer conceded that this criterion was never adequately formulated,[8] little argument was ever offered in its defense, and it is cognitively meaningless – and so not true, since neither true nor false – according to its own standards. But, ignoring these problems, if it is combined with the premise that judgments about what is morally good/bad, right/wrong, just/unjust and virtuous/vicious are neither analytic nor empirically verifiable, it validly implies that these judgments are neither true nor false.

Ayer accepted that argument. He thought that moral judgments were neither analytic nor empirically verifiable. Sentences like “Eve’s beating her baby today was wrong” or “Causing pain for fun is wrong” are not analytic since, if true, are not true in virtue of meaning (on nearly any view of meaning), are not formal truths or definitions, are not contradictory to deny and can be refuted by an empirical fact such as that Eve never beat her baby. If analytic, they would have none of these features.

Ayer’s arguments that moral judgments were not empirically verifiable depended on his arguments against ethical naturalisms and non-naturalisms.

Naturalists defined moral terms in naturalistic terms. They claimed that moral expressions are synonymous with empirical expressions and that, therefore, moral properties just are natural properties. Utilitarian-naturalists claimed that “x is right” means “x produces the most happiness”; subjectivist-naturalists reduced judgments of rightness to judgments of individual or collective approvals. Naturalists argued that since these latter judgments are verifiable, moral judgments are verifiable as well. 

Ayer argued that naturalistic definitions were not analytic since it isn’t self-contradictory to deny them and it is an “open question” for competent speakers whether acts with some proposed natural property (e.g., maximizing pleasure) have some moral property (e.g., being right). If the naturalists’ various definitions were correct, these questions would be as “closed” as the question are “right actions right?”[9]  That question’s answer is “obviously, yes!” but answers to questions whether acts with the proposed natural property are right are, at least, not obvious, or, according to some critics of naturalisms, obviously not what any naturalists thought.

Ayer concluded that naturalist definitions were not “consistent with the conventions of our actual language”[10] and, thus, their analytic bridges to reduce moral judgments to empirical judgments were burned down. His arguments here did not depend on positivism; Moore made the same arguments. But Ayer could have also observed that naturalistic definitions (or, sentences where these definitions are offered) are not empirically verifiable either and concluded that, according to positivism, they are neither true nor false. However, he seemed content with these definitions being only false.

Non-naturalists, like Moore and Ross, whom Ayer called “absolutists,”[11] claimed that moral terms were either indefinable or definable only in other moral terms: e.g., “x is right” is synonymous with “x produces the most good,” but “good” is either indefinable or definable only in other moral terms (e.g., “worth having for its own sake”[12]). They claimed moral judgments were synthetic but that moral properties could be verified by non-empirical intuition. 

But Ayer rejected appeals to intuition, calling it “mysterious.”[13] Many claim to intuitively be able to “see” that something is morally good (or bad). However, there are disagreements and since there is no empirical test to adjudicate between competing moral visions, Ayer thought these appeals were “worthless.”[14] He argued that since there is no moral intuition, no terms’ meanings are such that they stand for properties that can be verified non-empirically. He concluded that non-naturalists’ epistemology, semantics and metaphysics were mistaken.

Unlike his arguments against naturalism, these arguments are distinctly positivistic.[15] But, even if one rejected positivism, one might find appeals to intuition independently suspicious and draw not merely a skeptical conclusion (i.e., “There is goodness, but unfortunately no belief that something is good is ever epistemically justified or known”) but positivist-like semantic and/or metaphysical conclusions. As we shall see, Hare, Mackie and Harman do just this.

3.3. Ayer’s Ethical Emotivism.

If moral terms have cognitive meaning (i.e., if “cognitivism” is true and so moral judgments are either true or false[16]), their meanings must be the same as either naturalistic or non-naturalistic expressions’ meanings. Ayer thought he refuted both possibilities (and if moral terms lack cognitive meaning, that’s another reason to think that moral judgments aren’t analytic, i.e., true in virtue of cognitive meaning). 

His final assessment was that ethical concepts are “pseudo-concepts,”[17] but he did not think that they were meaningless, in the sense of being incomprehensible gibberish. His proposal was that moral terms were meaningful but that their meaning was of a different kind: emotive, not cognitive. The meanings of these terms are such that when people use them, to make moral judgments, they are expressing (not describing) states of mind that are “simply expressions of emotion which can be neither true nor false.”[18] He suggested that moral judgments “are calculated also to arouse feelings, and so to stimulate action.”[19] We might say that he thought there meanings are such that their sincere use had these functions.

He illustrated his position this way:

If . . I . . say, “Stealing money is wrong,” I produce a sentence which has no factual meaning—that is, expresses no proposition which can be either true or false. It is as if I had written “Stealing money!!”—where the shape and thickness of the exclamation marks show, by a suitable convention, that a special sort of moral disapproval is the feeling which is being expressed.[20] It is clear that there is nothing said here which can be true or false. Another man my disagree with me about the wrongness of stealing, in the sense that he may not have the same feelings about stealing as I have, and he may quarrel with me on account of my moral sentiments. But he cannot, strictly speaking, contradict me. For in saying that a certain type of action is right or wrong, I am not making any factual statement, not even a statement about my own state of mind. I am merely expressing certain moral sentiments. And the man who is ostensibly contradicting me is merely expressing his moral sentiments. So there is plainly no sense in asking which of us is in the right. For neither of us is asserting a genuine proposition.[21]

Cognitivists, i.e., naturalists and non-naturalists, agree that strong feelings sometimes, if not often, accompany moral judgments and that we often voice moral judgments to try to influence feelings and behavior. But they deny that this is a necessary consequence of the meanings of moral terms; they maintain that one could sincerely make a moral judgment that fails to influence any feelings or behavior (including one’s own).  Since emotivists think that moral judgments just are the expression of feelings, they deny this possibility.

This is Ayer’s perspective on ethics. Later I will discuss Ayer’s response to the “only criticism which appeared to threaten it.”[22]

3.3.1. Ayer, Positivism and Epistemology.

While many objections have been raised to emotivism, I wish to raise a new objection to Ayer’s defense of it.  He said his emotivist analysis applied to aesthetic judgments, but my interest is what this perspective implies for epistemic judgments and definitions. On the face of it, it seems that they would have a similar fate. I will argue that this is a rationally unacceptable implication.[23]

Consider some particular epistemic judgments, e.g., some of Ayer’s. Surely he, like many people, thought of many of his beliefs were reasonable and justified for him, if not sometimes known. Using his preferred terminology, he might have said he had “the right to be sure” of them.[24] But it seems that these epistemic judgments (e.g., “I am justified in believing that the lights are on”, “It is reasonable to believe that I have hands”, “As a positivist, I have a right to be sure of the emotive account of ethics”, etc.) are not analytic since they are not definitions, tautologies, or formal truths: by the standards Ayer accepted in ethics, investigating the meanings of the words or linguistic conventions would not reveal their truth values. And, at least prima facie, they don’t seem empirically verifiable either: scientific observation might one day, presumably, reveal that one has some belief but it wouldn’t show its epistemic status.[25]

These considerations seem applicable to any epistemic judgment that some belief is justified or reasonable, or is something that one should or ought believe, and so on. So it is plausible to think that, on positivistic grounds, particular epistemic judgments are – like ethical judgments – also neither true nor false.[26]  Since epistemic platitudes, such as Ayer’s suggestion that “certain standards of evidence ought always to be observed in the formation of our beliefs,”[27] are neither analytic nor empirical either, it’s presumably not true that you ought to follow your evidence, and it’s not false either. 

One might resist these conclusions by arguing that they follow only if one is sufficiently inattentive to the meanings of epistemic terms. Perhaps, given true definitions or analyses of epistemological terms, epistemic judgments meet the positivistic criterion for cognitive meaning. 

While the question of moral terms’ meanings has been much discussed, there has been little comparable discussion about the meanings of epistemic terms. But if epistemic terms have cognitive meaning, those meanings are either naturalistic or non-naturalistic expressions. But then they fall prey to Ayer’s objections to ethical definitions. If ethical definitions were of no help in making moral judgments analytic or empirically verifiable (and hence either true or false), epistemological definitions are of no help either. If, however, these objections do not refute epistemological definitions, then they likely do not refute moral definitions either.

3.3.2. Against Naturalistic Epistemological Definitions.

Let us first consider some naturalistic epistemological definitions or analyses. While any would do, let use Ayer’s. He said that “we define a rational belief as one which is arrived at by the methods which we now consider reliable” and that “to be rational is simply to employ a self-consistent accredited procedure in the formation of all of one’s beliefs.”[28]

            These analyses were likely attractive to him since they identified epistemic judgments with psychological-causal-logical judgments that (ignoring possible grave difficulties in identifying “methods” and “procedures” and understanding “reliability”) were, at least in principle, empirically verifiable. In this way they were like naturalistic definitions of ethical terms. But then they were, therefore, also subject to the exact same objections, which Ayer did not seem to realize or press against them.

We might grant that the presence of these psychological-causal-logical states is in principle empirically verifiable: perhaps brain scanners will one day be able to reveal them. However, to deny these naturalistic definitions does not seem to utter a contradiction. Although this is not fully clear since Ayer does not define “reliable,” it does not seem that one contradicts oneself by saying “this belief is rational but it was not brought about by a method which we now consider reliable.” Linguistic conventions surely were never so much in Ayer’s favor to make that claim analytic.  These definitions are also subject to “open question” arguments, since one can sensibly (or quite confusedly) ask, “This belief has been formed by the employment of a ‘self-consistent accredited procedure,’ but is it a rational belief?”  If “rational” meant that, then there wouldn’t be a question here. But surely there is, so this definition, even if true, is not analytically true.

Ayer thought that these tests show that naturalistic definitions of ethical terms are not analytic, that these definitions are false, and, therefore, that moral properties are not identical to natural properties and that moral judgments are not empirical judgments.[29] Unless epistemic definitions are an exception to the semantic rules that Ayer accepted in the context of ethics (and there is no reason to think that they are), these tests also imply that naturalistic epistemological definitions are also not analytic, are false, and so epistemic properties are not natural properties and epistemic judgments are not empirical judgments either.[30]

So, given what Ayer said about ethical naturalism, he should have rejected his own naturalistic epistemological definitions. Although it is not clear what he thought about this, if he thought epistemic judgments were empirical judgments, this view was inconsistent with his reasoning against ethical naturalism (and if he thought recognized inconsistency should be avoided, that too is inconsistent with his principles). Since naturalistic epistemological definitions are also not empirically verifiable, it seems that, on positivistic principles, they should be considered not merely false, but neither true nor false. That also seems follow from Ayer’s position, which he also did not realize.

3.3.3. Against Non-Naturalistic Epistemological Definitions.

If naturalistic epistemological definitions are mistaken since they are not analytic, then a non-naturalistic definition must be correct, if any are. But Ayer’s objections to ethical non-naturalisms apply equally well to epistemic non-naturalisms also.

            If what a person believes can in principle be empirically verified, many aspects of beliefs are empirically verifiable as well. The various effects of those beliefs (i.e., causal properties of those beliefs) are presumably verifiable, e.g., that it makes someone happy or sad, or is financially productive. Beliefs’ psychological properties of being held with a certain level of strength, confidence or doubt are empirically verifiable. Formal, logical properties of a belief can also be identified, i.e., logical relations to other propositions, although positivists would likely claim that these relations can be identified because they are analytic.[31]

But whether some belief is justified or reasonable, or whether there is sufficient evidence for some belief, does not seem to be an empirical matter: observation and experiment do not show the epistemic status of a belief.[32] At best, lab results could show that some belief-forming process has some level of statistical reliability. That beliefs formed by these processes (or processes with some specified level of reliability, or probability, or any other candidate for a “natural” feature) are justified, or ought to be held, however, is not empirically confirmable. Reliabilism, as a philosophical theory, cannot be confirmed by science; according to arguments parallel to those Ayer gave about ethical naturalisms, it cannot be confirmed by an investigation of our epistemic language either, so positivism suggests that it’s neither true nor false. Scientific results do not show if some experience or belief is evidence, or if some evidence is sufficient evidence, if some proposition ought to be believed, or if some explanation is more reasonable than another. In fact, scientific thought and practice depends on antecedent capacities for epistemic assessment that do not seem to be ultimately confirmable in any straightforwardly empirical manner.[33]

Empirical information is often relevant to epistemic evaluation (e.g., whether S believes p is relevant to whether S’s believing p is justified). But identifying the epistemic status of belief seems to be a matter of something that might be best described as non-empirical intuition. And Ayer rejected views that appealed to intuition since he thought there simply is no such thing. He thought intuition was especially troublesome in ethics since there is no test to decide between conflicting intuitions other than further intuitions: there is no empirical tie-breaker. But the same is true in epistemology: there is no empirical test to adjudicate competing intuitions about what’s reasonable or justified, or whether some evidence is sufficient evidence, or if some belief-forming-process is the process that justifies some belief. If intuition is “worthless” in ethics, as Ayer thought, it is equally worthless in epistemology.

Ayer rejected ethical non-naturalism because of its intuitionism; he concluded that moral terms do not have the same meanings as non-natural terms, there are no non-natural moral properties and, thus, that moral judgments are neither true nor false.  It seems that analogous conclusions would follow for epistemology as well: any theory that implies that epistemic terms mean terms that are non-empirically verifiable is equally worthy of rejection from positivistic perspectives. In case he did, Ayer should not have thought that epistemic terms’ meanings are non-natural expressions. He should have rejected non-naturalistic definitions and analyses that claimed that epistemic terms are indefinable or definable only in other epistemic terms (e.g., “a belief is rational” means the same as “there are good reasons for that belief” or “a belief is justified if, and only if, it is supported by the evidence”). 

3.4. Ayer’s Epistemic Emotivism.

Since naturalistic definitions fail to be analytic and non-naturalistic definitions presuppose intuition, it seems that Ayer should have rejected them both. He should have concluded that epistemic terms lack cognitive meaning altogether and so epistemic judgments are neither true nor false. These results are not idiosyncratic to Ayer’s epistemology: on positivistic assumptions, contemporary epistemologies – both naturalistic and non-naturalistic – would result in epistemic evaluations being “cognitively meaningless” also.

Unless epistemic judgments are declared completely meaningless, it seems likely that, on positivism, their meaning would also be emotive. There are other options, however since there are other non-cognitivist interpretations of epistemic language that are, in some sense, “consistent” with epistemic sentences lacking truth values.[34]

But on epistemic emotivism, to sincerely judge that, e.g., some belief is justified (or unjustified) is to express one’s favorable (or unfavorable) feelings towards that belief and to attempt arouse the feelings of others regarding that belief and stimulate them to believe (or disbelieve) it. The view would have it that like other “statements of value,”[35] the presence of an epistemic term in a proposition adds nothing to its factual content. So, in saying “Your belief that p is unjustified and unreasonable” one might not saying anything more than ‘You believe p!!??’ in, like Ayer suggested about moral evaluations, “a particular tone of horror, or written it with the addition of some special exclamation marks” to express one’s feelings of disapproval and, perhaps, to try to command the believer to cease believing p.[36]

If epistemic emotivism is true, a seven-day creationist’s saying, “evolution is not supported by the evidence; it is unreasonable to accept the theory of evolution,” would not be an expression of a proposition; rather, it would be an expression of disapproval of belief in evolution and an attempt to persuade others to reject the theory. An evolutionist’s response that creationists are irrational and that, in fact, creationism is a well-justified theory would be equally emotive. Neither party would be attempting to state facts about the epistemic quality of their respective beliefs or attribute epistemic properties.[37] To judge that one knows that one knows something, or that one is justified in thinking some belief is justified, would seem to be expressions of complex feelings: some kind of approvals of one’s approvals. That is an interesting result.

This interpretation of epistemic judgments is strongly suggested by Ayer’s positivistic critique of ethics. Since most philosophers think that epistemic judgments are propositional attitudes, and epistemic emotivism denies this, it is at odds with common assumptions. If it is more reasonable to accept the common, realistic view about the nature of epistemic judgments, then it’s more reasonable to reject epistemic emotivism. This, in turn, might provide strong reasons to reject ethical emotivism and at least some arguments that are given in its favor.

Ayer might have agreed with this claim. As far as I know, not even Ayer and other positivists accepted epistemological emotivism, or even noticed that their broader, positivistic perspective suggested it. Perhaps this implication, had positivists seen it, would have led many to change their view about positivistic principles. Perhaps for many non-positivists, is shocking enough – due to its tension with other reasonable beliefs – to serve as (yet another) reductio of positivistic principles.[38] 

3.4.1. Epistemic Emotivism Undercuts Ethical Emotivism.

For now, I wish to note this emotive consequence for epistemic judgments has interesting and important implications for arguments from positivism to ethical emotivism. This is because epistemic emotivism has implications for what to think about all arguments and reasoning.[39] These implications are likely troublesome for positivists and those who reject ethical cognitivism for Ayer’s reasons.

To see this, consider the main argument under consideration in this chapter:

(1)   A sentence states a proposition if, and only if, it is either analytic or empirically verifiable.

(2)   Moral and epistemic sentences – particular judgments (and, clearly, naturalistic definitions) – are neither analytic nor empirically verifiable.

(3)   Therefore, moral and epistemic sentences do not state propositions.

Premise (1) is “paradoxical” in that if it is true, it implies that it is neither true nor false. Perhaps it was intended to express feelings of approval for analytic and empirically verifiable sentences and disapproval for all others, but this is doubtful. I will ignore these problems and, for the sake of argument, accept (1) as true. And I have argued that (2) as true as well, especially on positivistic assumptions. It has long been recognized as true for moral judgments, and I observed that it seems true for epistemic judgments also.

So, we have a recognized-as-valid argument with premises that, for the sake of argument, we accept as true. Now, the problem: should we accept the conclusion?  Would we be justified in doing so? That is, are the sentences “We should accept this conclusion” and “We are justified in believing (3)” true?  Neither of these sentences is analytic or empirically verifiable. So insofar as (1) can have implications, it seems to imply that these sentences, as epistemic evaluations, are neither true nor false.[40] A sentence even like “someone who believes an argument to be valid, thinks the premises are true, sees that the premises entail the conclusion, and is more confident of the premises than the denial of the conclusion ought to accept the conclusion” seems to be neither true nor false either: it is clearly not empirical, does not seem to be analytic, is not a logical truth and reflection on the meanings of its terms does not reveal it to be true. Given all this, positivism implies that these sentences are not true, since they are neither true nor false.

From positivistic perspectives, naturalistic epistemological definitions that might be offered (in conjunction with empirical premises describing psychological states) to reduce these epistemic judgments to empirical judgments would be refuted by open-question arguments; definitions that wouldn’t be refuted this way would require a kind of intuition that they found too mysterious. Thus, according to positivism, it is neither true nor false that anyone epistemically should (or should not) accept ethical emotivism. A response that “you should be an ethical emotivist but you should resist these arguments for epistemic emotivism,” would be neither true nor false either.

So, if epistemic emotivism is true, then it is not true that it ought to be accepted or anyone is justified or reasonable in believing it. If positivist-based epistemic emotivists would judge that they are reasonable or justified in their views, they would be saying something that, on their view, is not true. This is an interesting consequence, one that, probably, few positivists and ethical emotivists would have willingly accepted. Whether they should have resisted this conclusion and, if so, how they might have done so will be discussed below.[41]

            This interesting consequence extends beyond particular epistemic evaluations to evaluations about which inferences ought to be accepted. Logical relations among propositions are one issue and a fully legitimate one, on positivistic assumptions.[42] Discerning the logical consequences of one’s beliefs is surely part of what’s involved in reasoning, and it is the basis of reasoning. But reasoning involves attempting to see these logical consequences and attempting to discern which consequences we should accept. What you should conclude (given your experiences and other things you believe), how you ought to reason, what you are justified in believing, etc. are separate kinds of questions from the question of what are the logical consequences of a set of propositions. Not all epistemic judgments appeal to logic, and reasoning and evaluating arguments is more than displaying sets of statements in various logical relations. And seeing these logical relations does not establish what one, epistemically, should believe: it is an open question whether some proposition(s) in some logical relations to others ought to be accepted or is justified, even when these logical relations – these naturalistic features – are recognized. This is true on positivism, and it seems true on non-positivistic assumptions also.

Reasoning typically involves presumptions that “given this, you ought to believe that” and “you should drop this belief if you want to retain that belief”[43] and “it’s better to believe this than that” and so forth. In practice, it seems to presuppose an imperative of consistency, a presumption that it’s better to have consistent beliefs than inconsistent ones and that, at least sometimes, if one recognizes inconsistency, it should be resolved.

Typically, these epistemic “should’s”, “ought’s” and other value judgments are understood realistically. Few think that we are merely expressing our emotions when talking about good and bad reasoning, and what we should believe.[44] In cases where we might be expressing emotions in making epistemic judgments, few of us think that we are only expressing our emotions. Perhaps we are only expressing our emotions, but many would deny that and think that their denial is reasonable. There are reasons to think that epistemic judgments are sometimes true. Some of these are analogous to some of the reasons to given to think that moral judgments are sometimes true. Let me briefly present just a few of these reasons, in no particular order of strength or importance.

First, to many, it seems that epistemic error is possible. Many people think that there are propositions that they used to believe, and even currently believe, but that they were unreasonable or unjustified in believing them, given the evidence that they had at the time. This judgment suggests that beliefs can be unjustified (as well as justified), but we can fail to see this. It also suggests that epistemic judgments are descriptive, i.e., are attributions of epistemic properties, since we can think that some of our own epistemic judgments have failed to describe our own epistemic situations, and that they are made true by things other than our own attitudes toward them. Other examples might reveal that it seems we can accept general epistemic or intellectual standards that we later come to regard as quite mistaken.

These common insights seem to suggest that there are stance-independent epistemic truths. Epistemic emotivisms, and other epistemic irrealisms, might have a very hard time making sense of these phenomena: perhaps they could appeal to expressions of approval (and disapproval) for various expressions of approval (and disapproval), but, to many, this seems to be a strained description of the phenomena. Perhaps such a theory could be worked out, but I suspect an epistemic realist has stronger reason to accept his or her realistic interpretation of the phenomena of epistemic error.

Second, it seems that one can make epistemic evaluations without any emotional involvement. And it seems that any epistemic evaluations made initially with heated emotions can be restated later, in a cooler frame of mind, but the same thing is said: no meaning of the epistemic evaluation is lost. Epistemic realists have a simple explanation of how this can be, since they hold that, for any emotional factors related to epistemic evaluations, they are only contingently related. But on epistemic emotivism, these phenomena are either impossible since there are no emotion-free epistemic evaluations, or we are saying different things when we “cool down” and “heat up” in our evaluations. Both of these claims seem false, and this provides reason to reject epistemic emotivism. Again, while an epistemic emotivist surely has things to say here in defense of his or her point of view, I submit that these responses are weaker than the reasons that can be given in favor of the realist understanding of these phenomena.

Third, and related to the previous points, it seems that “a-epistemicists” are possible. These are believers who make epistemic and other intellectual evaluations, but completely lack any motivation or affective “pull” towards complying with their own evaluations: e.g., they judge that some of their beliefs are completely irrational and baseless, but this in no way motivates them towards changing their views. This possible phenomenon will be characterized in greater detail later in this chapter, but what’s important is that epistemic emotivisms would usually imply that such an “a-epistemicist” is impossible. However, if there is reason to think that there could be, and even are, such believers, this provides reason to think that epistemic emotivism is false. Again, while an epistemic emotivist surely has things to say here in defense of his or her point of view, I submit that these responses are weaker than the reasons that can be given in favor of the realist understanding of these phenomena.

Fourth, epistemic evaluations seem to behave like other descriptive language: they can figure into truth functional operations and can be part of seemingly valid arguments. So, this argument seems valid:

(1)   If each person should believe what’s justified for him or her, then Sally should believe what’s justified for her.

(2)   Each person should believe what’s justified for him or her.

(3)   Therefore, Sally should believe what’s justified for her.

Epistemic emotivism might make it a challenge to understanding why this argument – or other arguments with epistemic terms – is valid, just as ethical emotivism makes it a challenge to understand why an argument like this is valid also:

(4)   If lying is wrong, then it’s wrong to get your brother to lie.

(5)   Lying is wrong.

(6)   Therefore, it’s wrong to get your brother to lie.

The problem is that, if epistemic or ethical emotivism is true, then the meanings of (2) and (5) differ from the meanings of these phrases when used in (1) and (4) because, put most simply, these are different kinds of assertions. But if this is so, then these arguments are invalid, due to, put most simply, equivocations in the premises. If this is so, then the arguments are invalid. However, they seem valid, so it seems that the various phrases, when stated in the differing premises, have the same meaning. Since epistemic and ethical emotivisms imply that they don’t, this provides reason to think that these emotivisms are false.

Again, while both ethical and epistemic emotivists have responses to this problem, it’s safe to say that these responses are very complicated and that none have been met with much positive response. Perhaps this is due to stubbornness on realists’ part, but I submit these negative evaluations of these irrealistic responses to this problem are weaker than the reasons that can be given in favor of the realist understanding of these phenomena. Thus, it is more reasonable to accept the realist explanation of why arguments with epistemic and ethical terms can be, and often are, logically valid.

            A final point in favor of epistemic realism is that it just seems true. Judgments about what’s reasonable, justified, known, and should be believed just sometimes seem to be true: intuitively they seem true, and considerations given above and below are reasons that support this judgment. Thinking about our options can help. If epistemic evaluations are not ever true, then are they always false? Maybe this is the way things are, but this does not seem true either, and many have thought that there are strong reasons to resist epistemological skepticism, which this view might be a species of. Perhaps there are no such good reasons to reject skepticism, but many people would think that we should think that only if we have been given strong reasons to think so. But if this happens, then strong reasons have been given to think that there are no strong reasons. But this is a self-contradictory position that is, for many, unbelievable.

Epistemic judgments might be neither true nor false, but this does not seem true either. When ethical emotivists respond to the problem above with claims that when one is expressing the state involved in asserting a major premise of such an argument, and then one expresses the different state in asserting a minor premise, then one should express some final state also. They make all kinds of evaluative judgments in reasoning, and it seems that these are attempts to state truths, not merely expressions of their preferences for people accepting various patterns of reasoning (or, to describe reasoning in a manner without any evaluative language, accepting various sentences when displayed in various relationships to each other).

It seems that Ayer accepted this common view about reasoning: it does not seem that he was merely interested in noting various views about the nature of ethical judgments that are consistent with positivism, yet not claiming that some of them (like emotivism) truthfully should be accepted. But positivism seems unable to accommodate these judgments, since none of these claims about reasoning and what we should believe or what is reasonable meet their criteria for being cognitive meaningful. Ethical emotivists have been criticized for being unable to account for the role of reason in ethics, since they offered no mechanism to explain which emotive responses are appropriate (and typically claimed that there were no truths about which feelings “fit” a situation) and how one might give reasons for one’s ethical perspectives. But my arguments run deeper in that I’m arguing that, if positivism is the basis of their emotivism, then they are unable to account for good and bad reasoning simpliciter. 

This is because reasoning involves more than observing the natural facts that some propositions of interest are in some logical relations to other propositions of interest; it involves making these observations and then making judgments about what we should believe in light of them, or whether we ought to accept some consequence of our views, or whether we are justified in accepting what we have believed in light of this newly seen implication. Since positivism implies that all such judgments are neither true nor false, it seems to lack the resources to account for what’s involved in reasoning: it implies that the kinds of thoughts essential to reasoning are neither true nor false.  So, on their view, argumentation is based on emotions, never true epistemic principles about how we should respond to recognitions of logical truths, and so arguments for ethical emotivism are based on emotional expressions, not any true principle about reasoning. Ethical emotivists might express strong feelings about believing emotivism and try to arouse our sentiments so we might accept it, but it’s not clear how those are reasons to accept it, especially for those who don’t accept the view. And, at least if positivism is the motivation for ethical emotivism, there’s no truth to the claim that we should be reasonable, or ought to have intellectually defensible views anyway.[45] Those are also merely expressions, and there’s no truth to the claim that one should accept them or deny them. Either is equally untrue.

If positivism undercuts the truth of any epistemic judgments, including judgments about what is and isn’t good reasoning – again, since reasoning involves more than observing the natural facts that some propositions of interest are in some logical relations to other propositions of interest – this is yet another reason to reject positivism and undercuts a historically influential motivation for moral non-cognitivisms and irrealisms (that I suspect is still lingering in some “naturalistic” and “scientistic” perspectives). This is because positive implies something that is false and something, as I argued above, that we are reasonable and justified in regarding as false, viz. that there are no literally true epistemic evaluations and never any truths about how we ought to reason.

Below I will consider some ways that positivism-motivated epistemic emotivists might try to resist these arguments. First, however, I wish to sketch my general argument against arguments for moral irrealism, some of which I have developed above. Although I have repeatedly acknowledged that epistemic emotivism, or any other epistemically irrealistic view, might be true and more common views about the semantics, metaphysics, psychology, logic and epistemology of epistemic judgments false and rationally indefensible, I want to make it clear what I concede to be true of possible arguments for epistemic irrealisms and the exact nature of my criticisms.


3.4.2. Criticisms of and Concessions to Arguments for Epistemic Irrealism.

Throughout this work, I observe that we can understand moral irrealists as arguing for their preferred version of moral irrealism using a version of the argument stated with what we might call the general argument schema for moral irrealism:

(1)   If a judgment has features Φ, then it is not objectively true.[46]

(2)   Moral judgments have features Φ.

(3)   Therefore, moral judgments are not objectively true. 

From (3), positive reasons are offered to try to establish some kind of moral non-cognitivism, expressivism, prescriptivism, nihilism, relativism or other meta-ethical theory which, if true, implies that moral realism is false.

While arguments for moral irrealism are rarely formulated as simply and explicitly as this schema, the schema can help us clearly identify the basic reasons each classical moral irrealist gives for his position.[47] For each, we can ask what his basic “Φ” is: what features do moral judgments have such that he thinks that, because they have these features, an irrealist understanding of them is warranted? Each of these figures’ Φs can be used to state an instance of the general argument schema for moral irrealism.

These are simplifications that overlook important details which will be developed in later chapters, but, as we have seen, Ayer’s basic Φ is it’s that moral judgments are neither analytic nor empirically verifiable; he also offers some non-positivistic-based considerations which, as we will see, some later philosophers develop in support of their own views. We will see in the concluding section of this chapter that Stevenson’s Φs include that moral judgments have a motivational quality, that moral discourse has an emotional, dynamic aspect, and a Hume-motivated puzzlement about what a moral truth-maker could be like.

In later chapters, we will see that Hare’s Φs include the fact that people can disagree about their moral evaluations even when they accept different naturalistic and non-naturalistic understandings of the meanings of moral terms: they are not talking past each other, as these theories might imply. Mackie’s Φs include claims that moral judgments are intrinsically motivational, that moral properties are commonly thought to be supervenient on natural properties, and the claim that there is fundamental, intractable disagreement about what has which moral properties, i.e., disagreements about which moral evaluations are correct. And Harman’s main Φs pertain to moral disagreements and the judgment that moral properties don’t seem to be appealed to in order to explain our moral beliefs and attitudes or any other clearly natural phenomena. Contemporary moral irrealists (some of whom are also epistemic irrealists), like Blackburn, Field, Gibbard, Timmons, and Wright often accept and build upon many of these Φs in for their arguments against moral (and, sometimes, epistemic) realism. I noted this in my first chapter and will look at Gibbard’s and Field’s arguments in my final chapter.

I argue that all these arguments for moral irrealism have a problem: the exact nature of my criticism, and some interesting challenges in arguing for it, will be explained below.

My main objection is against the various first premises – instances of (1) – of this argument schema immediately above, so let me make a few remarks about the second premises of this kind of argument. For some of these alleged features of moral judgments, i.e., claims stated in various instances of premise (2), it’s very plausible to believe that moral judgments have them: e.g., it’s very plausible to think that Ayer is right at least in thinking that particular moral judgments are neither analytic nor empirically verifiable, and so this version of premise (2) is true. Some features expressed in alternative premise 2s, however, are at least controversial: e.g., it is not at all clear that moral judgments have an intrinsically motivational quality about them and so, for a variety of conceptual, phenomenological and empirical reasons, evaluating a premise like that is more challenging than, say, Ayer’s premise.

In face of these challenges, however, my general response is to concede that it is plausible to think that moral judgments have such features. So, for each given instance of (2) from the meta-ethical literature that I discuss, I either sincerely accept it or, for the sake of argument, I concede it as true. Thus, I agree with moral irrealists on nearly all points about what it’s like to make moral judgments or what moral judgments are like.

Since I find a major fault with arguments for moral irrealism, my objection then must be with instances of premise (1) of the general argument schema above, that if a judgment has features Φ, then it is not objectively true. I intend to argue that most actual, given instances of this premise are problematic:[48] each major premise has implications that we have better reason to regard as false, and so reject, than to accept as true. This defect renders the entire argument epistemically defective, since it implies that at the premise is either false or rationally unacceptable, or both.[49]

Before I turn to this objection, I will first discuss one response that I will generally not develop or rely on, as it is rather weak. Some instances of premise (1) could be challenged by claiming that no good reason, or inadequate reason, is given in their defense. A “lazy” version of this response could come from someone of a philosophical temperament who takes philosophical arguments seriously only when positive reasons have been given in defense of all premises of an argument. Someone like this might claim that, as far as she can tell, too rarely is positive defense given for the various instances of premise (1).

As a matter of fact, this is often true since the ideas motivating instances of the premise are often only assumed and not explicitly defended: e.g., both Mackie and Harman seem to only assume that if there were “objective” moral properties or facts that made moral judgments true, then more people would detect these properties and moral evaluations would converge, especially when people agreed on (what they agree is) the relevant empirical information. Since there isn’t such agreement, Mackie and Harman conclude that there are no moral properties.

So their version of premise (1), roughly stated, that if a judgment has the features pertaining to a specific kind of disagreement (a kind of disagreement that they think is not better explained in some other manner), then it is not true, or probably not true, is based on an assumption for which they provide little or no explicit defense. Their premise is not supported by the kind of discussion that would be best to defend it with, such as a fully general discussion of what kinds of properties are highly “see-able,” which kinds of properties (if any) are not, what we should think about the ontological status of the various kinds of properties that have low or no “see-ability,” and an application of this all to the metaphysics of moral properties. Something like this would be needed for an optimal defense of their assumption behind their premise (1). As it stands, this defense is missing, so our “lazy” philosopher might see no reason to take this argument seriously. She might not have any positive objection to the premise; it just doesn’t seem true to her so she doesn’t accept it. 

            My response to instances of premise (1) is not as lazy as this. Although it does seem that too often too little positive reasons are given in favor of the various versions of this premise, I intend to argue that we should regard the various instances of these arguments’ major premise are false. Someone who understands my arguments would be justified or reasonable in rejecting (1), ought to reject (1), and so should think that the typical arguments for moral irrealism are not sound. To defend such claims, I argue that instances of premise (1) have false implications that, when recognized as such, reveal that the premises are false and so ought to be rejected.

To try to show this, I develop various arguments for epistemic irrealisms that are parallel to the arguments for moral irrealism. Epistemic irrealism implies that epistemic judgments – e.g., that some belief is epistemically reasonable, justified, known, ought to be held, should be rejected, and so on – are never objectively true. To develop an argument for such a view, we can construct a parallel general argument schema for epistemic irrealism building on premise (1) above from the argument for moral irrealism above:


(1) If a judgment has features Φ, then it is not objectively true.

(4) Epistemic judgments have features Φ.

(5) Therefore, epistemic judgments are not objectively true. 

From (5), considerations might be offered to attempt to encourage the adoption of epistemic expressivism, nihilism, relativism or other meta-epistemological theory that, if true, implies that epistemic realism is false.

            I have, and will, argue that instances of the argument from (1, 4, and 5) are unsound and use this result to argue that the analogous arguments against moral realism are also unsound. This argument about these arguments can be presented as the following:

(6)   If the discussed arguments for moral irrealism are sound, then analogous arguments for epistemic irrealisms are sound.[50]

(7)   Analogous arguments for epistemic irrealism are not sound.

(8)   Therefore, the discussed arguments for moral irrealism are not sound.

Conclusion (8), if shown, does not provide positive support for moral realism; it is only defensive in showing that common arguments against moral realism are unsuccessful.

To defend this argument (6, 7, and 8), I will need to defend premise (7): my claim that these arguments for epistemic irrealism are not sound. To do that I need to argue that the relevant instances of either premise (1) or premise (4) in the schema for arguments for epistemic irrealism are false.

I accept instances of premise (4) as true: epistemic judgments have the features that moral irrealists find problematic about moral judgments. Like premise (2) above, which describes alleged features of moral judgments, I accept instances of premise (4) for features Φ that moral irrealists appeal to in making their case for their moral irrealisms. Moral irrealists say, “Moral judgments are like this.” I reply, either sincerely or for the sake of argument), “OK, but epistemic judgments are like that also.” These agreements are stated through instances of premise (4).

Thus, my objection is to the shared premise (1), that if a judgment has features Φ, then it is not objectively true. If that premise is false, then arguments for epistemic irrealism and for moral irrealism are unsound. Those who should believe that this premise is false should also believe that the arguments are unsound; those who are justified in rejecting it are justified in rejecting the arguments against these realisms as unsound.

To object to premise (1), I argue that epistemic judgments have features Φ yet they are sometimes objectively true. They are “in the same boat” with moral judgments, in terms of many of the metaphysical, semantic, psychological, and epistemic features that moral irrealists have found noteworthy about moral judgments. However, I claim that having these features is not sufficient for them never being objectively true, and so premise (1) is false.

To do this, I show what’s entailed by the falsity of epistemic realism, or the truth of (5) above, and claim that we should not accept these entailments: we should regard them as false. Many of these entailments, and the argument against them, might be summed up with this complex argument:

(9)         If no epistemic judgments are ever objectively true, then either there are no epistemic beliefs, or/and there are no epistemic propositions, or/and there are no epistemic properties, or/and there are no epistemic facts, or/and there are no accurate representations of things as they epistemically are (or/and, we never even attempt to represent things epistemically) or/and the only epistemic “truths” are those relative to different epistemic frameworks, none of which are any better or worse than any other.

(10)     But these consequences are all mistaken; we have better reason to reject them than accept them.

(11)     Therefore, some epistemic judgments are objectively true.

The most important consequence of the truth of (11) is that (1) is false, and if that’s so, then the typical arguments for moral irrealism are not sound. Recognizing that they have false premises provides strong reason to reject these arguments. If one is antecedently fond of moral realism, at least one might reasonably think that there are no good objections to the view.

            How, however, do I defend (10)? That is, how do I show that there are epistemic properties or facts that serve as truth makers for epistemic claims, and even that there are epistemic claims, i.e., epistemic beliefs and other propositional attitudes? How do I show that epistemic discourse is not (merely) emotionally expressive and not founded on false metaphysical, semantic, psychological and epistemic assumptions? After all, it could be true that some version of epistemic irrealism is true and some argument for that conclusion sound. The claim that there are no epistemic properties and all epistemic and intellectual evaluations are purely emotive is not easily seen to be self-contradictory: contemplation of the claim itself does not inhibit belief. And there’s no contradiction in thinking that an epistemic platitude along the lines of if you realize you have contradictory beliefs, then you should abandon some belief to eliminate perceived inconsistency is likewise never true (because it’s emotive or there are no truth makers for it, or whatever). Someone might accept some kind of epistemic irrealism, with full awareness of its implications, even its implications for the epistemic evaluations of her belief in the theory itself (and everything else she believes): she might call it “reasonable” or “known” but recognize that, on the theory, perhaps, only emotions have been expressed, no truths stated. She could accept these implications even her belief that epistemic irrealism is true and some argument in its favor is sound: there’s no flat out contradiction in doing so and she could interpret all forthcoming negative epistemic appraisal in terms of her own theory – as objections that are never true, and are founded on untrue presuppositions – and not be troubled.

So it might very hard to convince some epistemic irrealist that (10) is true, that her view is mistaken and that she should reject whatever argument she accepts in its favor. An epistemic irrealist might think, or feel, that she has good reasons for her view. If asked, “Why believe epistemic irrealism?” an epistemic irrealist might note that it follows from her firmly held principles and beliefs (one of which might be premise [1]), it coheres well with her other beliefs, she thinks is has resulted from a reliable belief-forming-process and/or denying it would seem stranger than accepting it.

All these claims might be true, but, if epistemic irrealism is true, then – if the recognition of various is-ought and natural-evaluative “gaps” in moral discourse provide reasons to think that naturalistic moral realisms are false – it is also not true that if any of these “natural” facts obtains, then the belief is reasonable or ought to be held, since that gap is there also. An insightful epistemic irrealist might see this and, perhaps, embrace it as the way the world is: just as the world is devoid of moral value, it is devoid of intellectual or epistemic value also; all there is are natural features of beliefs and inferences, but no features are objectively more epistemically valuable or desirable, as epistemic realists presume they are.[51] An epistemic irrealist might respond to nearly any objection, any observation of what seems, from common views, to be a crazy implication of the view, with acceptance and, perhaps, the claim that the “epistemic world” is not the way many have thought it is.[52]

            This is a general statement of a possibly true view. I suppose it’s sustainable, at least in the sense that an advocate of the view would be able to utter many words in response to objections and the discussion could continue for quite a long time. One could truthfully observe that if it is true, then nearly all philosophers’ understanding of the nature of reasoning and epistemic evaluation is deeply mistaken, including the understanding of many traditional radical skeptics. An advocate of this kind of view could accuse his critics of begging the question, or merely assuming a position and not giving any reasons in its favor (not that there’s anything wrong with that, on this view). How then can we argue against it?

To do this, we begin where, as a matter of fact, we are and with what we have to work with. First, we note that – among those who have thought about such views – few people actually believe this kind of view. In itself, this doesn’t count for much, since people can be prejudiced and mistaken, but what does count is that the view seems clearly false in its implications and in its contribution to a picture of the way the world is. To most, it seems clear that there are better and worse ways to reason, that some mental states have greater intellectual value or worth than others, and we can have beliefs about these states of what’s reasonable, or justified, or known, or should be believed, from an intellectual point of view. Emotions sometimes accompany these epistemic evaluations, but they aren’t the evaluations themselves. The evaluations are based in evidence, strong evidence, of what clearly seems true and nothing that seems false in comparably clear or compelling ways.  This provides good reason to think that epistemic irrealisms are not true, the arguments for them unsound and, in particular, that premise (1) is false. We can see that premise (1) is rationally unacceptable for us by either considering it directly in light of counterexamples, or we can see that other kinds of reason beliefs entail that it is unreasonable.

            An epistemic irrealist might respond that if his theory is true (and, I concede that it could be true), then all these common judgments are mistaken and we are radically in error. In this way, the epistemic irrealist is like a skeptic who notes that if a skeptical hypothesis is true, then we lack knowledge or justification. We should agree, but note that we have strong evidence that his theory is false and that he has given, at best, little evidence or positive reason to think his theory and all its radical implications are true. He might then reply that, on his view or if his view is true, we have not said anything that’s objectively true. Again, that’s true, but for an epistemic realist to reasonably be convinced that epistemic irrealism is true (and for her to, perhaps paradoxically, reasonably come to believe that all talk of reasons and other epistemic evaluations is, objectively, never true[53]), she needs to be given reasons that are stronger than the reasons she already has for epistemic realism and its metaphysical, semantic, psychological and epistemological presumptions.

It’s possible that an epistemic irrealism is true, but, for most, possible only in a logical or metaphysical sense. Relative to what most people reasonably believe, it is not a reasonable option. They have reasonable beliefs about epistemic matters, they reasonably believe that these beliefs imply that epistemic irrealisms are false, are reasonable in believing (at least in this case) a principle along the lines of that if you reasonably see that something is inconsistent with what you reasonably believe, and especially if you see that there is little to no antecedent positive reason to accept that something, then you should not accept it, so epistemic irrealisms are not something they should accept. Again, this might all be mistaken – there is no reason to think the world must have epistemic value in it – but we have good reason to affirm the common sense semantic, metaphysical, psychological and epistemic foundations of epistemic judgments, and no good reason to deny it.

To apply this theme to positivism, the main view under consideration in this chapter, I maintain that we, and I suspect this includes most positivists, had (and had) stronger reasons to believe that there are reasons, and justified beliefs, and ways we ought to reason, than the reasons ever given to accept positivistic principle and its irrationalist implications for epistemic evaluations. It’s possible that positivism is true and that its suggested epistemic emotivism is true, but nearly everyone (positivists included) has or had stronger reasons to deny than affirm it. And if this is so, this importantly undercuts the premise needed to defend a positivistic-based moral irrealism.

3.4.3. Non-Positivistic-Based Epistemic Emotivism.

Before I consider objections I will briefly discuss some non-positivistic bases for emotivism, both ethical and epistemic. Ayer also thought ethical emotivism was “valid on its own account,” apart from positivism. He didn’t say why he thought this, but I will briefly mention some of the considerations he gave for ethical emotivism and note that they apply to epistemic judgments as well. Some of these overlap with positivistic reasons, but one needn’t be a positivist to accept them.

Ayer calls moral (and aesthetic) judgments “statements of value” and claims that they “cannot with any show of justice be represented as hypotheses, which are used to predict the course of our sensations.”[54] He argues that naturalist definitions are not analytic, so moral judgments can’t be reduced to empirical judgments. While some say that moral judgments can be verified by “intuition,” Ayer rejects appeals to intuitions since they can conflict and there seems to be no empirical test to adjudicate between them. He notes that what seems intuitively, certainly right or good to someone may seem doubtful, or even obviously false, to another. He observes that moral debates typically are more contentious and emotionally expressive than debates about empirical matters.[55] Someone might conclude, given all this (and without any thought of positivism), that people are just expressing their emotions when making moral judgments: perhaps this would seem to be the best explanation of the phenomena. 

But all these considerations apply to epistemic judgments as well: judgments about what’s reasonable or justified don’t predict sensations, aren’t empirically verifiable, and seem to be the result of “intuitions” which, when conflicting, can’t be resolved by empirical tests. Epistemic terms are hard to define in empirical terms: any claim that some naturalistic expression is synonymous with an epistemic expression will likely be subject to “open-question” objections; Ayer, at least, thought that this refuted the definitions. Epistemological definitions that are not said to reveal an analytic or synonymous relationship between terms are subject to the same kind of objections leveled against non-analytic ethical definitions, e.g., that they have false implications, that meeting their condition(s) is neither necessary nor sufficient for having some evaluative status, that they do not correctly identify the features that explain something’s evaluative status, and so on.

Finally, debates about whether it’s rational to believe something, something is genuinely known, or some evidence is sufficient evidence (and even if something is evidence) are often contentious, inconclusive and feelings can run equally high. Epistemic judgments also seem to be a kind of “value judgment” anyway: to say that a belief is justified or reasonable is to impart some kind of positive value to it.[56] There are judgments of epistemic virtue and vice, judgments that knowledge is intrinsically good, and assumptions that knowledge is better than “mere” (i.e., unjustified) true belief, all value judgments in themselves. As suggested above, the presumption that there is good and bad reasoning seems to presuppose value judgments as well. So, if these general, non-positivistic-based considerations support ethical emotivism, as Ayer believed they did, they support epistemological emotivism as well. 

3.4.4. Conclusions on Ayer’s Epistemic Emotivism.

Thus, I have argued that if Ayer’s reasons given for ethical emotivism are genuinely good reasons to accept it, then they are also good reasons to accept epistemic emotivism as well. But few are willing to think that there are (or, perhaps, even could be, in light of the views implications for itself) good reasons to accept epistemic emotivism. And if an Ayer-inspired epistemic emotivism is true, to say that “there are good reasons to believe that view” is merely to express one’s feelings about believing it. That might be the truth about what epistemic evaluations are. But most philosophers, even positivists, think that epistemic emotivism should be rejected, and that’s not just how they “feel” about it. They think this is a truth, a literal truth, and that it is quite reasonable to accept it. I have argued above that they are reasonable in rejecting it: the evidence in favor of it is stronger than the evidence against it.

Since epistemic emotivism (or some other non-cognitivist interpretation of epistemic discourse) seems to follow from positivism, positivists and so motivated ethical emotivists might have a dilemma on their hands: accept the consequence and go emotivist more broadly to include epistemic judgments (including the epistemic and non-logical judgments involved in reasoning), or drop positivism. If epistemic emotivism is true, it’s not true that they should respond to this dilemma in any way. But a more global emotivism seems quite implausible and would be very difficult, if not impossible, to defend in a plausible manner (due to the fact that epistemic emotivism makes all epistemic judgments emotive), so for many that’s an unattractive response. Dropping positivism wouldn’t require dropping ethical emotivism since it could be retained on independent grounds. But many of these reasons would likely suggest epistemic emotivism as well, as I argued regarding Ayer’s non-positivist motivations for emotivism.[57]

The easiest response, the one that seems suggested by the best reasons, is to drop positivism, especially since there is no good reason to accept it in the first place. But this eliminates a traditional justification for ethical emotivism. But since other justifications for ethical emotivism likely suggest epistemic emotivism, those justifications should be dropped as well. So, I conclude that one should reject both ethical and epistemic emotivism and the cases Ayer offered in their favor, as well as the arguments that can be developed from his writings. 

3.4.5. Some Objections and Replies.

As far as I can tell, there is no hope for retaining positivism and Ayer’s case for ethical emotivism but resisting the conclusion that epistemic judgments are neither true nor false also: the positivistic principle clearly supports epistemic judgments being non-cognitive. The only response, besides agreement, then might be that epistemic emotivism is as palatable as ethical emotivism. Positivists thought the latter wasn’t a problematic position, so why should they think the former is problematic?  But it is very difficult to see how epistemic emotivism isn’t anything but a very difficult position to hold. This is clear in light of the discussion above, where I argued that epistemic emotivism undercuts the possibility of truthful epistemic support for itself (and any other view, including ethical emotivism) and, more importantly, conflicts with so much that seems true about epistemic evaluations. Adopting epistemic emotivism makes it even more difficult to defend ethical emotivism also: adopting epistemic emotivism makes defending ethical emotivism from objections all the more challenging. 

To see this, let us discuss the only objection to ethical emotivism that Ayer discussed in Language, Truth and Logic. This objection was that if it is true, there are no moral disagreements since there are no moral propositions for one party to accept and another reject. Since people seem to disagree, and if moral propositions are needed for disagreement, the objector concludes that emotivism is mistaken. Ayer responded, surprisingly, that people do not disagree about moral matters, strictly speaking.[58] He claimed that “we do not and cannot argue about . . the validity of . .  moral principles. We merely praise or condemn them in light of our own feelings.”[59] He claimed in all cases of what are intuitively called moral disagreements, the dispute is about cognitively meaningful questions of logic or empirical matters of fact and their relevance to the moral question.

Positivism would seem to preclude their being truths about what empirical information is relevant to a moral issue,[60] but it will allow there being truths about empirical information and logical relations. But if epistemic emotivism is true, Ayer’s even diminished account of moral disagreement is mistaken because we do not, strictly speaking, disagree on what we should think about logical and factual matters either: when someone says, “No, you are not believing what you should about the relevant empirical matters,” one is never saying something that is true. Insofar as empirical and logical correction involves making epistemic judgments

Ayer claims that much moral debate consists in getting people’s particular judgments to fall in line with the principles they accept.  He claims that if “a man has certain moral principles . . he must, in order to be consistent, react morally to certain things in a certain way.”[61] Ayer’s remark here is surely more than the trivial claim that one has to be consistent to be consistent; he seems to be thinking that revealing some inconsistency to someone should result in her changing her moral views: she should do something to resolve the inconsistency. But that imperative does not appear to be an analytic or empirical truth, so it is presumably neither true nor false. If it is a moral principle, Ayer has only praised it in light of his own feelings. So logical correction, in terms of how one ought to reason, is not a cognitively meaningful resource for Ayer to appeal to.

Empirical correction is the “attempt to show that [someone] is mistaken about the [non-moral] facts of the case . . has misconceived the agent’s motive . . has misjudged the effects of the action . . or has failed to take into account the special circumstances in which the agent was placed.”[62] But this activity appears to presume epistemic standards, that, given certain experiences and other beliefs, someone ought to have some specified belief. As I argued above, these kinds of judgments are cognitively meaningless on Ayer’s principles. There is no truth to claims that someone’s beliefs should change if new information is presented or that there is something objectionable with someone dogmatically refusing to consider this new information.

Ayer claimed that, “one really never does dispute about questions of value.” But insofar as judgments about how one ought to reason, that one should be consistent, and that one should accept some empirical judgments are neither true nor false, there are no cognitive disputes about them as well, on his principles. People can disagree, in the sense that they believe inconsistent propositions. But on his view, for any proposition p, when someone says “you should believe p” and someone replies, “no, you should not believe p” they do not, strictly speaking, contradict each other: they are expressing contrary emotions, but nothing more.

That’s a more surprising, and damning, implication than what follows from ethical emotivism alone. Thus, his response to the objection that emotivism cannot account for moral disagreement fails: it can account for it only by appealing to other non-cognitive factors. And these we also do not and cannot argue about: we merely praise or condemn epistemic standards and principles about reasoning in light of our own feelings.

            It’s unlikely he would have accepted this consequence, but he might have. He might have replied that, yes, strictly speaking, all evaluative or prescriptive judgments – about what one should, ought or must believe, reason, or react – are neither true nor false. However, he might have responded that this was no objection to the view because he had no need for epistemic language. He could observe that it is fully legitimate on his principles to make descriptive statements, e.g., about what people believe if he’s careful to not describe these statements in epistemic terms, e.g. “justified” or “rational” or about what one “ought” or “should” believe and so on. And he has logic, although perhaps not non-deductive logic, since judgments of probability might also be rendered non-cognitive by positivistic principles.

So, in argumentative contexts, he can point out logical relations between propositions and what follows and does not follow from what people believe: he can note that some propositions are inconsistent with someone’s beliefs or are entailed by them.  He can observe how confident someone is in some beliefs and report that, oftentimes, when people are confident in some belief and see its logical consequences, they are confident in these consequences also. He might also make the empirical claim that some belief was, or was not, “arrived at by the methods which we now consider reliable” or that “a self-consistent accredited procedure” was employed in the formation of some beliefs. If he is talking to positivists, he can note that his kinds of positions fit well with their views; in fact, for some people he might observe that his positions are their views and that denials of these positions are inconsistent with their views and do not all cohere with what they think. He can attempt to find out what people’s various goals are and make suggestions for what kinds of beliefs would best meet these goals.

Ayer might claim that making these kinds of observations is all that is needed for argumentative discourse: evaluative epistemic talk can be abandoned with no loss.  However, it seems clear that there would be a loss. All these suggested methods of engaging in argumentative discourse without evaluative language consist in making naturalistic claims or observations about beliefs. These moves are thereby repeatedly subject to “open-question”-related concerns. Competent speakers could easily understand these claims have no clue what is supposed to follow from them: they might ask, “Why is Ayer telling me this, even if what he says is true? What is supposed to follow, in terms of what I believe, from my recognition of this natural feature?” If they are logically inclined, they might think that these remarks above are supposed to be premises in some argument.

In most or all cases, however, a second premise, linking the first to the sought conclusion, is left tacit. This premise will either have an evaluative consequent (e.g., “if some belief has some natural feature(s), then it is justified”) or it will not (perhaps the consequent will be that the belief is true, or probably true). If there is an evaluative antecedent, then evaluative language has been admitted and so this response does not avoid evaluative language: it fails. If, however, the linking premise and conclusion are devoid of evaluative language, we might be left with another open question: what is supposed to follow, in terms of what we believe, now that we recognize these propositions in this logical relation to each other? If someone thought that, given her understanding of Ayer’s claims, she should then believe or do something then we have slipped back into illegitimate “shoulds.”

Even if Ayer pointed out some inconsistency, the question of what to believe seems unavoidable: should they keep a premise, reject a conclusion, or reject Ayer’s judgment that there is an entailment?  That’s a question that naturalistic descriptions alone couldn’t answer and Ayer’s positivism precludes there being a true answer to. Perhaps Ayer would hope that people care about avoiding inconsistency and that this would typically lead them to think as he wished, but surely even if a person lacked such concerns, epistemic questions about what she should believe, i.e., what is reasonable or justified, remain. Perhaps he could offer a theory about good reasoning (perhaps “good” is too evaluative, so this should be a theory of “sought” or “desired” reasoning) to the effect that person A has reasoned well just in case A has come to believe whatever B wishes her to. This theory might be couched in entirely naturalistic terminology, but, unfortunately, it would be a theory of persuasion or manipulation, not a theory of what would, to most people, be called good reasoning, since it would countenance even the most irrational and despicable methods of inducing belief change as “reasoning.”

            In sum, it seems that epistemic language is practically unavoidable: attempts to avoid it will often rest on evaluative presumptions typically along the lines of a belief’s having some natural features yields some evaluative status, perhaps its being the case that it ought to be believed. Or attempts to avoid epistemic language will depend on our contingent desires that our beliefs have various natural features, e.g., that they be true, that they be part of a coherent set, that they be produced by reliable processes, and so on.

For practical reasons (i.e., boredom, frustration and the sense that we have better things to do), we might abandon attempts to argue with someone who refused to acknowledge any epistemic standards, and, as Ayer observes about some ethical disagreements, in some contexts might even resort to mere abuse.[63] But we would attribute failure of acknowledgment as a failure to see the truth about epistemic standards.  And it’s not merely that we would do this; rather, from an epistemic point of view, it’s true that we should do this. To deny this is, fundamentally, to accept a view that is unable to accommodate good reasoning, the common presumptions of epistemic judgments and seems to undercut whatever epistemic support it might have (which, if epistemic emotivism is true, is either none or only someone’s emotional support).

            Thus, I have argued that Ayer’s case for ethical emotivism has deep and troubling implications for the cognitive status of epistemic discourse. I conclude that, for a variety of reasons, the standard, cognitivist, fact-stating interpretation of epistemic judgments truthfully should be maintained and Ayer’s principles should be rejected. Thus, his case in favor of ethical emotivism is undercut.

3.5. C.L. Stevenson’s Ethical and Epistemic Emotivisms.

I now turn briefly to C.L. Stevenson’s ethical emotivism.  Fortunately, Stevenson was not a positivist, so his perspective lacked that semantic and metaphysical baggage.  His emotivism was much more developed and subtle than Ayer’s, but the basic position was similar. He thought that ethical sentences typically express attitudes and invite others to share those attitudes.  By “attitudes” he meant, “tendencies to be for or against something, as typified by liking, disliking, approving, disapproving, favoring, disfavoring, and so on.”[64] On his view, attitudes contrasted with “beliefs.” So, e.g., if A says that “X ought to be done” but B responds that “X ought not be done,” then – according to Stevenson – they disagree in attitude, not in belief (however, this disagreement could be a consequence of a disagreement in non-moral belief, but not necessarily since Stevenson allows for fundamental disagreements in people’s attitudes, i.e., what they approve of and so on).

            Cognitivists agree that, sometimes (if not typically) when people judge something to be right or good, they have certain emotive attitudes about that thing and, were they to express their attitudes orally, they would often be trying to influence others. But Stevenson thought that the “function” of ethical sentences was the service of this expression: there wasn’t a mere contingent connection. On his final analyses of “good,” to say something is good is to say that one approves of something, but said in such a way that one’s expression of approval would evoke favorable feelings in their hearer.[65]

This position might be illustrated by this case: suppose someone asks if some action was right. Someone might respond, “Well, I approve of people doing that!” This could be said in a way to express one’s approval of what was done and attempt to influence others. This expression of feeling is a consequence of moral terms’ meanings. Stevenson said, “The emotive meaning a word is a tendency of a word, arising through the history of its usage, to produce (result from) affective responses in people. It is the immediate aura of feeling which hovers about a word.”[66] Given these meanings, moral judgments are not beliefs, since they have, strictly speaking, no emotive influence in themselves; rather, they are non-cognitive expressions.

Stevenson’s analyses have been criticized as not capturing the correct meanings of moral terms, but I will not review those criticisms here.[67] I wish to only argue that his considerations in favor of emotivism also suggest epistemological emotivism. As I argued earlier about Ayer’s emotivism, this is an undesirable consequence in itself and had problematic implications for the rational acceptability of emotivism. Stevenson’s position (and its defense) has analogous difficulties. 

            Much of Stevenson’s discussion consists of detailed descriptions of what it’s like to engage in moral discourse and debate; e.g., what someone would say, how feelings might be roused, how one’s interlocutor might respond, how confidence levels might change, when disputants might be satisfied and end the dispute, and so on.  Thus, ethical discourse has what he calls a “dynamic” aspect. 

If these observations were intended to provide support for emotivism (and if they are not, then they are merely descriptive moral psychology), they seem equally applicable to disagreements about what’s epistemically justified or rational as well.  One could do a phenomenology of epistemic disagreements and observe that heated emotions often get expressed in debates over whether it’s reasonable to believe empirical claims (about, for example, the causes of diseases, the utility of various research methods, the age of the earth, consequences of tax cuts, philosophical topics and many much more mundane issues). And calling a belief “irrational” or “unjustified” can be an attempt to influence others and “invite” them to share one’s belief. There are, of course, many more “colorful” and obviously expressive terms of epistemic appraisal as well. If the fact that emotions often accompany a kind of discourse and it sometimes has a persuasive effect suggests an emotivist analysis of that kind of discourse, then epistemic language seems fit for that analysis as well. 

Stevenson’s main explicit argument for his emotivism was based on a claim about moral terms’ alleged “magnetism.” He asserted that, “A person who recognizes X to be ‘good’ must ipso facto acquire a stronger tendency to act in its favor than he otherwise would have had.”[68] Following Hume’s theory of motivation, he thought this showed that moral judgments are not beliefs since beliefs do not, in themselves, provide motivation to act or influence one’s emotions. He thus concluded that moral judgments were emotive attitudes, not beliefs (or solely beliefs).

In a thorough discussion of these motivational issues, Russ Shafer-Landau concludes that, “We are misled if we move from the obvious fact that moral judgments are usually motivating, to the stronger claim that they cannot fail to be.”[69] He suggests that Stevenson was misled in just this way. I agree and I might only add a brief response to arguments from motivation: I find them unmotivated. I just don’t find any necessary, “internal” connection between moral judgment and motivational or affective considerations and find “amoralists,” agents who make moral judgments yet are not affective moved or motivated by them, quite conceivable (and, in fact, actual).[70] Furthermore, explanations of moral motivation that acknowledge only a contingent relation between moral judgments and moral motivation seem much more plausible, all things considered.

However, for my purposes, it’s only important to note that it seems that if someone were impressed by these “internalist” considerations in ethics, she might plausibly also think that there’s some kind of necessary connection between epistemic judgments like ‘p is not justified for me’ and one’s affective states as well. Consider someone, e.g., a religious believer, or a scientist, who comes to think that he doesn’t have particularly good reasons for some of his beliefs. Someone convinced that the term “morally good” has a “magnetism” might easily, and plausibly, also think that this judgment of epistemic badness would necessarily be accompanied by a desire or motivation to not have that belief. Similarly, someone might think that judging a belief to be justified entails a desire to believe it.

If one thought either of these, then one might adopt something like a Stevensonian analysis of epistemic judgments and conclude that they too are not beliefs and that the meanings of epistemic terms is emotive. This position seems as motivated as the more common position regarding moral judgments. Thus, if motivational concerns suggest ethical emotivism, they suggest epistemic emotivism as well.

Finally, Stevenson considers a disappointed reader who wants to know the truth about whether something is morally good: she doesn’t want mere correction about her judgments of various empirical facts or be subject to emotional persuasion, which is all Stevenson’s position has to offer. In response, Stevenson asks:

What is this truth to be about?  For I recollect no Platonic Idea, nor do I know what to try to recollect.  I find no indefinable property nor do I know what to look for.  And the ‘self-evident’ deliverances of reason, which so many philosophers have mentioned, seem on examination to be deliverances of their respective reasons only (if of anyone’s) and not of mine. I strongly suspect, indeed, that any sense of ‘good’ which is expected to both unite itself in synthetic a priori with other concepts and to influence interests as well, is really a great confusion.[71]

Later, he rejects what he calls the “ethical analogue of a fact,” stating that he “find[s] nothing ‘out there’ for our attitudes to represent,” copy or be faithful to.[72] Basically, he is saying that he cannot understand what a moral fact or property would be like or what a truth-maker for a moral judgment could be.[73] 

His position seems, in some ways, to echo Hume’s.  Recall his famous passage:

Take any action allow’d to be vicious: Wilful murder, for instance. Examine it in all lights, and see if you can find that matter of fact, or real existence, which you call vice. In whichever way you take it, you find only certain passions, motives, volitions and thoughts. There is no other matter of fact in the case. The vice entirely escapes you, as long as you consider the object. You never can find it, till you turn your reflexion into your own breast, and find a sentiment of disapprobation, which arises in you, towards this action. Here is a matter of fact; but ‘tis the object of feeling, not of reason. It lies in yourself, not in the object. So that when you pronounce any action or character to be vicious, you mean nothing, but that from the constitution of your nature you have a feeling or sentiment of blame from the contemplation of it.[74]

Despite their differences, both Stevenson and Hume report that can only “see” natural properties; they can’t find moral ones. But it seems that if one cannot find moral properties, epistemic properties, and facts about what one should and shouldn’t believe, should be hard to find as well. This is because, to use Stevenson’s terms, it seems equally hard to see what might be “out there” for our epistemic attitudes (or on the common, cognitivist presumption, our beliefs about what we should believe) to represent, copy or be faithful to. But, if that’s so, it’s not clear what could make it such that anyone rationally ought to accept Stevenson’s emotivism, or Hume’s views, or any other view. That too would seem to be a matter of feeling, not reason.

Thus, Stevenson’s arguments from the phenomenology of moral debate, the ‘magnetism’ of good’ and the mysteries of moral truth and properties suggest analogous arguments concerning epistemology: that there are no epistemic truths or properties and epistemic discourse is emotive. As I argued earlier, this is a difficult implication in itself, since it implies a view about the nature of epistemic evaluation that, for many reasons, seems false. And, like epistemic emotivism implied for Ayer’s views, it makes it the case that it’s untruthful to say we should believe the view, or any other view. Thus it again appears that the assumptions supporting ethical emotivism undercut their very defense. In light of, especially, these meta-epistemological considerations, I believe we should reject the premises that motivate what would be Stevenson’s epistemological emotivism, and doing so leads us to reject his arguments for his ethical emotivism also.

3.6. Conclusion: Brief Remarks on Gibbard.

I have argued that the reasons Ayer and Stevenson offered in favor of their ethical emotivism suggest analogous epistemic emotivisms. This has important general epistemic consequences: first, since epistemic emotivism provides a possible way to understand the nature of epistemic evaluations, but a way that seems false and contrary to good reasons; second, it has interesting implications for the epistemic status of ethical emotivism since it implies that it’s not true that it’s a justified view. 

As far as I know, Ayer and Stevenson did not realize that their arguments for ethical emotivism suggested these consequences. They seemed content confining emotivism to ethics and aesthetics, but I have argued that this is difficult since it naturally seems to spill into judgments about epistemology, probability, and critical reasoning. Insofar as Ayer and Stevenson, and most of their defenders, probably would not have been inclined to accept these kinds of emotivism, and for good reasons, this suggests that their cases in favor of ethical emotivism should be rejected.

            Alan Gibbard, on the other hand, explicitly accepts both kinds of emotivism, or expressivism, that he calls “norm expressivism.” Briefly, his view is that “[t]o call something rational is to express one’s acceptance of norms that permit it. [It is] . . not to attribute some particular property to that thing—not even the property of being permitted by accepted norms.”[75] He claims that since to call something rational is to endorse that thing, these judgments are expressive, neither true nor false. He claims that judgments with this “endorsing” aspect are not well captured by naturalistic or non-naturalistic (or intuitionistic) theories of moral judgment; he thinks his kind of expressivism does.

In my final chapter, I argue that the considerations Gibbard offers in favor of his “norm expressivism” are unconvincing and that, on Gibbard’s own view, it is very difficult to understand why anyone should accept the view.  Either it is true that someone – given his or her understanding of the theory – should accept it, or it is not.  If it is not, then, of course, it is not true that it should be accepted. But if it is true, that would seem to be inconsistent with the theory itself. Perhaps judgments about how logical inconsistencies ought to be avoided are mere expressions of acceptances of norms that don’t permit some kinds of inconsistencies: perhaps there really are no truths like these. Perhaps judgments that these kinds of principles should be accepted are also expressions, as Gibbard says they are. And perhaps his views about endorsement provide some reason to reject all varieties of realism.

However, as I have argued, and will argue at greater detail later, it seems that we have better reasons than not, all things epistemic considered, to reject epistemic expressivisms than accept them. While Gibbard’s view may avoid some of the objections to Ayer and Stevenson’s theories, I will argue that it is not true that one ought to accept it. Gibbard’s own norms might “permit” him to believe it, but that’s no reason why anyone should to accept it, or come to share Gibbard’s norms. Epistemic emotivism is as unwieldy in Gibbard’s hands as with earlier emotivists. I argue for this in my final chapter.



[1] In an earlier version of this chapter, I believe I was too quick to dismiss epistemic emotivism as just obviously not true. See Nobis (“Ayer and Stevenson’s Ethical and Epistemological Emotivisms”). I now believe that stronger reasons can, and should, be given against it.

[2] See Ayer (Language, Truth and Logic 20). 

[3] See Ayer (“Editor’s Introduction” 10).

[4] See Ayer (Language, Truth and Logic 5).

[5] See Ayer (Language, Truth and Logic 13).

[6] See Ayer (Language, Truth and Logic 79).

[7] See Ayer (Language, Truth and Logic 16, 41).

[8] See Ayer (The Central Questions of Philosophy 27).

[9] Or, ‘If there are (or were) any right actions, are right actions right?’

[10] See Ayer (Language, Truth and Logic 105).

[11] See Ayer (Language, Truth and Logic 106).

[12] See Moore (Philosophical Papers 94).

[13] See Ayer (Language, Truth and Logic 106).

[14] See Ayer (Language, Truth and Logic 106).

[15] Non-naturalistic definitions are not empirically verifiable, but Ayer did not argue they were not analytic.  He did not argue that their denials weren’t self-contradictory or that they failed open question tests.  So, he had no argument that they were neither true nor false.  Since the definitions do not in themselves imply that there is non-empirical intuition, Ayer had no argument that they were false either.  His argument was more at non-naturalistic perspectives as a whole; in particular, the moral epistemologies they conjoined with their definitions.

[16] Here I am ignoring possible vague moral judgments.

[17] See Ayer (Language, Truth and Logic 107).

[18] See Ayer (Language, Truth and Logic 103). 

[19] See Ayer (Language, Truth and Logic 108).

[20] Since “stealing” has a moral connotation, Ayer should have said something more neutral like “Taking money” or “acquiring money” or something like that.

[21] See Ayer (Language, Truth and Logic 107-108). Ayer did, however, think that moral judgments have some descriptive content when they entail an empirical proposition.  In the preceding paragraph he writes, “[I]f I say to someone, ‘You acted wrongly in stealing the money,’ I am not stating anything more than if I had said, ‘You stole the money.’ In adding that this action is wrong I am not making any further statement about it.  I am simply evincing my moral disapproval of it.”

[22] See Ayer (Language, Truth and Logic 112).

[23] See Kvanvig (The Value of Knowledge 173) for an endorsement of arguments like those I provide below. 

[24] See Ayer (The Problem of Knowledge 31-35).  Ayer analyzes knowledge into the conditions that “first that what one is said to know be true, secondly that one be sure of it, and thirdly that one should have the right to be sure” (p. 35, my emphasis). 

[25] Science might reveal all sorts of natural features of the belief, such as its deductive and non-deductive logical relations to other propositions (and, lets suppose, experiences) as well as, perhaps, which belief forming “process” it has resulted from. This is interesting information, but an additional premise must be added to propositions stating that information to yield a conclusion about the beliefs justificatory status or whether it ought to be held, and that premise will not be empirically verifiable. I develop this idea below.

[26] Ayer says of particular ethical judgments that “inasmuch as they are certainly neither definitions nor comments upon definitions, nor quotations, we may say decisively that they do not belong to ethical philosophy.  A strictly philosophical treatise on ethics should therefore make no ethical pronouncements” (Language, Truth and Logic 103).  Perhaps these remarks would apply to particular epistemic judgments as well (e.g., whether a belief is Gettier case, whether victims of ‘evil demons’ have justified beliefs), and so a “strictly philosophical treatise” on epistemology would contain no particular judgments about the epistemic qualities of particular beliefs or believers. While there is no point in debating what is “strictly philosophical,” insofar as current epistemologists use particular judgments about actual or hypothetical particular cases to evaluate epistemic definitions (or analyses or principles), Ayer’s suggestion would, surprisingly, seem to suggest that this aspect of epistemological inquiry is not strictly philosophical. And, if my arguments are sound, positivism implies that these epistemic judgments – particular and general – are neither true nor false anyway.

[27] See Ayer (Language, Truth and Logic 100).

[28] See Ayer (Language, Truth and Logic 100), emphasis mine. Presumably, these weren’t mere stipulations. 

[29] This reasoning was founded on the false assumption that if the claim that “A’s are not B’s” is not analytic then A’s are not B’s.  For the sake of argument, here I ignore this.

[30] That seems correct, apart from any concerns about Ayer: even if reliabilism and other naturalistic epistemic theories were true, would not be analytically true or true by definition.

[31] This appears to presuppose some kind of non-empirical intuition that is likely hard to account for on positivistic perspectives, but I will not press that point.

[32] This isn’t at all to say that scientific information or, especially common-sense empirical information is not or could not be relevant to whether some belief is justified or whether it should be judged as justified.

[33] Some might disagree by arguing that, e.g., a successful track record in terms of yielding truths amounts to an empirical defense of the justifiability of a method. This kind of argument would surely appeal to empirical observations, but they would never be the whole story. This is because such an argument would probably rest on a premise to the effect that if some method has yielded many truths, then it is justified and its implications ought to be accepted. If there were evidence for (a more plausible version of) this premise, it would be non-empirical or not wholly empirical.

[34] Should positivists only accept views that are consistent with their position (insofar as propositions can be consist with the principle of verification, a sentence that’s seems to fail to be a proposition, on positivistic)?  While positivists might wish to answer, “Yes,” but below I argue that they truthfully cannot.

[35] See Ayer (Language, Truth and Logic 102). 

[36] See Ayer (Language, Truth and Logic 107, 108). 

[37] Perhaps an epistemic emotivist would allow that they are, at least, attempting to state the truth about the epistemic qualities of their beliefs, but would think that they never succeed.

[38] Later I investigate in greater details whether this reductio could be resisted and whether epistemological emotivisms or expressivisms could be plausibly maintained and defended.  I will argue that it can’t.

[39] I say seems to have implications: although a statement of epistemic emotivism might entail other propositions, claims like “those who accept epistemic emotivism ought to accept these implications” are neither true nor false, according to epistemic emotivism. 

[40] However, I argue below that, on positivism, any principle like ‘if someone recognizes that p entails q then she ought to believe q’ (or, ideally, a more refined and plausible principle) is neither true nor false.

[41] In my introduction chapter, I note that the claim that someone, from an intellectual point of view, should (or should not) resist some argument is a kind of epistemic or intellectual judgment. Throughout this work I argue that reasons to think that moral judgments are never true suggest that judgments like these are never true also. Thus, if, e.g., some positivist were to think that that he or she ought to respond to objections like mine, we need to ask if this judgment is true or not. If it is true, then, I argue this implies that positivistic principles are false. If it’s not true, then perhaps it is a kind of non-cognitive expression. I am arguing, however, that most people have better reasons to reject this kind of view than accept it.

[42] Ayer’s analysis of judgments of probabilistic logical relations has been subject to the same objections I have raised to his treatment of epistemic judgments, so perhaps all he can truthfully make are judgments about entailments and their absence. R.F. Atkinson persuasively argues that the positions Ayer takes in ethics and probability are inconsistent.  See Atkinson (“‘Good’ and ‘Right’” 242-246).

[43] Some might suggest that what can be called “epistemic imperatives,” such as these I suggest here involving reasoning, are hypothetical imperatives: if you want X, believe p (or reason in various ways); you want X, so believe p (or reason in various ways). They might claim that this somehow eliminates appeal to evaluative language. But this does not because we are still left with the question whether some principle like ‘if you want X and Y is a means to X, then do Y’ is a proposition, if it is true, and, if so, what makes it true: why should you satisfy your desires and take the best means towards doing so?  Hypothetical imperatives presuppose all the metaphysical and semantic baggage that categorical imperatives do; to avoid a regress, they seem to presuppose a categorical imperative also such as, perhaps, that one’s desires should be satisfied (irrespective of whatever one’s desires are).

[44] Allen Gibbard and Hartry Field are notable exceptions: they recognize that the arguments of ethical emotivism, and other non-realist meta-ethics, have implications beyond what Ayer and Stevenson saw for them. I discuss their views in my final chapter.

[45] What if someone desires that she be able to defend her views, give considerations in their favor, respond to objections, etc.?  On positivism, that’s not a good thing, or admirable, or an attitude that people ought to strive to have. Those claims are all neither true nor false.

[46] Although I do so, I hesitate using the phrase “objectively true,” since that would seem to suggest that there is a plausible alternative kind of truth, or that there are at least two ways something can be true, objectively and non-objectively. Some (e.g., Harman) have proposed “relative truth” for a non-objective alternative. By this, they mean to say that some judgments are true relative to some other propositions – they are implied by them, or are somehow supported by them – but are not implied or supported by others, and so are false relative to them. To me, it seems misleading to describe this as a way of being true: it would be better to just say that no moral judgments are true, yet – perhaps in conjunction with empirical propositions – some are implied or supported by other moral propositions (e.g., more general moral principles) while they inconsistent with others. This avoids possible confusing concerns about whether some truth is true “objectively” or merely “relatively.”

[47] Although this argument schema is that of a deductively valid argument, not all the arguments I discuss below and in subsequent chapters are deductively valid. For my immediate purpose here, however, this does not matter. In this section, I also evaluate arguments as “sound” and “unsound,” but these evaluations should be understood here as pertaining to non-deductive arguments also.

[48] I do not argue that all possible instances of premise (2) or even every actual articulated instance of premise (2) is problematic. I mainly only examine versions of (2) that are commonly discussed or articulated by philosophers who I have called “classical” moral anti-realists.

[49] Below I discuss how a premise’s being false is not, strictly speaking, a reason to reject it (or makes it such that it is not justified or that it should be rejected), just as a premises being true is not, strictly speaking, a reason to accept it (or makes it justified or that it should accepted). Rather, the relevant consideration is that it seems false (or seems true), “on balance,” or “overall,” or more clearly false than true. The exact phenomenon is not easy to describe.

[50] In this immediate discussion will assume this premise is true, since the two kinds of arguments are structurally parallel and concern the same properties. Earlier I discussed how the fact that moral and epistemic properties have some different features (as well as some shared, common features) would not make an obvious difference to my arguments, since these differences do not change the fact that moral anti-realists suggest that having these (shared) features is sufficient for irrealism. Later I discuss how a moral anti-realists sense that he or she ought to revise his arguments, that he would be justified in accepting them only if he re-tinkered with them to meet my objections, might be based on intellectually evaluative and deontic assumptions that are ultimately inconsistent with his or her premises given for moral irrealism.

[51] I presume an epistemic anti-realist would be a moral anti-realist also. It would certainly be surprising for someone to recognize moral values yet disavow epistemic ones. Earlier I discussed attempts to “ground” epistemic values in the contingent desires we have, thus turning epistemic evaluations into a kind of hypothetical imperative. (This proposal could be involved in trying to argue from the truth that (a) some methods of belief formation yield typically truths to a conclusion (c) that we ought to use these methods.) I briefly argue that any hypothetical imperative is true only if a categorical imperative (perhaps that one’s desires ought to be satisfied) is true also, so the move to hypothetical imperatives does not avoid any axiological and metaphysical mysteries it is designed to avoid.

[52] Interestingly, moral realists are sometimes accused of believing in a bizarre ontology. It seems to me, however, that epistemic anti-realists would believe in an even more bizarre world: a world without intellectual or epistemic values (or, actually, a world where we vividly believed there are no intellectual values, nothing that is epistemically better or worse) would be quite strange, if we always told the truth about our pursuits regarding belief.

[53] My guess is that some epistemic anti-realist could engage in some kind of gyrations upon prompting by those who would think that troubles with this claim reveal that epistemic realisms are false and rationally unacceptable.

[54] See Ayer (Language, Truth and Logic 102).

[55] In Logical Positivism, Ayer claims, “[Moral] arguments do not work in the way that logical or scientific arguments do” (p. 22).  That might be true, but if my arguments are sound, then logical and scientific arguments do not work the way Ayer thinks they do either.

[56] Allan Gibbard, quoting Hilary Putnam states, “The terms that ground our conception of rational acceptability—‘coherent’, ‘simple’, ‘justified’, and the like—are often used as terms of praise,” and they “have too many characteristics in common with the paradigmatic value terms for us to deny that that is what they are.” See Gibbard (Wise Choices, Apt Feelings 32).

[57] One way to avoid arguments for ethical non-cognitivisms being converted to argument for epistemic non-cognitivisms is to make the major premises so ‘narrow’ and non-general that they apply only to moral judgments. This, of course, would make the arguments weaker since they would appear more and more ‘question-begging.’

[58] See Ayer (Language, Truth and Logic 110-112).

[59] That seems mistaken since it seems we can reason about moral principles since it seems that they can figure into truth-functional logical operations.

[60] This is important insofar as many discussions of moral issues involve claims to the effect of “if such and such are the non-moral facts, then this ought to be done.” If these kinds of claims are never true, as positivism implies, then moral argumentation and discussion – insofar as these kinds of claims are commonly made – is founded on many claims that are never true.

[61] See Ayer (Language, Truth and Logic 111), emphasis mine.

[62] See Ayer (Language, Truth and Logic 111).

[63] Ayer notes that this is how we might respond to someone who fails to agree with us about some moral question, in spite of empirical and logical correction (Language, Truth and Logic 111).

[64] See Stevenson (“Ethical Fallibility” 199).

[65] See Stevenson (Ethics and Language 206-226). Some commentators (e.g., Hare [Sorting Out Ethics 103-104]) have suggested that Stevenson’s initial incorporation of a naturalistic analysis, i.e., that the speakers likes (or dislikes) whatever is being evaluated, was unfortunate for the theory for all the reasons, discussed below, that naturalisms are problematic. Charitable critics have, therefore, understood Stevenson’s theory as a pure expressivism, ignoring the naturalistic condition, which Stevenson eventually eliminated from his analysis.

[66] See Stevenson (“Emotive Meaning” 21).

[67] For a summary of these criticisms, see Feldman (Introductory Ethics 223-231).

[68] See Stevenson (“Emotive Meaning” 13). Emphasis mine.

[69] See Shafer-Landau (Moral Realism: A Defense 161).

[70] Michael Smith asserts that, “all else being equal, to have a moral opinion simply is to find yourself with a motivation to act.”  Unfortunately, he does not explain what might not be equal, but I simply deny his assertion: I lack this intuition and my moral experience suggests otherwise. See Smith (“Realism” 400).

[71] See Stevenson (“Emotive Meaning” 30-31)

[72] See Stevenson (“Ethical Fallibility” 212, 214).

[73] Stevenson also says that he cannot understand what an intrinsically motivating fact would be. That might be incomprehensible, but cognitivists or realists needn’t think that anyway.

[74] See Hume (A Treatise of Human Nature 468-469).

[75] See Gibbard (Wise Choices, Apt Feelings 7, 8).