If you find a broken link, or have any questions about the Interlocutor, please feel free to contact
Theological Noncognitivism Examined
noncognitivism is usually taken to be the view that the sentence God
exists is cognitively meaningless. There are various definitions
of cognitively meaningless, but the two most common are
probably the following:
Let us first consider (D1). The grounds on which many noncognitivists take the sentence God exists to be cognitively meaningless is that it expresses an unthinkable proposition. That is, according to many noncognitivists, under many (perhaps even most) interpretations of the given sentence, although it expresses an idea, that idea is incoherent and so cannot be entertained in thought. For instance, it might be interpreted to mean that there exists a being who is simultaneously both one person and three persons, a notion which is not only incoherent but self-contradictory. Some such noncognitivists might nonetheless refer to the given sentence as expressing nothing that is either true or false. But that would be wrongheaded, for every proposition is either true or false and so if God exists expresses an unthinkable proposition, then, necessarily, it expresses something that is true or false. However, most noncognitivists who advocate (D1) are likely aware of this inconsistency and thus refer to the given sentence as expressing a (necessarily) false proposition. But that is also problematic, as one who regards the sentence God exists as expressing a false proposition is ordinarily viewed as an atheist, since an atheist is one who denies that God exists and so regarding the given sentence is tantamount to denying Gods existence.  It would seem more appropriate, then, for such people to call themselves atheists than noncognitivists. So much, then, for noncognitivism where cognitively meaningless is defined as expressing an unthinkable proposition.
Let us turn now to (D2). According to that definition, a sentence is cognitively meaningless iff. it does not express a proposition. I take it that most noncognitivists would say that a given sentence S fails to express a proposition iff. S either contains a term for which no definition whatever is supplied (or for which no definition is understood to be employed) or for which an unintelligible definition is supplied. Thus the important question now becomes: how often, if ever, do theists utter the sentence God exists such that at least one of those conditions obtains (as regards that sentence)? In order to answer that question, it would be helpful to first generate a list of common definitions of the word God:
For the sake of economy, let God-1 exists henceforth (purport to) express the proposition that there exists a being with all the properties enumerated in (G1), God-2 exists henceforth (purport to) express the proposition that there exists a being with all the properties enumerated in (G2), etc. Likewise, let noncognitivism-1 denote noncognitivism with respect to the sentence God-1 exists, etc., where noncognitivism is taken in the (D2) sense.]
(G1) is the definition of God supplied in Random House Dictionary and the vast bulk of theists would surely go along with it. For most of them regularly use God in the given way, i.e., to refer to the one Supreme Being, the creator and ruler of the universe. (That is how the present definition got into the dictionary in the first place.) But what do they mean by the one Supreme Being and the creator and ruler of the universe? That is a purely empirical question, obviously, but it would be easy to hazard a reliable guess: as most theists take God to be eternal, all-powerful, all-knowing, and all-loving, probably most of them mean the one being who is supposed to possess all those properties. The question which of those properties, if any, is coherent I shall defer until I take up a discussion of (G2), below.
respect to the property of being the creator and ruler of the universe,
it is difficult to see how even the most steadfast noncognitivst could
regard it as incoherent.
After all, there is no unclarity
in the concept of a being who creates a cosmos per se, for nearly everyone understands perfectly well what it means
to create something and a cosmos is just a very big thing (or collection
of things). Likewise with the concept of ruling something. Indeed, I
am not aware of any noncognitivist who objects to sentences like God
created and rules the universe as unintelligible save on the grounds
that God refers to a being who is disembodied and/or outside
time and the notion of such a beings creating something is incoherent.
But (G1) does not ascribe to God either of those properties and so the
given objection is here irrelevant. Furthermore, a (rather peculiar
and unsophisticated) theist might view God as a physical entity who
resides in some kind of eternal hyper universe of which our cosmos is
part and who triggered the Big Bang by, say, pressing a button on a
highly complex machine. Fantastic though it is (and decidedly anthropomorphic
though such a deity would be), there is nothing at all conceptually
problematic about that view. It is every bit as empirically confirmable
in principle as the proposition that massive wombats live in the northern
hemisphere of Pluto. Such a theist could therefore utter the sentence
God created and rules the universe without uttering anything
metaphysical and thus perhaps unintelligible. It follows that (G1) is
a perfectly intelligible definition of God provided the
definition of the one Supreme Being is itself intelligible.
So, if that provision is met (as it shall be shown to be), then noncognitivism-1
is a mistaken view.
Some theists might complain that (D3)-(D5) do not reflect the meanings of the given terms as they typically use them, but probably most of them would accept those definitions. Most theists, like most theologians, would probably agree that Gods actions are constrained by (at least) logic and consistency with His other essential attributes. In other words, most would probably agree that God cannot, e.g., draw a square circle inasmuch as that is logically impossible; and most would probably agree that God cannot commit evil acts inasmuch as that would contradict His omnibenevolence. Furthermore, though some theists might take omniscience as the property of possessing maximal procedural and experiential knowledge in addition to maximal propositional knowledge, most of them would likely take the last as sufficient. Similarly, while some might say that omnibenevolence involves more than just loving everyone maximally, most of them would probably say that that property is enough on its own.
Although (D3) & (D4) give rise to certain oddities and perhaps even paradoxes, they are free of many of the defects that plague more traditional, imprecise definitions of the given terms.  (For example, the definition of omnipotence as having all power and the definition of omniscience as having all knowledge.) In any case, they clearly do not turn God-2 exists into patent nonsense, and that is all that we are concerned with here.
Matters are not quite so clear, however, with respect to (D5). What might be meant by the expression maximally loving? One popular definition is this:
(D6) Maximally loving = for any person P, P is maximally loving if, and only if, P always does everything in Ps power to protect everyones best long-term interests (out of concern for others welfare)The problem here is how to construe (D6) in such a way that sentences of the form x does that which is in everyones best long-term interests (out of concern for others welfare) express more than mere value judgments, as it might be objected that value judgments do not constitute propositions. However, we must also construe (D6) in a way that is consistent with how most theists would interpret does that which is in everyones best long-term interests (out of concern for others welfare), else the project is moot. First, though, we must consider whether (D6) is a definition that most theists would accept at all.
I think that it is, for if we were to ask the average believer what he means when he says that God is all-good, he would likely reply along these lines: I mean that God always does what is right (or good, just, etc.), that His nature is perfectly righteous and so His actions cannot fail to be ethical. And while probably he would reject a utilitarian theory of ethics in favor of the divine command theory, surely he would agree that part of what it means to be perfectly righteous and to act ethically is to never needlessly (intentionally) inflict harm upon anyone.  He would thus surely agree that always acting in a way which protects everyones best long-term interests is a necessary condition for possessing omnibenevolence. But it is quite unlikely that he would also view it as a sufficient condition for that, and that is the reason for including the qualifier out of concern for others welfare. That is, somebody could invariably avoid causing gratuitous harm and yet at least sometimes do it for the wrong reason, i.e., not do it out of altruism. Any manner of selfish motive might impel him to appear magnanimous: wanting to win the praise and admiration of others, wanting to avoid obstacles to his own happiness, and so forth. From the standpoint of the average theist, then, always acting in a way which protects everyones best long-term interests is alone insufficient to make one omnibenevolent. There is more to it than just that, in other words, and the additional factor in question is so acting out of concern for others welfare. As (D6) takes that factor into account, I think most theists would agree that it accurately captures their idea of maximal love.
I turn now to the task of assigning the verb phrase does that which is in everyones best long-term interests (out of concern for others welfare) a non-morally-evaluative (or amoral) analysis. The most obvious candidate is the following: for any person P, if P performs that act which ultimately provides everyone with a greater degree of happiness (or comfort, well-being, etc.) than could any other act (and does so regardless of Ps own desires), then P does that which is everyones best long-term interests (out of concern for others welfare). This analysis, though obvious, is problematic. For any given act is bound to provide at least one person with a lesser degree of happiness than the degree of happiness with which another act would provide him, but which provides somebody else with a greater degree of happiness than could any other (conceivable) act: one mans heaven is another mans hell. The present analysis thus appears to turn omnibenevolence into a self-contradictory attribute, like square circularity. According to that analysis, then, God-2 exists expresses an unthinkable (hence necessarily false) proposition and so renders noncognitivism-2 an incorrect view.
However, perhaps a slightly different analysis would better reflect how most theists would interpret the verb phrase at hand. Consider, for example, the following: for any person P, if P performs that act which ultimately brings about a greater degree of happiness (or comfort, well-being, etc.) among people generally than could any other act, then P does that which is in everyones best long-term interests. Although this analysis may strike some as rather counter-intuitive and off the track, it may well accurately reflect what most theists mean when they say that God is omnibenevolent. They might argue that while pleasing most people does not immediately please everyone, in bringing about the greatest possible measure of (overall) happiness it ipso facto creates the ideal world, and the creation of that world is certainly in everyones best long-term interests whether he realizes it or not. At any rate, it should be plain that on the given analysis there is nothing obscure about the property of omnibenevolence, for the notion of bringing about maximal happiness seems quite clear. So, on this analysis, too, noncongitivism-2 is incorrect. Atheism might be in order, but certainly noncognitivism isnt.
I cannot think of any other analysis of the given verb phrase with which most theists might agree. However, I suspect that nearly any such analysis would at least be intelligible and so would not render God-2 exists cognitively meaningless per (D2). Let us therefore here conclude our discussion of (G2) and move on to the divine attributes referenced in the subsequent definitions of God.
Since most or all noncognitivists would regard God-3 exists and God-4 exists to express (clearly thinkable) propositions to the extent that love exists and the universe exists express (clearly thinkable) propositions, let us skip those and move on to (G5): the Ground of Being; the Source of Everything. This is a definition of God that is quite popular among New Agers and other liberal believers. But what do they mean by the terms the Ground of Being and the Source of Everything? We have here yet another wholly empirical question. However, it is once more possible to venture a reasonable guess. Likely the vast majority of people who use such expressions intend to thereby refer to the entity without which it is supposed that nothing could exist. Put another way, by such expressions is usually meant the entity whose existence is supposedly necessary for existence itself.
There is a certain temptation here to dismiss the notion of such an entity as nonsensical rubbish and thus invoke noncognitivism-5. But we must be careful. Does the sentence God-5 exists express no proposition at all, or does it instead express an unthinkable proposition? The correct answer, I believe, is the latter. For our inclination to write off God-5 exists as unintelligible gibberish arises not from any incapacity to comprehend the language of that sentence, but rather, from our inability to conceive what it would be like for God-5 to exist. That is, the proposition expressed by God-5 exists entails that God is responsible for the existence of (at least) everything except God himself.  But we cannot comprehend that to the extent that we cannot comprehend how something could be responsible for the existence of space and time For the concept of causation is inherently spatio-temporal and so the idea that x should cause y without occupying points in spacetime is unthinkable.  In other words, space and time are both built into the very notion of causation: to say that x causes y is to imply that x temporally precedes y and that x is spatially contiguous to y. No other analysis of causation is intelligible. Theists might dispute this, but certainly noncognitivists wouldnt. After all, most of them are staunch materialists whose esteem for the scientific method would be difficult to overstate. So, when they say that God-5 exists does not express any proposition whatever, what they really mean to say is that God-5 exists expresses an unthinkable and so necessarily false proposition: it is inconceivable that something should be the cause of everything besides itself, including space and time. Therefore, they are mistaken in adhering to noncognitivism-5. They should instead espouse atheism-5.
We come next to (G6): the personal, omnipotent, omniscient, omnibenevolent, eternal ruler and creator of the universe. That this definition is intelligible should not be controversial. We have already seen that all the properties mentioned therein save for that of being personal are coherent, and few properties are more readily graspable than that, as it is a property that we each possess. It might be objected, however, that while that property is perfectly comprehensible on its own, it becomes problematic when combined with the other properties mentioned in (G6). But why should it? Can we not easily imagine a person who is able to perform any act that is both conceivable and compatible with all of his other essential characteristics, who has all and only true propositions as beliefs, etc.? If not, then it is hard to see to what sort of entity, if not one of a personal ilk, we might ascribe those properties. Thus, if the noncognitivist grants the intelligibility of all the properties mentioned in (G6) and yet nonetheless insists that God-6 exists is cognitively meaningless, then it is incumbent upon him to explain how that sentence might fail to express a proposition.
Let us now consider (G7): the personal, omnipotent, omniscient, omnibenevolent, nonmaterial, eternal creator and ruler of the universe. Here the property of questionable coherence is that of being nonmaterial. It is a property that the overwhelming majority of theists ascribe to God, if only implicitly, e.g., by referring to him as a Spirit. Noncognitivists, of course, complain that that ascription is meaningless, especially when coupled with other properties commonly ascribed to God, e.g., personhood and benevolence. As Kai Nielsen writes:
What Nielsen seems to be suggesting here is that God-7 exists is a type crossing (or category mistake), for it ascribes to one type of thing (persons) a property (being bodiless) that is ascribable only to a different type of thing (non-persons), if indeed it is ascribable to anything at all.  He further implies that being personal entails performing actions and yet nonmaterial beings cannot perform actions, for it is impossible to perform an action without a body. Whether he is right about that latter point is irrelevant to our discussion here. All that is important is that we recognize Nielsens implicit admission that God-7 exists expresses a proposition, albeit obviously unthinkable, to the extent that type crossings express unthinkable propositions. Therefore, if he regards noncognitivism-7 as warranted (as I believe he does), then he is mistaken.  Clearly, the proper stance to take here is, instead, atheism-7. 
The next definition on the list is (G8): the personal, omnipotent, omniscient, omnibenevolent, nonmaterial, atemporal creator and ruler of the universe. Is the concept of an atemporal (or timeless) person coherent? That is hard to say, but certainly nothing like the ordinary concept of personhood is equipped to accommodate such an obscure predicate as timelessness. As Nielsen points out, it is only persons that can be coherently supposed to perform actions. However, performing actions, no less than causing effects, cannot be viewed as occurring outside time without so stretching the definitions of perform and action as to render those terms wholly vacuous, viz., devoid of intelligible content.  To perform an action is to have, prior to performing it, the intention to perform it, and then, after performing it, no longer have that intention. But that requires change and change requires (the passage of) time; to posit an action that takes place sans time is therefore absurd. So, it makes no sense to say of an atemporal being that it performs actions, performing actions being every bit as inherently temporal as causing effects. Yet the concept of a person who never performs actions seems woefully nebulous. To be certain, part of what it means to say x is a person (or x is personal) is that x has intentions and fulfills those intentions, being purposeful in character. However, as was just shown, atemporal beings cannot fulfill intentions, since to do so requires existing within time and atemporal beings by definition exist outside time. Therefore, that any sentence of the form x is an atemporal person (or x is atemporal and x is a person) should express a thinkable proposition is dubious at best.  Besides, as it is already clear that causing effects requires existing within time and creating the universe involves causing an effect, since (G8) defines God as both atemporal and the creator of the universe, there can be no doubt that God-8 exists expresses an unthinkable and so necessarily false proposition. It follows that noncognitivism-8 is an incorrect view. As with all the previous definitions of God, it is atheism that ought to be embraced here.
There are two conceptions of God which, though less common than (G1)-(G8), are nonetheless sufficiently prevalent that a brief consideration of them is in order. Call them (G9) & (G10):
That God is indefinable is a view endorsed by mystics and perhaps some New Agers as well. They claim that God can be recognized only via mystical experiences, which are themselves verbally inexpressible. That is, such experiences can be exhibited through behavior but not intelligibly spoken about. Is this view coherent? Or, put another way, does God-9 exists express a proposition? I am inclined to say that it does, namely, a necessarily false proposition. For, first of all, it can be taken to express the proposition that there exists something (which we call God) the term for which cannot be defined. But every term can be defined at least ostensively or stipulatively. The notion of a totally indefinable term is therefore incoherent. As regards God-9, surely even the mystic would agree that that term could be stipulatively defined as that whose presence is felt during a mystical experience, and, furthermore, that mystical experience could be stipulatively defined as that kind of experience during which one experiences such-and-such sensations (or apprehends such-and-such propositions). Given such (stipulative) definitions, it would be an open question whether God-9 exists expresses a true proposition, but that it would express a proposition is clear.
Now consider (G10). This is a definition of God that is sometimes employed by theists of a philosophical bent. They say that God cannot be understood in positive (descriptive) terms and so can be accurately described only through the use of negations, e.g., God is not red, God is not small, etc. I am not entirely sure what to make of this view of God. My inclination, however, is to take God-10 exists to express a necessarily false proposition inasmuch as there exists something to which no positive properties can be ascribed expresses an unthinkable proposition. There are several reasons for that. First of all, it entails that it is not the case that if God acts, then God always acts in accord with the laws of logic, since to always act in accord with the laws of logic is to have a positive property. But that x should act and yet fail to act in accord with the laws of logic is unthinkable. The theist who advocates (G10) could simply object that God never acts, but then he owes us an account of what sort of being never acts. (Incidentally, he must also explain what sort of being never thinks, has emotions, etc.) Second, there seems no reason to exclude (all) relational properties from the class of positive properties, and yet it is not the case that God is less powerful than humans, which entails that God has the (presumably positive and therefore forbidden) property of being at least as powerful as humans, expresses a proposition to which surely even theists who employ (G10) would assent. If indeed they would, then God-10 exists would entail a contradiction and so would be unthinkable, hence necessarily false. Third, and finally, it is impossible to conceive something the description of which is wholly devoid of positive ascriptions. For instance, suppose you are asked to imagine a box that is not square, that is not tall, that is not made of cardboard, etc. How can you imagine such a box if you are only told what it is not and never told what it is? Until some positive ascription is supplied, your mind is but a blank slate, as it were; you are thus merely fooling yourself if you think you have thought the proposition that the given box exists. One cannot imagine a bunch of is nots in isolation from some is.
We have now examined all ten definitions of God supplied above. Perhaps there are other common definitions of that term that could not be dealt with analagously, but I am not aware of any. At any rate, (G1)-(G10) are certainly the definitions that are most often employed by both philosophers and laypeople. I must conclude, therefore, that if there is any version of noncognitivism that is true, then it has relevance only to the most esoteric forms of God-talk.
The question arises what of extra-theological significance, if anything, follows from that conclusion. I would say that much does, in particular certain implications for the philosophy of language. One important such implication is that no adequate theory of sentential cognitivity can take the cognitive meaninglessness of S to consist in Ss failure to express a proposition. For if I am correct in asserting that there are few if any theological sentences with regard to which noncognitivism-2 is correct, viz., few if any theological sentences which fail to express propositions, then it is because sentences which fail to express propositions are a rarity generally, assuming such sentences exist at all. But surely no theory of cognitivity on which nearly any given sentence comes up cognitively meaningful can be considered adequate; the whole purpose of constructing such a theory is presumably to afford a clear demarcation between those sentences with cognitive meaning and those without it, and so presupposes, ostensibly rightly, that the latter are more than but a figment of certain philosophers imaginations.
But neither will do any theory of sentential cognitivity per which the cognitive meaninglessness of S consists in Ss expressing an unthinkable proposition. For otherwise we encounter certain apparent absurdities, chief among them that cognitively meaningless sentences make perfect linguistic sense and that there is point to (loose, economical) talk of their truth-values.  Both of those theses follow straightforwardly from the idea that cognitively meaningless sentences express propositions, which is of course entailed by the sort of theory in question.
Perhaps the best approach, then, is to simply abandon the notion of sentential cognitivity and instead view linguistic aberrations in terms of syntax and thinkability only. Some linguistic items- e.g., hence if but horses and die- are ungrammatical and so do not even constitute sentences. These philosophers neednt worry about; the constructions of such sequences might interest the psychologist, but they themselves are epistemically insignificant. Other such items- e.g., the theory of relativity is blue- are perfectly grammatical but do not express anything that might be entertained in thought. These typically involve sortal confusion and have long troubled philosophers of language. But whatever solution those philosophers might decide upon, it is hard to locate any justification for invoking the idea of sentential cognitivity. Or, at the very least, it is hard to see what good is accomplished by retaining our present vernacular, whereby either cognitively meaningless sentences are virtually fictive or else syntactically flawless sentences might be described using language that is normally reserved for ungrammatical gibberish. By substituting ungrammatical or incapable of conveying knowledge for cognitively meaningless, we could perhaps avoid much of that linguistic confusion that too often hampers dialogues concerning matters of metaphysics.
 See Drange, Atheism, Agnosticism, Noncognitivism, http://www.infidels.org/library/modern/theodore_drange/definition.html.
 Unless, of course, it were taken as the property of creating not only the four-dimensional spacetime continuum that comprises the known universe, but space and time themselves. I touch on that matter below.
 For instance, many of the problems traditionally associated with omnipotence, such as the so-called paradox of the stone and the alleged incompatibility between omnipotence and omnibenevolence, do not arise on (D3). That is because omnipotence is there defined in such a way that logically impossible acts (e.g., ones lifting a stone than one cannot lift) do not fall within the range of acts that omnipotent beings can perform. It is also defined in such a way that the conditional if x is omnipotent, then x can commit evil acts does not hold, since if x is all-loving, then x could be omnipotent on (D3) even if x could not commit evil acts. However, other problems remain. For instance, on (D3) there might be an omnipotent being who can only scratch his left ear, for if xs sole essential attribute is the ability to do nothing save scratch xs left ear and x can indeed scratch xs left ear, then, according to (D3), x is omnipotent. And that result is counter-intuitive. Equally odd is the result that humans can perform certain acts (e.g., torturing babies for fun) that an omnipotent being perhaps could not, given that an omnipotent being might also be all-loving and therefore incapable of committing evil acts. Albeit problematic, (D3) may well be the best possible definition of omnipotence, and certainly many theists would go along with it despite the given difficulties.
 He would of course deny that Gods sending unsaved sinners to hell is gratuitous to the extent that such folks deserve to be eternally damned; for God to do other than sentence them to eternity in hell, then, would be the real injustice. Furthermore, he would contend that God sends unsaved sinners to hell not out of spite or vengeance, but rather, out of love for them. That is, according to the average believer, the unsaved sinner wants to spend eternity in hell and so by sending him there God is only giving him what he desires: everlasting separation from Gods kingdom.
 Obviously, if that sentence were construed to express the proposition that God is the cause of absolutely everything, including himself, then it would entail a contradiction, namely, that God existed before he existed. In that case, noncognitivism-5 would of course be false, since self-contradictory propositions are necessarily false.
 The same is true of the idea that x should sustain y without occupying points in spacetime, since to sustain something is to do nothing more than cause it to remain in existence (and perhaps remain in a certain state).
 Kai Nielsen, An Introduction to the Philosophy of Religion (New York: St. Martins Press, 1982), p. 36.
 So far as I know, there is nothing like a universally accepted definition of type crossing (or category mistake), but the most precise that I have seen formulated in the literature and the one that seems to best capture the essence of the sort of sentence at hand is as follows: a sentence which ascribes to something, x, a property with which the type associated is a class to which x does not belong (Drange, Type Crossings, p. 203). A simpler if somewhat less exact definition would be a sentence which says of one type of thing something that has application only to a different type of thing. Popular examples in the literature include the theory of relativity is blue and Socrates is a prime number.
 See Martin, Atheism, pp. 45-46.
 Again, theists might deny that the property of being personal is incompatible with the property of being bodiless, but it is exceedingly unlikely that any noncognitivist would do so, for it is a staple of noncognitivism that metaphysical notions like that of a disembodied entity are incoherent. And since it is only with noncognitivism that we are here concerned, there is no need to explore the issue whether personhood and nonmateriality really are incompatible. (I would argue, however, that they are.)
 The astute reader will observe that it is impossible to perform an action without also causing an effect. However, the reverse is not true: for x to cause y does not necessitate that x perform an action, since x might be a mere force of nature (e.g., wind) and performing actions is usually taken to constitute a deliberative undertaking of which only sentient creatures (or personal agents) are capable. Where the cause is non-sentient (or non-personal), then, rather than say of it that it performs an action, it would be more appropriate to say of it that it brings about a state of affairs (in a non-intentional way).
 It should be noted that all these same considerations apply equally well to thinking thoughts, satisfying desires, etc. That is, all of those necessarily take place within time and so cannot be consistently predicated of atemporal beings.
 I include the parenthetical qualifiers loose and economical because I do not regard sentences as truth-bearers in any precise sense of the term truth-bearer. I reserve that role for propositions.