ISSN 1393-614X Minerva - An Internet
Journal of Philosophy Vol. 13 2009
Euthyphro and the Open Question
In his excellent introduction to metaethics, Alexander Miller argues that there are affinities between G. E. Moore's open-question argument and Socrates’ argumentation in Euthyphro dialogue. Miller is also led to ask how Moore's argument can be disdained without being unsympathetic to Socrates’ argument. This paper answers to Miller's question by showing that the two arguments are quite different. It is also argued that the two arguments merit different assessments: one may well appreciate Socrates’ reasoning and yet be unconvinced by Moore's.
In Principia Ethica, G. E. Moore argues that moral concepts, such as the concept of good or that of evil, cannot be reductively analyzed in terms of naturalistic concepts, such as the concept of desire or that of aversion. Moore's argument to this effect has been coined 'the open-question argument'. In Plato's Euthyphro, Socrates argues, against Euthyphro, that the concept of good cannot be reductively analyzed through the concept of God's love. In his excellent introduction to metaethics, Alexander Miller (2004, 15n4) is struck by the impression that there are affinities between the open-question argument and the argument against Euthyphro. Moreover, because the former argument may be disdained without being unsympathetic to the latter, Miller is led to ask whence the differing assessments, especially if the two arguments are really identical. So, are they?
Let me give some further motivation for the question. First of all, it is clear that there are strong theoretical affinities between Moore's and Plato's views; Moore's ethics can be classified as Platonistic (Moore 1991, lecture V; Regan 1991, xxxi-xxxii), and both Plato and Moore believe that 'good' is an unanalysable concept. Furthermore, both Moore and Socrates of Euthyphro set forth an argument to the effect that an attempt to define good in a given way goes amiss. Besides, both Moore and Socrates can be understood as pursuing real definitions as opposed to nominal definitions. Given these grounds, the question quite naturally arises as to what the relation between the open-question argument and the argument against Euthyphro is.
In this paper I argue for the view that the two arguments are quite different. Moreover, I argue that they do merit differing assessments: one may well appreciate Socrates’ argument against Euthyphro and yet be wholly unconvinced by Moore's argument against naturalism.
I shall begin by reconstructing the argument against Euthyphro and then the open-question argument. With the help of these reconstructions, I hope to bring the two arguments structurally as close to each other as possible. An important difference becomes clear, though. The disparity pertains to conceptual substitution – it is governed by different principles in the two arguments, respectively. Finally, I argue that a more fundamental divergence is there to be found as well. Namely, the so-called paradox of analysis does not have to be a problematic issue for a proponent of the argument against Euthyphro in a way that it is for the defender of Moore's open-question argument.
1. A Reconstruction of the Argument Against Euthyphro
Let me now formulate Socrates’ argument against Euthyphro. The general outline of my treatment of the argument against Euthyphro is due to Mark Johnston account (see Johnston 1993, 118-119), discussed in Miller 1995, on which I will heavily rely here. The basic assumption invoked in the present reading is that both Socrates and Euthyphro agree that something is good if and only if it is loved by God; what they disagree about is the correct interpretation of this fact (cf. Miller 1995, 858). In a nutshell, Euthyphro proposes that the above equivalence is conceptually necessary, whereas the argument against Euthyphro purports to show that it is not (cf. Miller 1995, 858-859).
According to Socrates, then, the following equivalence is not conceptually necessary:
(EE) For all x, x is good if an only if x is loved by God.
An important role in the argument against Euthyphro is played by what can be called the Euthyphro dilemma (Plato 1997, 9e):
Does God love x because x is good, or is x good because God loves x?
As an answer, Euthyphro assents to the following claim (Plato 1997, 10e):
(EG) For all x, God loves x because x is good (and it is not the case that x is good because God loves x).
The 'because' here is the 'because' of explanation, as distinguished from mere conceptual articulation. Socrates and Euthyphro agree that (EG) does not articulate the concept of God's love. Instead, (EG) is genuinely explanatory: God's love for x can be explained by appeal to x's goodness.
Obviously enough, for a statement to be a potentially genuine explanation, it must not be trivially true; a statement that cannot be informative due to its literal meaning, will not qualify as a potential explanation. Following Johnston, Miller (1995, 858) calls this type of statement an 'explanatory solecism'. For instance, the statement "Donald loves Peg because Donald loves Peg" is an explanatory solecism, since it is trivially true; its literal meaning makes it uninformative.
Now, Socrates argues that, when read as a conceptually necessary truth, (EE) is in tension with (EG). The two statements are in tension, since the following substitution principle holds (cf. Miller 1995, 859):
(SS) Substitution of conceptual equivalents cannot turn a genuinely explanatory statement into an explanatory solecism.
According to the present interpretation, Socrates argues that since (EE) allows a substitution that turns a genuinely explanatory statement into an explanatory solecism, (EE) cannot be conceptually necessary. Namely, the substitution turns (EG) into
(ES) For all x, God loves x because God loves x.
2. A Reconstruction of the Open-Question Argument
I shall next propose a reconstruction of Moore's open-question argument, presented in §§12-13 of Principia Ethica (Moore 1954). First of all, I believe Moore, while arguing that good resists naturalistic analysis, is trying to show that the predicate 'good' is not conceptually equivalent to any naturalistic predicate 'N' (e.g. Moore 1942, 661). He thus argues that the following equivalence is not a conceptually necessary truth:
(EM) For all x, x is good if and only if x is N.
Now, according to Moore, the following question is genuinely open:
(OQ) Is an x which is N also good?
Obviously enough, there are open questions and closed questions. A clear case of a closed question would be a question that can be answered just by considering its literal meaning. For instance, the question "Are all black ravens black?" is obviously closed, since the literal meaning of it already gives the answer. The question "Are all black ravens greedy for glittery things?", in turn, is an open question, since answering that question requires information not due to merely understanding that question. In distinguishing open questions and closed questions, Moore emphasizes the idea of intelligibility. What he calls closed questions are those that are unintelligible, whereas the open questions are intelligible; it makes sense to ask an open question, but it does not make sense to ask a closed question (Moore 1954, §13).
The gist of Moore's argument, I find, is that when (EM) is read as a conceptual truth, it is in tension with (OQ). The tension surfaces when the following substitution principle is accepted:
(SM) Substitution of conceptual equivalents cannot turn a genuinely open question into a closed one.
According to the interpretation presented here, Moore argues that since (EM) allows a substitution that turns a genuinely open question into a closed one, (EM) cannot represent a conceptual truth. Namely, the substitution turns the open (OQ) into the closed question (CQ):
(CQ) Is an x which is good also good?
3. The Open-Question Argument and the Paradox of Analysis
The view that the open-question argument falls prey to the paradox of analysis has been set forth by various scholars (see e.g. Fumerton 1983, 477-479). To my knowledge, however, the comparison between the open-question argument and the argument against Euthyphro from the point of view of the paradox of analysis is missing. This is what I shall focus in on next.
Now, take any analysis of the form "A is B", where A is what is analyzed and B what is offered as the analysis. Suppose 'A' and 'B' have the same meaning. The analysis is then correct, but only expresses a trivial identity statement "A is A". However, if 'A' and 'B' do not mean the same, the analysis does not seem to be correct. Therefore, it seems that no analysis can be both correct and informative. What we have here is the paradox of analysis.
A crucial concept invoked in the paradox of analysis, obviously, is the concept of analysis. What does it mean to provide an analysis? In particular, what does Moore say about this? I read him as holding that the pursuit of defining good is the pursuit of finding a real definition of good, as distinguished from a nominal definition (e.g. Moore 1954, §6). Accordingly, Moore is dealing with the question of how and whether good can be metaphysically analyzed. However, Moore (e.g. 1942, 661) understands analysis as conceptual analysis. Therefore, I read Moore as holding that to provide a correct metaphysical analysis is to provide a correct conceptual analysis.
Now, the equivalence
(EM) For all x, x is good if an only if x is N
qualifies as a correct metaphysical analysis of good only if the equivalence is metaphysically necessary. However, Moore's conception of analysis requires that the equivalence represents a correct metaphysical analysis of good only if the equivalence is also conceptually necessary. That is, such equivalence should be true simply as a matter of meaning of the terms involved in it.
This being the case, an equivalence such as
(K) For all x, x is water if and only if x is H2O,
supposing that it represents a correct analysis, should not state anything that is not involved in
(K') For all x, x is water if and only if x is water.
So, correct analyses should be trivially true ― hopelessly uninformative ― since they do not state anything that is not involved in some uninformative statement such as (K'). Nevertheless, correct definitions do not always seem trivial. For instance, it is quite plausible to hold that if (K) is true, it is necessarily true (Kripke 1972, 314, 320-321; Putnam 1975, 233), and hence it may be held to give us a correct analysis of what it is to be water, by telling us that water is the same stuff as H2O. Furthermore, (K) is obviously informative; it takes a lot of scientific effort to discover that the equivalence holds.
Let me now formulate the way in which I believe the paradox of analysis is a serious problem for a proponent of the open-question argument. To start with, Moore can be read to claim that correct analyses are to be represented by means of conceptually necessary equivalences, as noted above. And conceptual necessities coincide with metaphysical necessities. The crucial question now becomes "How can there be any necessary equivalences that are not trivially true?" I shall next argue that the substitution principle (SM), coupled with Moore's conception of analysis, prevents there from being such equivalences.
In order to make my point clearer, let me first make a distinction between a question and its corresponding equivalence. Consider the following question:
Q: Is an x that is A also B?
Now, for each Q-type question we can formulate the following kind of equivalence:
E: For all x, x is A if and only if x is B.
I shall call E the 'corresponding equivalence' of Q, and Q the 'corresponding question' of E.
It is easy to see that, for each Q-type question, there is an equivalence E that allows a substitution that results in a closed question. Namely, E turns Q into
Q': Is an x that is A also A?
Keeping in mind that a necessary equivalence always closes its corresponding question, we see that the genuine openness of a question straightforwardly implies that its corresponding equivalence is not necessary. In particular, the equivalence is then neither conceptually nor metaphysically necessary. Hence, the open-question argument presupposes that an equivalence can represent a correct analysis only if its corresponding question is closed and hence unintelligible. However, the intelligibility of a question is a prerequisite for the informativeness of its corresponding equivalence. Therefore, Moore seems committed to the view that the substitution principle (SM), invoked in the open-question argument, prevents there from being informative necessary equivalences.
4. The Argument Against Euthyphro and the Paradox of Analysis
I shall next argue that the argument against Euthyphro can be defended without falling prey to the paradox of analysis. I will not be claiming that Socrates would have provided such defense; my treatment will be, from an ancient point of view, unquestionably anachronistic. However, I hope to succeed in pointing out that there is ample room for a theory whose proponent might put the argument against Euthyphro into good use without being committed to the claim that there cannot be any non-trivially but necessarily true equivalence concerning good.
One crucial tenet in the moral semantics in question is the distinction between metaphysical necessities and conceptual necessities. The above example of water as stuff composed of H2O molecules is a case in point. 'Water', as understood here, is a rigid designator; it designates the same kind of stuff – stuff composed of H2O molecules – in each possible world in which it exists. The equivalence
(K) For all x, x is water if and only if x is H2O
is thus metaphysically necessary. However, the equivalence is not conceptually necessary, since 'water' and 'stuff composed of H2O molecules' are not synonymous.
Notice that a proponent of the above sketch of semantic theory for natural kind terms may formulate an argument that closely resembles the argument against Euthyphro, in order to show that the concept of water resists analysis into the concept of stuff composed of H2O molecules. Importantly, one may argue in this fashion and still hold that the equivalence (K) is necessarily true, without committing herself to the view that the equivalence is trivially true. One might proceed as follows.
First, one might single out the purported conceptual analysis, represented by the equivalence (K). Then one could accept the following genuinely explanatory statement:
(EW) For all x, x is water because x is composed of H2O molecules.
One could then argue that (K) and (EW) are in tension, since the substitution principle (SS) holds. To repeat:
(SS) Substitution of conceptual equivalents cannot turn a genuinely explanatory statement into an explanatory solecism.
According to the present argument, (K) allows a substitution that turns the genuinely explanatory (EW) into the explanatory solecism:
(EH) For all x, x is water because x is water.
Hence, (K) cannot be conceptually necessary. However, one may still hold that even if (K) is not conceptually necessary, water and stuff composed of H2O molecules are, so to speak, metaphysical equivalents. Finally, one may point out that substitution of non-synonymous terms that nevertheless refer to metaphysical equivalents does not preserve meaning, so no harm is done to (K)'s informativeness. To wit, the substitution would trivialize (K) only if (EW) and (EH) meant the same.
Along quite similar lines, the proponent of the argument against Euthyphro may criticize the attempt to analyze the concept of good in terms of that which is loved by God. She may claim that moral terms, such as 'good', behave in an important sense like natural kind terms, such as 'water'. In particular, she may argue that the equivalence
(EE) For all x, x is good if an only if x is loved by God
is metaphysically but not conceptually necessary. Very well, she may say, (EE) turns a genuinely explanatory statement into an explanatory solecism by substitution, but this is not a problem for (EE)'s informativeness, since meaning is not preserved by this substitution. Hence, the argument against Euthyphro can in principle be accepted without being committed to the view that the equivalence (EE) is trivially true. It can therefore be argued that the paradox of analysis is not an issue for the proponent of the argument against Euthyphro as it is for Moore's open-question argument.
5. Tracking Moral Truths
Obviously, something more has to be said as to how the proponent of the argument against Euthyphro can work out the distinction between metaphysical and conceptual necessities. Mere analogy between moral terms and natural kind terms is not enough.
In fact, the analogy between moral terms and natural kind terms is not quite tenable in the present context. However, this fact helps to steer in the right direction when trying to describe how the proponent of the argument against Euthyphro might understand the metaphysically necessary equivalence (EE). The analogy ultimately fails because (EE), when read as a metaphysically necessary truth, does not state that the properties being good and being loved by God are identical; however, a metaphysically necessary equivalence involving natural kind terms, according to the view discussed here, is read to express the view that some properties, such as being water and being composed of H2O molecules, are identical. The question now arises as to how the proponent of the argument against Euthyphro explains the metaphysical necessity of (EE). Let me now sketch an answer.
First, consider a distinction, due to Crispin Wright, between views according to which the most informed opinions are extension-reflecting and views according to which the most informed opinions are extension-determining. Take, for instance, Socrates’ belief that piety is good. On an extension-determining view, the most informed opinions constitutively determine facts that confer truth or falsity upon Socrates’ belief. On an extension-reflecting view, in turn, the most informed opinions track mind-independent facts that confer truth or falsity upon Socrates’ belief (cf. Miller 1995, 859-860.) This distinction helps to understand how the proponent of the argument against Euthyphro may account for the metaphysical necessity of (EE). She may claim that God is an infallible detector of good (Miller 1995, 858). (EE) may hence be understood as a thesis expressing the view that in all possible worlds God tracks moral truths without any error and loves all things that are good independently of his judgment. However, the property of being loved by God is not identical with nor is it constitutive of the property of being good. (EE) expresses an extension-reflecting view of God’s moral beliefs, not an extension-determining one.
To me the above view seems very plausible, if the theistic framework involved in it is accepted in the first place. When God is brought into the picture in moral epistemology, it is very plausible to hold that he is morally omniscient. For if God exists, his beliefs certainly are the most informed ones there could be. And these beliefs of an omniscient being infallibly track moral facts that make those beliefs true. Hence, I find it very plausible to hold that if the equivalence (EE) is true, it is true in all possible worlds. Furthermore, the claim both Socrates and Euthyphro agree on as an answer to the Euthyphro Dilemma, i.e. the thesis (EG) that concerns the ontological priority between God's love and goodness and includes a commitment to moral realism rather than moral anti-realism, is very arguable as well. For instance, if God's judgment of a thing's moral status was the cause of that status rather than its effect, the root of morality would seem to boil down, someone might argue, to a sheer caprice or whim of a divine agent. Furthermore, as a result, moral standards would appear nothing but wholly arbitrary. Again, someone could find it hard to think highly of the goodness of God himself, for an agent who managed to live up to a self-made standard would not strike us as a particularly admirable being.
I am not claiming that the thesis (EG) is unproblematic. One might, of course, argue for the opposite. For instance, one might perhaps point out that the judgments of an omnipotent God cannot be regulated by moral standards that exist prior to those judgments. One might also consider God's unlimited freedom and independence as grounds for the view that a thing's moral status is dependent upon God's judgment rather than the other way around. What I am claiming, however, is that there is ample room for such moral semantics, epistemology, and ontology that enable one to accept the argument against Euthyphro without being committed to the claim that the equivalence (EE) is trivially true.
6. Euthyphro and the Open Question
Two important purportedly necessary equivalences have been discussed in this paper:
(EE) For all x, x is good if an only if x is loved by God;
(EM) For all x, x is good if and only if x is N.
The argument against Euthyphro aims to show that (EE) is not conceptually necessary, whereas Moore's open-question argument purports to draw the same conclusion for (EM). These arguments can be read to bear structural similarity, but, I have argued, there is a crucial difference: Moore's open-question argument is acceptable only if (EM) is trivially true, whereas the argument against Euthyphro can be accepted without being committed to the view that (EE) is trivially true. Hence, the paradox of analysis is an issue for Moore's argument in a way that it is not for the proponent of the argument against Euthyphro. One may, then, well be sympathetic to the latter argument without being at all convinced that the open-question argument is plausible.
Boyd, R. 1988. How To Be a Moral Realist. In Essays on Moral Realism, ed. G. Sayre-McCord, 181-228. Ithaca: Cornell University Press.
Fumerton, R. 1983. The Paradox of Analysis. Philosophy and Phenomenological Research 43: 477-497.
Horgan, T and M. Timmons. 1992. Troubles for New Wave Moral Semantics: The 'Open Question Argument' Revived. Philosophical Papers 21: 153-175.
Johnston, M. 1993. Objectivity Refigured: Pragmatism without Verificationism. In Reality, Representation & Projection, ed. J. Haldane and C. Wright, 85-130.
Kripke, S. 1972. Naming and Necessity. In Semantics of Natural Language, ed. G. Harman and D. Davidson, 253-355. Dordrecht: Reidel.
Langford, C. H. 1942. The Notion of Analysis in Moore's Philosophy. In The Philosophy of G. E. Moore, ed. P. A. Schilpp, 321-342. La Salle, Illinois: Open Court.
Miller, A. 1995. Objectivity Disfigured: Mark Johnston's Missing-Explanation Argument. Philosophy and Phenomenological Research 60: 857-868.
Miller, A. 2004. An Introduction to Contemporary Metaethics. Cambridge: Polity Press.
Moore, G. E. 1954 . Principia Ethica. Cambridge: Cambridge University Press.
Moore, G. E. 1942. A Reply to My Critics, §11: Analysis. In The Philosophy of G. E. Moore, ed. P. A. Schilpp, 660-667. La Salle, Illinois: Open Court.
Moore, G. E. 1991. Elements of Ethics, ed. T. Regan. Philadelphia: Temple University Press.
Plato 1997. Euthyphro. In Plato: Complete Works, ed. J. M. Cooper. Indianapolis: Hackett.
Putnam, H. 1975. The Meaning of "Meaning". In his Mind, Language, and Reality. Philosophical Papers, vol. 2, 215-271. Cambridge: Cambridge University Press.
Sharvy, R. 1972. Euthyphro 9d-11b: Analysis and Definition in Plato and Others. Noûs 6:119-137.
Sayre-McCord, J. 1997. 'Good' on Twin Earth. Philosophical Issues 8: 267-292.
Regan, T. 1991. Editor's Introduction. In Elements of Ethics, ed. T. Regan. Philadelphia: Temple University Press.
 Notice that Miller is not the only one who is aware of the similarities between Moore's and Socrates’ arguments. Richard Sharvy (1972), for one, discusses these similarities.
 It should be noted that this is not the only possible solution to our question. For instance, Richard Sharvy argues that, despite some superficial similarities between Moore’s argument and Socrates’ argument, these similarities disappear when Socrates’ argument is viewed as essentially involving principles about formal causation rather than principles about substitution (see Sharvy 1972).
 The ‘because’ of explanation is to be distinguished from the ‘because’ of conceptual articulation. According to the present distinction, for instance, the statement “John is a bachelor because he is an unmarried male of marriageable age” is a conceptual articulation of John’s bachelorhood (see Miller 1995, 858), whereas the statement “John is a bachelor because he prefers to be single” (at least potentially) explains why John is a bachelor.
 Incidentally, Moore (1942, 661) continued to insist that analysandum and analysans the same concepts even after Langford (1942, 322-323) had set forth the paradox of analysis to him. As a reaction to Langford’s criticism, Moore (1942, 666) stated that, in a correct conceptual analysis, analysandum and analysans are different expressions of the same concept. It is not easy to see how the paradox of analysis could be avoided this way, and Moore (ibid.) in fact admitted that he has no clear solution to the paradox.
 The semantic advances due to Kripke and Putnam have motivated some philosophers to develop the so-called new wave moral realism in which moral terms behave, pretty much at least, like natural kind terms. For instance, according to the naturalistic moral realism of Richard Boyd (see his 1988), moral terms rigidly designate natural properties that causally regulate the correct use of these terms, without there being any synonymy between moral terms and naturalistic terms (see also Horgan & Timmons 1992, 159 for a terse summary of Boyd’s moral semantics). Jeffrey Sayre-McCord, for one, has defended a meta-ethical rigid designation theory without any ineradicable naturalistic strand. In Sayre-McCord’s thinking, moral terms resist analytic definitions but rigidly designate moral kinds that causally regulate the correct use of moral terms; furthermore, according to Sayre-McCord, it is irrelevant whether moral kinds coincide with natural kinds (Sayre-McCord 1997, 269-270, 285). The view I am attributing to the proponent of the argument against Euthyphro differs in an important sense from both naturalistic and non-naturalistic new wave moral realism, though. To wit, new wave moral realism includes the idea that the property of being good is identical with some purported property P, without there being any synonymy between the terms that refer to good and P, respectively. This does not hold good for the view I am here attributing to the proponent of the argument against Euthyphro; she does not claim that the properties being good and being loved by God are identical.
 I would like to thank a referee of Minerva for helpful comments. My work on this paper has been financially supported by the Academy of Finland (grant 8114178).
Copyright © 2009 Minerva
Dr. Timo Kajamies is a post-doctoral researcher at the department of Philosophy, University of Turku, Finland.