In my last two posts, I argued that Erik Wielenberg fails to show that Godless Normative Robust Realism (GRNR) avoids some of the standard objections to the autonomy thesis. This brings me to Wielenberg’s third claim III, Wielenberg suggests that GRNR is prima facie preferable to various theistic accounts of axiological properties. Several authors have defended accounts of the nature of moral goodness which identify goodness with certain types of relationships to God. Robert Adam’s has defended a platonic conception of goodness, where God is the paradigmatic good, and all finite things are good or bad depending on how they resemble Gods nature. Mark Murphy has defended an Aristotelian position whereby goodness consists in “being like God in ways that belong to the kind to be like God”[1]. Linda Zagzebski, by contrast, has appealed to Gods motives[2] and, and Thomas Carson has defended an account of Goodness whereby goodness is what God prefers[3]. 
Wielenberg refers to all these positions under the umbrella term “theological stateism” (TS). Wielenberg argues that prima facie GRNR is preferable to TS because, unlike TS, GRNR is consistent with the intuition that certain things have intrinsic value:
“I suggest that among our common-sense moral beliefs are the belief that some things distinct from God are intrinsically good: for example, the pleasure of an innocent back rub, or the love between parent and child,” …“[B]ecause non-theistic robust normative realism allows for the intrinsic goodness of things distinct from God, that theory fares better in this respect than its theistic alternatives.”[4]
By “the intrinsic value of a given thing” Wielenberg means the “value it has, if any, in virtue of its intrinsic properties.” Something’s extrinsic value by contrast “is the value it has in virtue of how it is related to things apart from itself.”[5]
Can Theological Stateism accommodate Intrinsic Value?
Wielenberg’s argument relies, crucially, on the claim that TS entails that nothing distinct from God is intrinsically good. His argument for this claim is as follows:
A second noteworthy aspect of Adams’s view is its implication that no finite thing is intrinsically good (or evil) since the goodness (and badness of things) of all finite things is dependent upon their relationship to God. Craig follows Adam’s in holding that finite goodness=resemblance to the necessarily existing divine nature. (emphasis mine)[6]
Later he cites Mark Murphy’s view as an example;
Back in section 2.2, I noted that Adam’s theory implies no finite thing is intrinsically good (or evil) since, on Adam’s view, the goodness (and badness) of all finite things is partly determined by how they are related to God. Consequently, Adam’s view holds that nothing distinct from God is intrinsically good. Murphy holds also holds that the goodness of things distinct from God consists in their standing in certain relationships to God: their goodness is extrinsic rather than intrinsic because it is explained not merely by intrinsic properties but also by certain properties of God.[7]
According to Wielenberg then reason TS entails that “the value of all finite things is dependent upon their relationship to God” (emphasis mine)[8]. However, if a finite thing has intrinsic value, it is valuable “in virtue of its intrinsic properties” and not “in virtue of how it is related to things apart from itself.” Consequently, TS entails that that “nothing distinct from God is intrinsically good.”
For this argument to be valid, the kind of dependence that TS postulates to hold between God and the goodness of finite things must be the same kind of dependence that Wielenberg’s definition of intrinsic value rules out. Consequently, the phrase “in virtue” in Wielenberg’s definition, must refer to the same kind of dependence relationship which TS entails exists between God and the value of finite things. However, a careful examination of Wielenberg’s definition, however, shows this is not the case.
Earlier in Robust Ethics Wielenberg spends some time elaborating what he means when he defines intrinsic value something has in virtue of its intrinsic properties. On pages 9-15 he distinguishes between two[9] different types of supervenience relationship which can hold between finite things and their evaluative properties. The first is what Wielenberg calls “reductive supervenience” (or R supervenience) this is where a moral property M supervenes upon a base property B because the moral property is identical with the base property. The second is what Wielenberg refers to as “De-Paul supervenience” (D- supervenience). If a moral property M D supervenes upon a base property B then “M is not identical with, reducible to or entirely constituted by B” instead “B’s instantiation makes M be instantiated”.
D supervenience and R supervenience are distinct types of relationship. M can R supervene upon B without D supervening upon B and M can D supervene upon B without R supervening upon it. GRNR is itself an example of this distinction. According to GRNR moral properties are sui generis irreducible non-natural properties that D supervene upon natural properties. Consequently, GRNR presupposes that moral properties can D supervene upon natural properties without R supervening upon them.
Not only do are D and R supervenience distinct relationships. It appears they are mutually exclusive. If B is reducible to or identical to M, it B cannot be a distinct property from M, which M causes to exist. That would involve self-causation. Similarly, if B is distinct from M and B makes M be instantiated, B is not reducible to or identical with M.
When Wielenberg defines a finite thing’s intrinsic value as the value, it has in virtue of its intrinsic properties, he uses the phrase “in virtue” to refer to a relationship of D supervenience.
If there are entities distinct from God that possess intrinsic value, then Craig is mistaken. I think there are such entities. As I suggested in chapter 1, some finite things pass the isolation and annihilation tests which, which suggests such things are intrinsically valuable. The intrinsic value of such entities D supervenes upon some set of their intrinsic properties and not on how they are related to other things.[10]
Earlier he writes:
In my view, the most plausible way of understanding the “in virtue” relationship which I earlier claimed holds between the intrinsic properties of certain things and their intrinsic value is making. To claim that a given thing is intrinsically valuable is to claim that some of that thing’s intrinsic properties make it valuable… More generally, I think that moral properties indeed all moral properties D supervene upon non-moral properties.(emphasis mine)[11]
The problem is that proponents of TS are not committed to denying that moral properties D supervene on the intrinsic properties of finite objects. When people like Mark Murphy or Robert Adam’s or William Lane Craig contend that the goodness of all finite things is dependent upon their relationship to God, they are not claiming that goodness D supervenes upon this relationship. They are claiming it R supervenes upon this relationship. Consider how Wielenberg himself describes Adams’s position.
A noteworthy feature of Adams’s view is its implication that no finite thing is intrinsically good (or evil) since the goodness (and badness of things) of all finite things is dependent upon their relationship to God. Craig follows Adam’s in holding that finite goodness=resemblance to the necessarily existing divine nature... Murphy holds also holds that the goodness of things distinct from God consists in their standing in certain relationships to God: their goodness is extrinsic rather than intrinsic because it is explained not merely by intrinsic properties but also by certain properties of God. (emphasis mine)[12]
It is true that according to Adams the goodness of finite things is “dependent” upon their relationship to God. However, but in the highlighted sentence, Wielenberg spells out the kind of dependence Adam’s has in mind. The dependence Adams refers to is one where “finite goodness=resemblance to the necessarily existing divine nature”. Adam’s is, therefore, saying that the goodness of finite objects R supervenes upon the divine nature. Wielenberg notes this when a few paragraphs earlier he describes Adam’s (and Craig) as holding that:
[S]ince the Good just is God, the existence of God cannot explain or ground the existence of the good. In the context of Adams’s view, the claim God serves as the foundation of the Good is no more sensible than the claim that H20 serves as the foundation of water. Indeed, once we see that on Adams’s view the good = God, we see that Adams’s theory entails that the Good has no external foundation, since God has no external foundation.[13]
Similarly, when Wielenberg discusses Murphy he notes that according to Murphy the goodness of things distinct from God “is explained”, not merely by intrinsic properties, but also by certain properties of God. However, he clarifies that, by this, Murphy means the goodness of finite things “consist in” their relationship to God So again, Murphy is talking about R Supervenience, not D Supervenience.
So, contrary to Wielenberg TS does not entail that things distinct from God cannot have value in virtue of, their intrinsic properties. TS entails that the value of something distinct from God R supervenes upon its relationship with God. However, that does not entail its value D supervenes upon this relationship. So TS is not incompatible with the common sense intuition that finite objects can have intrinsic value.
On the other hand, If the objection is that TS entails that the value of finite things cannot R supervene upon their intrinsic properties, then GRNR fares no better. According to Wielenberg, the proponent of GRNR holds that moral properties are “real and sui generis; they are non-natural and not reducible to any other sort of property” (emphasis mine). If moral properties are not reducible to other properties, then they do not R supervene on the intrinsic properties of finite things. In fact, on GRNR and the goodness of every distinct and finite thing will R supervene, not upon its intrinsic properties but on its relationship to distinct irreducible non-natural properties and so finite things will have only extrinsic value. This objection, therefore, is a non-starter.
[1] Mark Murphy God and The Moral Law: A Theistic Explanation of Morality (Oxford: Oxford University Press) 155
[2] Linda Zagzebski Divine Motivation Theory (Cambridge: Cambridge University Press, 2004)
[3] Thomas Carson Value and the Good Life (Notre Dame Ind: University of Notre Dame Press, 2000)
[4] Erik Wielenberg Robust Ethics 84
[5] Ibid. 2
[6] Ibid. 44
[7] Ibid.
[8] Ibid.
[9] Wielenberg actually spells out three relationships, but mentioning the third would only complicate this discussion unnecessarily. The proceeding argument applies equally well when the third type of supervenience is taken into account.
[10] Ibid. 44
[11] Ibid. 13
[12] Ibid. 44
[13] Ibid. 43
Tags: Erik Wielenberg · God and Morality · Intrinsic Value · Linda Zagzebski · Mark Murphy · Moral Realism · Robert Adams · supervenience · Thomas Carson23 Comments
A common objection to belief in the God of the Bible is that a good, kind, and loving deity would never command the wholesale slaughter of nations. In the tradition of his popular Is God a Moral Monster?, Paul Copan teams up with Matthew Flannagan to tackle some of the most confusing and uncomfortable passages of Scripture. Together they help the Christian and nonbeliever alike understand the biblical, theological, philosophical, and ethical implications of Old Testament warfare passages.
“For this argument to be valid, the kind of dependence that TS postulates to hold between God and the goodness of finite things must be the same kind of dependence that Wielenberg’s definition of intrinsic value rules out. ”
I don’t see the force of this. I read Wielenberg as arguing that for an object to have intrinsic value, it must be the case that its value depends on its intrinsic properties. But “depend on” can be given different readings. In particular, to say that the intrinsic value of M depends on its intrinsic properties might mean,
(a) M’s value R-supervenes on M’s intrinsic properties
or
(b) M’s value D-supervenes on M’s intrinsic properties.
If either (a) or (b) is true, then M’s intrinsic value depends on its intrinsic properties.
The problem is that, on theistic accounts of value, for all finite objects, neither (a) nor (b) is true.
Wielenberg is committed to claiming that D-supervenience is the appropriate account of the dependence relationship between an object’s moral properties and its intrinsic natural (non-moral) properties. And so he argues that, on TS, a finite object’s value does not D-supervene on its intrinsic base properties. However, it is also true that, on TS, a finite object’s value does not R-supervene on its intrinsic base properties.
This makes your closing paragraph misdirected as a criticism of Wielenberg. It is true that, on GRNR, value does not R-supervene on natural properties. But that is irrelevant since, on GRNR, value does depend on an object’s intrinsic (natural) properties since its value D-supervenes on those properties. The problem for TS is that, on TS, a finite object’s value neither R-supervenes nor D-supervenes on its intrinsic properties. Since neither dependence relation is true on TS, it is not clear how a defender of TS can maintain that value depends (In any way) on an objects intrinsic properties.
Jason, you write:
I think that’s what Wielenberg would have to argue for his argument to be sound. The problem is that I don’t think he does argue or justify that claim that “on TS, a finite object’s value does not D-supervene on its intrinsic base properties.” What he does is point out that on TS moral properties depend on extrinsic properties. The problem is that in the examples he uses to show this, the dependence at issue is R-supervenience. He doesn’t provide any argument or reason for thinking that according to TS moral properties don’t D supervene upon intrinsic base properties.
Moreover, I am inclined to think that the claim that on TS action’s moral properties can’t D supervene on its intrinsic properties is false. D supervenience is a relationship where the moral property M and the B Property B are distinct (non-identical properties) which modally co-vary and where the instantiation of M depends on B.
But consider an example Wielenberg himself uses. The divine command theory of William Paley. Paley writes that because God is essentially benevolent one can conclude:
George Berkeley made a similar and more dense and precise version of this argument several decades earlier:
As I read these writers, both contend that moral wrongness and rightness consist in agreement and disagreement respectively with Gods commands. So they plausibly can be read to be claiming that claim that moral properties R supervene upon Gods commands.
However, they also think God commanded certain actions because those actions had certain natural properties. Certain actions or rules have a tendency to promote human happiness, which is on their accounts a natural property. Hence, there will be distinct natural properties involving somethings tendency to produce happiness the universal, which modally co-vary with Gods commands and God issued his commands because those actions had those properties.
So I am inclined to think both GRNR and TS are on par here. Both think moral properties aren’t reducible to natural properties. However, both think the distinct moral properties things have depend upon their natural properties in some non-reductive sense.
Examples like this suggest TS doesnt entail that moral properties can’t depend on the natural properties of actions or rules in some non reductive sense of dependence.
Consider the following general claim about supervenience:
(S) If property A supervenes on property B, then there cannot be a difference in property A unless there is a difference in property B.
I think that TS implies that this principle cannot be true for A = moral properties and B = intrinsic (non-moral) properties. TS says that moral properties depend on states of God. One implication of TS is that if God does not exist, then objects do not have moral properties. But the absence of God would not necessitate any difference in an object’s intrinsic non-moral properties. Thus, according to TS, there can be a difference in the moral properties of an object without a difference in the object’s intrinsic (non-moral) properties.
Consider two possible worlds:
W1: God exists, and object O exists. God approves of O (or occupies the relevant state, whatever it happens to be, with respect to O). So, on TS, O is good in W1
W2: God does not exist but object O exists. O has the same intrinsic (non-moral) properties in W2 that it has in W1. Since God does not exist, God does not occupy any state with respect to O. So, on TS, O is not good in W2.
On TS, O has different moral properties in W1 and W2. (S) tells us that if O’s moral properties supervene on O’s intrinsic (non-moral) properties, then any difference in O’s moral properties must correspond to a difference in O’s intrinsic (non-moral properties). TS says that O’s moral properties differ in W1 and W2 but there is no difference in O’s intrinsic (non-moral) properties in W1 and W2. Thus, on TS, O’s moral properties do not supervene on O’s intrinsic (non-moral) properties.
Jason,
I think a problem is that TS is not committed to “Possibly, God does not exist, and possibly, God exists”. In fact, if TS is defined in relation to all of the views that Wielenberg criticizes, it seems to me that TS is committed to the necessary existence of God, so the reply would be that W2 is not a possible world.
Here’s a potential modification that might get around that problem: in possible world W1, God commands that Joe does X. In possible world W2, God does not so command. So, what’s the value of Joe’s not doing X? It seems to me on TS, there would probably be a greater negative value on W1 than on W2 (though I’m not entirely sure since “TS” refers to more than one theory and is as a result a bit ambiguous).
I suppose a version of TS might hold that it’s not possible that God commands are different in that manner, but that would seem to present problems for other beliefs defenders of TS have (i.e., involving Christianity).
An alternative theistic reply might be that the value of things does not include the value of behaviors. But it’s hard to see why, if it involves the value of things like love, or pleasure, or pain (value can be negative) – and if it doesn’t, that’s enough to block Wielenberg’s examples.
Another alternative would be to modify Wielenberg’s argument and hold that on TS, the value of any finite object is not an intrinsic property of it, but a relational property, whereas on what he calls “GRNR”, an object’s value can be an intrinsic property of it (of course, one would have to modify Wielenberg’s definition of “intrinsic value”, and perhaps “GRNR”). But I think the argument would has a better chance if it’s not linked to “GRNR” (the quotes are because I disagree with the terminology, as it would imply that other, in my view much more probable metaethical views are somehow not “robust”).
Angra,
“God exists necessarily” can be given at least two readings:
(I) God exists in every possible world.
(II) God exists in every logically possible world.
Do you think that TS is committed to (I), (II), or (I) and (II)?
A world in which God does not exist, even if it is not a possible world, is still a logically possible world. A logically possible world is just a world in which there are no logical contradictions. The statement “God does not exist but other objects exist” is not a logical contradiction. Given that, W2 is a logically possible world.
My argument does not require that W2 is a possible world, only that it is a logically possible world. I admit that I used the term ‘possible world’ but the argument can easily be fixed by replacing the occurrences of ‘possible world’ with ‘logically possible world.’
Jason,
I meant in the sense of metaphysical necessity, often understood in terms of possible worlds (though there are philosophers who reject that interpretation); as long as one accepts possible worlds, it would be (I).
I don’t know that they’re committed to (II) – some might be, but I don’t know that.
If your argument is about logically possible worlds, a reply would be that:
1. Even if TS is not committed to the logical impossibility of God’s non-existence, it’s not committed to the logical possibility of it, either.
2. Granting TS is committed to the logical possibility of God’s non-existence, your argument shows at best that TS is committed to (S) being false for the relevant properties, but only if “cannot” is understood in terms of logically possible worlds. Yet, the supporter of TS is probably unconcerned by that.
3. It seems to me that a similar argument would work in the case of “GRNR”, for the following reason:
Let O be a good thing that actually exists. Let W3 be a logically possible world in which there are no moral properties (so a moral error theory is true), but object O exists. On “GRNR”, O has different moral properties in the actual world A and in W3.
While you could argue that the supporter of “GRNR” is not committed to the logical possibility that there are no moral properties, I think they probably are so committed because they reject any sort of analytic reductionism – and in any case, if that reply works, then the supporter of TS might also reply that they’re not committed to the logical possibility that God doesn’t exist.
The first modification I suggested does seem to avoid this objection the context of Wielenberg’s argument (i.e., without modifying it), since the modified argument can be made in terms of metaphysical rather than logical possibility.
Angra,
What you say about W3 is interesting and compelling. However, after reflection, I am not convinced that my argument is undermined.
Consider the following statement, which, at first glance, might seem to be a consequence of (S):
(R) For any two worlds composed of the same concrete objects, any difference in an object’s moral properties requires a difference in the object’s non-moral properties.
Let’s assume that W2 and W3 are composed of the same concrete objects. For W2 and W3, (R) is false for all concrete objects. But that (R) is false for W2 and W3 is a logical consequence of the fact that Error Theory is true in W3. Because of this, I do not think that the failure of (R) for W2 and W3 is relevant for testing the truth of GRNR (or any other moral theory). Nor is it relevant for testing whether (S) is true on the theory. That is, when the failure of (R) is a logical consequence of the intrinsic nature of on of the two worlds, the failure of (R) does not count against (S).
So, sure, if you gerrymander the world so that, as a matter of logical consequence, (R) does not hold, then (R) will not hold. But so what? No moral theory can imply otherwise. Therefore, the failure or (R) for W2 and W3 is not the fault of GRNR (or any other moral theory, including TS).
On the other hand, TS implies that (R) fails for W1 and W2. But this is not a logical consequence of the intrinsic features of either W1 or W2.
Assume that W1 and W2 are composed of the same concrete objects (except that in W2 there is no God). It is not a logical consequence of the fact that God does not exist (in W2) that objects in W2 lack moral properties. Nor is it a logical consequence of the fact that God does exist (in W1) that objects do have moral properties. Thus, unlike with W2 and W3, the failure of (R) in this instance cannot be due to the the logical consequences of the intrinsic features of either W1 or W2. Thus, given that (R) fails here, it must be because of TS.
This strikes me as a very significant difference between TS and GRNR. When (R) fails on GRNR, it is because of logical consequences of the intrinsic features of (at least one of) the two worlds. (R) fails on TS in exactly those same instances. But on TS, (R) also fails in other cases.
Jason,
It seems to me that since God is a concrete object (some theists deny that, but supporters of the theories under discussion usually don’t as far as I know), W2 and W1 are different with respect to concrete objects, and so they don’t show that (R) fails on TS.
Regarding whether (R) holds on “GRNR”, after reflection, I don’t think that the fact that I proposed a moral error theory in my example is a problem for my objection.
However, in any event, that example is not required.
I will argue that on “GRNR”, (R) is false if you consider logically possible worlds, for the following reason: “GRNR” holds that there is no analytic reduction of moral statements to non-moral ones, so there is – for example – a logically possible world W5 that is entirely the same as our actual world A with respect to all concrete objects (that’s logically possible), but in which Hitler was a good person, and his deliberate choice to kill Jews and others in gas chambers was a morally praiseworthy choice.
We may even stipulate that W5 is the same as A with respect to all non-moral properties.
The fact that there are no moral properties in W3 (and thus the fact that (R) is false for W2 and W3) is a logical consequence of the identifying features of W3. That is how W3 is identified: as a world that is just like this one except that there are no moral properties. And similar things apply to W5. The fact that, in W5 Hitler’s moral properties are other than they are in the actual world is a logical consequence of the fact that W5 is a world in which Hitler is good (and etc.). So, the fact that (R) fails for W2 and W5 is a logical consequence of the identifying features of W5.
We’ve gotten this result (that is a failure of (R) on GRNR) only because we are specifying a world where it is logically guaranteed that (R) fails. But this tells us, importantly, that (R) fails on all moral theories for W2 and W3 and for W2 and W5. Since it is a logical consequence, the failure is not the fault of the theory. The failure is a result of the nature of W3 and W5. On the other hand, on TS, R fails for W1 and W2. And this failure is not a logical consequence of the nature of either W1 or W2.
When we are testing whether (S) is true on a given moral theory, we have to ignore those worlds in which the nature of the world guarantees that (R) fails. Why? Because, since the failure of (R) is a logical consequence, the failure tells us nothing about the moral theory. But W2 is not a world in which the failure of (R) is logically guaranteed. That (R) fails for W1 and W2 on TS is a feature of TS.
Jason,
I don’t think we got the result that (R) failed on “GRNR” because I specified a world where it’s logically guaranteed that (R) fails. Rather, (R) is a claim about all logically possible worlds, and the conclusion that my two examples show is that if “GRNR” is true, then (R) is false. Other examples would show the same.
Is it the case that on all moral theories, (R) is false?
The answer seems to be negative. Take, for example, analytical functionalism. On a standard version of analytical functionalism, moral statements are analytically reducible to non-moral ones. In particular, given that our actual world A is clearly logically possible, every logically possible world that is nonmorally identical to the actual world would also be morally identical. It would follow that worlds like W2 and W5 are not logically possible, so said examples would not show that (R) is false on analytical functionalism.
Sorry, I meant W3 and W5, rather than W2 and W5.
Jason,
With regard to whether (R) is true on TS, I don’t think that the examples of W1 and W2 show that (R) is false on TS, because W1 and W2 are not identical with respect to concrete objects, as long as God is a concrete object. While some theists deny that God is an entity/being/thing, and presumably an object, theists who defend TS aren’t always in favor of that view. For example, they may well hold that God is the GCB, and as such, a being, and more precisely, a concrete one.
In order to use W2 and W1 to show that (R) is not true on TS, it seems to me you would have to show that TS implies that God is not a concrete object. But that seems to be a very difficult task.
(R) is ambiguous and poorly written. It should be dispensed with. However, I think I can capture the point I was trying to make with a different principle:
(C) If two worlds (W’ and W”) share an object (O) in common, then any difference in O’s moral properties in the two worlds requires a difference in O’s non-moral properties in the two worlds.
I think that (C) is probably just a way of stating (S) or at least is an implication of (S). My point has been that, on GRNR, (C) fails for some pairs of logically possible worlds. But, when, on GRNR, (C) does fail for some pair of worlds, it is because one of the worlds has identifying features that logically guarantee that (C) fails.
So, on GRNR, (C) fails for W1 and W3. These two worlds do share at least some objects in common, and at least some of the objects that they share in common have the same non-moral properties in these two worlds. But these same objects do not share the same moral properties in these two worlds. I am claiming that (C) fails in this instance only because W3 has, as an identifying feature, the property of having no moral properties. This feature logically entails that (C) fails for W1 and W3. Similar remarks apply to W1 and W5. Hitler exists in both worlds and has the same non-moral properties in both worlds, but he does not have the same moral properties. However, this result is a logical consequence of the identifying features of W5 (i.e., that it is a world in which Hitler is good).
I also claim that, when, on some moral theory, (C)’s failure is logically guaranteed by the identifying features of one of the worlds, then this failure is not the fault of the moral theory. Thus, such failures of (C) are not relevant to whether (S) is true on the moral theory in question.
On TS, (C) fails for W1 and W2. W1 and W2 share at least some objects and at least some of these have the same non-moral features. However, these same objects, on TS, do not have the same moral properties. This failure of (C) on TS is not a logical consequence of the identifying features of either W1 or W2. Thus, the failure is a result of TS. This failure is relevant to whether (S) is true on TS; it shows that (S) is not true on TS.
Anga,
I should have said thanks for helping me see that (R) is flawed.
Jason,
You’re welcome, and thanks for helping me improve my arguments as well.
I think your modified argument does avoid the sort of objections I raised earlier, but there are other objections that a defender of TS may raise; here’s one:
TS is a metaethical theory that posits that moral properties are the same as some relations to God, but the “sameness” involved is in terms of metaphysical possibility, not in terms of logical possibility. Just as a theory that water is H2O, or that the property of being red is identical to the property of reflecting such-and-such frequencies under such-and-such conditions is not committed to the claim that those identities are identities in the sense of logical possibility (in the sense we’re using that expression here), so TS is not committed to the claim that the identity of moral properties and some relation to God is an identity in the sense of logical possibility.
As long as that TS is committed to the metaphysical necessity of God’s existence (I’m not sure it’s true for every version of TS actually defended, but it’s surely true for some at least), TS implies that W2 is not a metaphysically possible world. But TS does not make claims about metaphysically impossible worlds, or at least not about metaphysically impossible worlds in which God does not exist – just as the theory about redness above does not make claims about logically possible but metaphysically impossible worlds in which there is no such thing as an electromagnetic frequency but there are red objects, and the theory of water above does not make claims about logically possible but metaphysically impossible worlds in which there is no such thing as hydrogen but there is water.
So, TS does not claim that in W2, moral properties are the same as some relations to God, and so it’s not committed to the claim that on W2, O is not good. In fact, TS is silent about whether O is good on W2.
I would maintain that any moral theory that is committed to (S) is committed claiming that (C) holds for any two pairs of logically possible worlds unless the failure of (C) for any two such worlds is logically guaranteed by essential features of one (or both) worlds.
Why?
Any metaethical theory that is committed to (S) implies either (a) that the supervenience of the moral on the non-moral is logical supervenience or (b) that it is not logical supervenience. We can probably rule out theories that assert (a) because of Moore’s Open Question Argument. And W3 and W5 certain present problems for them as well. A defender of such a theory will have to maintain that W3 and W5 are not logically possible. I can’t see how that can be done. In any event, let’s set such theories aside. The theories we are considering here (GRNR and TS) do not maintain (a).
Let’s look at GRNR. On GRNR moral properties do not logically supervene on non-moral properties. GRNR is committed to (S). What does this tell us about the implications of GRNR in all logically possible worlds? Well, given that the supervenience of the moral on the non-moral is not logical supervenience, we know that, with respect to any logically possible world that does not have moral properties, (C) will not be true for this world and any world that does have moral properties. And we know that worlds that are identified in virtue of having moral properties that systematically conflict (in the manner of W5) with those of the actual world will also produce a failure of (C). But these are the only logically possible worlds where (C) will fail, on GRNR.
Take any logically possible world that does not have essential features that guarantee (logically entail) a failure of (C) and (C) will be true for this world and the actual world (or any other possible that similarly lacks such features). The reason is that, unless these atypical consequences (lacking moral properties or having systematically reversed moral properties, etc.) are built into the nature of the world, GRNR will insist that objects with the same non-moral properties (as they have in the actual world) will have the same moral properties (as the actual world).
Here is what I am saying: If there is no logical reason for (C) to fail, then, given the commitment to (S), the theory must imply that (C) is true for any two worlds. It does not matter whether the worlds are metaphysically possible or merely logically possible, if there is no logical reason that (C) must fail, (C) will be true. If a theory does not imply this, then I don’t see how it can be committed to (S).
It does not matter that, on GRNR, the supervenience of the moral on the non-moral is not logical supervenience. The commitment to (S) still entails that (C) is true for any two (logically) possible worlds unless there is some logical reason that it fails.
Think of it this way: If (C)’s failure for some pair of worlds is not a logical consequence, then what is the reason for (C)’s failure? It cannot be that O has some different non-moral property in one world compared to the other. O has the same non-moral properties in both worlds. But then, if the failure of (C) cannot be traced to a difference in O’s non-moral properties, how can we maintain that moral properties supervene on non-moral properties?
On TS, (C) fails for W1 and W2. This failure is not a logical consequence of the features of W1 or of W2. It is true that TS is not committed to (C)’s truth for all logically possible worlds. But, if it is committed to (S), then, unless there is some other explanation for why this should not be so, if there are two logically possible worlds that have at least one object in common and this object has the same non-moral properties in both worlds, it must have the same moral properties in both worlds. The fact that TS denies this for W1 and W2 shows that (S) is false on TS.
(S) articulates the insight that, if an object’s moral properties supervene on its non-moral properties, then if there is a difference in O’s moral properties between two worlds, this must be accounted for in terms of a difference in O’s non-moral properties between these worlds.
The explanation for why (C) fails on TS for W1 and W2 has nothing to do with the logical consequences of the essential features of these worlds, not does it have to do with the non-moral features of O. Rather, it has to do with any entirely different object, namely God. Given this, I don’t see how TS can maintain that an object’s moral properties supervene on its non-moral properties.
Jason,
That’s a very interesting argument. After reflection, I don’t find it persuasive, but I don’t rule out that a modified argument might work. I’ll have to give it more thought.
The reason I don’t find it persuasive is that I think it’s not true that any moral theory that is committed to (S) is committed to claiming that (C) holds for any two pairs of logically possible worlds unless the failure of (C) for any two such worlds is logically guaranteed by essential features of one (or both) worlds (side note: I don’t find Moore’s argument persuasive, but I agree to leave those theories aside).
Consider, for example, “GRNR”. That’s a theory committed to (S). Let’s consider now a logically possible world W7 that shares an object O (let’s say, Donald Trump) with the actual world A, and in which it is not the case that there are sui-generis, irreducible non-natural properties (SGINNP for short).
Is “GRNR” committed to claiming that (C) holds for A and W7?
It seems not. “GRNR” holds that moral properties are SGINNP. In W7, there are no SGINNP, and so “GRNR” is false, and “GRNR” does not seem committed to claiming anything at all about whether Trump has the same moral properties in W7 in which “GRNR” is false as he does in A.
Yet, the worlds A and W7 are not defined in a way such that their essential features logically imply that Trump (or some other shared object) has different moral properties in A and W7:
A is merely defined by its actuality, without making any statements about Trump’s moral properties. And W7 is defined in terms of Trump’s existence and the non-existence of SGINNP. Further, W7 appears to be logically possible, and A clearly is.
As I mentioned, I can’t rule out that a modification might work, but I’m not sure either way.
“Is “GRNR” committed to claiming that (C) holds for A and W7?”
As I understand it, on GRNR, that moral properties are SGINNP is a conceptual matter. This is certainly the case on Parfit’s version on moral non-naturalism. This is indicated, for example, by the fact that Parfit produces what he takes to be deductively valid arguments against moral naturalism (such as the Triviality Objection and the Soft Naturalist’s Dilemma). If there are moral properties and they are not natural, then they must be non-natural. Purely a conceptual matter.
Given this, the fact that (C) fails for A and W7 is a logical consequence of the essential features of W7 (that it is a world in which there are no SGINNP).
Alternatively, we might say the following: If the fact that W7 is a world in which the lack of SGINNP does not rule our the existence of moral properties, then W7 is not a logically possible world. Why? Because the claim that there are moral properties and yet there are not SGINNP yields a contradiction. Since, as a conceptual matter, moral properties are SGINNP, it follows (deductively) that if there are no SGINNP, then there are no moral properties.
Jason,
I didn’t know Wielenberg’s “GRNR” agrees with Parfit’s version of realism on that. If it does, that avoids my counterexample, so good point. By the way, would you say that a version of TS that holds that it’s a conceptual matter that moral value is resemblance of God, avoids your objection in the same way because on such theory, the fact that (C) fails for W1 and W2 is a logical consequence of the essential features of W1 and W2?
In any case, here’s a counterexample for “GRNR” that does not have the same difficulty.
On logically possible world W12, consequentialism is true. On logically possible world W13, deontology is true. Let’s say that W12 and W13 share an object, O, and has the same nonmoral properties in both.
It seems to me that “GRNR” is not committed to the claim that there is no difference between the moral properties of O in W12 and W13. Yet, it does not appear to be a consequence of the definitions of W12 and W13 that O has different moral properties on those worlds. O might have different or the same moral properties, as far as I can tell, even granting “GRNR”.
Suppose consequentialism is true in the actual world (Wa). Now consider the logically possible world, W13 in which deontology is true. Suppose further that consequentialism and deontology, in at least some cases, do not agree in their implications concerning the moral properties of objects (including events). Suppose further that there is some object O that exists in both worlds and that consequentialism entails that O is good/right, deontology entails that S is bad/wrong. (C) fails for Wa and W13.
Why does (C) fail for Wa and W13? It seems to me that (C)’s failure is a logical consequence of two facts: (1) That deontology is true in W13 and consequentialism is true in Wa, and (2) That deontology and consequentialism do not agree in their implications concerning the moral properties of objects. The only way to ensure the failure of (C) for Wa and W13 is to insist on both (1) and (2). If (2) is false, then we won’t get the failure of (C) or any two worlds. And so, it looks to me that (C) fails as a logical consequence of the identifying features of W13.
You also ask: “would you say that a version of TS that holds that it’s a conceptual matter that moral value is resemblance of God, avoids your objection in the same way because on such theory, the fact that (C) fails for W1 and W2 is a logical consequence of the essential features of W1 and W2?”
Yes. I would agree with that. However, the claim that it is a conceptual matter that moral value is resemblance to God is false. I know that you are skeptical of the open question argument, but it seems to me that a version of that argument would show this. In any event, there are other ways to show that this conceptual claim is false.
Jason,
But those are not identifying features of the worlds. The definitions of the worlds do not provide enough information to ascertain whether the object will have different properties.
More generally, if T is any moral theory that holds that there is analytical reduction of moral statements to nonmoral ones, then according to T, there are logically possible worlds W and W’ such that W and W’ share some object O, and yet the moral properties of O are different in W and W’.
Now, suppose that Wa is the actual world, and W14 is a world that shares Donald Trump with the actual world. Then, T does not say whether (C) holds for Wa and W14. It might or might not, as far as T says. Here, T might be a version of DCT that holds that there is no analytical reduction, or it might be “GRNR”.
As an alternative, one can mirror your argument as follows:
We have W1 and W2 as you defined them:
W1: God exists, and object O exists. God approves of O (or occupies the relevant state, whatever it happens to be, with respect to O).
W2: God does not exist but object O exists. O has the same intrinsic (non-moral) properties in W2 that it has in W1. Since God does not exist, God does not occupy any state with respect to O.
Why does (C) fail for W1 and W2, assuming the part of TS that holds that moral properties require God, or are the same as some relation to God, etc.?
It seems to me that (C)’s failure is a logical consequence of the following facts:
(1) That God occupies the relevant state – say, S- with respect to O on W1.
(2) That God does not exist in W2, so it does not occupy state S.
(3) That in W1 and in W2, O has moral property M(S) if and only if God occupies the state S.
Those are properties of the definitions of the worlds as well. My point is that TS does not hold that moral properties require God’s relevant state in all logically possible worlds, but rather, in metaphysically possible worlds. In order to conclude that (C) fails for W1 and W2 on TS, you need to include in the definition of the worlds that in W1 and in W2, O has moral property M(S) if and only if God occupies the state S.
It does not follow from TS (side note: a version of TS may well also hold that God exists, as part of the theory, so in particular, that would entail that TS does not hold on W2).
I do agree that it’s not a conceptual truth that moral goodness is resemblance to God; I was asking whether you thought your argument worked against such theories, or against theories that are silent as to whether there is analytical reduction (I’ve seen DCT theories that are silent).
Jason, you write:
Ok so I guess my question is this when you use the word “cannot” here, are you referring to logical possibilities or something narrower like metaphysical possibilities?
It seems for your argument to work, you need logical possibility you said:
But, the fact that it’s a logical possibility that God exists and other objects do, doesn’t contradict the supervenience thesis unless the supervenience is claiming its logically impossible for intrinsic properties to exist without their non-moral properties.
But here is the problem as I see it. It seems to me that the reason you give for arguing that the world in which God does not exist is logically possible, undercuts (S) taken as a claim about logical possibilities.
Take two claims:
(i) God does not exist but other objects do
(ii) moral properties do not exist but other properties do
It seems to me that (ii) no more expresses a logical contradiction than (i) does. I fail to see why replacing the word “God” with the word “moral” changes the logical structure of that statement. On the face of it, neither of them expresses an explicit contradiction.
So I don’t know how you can maintain that there is a logically possible world in which God exists and other objects don’t without also denying (S) taken as a truth about logical possibility