Disclaimer: This is way outside of my wheelhouse. So the chances that this is just misguided are great. But it’s my month to post and I’m answering the editor’s (Half-Assed) Plea for Half-Assedness.
Many moral skeptics appeal to the following metaphysical thesis in supporting their view:
- Metaphysical Naturalism (NAT) – There are only natural properties.
And ethical non-naturalists are committed to the following metaphysical thesis:
- Metaphysical Plurality Thesis (PLU) – In addition to natural properties, there are non-natural properties.
Now, many think that NAT is more plausible than PLU on the grounds of parsimony. I think that this is a mistake.
Too see why, I’ll need to introduce some terminology. Let’s say that a theory, thesis, or view, T, has a positive ontological commitment to Fs if and only if T entails that there are Fs. And let’s say that a theory, thesis, or view, T, has a negative ontological commitment to no Fs if and only if T entails that there are no Fs.
Now, I take it that all those who accept the principle of parsimony hold that positive ontological commitments are, other things being equal, bad—that is, they make a theory less plausible, other things being equal. But there are three possible views concerning negative ontological commitments: (1) they too are bad/negative, (2) they’re neutral, and (3) they’re good/positive. Correspondingly, we have three versions of the principle of parsimony (POP):
- (POP-neg) Other things being equal, if T1 has fewer (positive or negative) ontological commitments than T2, then T1 is more plausible than T2.
- (POP-neu) Other things being equal, if T1 has fewer positive ontological commitments than T2, then T1 is more plausible than T2. But T1’s having fewer negative ontological commitments than T2 makes T1 neither more nor less plausible than T2.
- (POP-pos) Other things being equal, if T1 has either fewer positive ontological commitments or more negative ontological commitments than T2, then T1 is more plausible than T2.
Those who think that NAT is more plausible than PLU must eschew POP-neg. After all, NAT and PLU are on a par in terms of POP-neg: they both have the same number of ontological commitments: two. NAT has a positive ontological commitment to natural properties and a negative ontological commitment to the non-existence of non-natural properties, whereas PLU has a positive ontological commitment to natural properties and a positive ontological commitment to non-natural properties.
But it seems to me that POP-neg is the best of the three. That’s because a theory that has fewer commitments is less open to being falsified than a theory with more commitments. And a negative ontological commitment is just as much a commitment as a positive ontological commitment.
Now, perhaps, some would appeal to the history of science in defense of POP-pos over POP-neg. But I don’t think that the appeal works. What the history of science has shown is that when it comes to explaining events in space and time, those who have done so by expanding their positive ontological commitments have, other things being equal, fared poorly. But for the history of science to support POP-pos, there would have to have been cases where theories that expanded their negative ontological commitments fared better when other things were equal. But there are no such cases—at least, not that I know of. Of course, there were cases were other things were not equal. For instance, a theory that denied the existence of unicorns would be better able to explain the absence of certain types of fossils in the fossil record. But what the supporters of POP-pos would need is some case in the history of science of the following sort: a scientific theory denied that there were, say, Platonic forms even though this didn’t help it to explain anything and didn’t have any other advantage. Subsequently, it was proven that there were no Platonic forms. And because of this, this theory fared better than those theories that were otherwise the same but didn’t have this negative ontological commitment. But there are no such cases. So the history of science doesn’t support POP-pos.
Lastly, POP-pos seems committed to the absence of evidence for non-natural properties being evidence for the view that there are no non-natural properties. But that’s just obviously fallacious, isn’t it?
So I don’t see how people can defend NAT by appeal to parsimony and I don’t see any other reason to favor NAT over PLU. So I wonder why people appeal to NAT in defending moral skepticism. And I wonder why NAT is so popular among philosophers. I can understand being agnostic about whether there are any non-natural properties, but I don’t see the logic in denying non-natural properties on the grounds of parsimony.
Hi Doug, interesting post. Doesn’t it seem to you that there’s an important asymmetry between positive and negative ontological commitments? If someone posits an invisible intangible causally inefficacious ectoplasmic dragon living in my garage (constantly suffering misery at the lack of a large pile of gold for it to sleep on), that posit seems in worse epistemic shape than its negation, even if there’s no possible empirical evidence that could distinguish (and hence falsify) either hypothesis.
That said, I share your general sense that parsimony-based arguments against metaethical non-naturalism are misguided, but on different grounds as explained here:
http://www.philosophyetc.net/2013/03/the-possibility-of-moral-realism.html
Hi Richard,
You write: If someone posits an invisible intangible causally inefficacious ectoplasmic dragon living in my garage (constantly suffering misery at the lack of a large pile of gold for it to sleep on), that posit seems in worse epistemic shape than its negation.
Why? I think that we should have no credence with respect to the existence of this “dragon”. We should not think that there is a 50% chance that it exists. Nor should we think that it’s probable or improbable that it exists. After all, we have no idea whether such a thing exists. And we have no idea what the chances are that such a thing exists, for we have absolutely no knowledge or evidence with respect to such things. I don’t see that you have any evidence for thinking that it doesn’t exist or that it’s improbable that it exists? Has the history of science shown that such theories that posit such causally inefficacious entities fare worse? And should our credences not be a function of our evidence? And if you admit that you have no evidence with respect to the existence or non-existence of this “dragon,” then why have a higher credence in the proposition that it doesn’t exist than in the proposition that it does exist?
So I think that a theory that posits that it exists is bad. But I don’t see why a theory that posits that it doesn’t exist isn’t likewise bad. Both are making postulations on the basis of no evidence whatsoever.
Perhaps, every imaginable causally inefficacious entity exists. Perhaps, none exist. I just don’t know, as I have no evidence with respect to such things.
That depends on the content, right Richard? The ontological commitment to there being no objects is much more implausible than the ontological commitment to there being some objects. Likewise with numbers, in my view, though there may be good reason to deny that there are numbers. (your mileage may vary.)
I find the talk about negative ontological commitments confusing mainly because of how many double negatives are needed.
I guess my way of thinking about this is to ask: how many different kinds of properties are needed to make NAT and PLU true? I’m thinking that fewer kinds of properties are needed to make NAT true than PLU true and that this is what parsimony here consists of. For NAT to be true, we only need natural properties in the world but for PLU both natural and non-natural properties, which is one kind more. It’s not like in order for NAT to be true we need both natural properties and further additional properties of not having non-natural properties (even if I do acknowledge that here we get to controversial issues in truth-making theory for negative truths).
Of course, the interesting question is: why prefer theories that require fewer kinds of properties to be true? I think that something can be said for the metainduction from history of science you mention. Additionally, it seems like a good idea to focus on the problems of the specific new kinds of properties PLU posits.
Hi Jussi,
Consider the following three theories:
(T1 = NAT) It make two claims: (a) there are natural properties and (b) there are no non-natural properties.
(T2) It make only one claim: there are natural properties.
(T3 = PLU) It makes two claims: (a) there are natural properties and (b) there are non-natural properties.
Now, you say, “For NAT to be true, we only need natural properties in the world.” That’s false. We need more than just that there are natural properties in the world for NAT to be true. We also need it to be the case that there are no non-natural properties in the world. What’s correct, then, is that, for T2 to be true, we only need there to be natural properties in the world. But T2 is, other things being equal, more plausible than NAT/T1 (or PLU/T3), for it is less open to being falsified. But in terms of risk of being falsified NAT and PLU seem to be on a par.
Jussi: You also ask: “why prefer theories that require fewer kinds of properties to be true?” As I see it, the reason is that such theories can make fewer claims. But if this is right, there is just as much reason to prefer theories that require that there be fewer kinds of properties that don’t exist to be true. Thus, the problem with NAT is it requires that non-natural properties don’t exist and the problem with PLU is that it requires that non-natural properties do exist.
hmh. What if inspired by Armstrong, I formulate the views in the following way:
NAT: There are no natural properties, there are no other properties.
PLU: There are natural properties, there are non-natural properties, there are no other properties.
This seems to commit PLU to more different kinds of properties given that PLU makes an additional claim to NAT. Also, given that NAT rules out all other kinds of properties with the that’s all clause it also rules out non-natural properties and hence there is no need to rule them out separately.
I should also say that there is a big debate about whether negative claims (there are no non-natural properties) need negative facts as truth-makers. If they don’t as many people have tried to argue, then this allows us to formulate parsimony in ontological commitments. We are not interested in how many ontological claims are true but rather in how many types of things and properties they require as truth-makers. It is an open possibility that even if NAT and PLU make equally many ontological claims (negative and positive) one of these views requires fewer types of truth-makers.
Hi Jussi,
Doesn’t PLU, so formulated, entail that there are no non-natural properties? I’m assuming that the statement “there are no other properties” says different things in NAT and PLU, because of the indexical ‘other’.
Now imagine the possible world in which there are both natural properties and non-natural properties. If that’s the actual world, then won’t PLU be true and NAT be false? Of course, another possibility is that the actual world is one in which there are natural properties and no non-natural properties. And if that’s actual, then PLU will be false and NAT will be true. But this means that PLU and NAT are on par. In each case, there’s a possible world in which one is true and the other is false. So absent there being some reason to think that one of these possible world is more likely than the other, we should have no more credence in NAT than in PLU.
Now, the view that’s superior to PLU is not NAT, but T2, for there is no possible world in which PLU is true and T2 is false. But T2 is not just superior to PLU. It’s also superior to NAT. For there is no possible world in which NAT is true and T2 is false.
Of course, you claim that what matters is not whether there are worlds in which one theory is true and the other is false, but how many properties there needs to be in order to make a theory true. But you haven’t given us any reason to think that this is what matters. Maybe that’s unfair. Maybe your reason is that you think that negative claims such as “there are no non-natural properties” can be true even if the world is not such that there are no non-natural properties. But I find that puzzling. Can you explain why I should accept that.
Hi Doug, I was going to agree with everything you said, but then God Himself (in the form of an invisible and usually causally inefficacious ectoplasmic dragon) told me that NAT is more parsimonious. I trust you’ll take my word for this, having no reason to doubt it!
Hi Simon,
I know that this is a joke. But, seriously, I do have a reason to doubt your assertion.
You assert that God Himself (in the form of an invisible and usually causally inefficacious ectoplasmic dragon) told you that NAT is more parsimonious. There are at least two possible explanations for your making this assertion:
E1: God Himself (in the form of an invisible and usually causally inefficacious ectoplasmic dragon) told you that NAT is more parsimonious. This results in your believing this claim. And you assert this because you believe it and seems relevant to our discussion.
E2: You don’t believe it, but assert it because it’s funny.
E2 doesn’t make any claim about whether God exists. So it is more parsimonious on POP-neg, which is the view that I endorse.
Hi Doug
I don’t quite follow you here. I formulated PLU as:
PLU: There are natural properties, there are non-natural properties, there are no other properties.
How could this entail “that there are no non-natural properties”? PLU says that there are.
About this passage:
“Now imagine the possible world in which there are both natural properties and non-natural properties. If that’s the actual world, then won’t PLU be true and NAT be false? Of course, another possibility is that the actual world is one in which there are natural properties and no non-natural properties. And if that’s actual, then PLU will be false and NAT will be true.”
To me, this sounds like a good formulation of what is at issue with parsimony. We compare the world in which PLU is true to the world in which NAT is true and we count how many kinds of properties there are in these worlds. The fewer properties are needed to make the view true in the world in which the theory is true, the less ontologically committed the view is. I don’t quite follow what you say after this though.
Whoops, I meant NAT. NAT entails that there are no non-natural properties. So NAT has a commitment that PLU doesn’t have.
About the passage: I’m assuming that if there is a type of property P in the actual world, this makes the claim that there are no Ps false and the claim that there are Ps true. You seem to suggest that there is some big debate about this. But I wanted you to give me some reason to reject this assumption.
In any case, you claim that if there must be a type of property P in the actual world for a theory to be true, then this counts against the theory, other things being equal. But you seem to deny that if there must not be a type of property P in the actual world for a theory to be true, then this doesn’t count against the theory, other things being equal. But why the asymmetry? If both the presence of a certain type of property and the absence of that type property in the actual world can falsify a theory, then we should treat the commitment to the absence of that type of property (e.g., NAT’s commitment to there being no non-natural properties) on a par with the commitment to the presence of that type of property (e.g., PLU’s commitment to there being non-natural properties). You seem to want to take into account one sort of truth/false-maker (the presence of a type of property) but not another (the absence of a type of property).
I agree that parsimony is somewhat difficult to interpret, and I agree that considerations of parsimony taken in isolation cannot get one to the conclusion that non-naturalism is false. But non-naturalism scores very poorly with respect to the other theoretical virtues discussed by Quine in the Web of Belief (as does, e.g., Richard’s dragon hypothesis.). I think that what naturalists should insist on, here, is that the non-naturalist’s hypothesis has very little overall explanatory value, so that it should be rejected, i.e., it is rational to disbelieve it, i.e., it is appropriate to say that it is false. The non-naturalist’s hypothesis is in part the claim that non-natural properties exist, and so this means that it is rational to believe they do not exist and appropriate to say “non-natural properties do not exist.” This form of reasoning is not deductively valid; it is a form of induction that is itself supported (as Jussi suggests) by metainduction from the history of science.
Additionally, I’m not sure that the attitude that you believe we ought to have towards Richard’s dragon hypothesis is possible or coherent – or at the very least, stable. But it’s not my area of expertise.
I worry that we are confusing two different questions:
1. How to measure ontological commitments and parsimony of theories?
2. Why is a more parsimonious and less ontologically committed theory better?
I’ve addressed my comments more on question 1. I’ve tried to describe a simple measure by which we can say that some theories are more parsimonious and ontologically committed than others. I still don’t see an objection to this proposal and I find the whole notion of negative ontological commitments confused (I worry that you don’t quite mean the same as other people in the ontological debates that continued on from Quine).
I haven’t said much about why we should prefer theories that are more parsimonious in the way I have explained. Here there might be general reasons to be sceptical about less parsimonious theories based on metainductions. This reasons are probably fairly weak. The other type of reasons are specific issues to do with the type of properties we are positing: how do we know about them, how does our language refer to them, and so on. I’m personally still on the fence on these issues in metaethics.
One way to put the question is: who has the burden of proof? If I believe in some additional thing, say, magic and you don’t, is it my task to show that magic doesn’t exist or is the burden on you to show that it does? I’m inclined to think that the burden is on those who are committed to the existence of additional kinds of properties and things rather than on those who deny their existence.
Doug,
This is also out of my wheelhouse, so take what I say here with a grain of salt. However, it seems to me that the ideas underlying your argument are still more favorable to moral skepticism than realist non-naturalism.
Theoretical simplicity can be in terms of parsimony – roughly, number of beings posited – or elegance – roughly, how many claims are needed to articulate the complete theory. Elegance is not how many entailments the theory has, but rather how many things the theory needs to explicitly say to get all of its entailments. It seems to me that what makes POP-neg attractive to you is that it favors elegant theories. You are saying that we shouldn’t just count the entities posited, but rather count the claims the theory makes either way about existence. My not-super-in-depth reading on this topic suggests that there’s a better case to made for elegance, rather than parsimony, as a theoretical virtue.
Anti-realist naturalism has a good claim to greater elegance than non-naturalist realism. Probably the best versions of both theories have to make roughly all the same claims about the non-moral world; both include all the claims of the best science. The realist has to make a lot of additional claims about how the moral world works – not just what exists, but how it operates. The anti-realist has to say only one thing about the moral world: there is no such thing. Prima facie, that’s more elegant.
You might respond by saying that you only care about how many claims are made about what exists. To determine elegance, I’m counting all metaphysical claims – what exists, and how it works. Maybe there’s something wrong with that. Or maybe you could argue that the anti-realist’s error theory is going to add a lot of complexity.
(BTW, I’m not endorsing moral skepticism here, just saying it is the more elegant view)
Hi Jason,
You may be right that the non-naturalist’s claim that there are non-natural properties has little explanatory value. I certainly think that their claim does nothing to help us to explain any events in space and time. Whether there are other things (e.g., moral facts) that their claim helps us to explain is a matter that I won’t discus here.
But let me ask you: What does the naturalist’s claim that there are no causally inefficacious non-natural properties help us explain? It seems even more clear to me that the naturalist’s claim that there are no causally inefficacious non-natural properties has no explanatory value. So I’m not seeing how the naturalist’s claim that there are no non-natural properties fares better than the non-naturalist’s claim that there are non-natural properties in terms of explanatory value. And what other theoretical virtues do you think that the naturalist’s claim that there are no non-natural properties has? I don’t see any.
Hi Jussi,
I don’t think that we’re confusing the two questions. So regarding question 1, I’ve given three possibilities: POP-neg, POP-neu, and POP-pos. Now, you think that my way of drawing the distinction between these three is confused. But that’s probably not that important. For it’s clear enough that I hold and you deny that, insofar as we think that parsimony is virtue, parsimony should be understood such that NAT is no more parsimonious than PLU. And it’s clear enough that you think that we should think of parsimony such that NAT is more parsimonious PLU, but in doing so you’ve left it relatively unclear why we should think that parsimony is a virtue. You suggest that we might have inductive evidence from the history of science that suggests that parsimony (as you conceive of it) is a virtue. But can you give me one case where a theory that denied that there are causally inefficacious non-natural properties fared better than one that didn’t? And if not, then I don’t see why we have any inductive evidence for parsimony as you conceive of it being a virtue.
Hi Brian,
Good. I’m open to the possibility that theoretical elegance is a virtue that favors moral skepticism over non-natural realism. After all, the moral skeptic needn’t deny that there are non-natural properties. The moral skeptic could just be silent on the issue. But many moral skeptics are not silent. Many assert NAT and even use NAT in their arguments for moral skepticism. They argue along the following lines:
P1) No natural property could be normative.
P2) Moral properties are, by conceptual necessity, normative.
P3) There are only natural properties — that is, the only properties that are instantiated are natural properties.
C) Therefore, nothing instantiates any moral property.
So if you have an argument for moral skepticism that isn’t based on NAT, then nothing I say here constitutes a challenge to your view. But if you do argue for skepticism on the basis of NAT and you argue for NAT on the grounds that it is putatively more parsimonious, then what I say is a challenge to you.
Brian,
Good point. I think it is probably even slightly stronger. Many stereo typical anti-realist views don`t even need to say there is no such thing. They merely have to remain agnostic about it. Which is more elegant than bull-headed “there ain`t no such stuff“ antirealism, if by `elegant` we mean “needs less claims to formulate the complete theory. So moral skepticism comes out better than nominalist realism comes out better than realism realism on the criterion of elegance.
Jack, Doug,
One way of reading the claim I made about elegance is that naturalists can be silent about non-natural properties. That’s not what I intended to say. I meant to talk about the naturalist who denies that there is moral stuff. Such a naturalist needs only make one claim about the non-natural moral realm – there’s nothing in it – whereas most moral realists have to make lots of claims.
Elegance is not determined by looking at the number of entailments of one’s theory, since probably all theories entail infinite things. It’s determined by looking at the number of statements needed to get these entailments (I think; again, I’m not anything like an expert). Anti-realism entails the denial of all moral claims. But it gets all of this by saying just one thing: there are no moral properties. Any realist moral theory is also going to get a lot of entailments – what so and so should do in such and a case. To get these, it requires some number of core commitments: what is good, what is bad, or what counts as a reason, how to weigh conflicts, etc. So a realist has to say more than an anti-realist, even if the anti-realist has to say at least one thing about morality. So anti-realism is more elegant. Doug, your post looks like it likes elegance, rather than parsimony, as a theoretical virtue and that’s why I thought your ideas might really favor this stronger anti-realism.
Now that I say this, it occurs to me that some non-naturalist realist theories might be almost as elegant as anti-realism (some sort of non-naturalist utilitarianism, perhaps).
No, sorry, I did take you to be saying that naturalists deny non-natural properties. The point I was making is that given your characterization in terms of statements needed to get the whole theory, the agnostic antirealist has a more elegant theory than both the realist who denies there are non-natural properties (the antirealist in your lingo) and the realist who accepts. They just lose out on (metaphysical?) completeness, but completeness is a very debatable scientific virtue anyways. Unless, I guess, you had some fixed idea of what a complete theory had to say. But the notion of the completeness of a theory is murky. Consider, for example, someone like Van Fraassen who takes the explanatory demand on our scientific theories to be empirical adequacy. By your lights, his theory, which makes no commitments to the existence or not of unobservables, is more elegant than a theory that takes a stand. By his lights, it`s more complete.
Anyways, interesting stuff and I like the original point. I just wanted to pick up the cudgels for (moral) agnostic antirealism as a more elegant alternative than what I think of as a paradigmatically realist view—the denial of all non-natural properties.
Whoops. For “more complete“ read “equally complete“.
Hi Brian,
If I understand you correctly, a theory like utilitarianism is more elegant than a theory like Rossian pluralism, because it can get all of its entailed moral verdicts by making only one statement (the statement of the principle of utility) whereas the Rossian pluralist has to make many statements to get all of its entailed moral verdicts. But now it seems to me that you’re talking about explanatory power. Utilitarianism can explain more with fewer principles than Rossian pluralism can. But I’m talking about parsimony. My concern is with how many different types of things or properties that one must assume is present or absent. And, in any case, I don’t think that the naturalist’s claim that there are no natural properties has any greater explanatory power than the pluralist’s claim that there are non-natural properties.
I should have wrote: “the naturalist’s claim that there are no NON-natural properties.”
Hi Doug,
You ask, “What does the naturalist’s claim that there are no causally inefficacious non-natural properties help us explain?”
I think this is going about things in the wrong way (and this relates to Jussi’s previous point about burden of proof). We don’t directly evaluate this non-existence claim for explanatory power. We get it for free, by ruling out the positive claim. In this context, absence of evidence is indeed evidence of absence.
Our explanatory target is “the moral phenomena.” This includes everyone’s moral beliefs, the most credible appearances of moral knowledge and moral progress, everything psychology has discovered about the formation and modification of moral beliefs, and probably a lot more. We then have various competing explanations of this phenomena. A bunch of these look pretty powerful and do not posit non-natural phenomena. These include: nihilism; constructivism of the Rawls-Korsgaardian-Streetian sort; naturalist realism of the Boyd-Sturgeon-Brink variety; certain forms of relativism; and expressivism. In many of these cases, the explanatory value of these theories is augmented by conjoining them with psychological explanations of moral belief formation and revision.
Of course, another explanation of the same phenomena is non-naturalist realism, which posits entities of a basically unintelligible sort (even if I were a fictionalist, I wouldn’t know what exactly to pretend or imagine) and equally mysterious ways of knowing. This non-naturalist explanation is no damn good, judged in terms of generality, simplicity, refutability, modesty, and precision. We rule it out. We’re entitled to disbelieve it and to say that it is false.
Perhaps you would question whether this Quinean methodology does not unfairly rule out entities, properties, and relations that are by hypothesis outside the causal order. Maybe the idea is that it gets us to (e.g.) atheism, nominalism, and naturalism about morality (broadly construed) too quickly. I would counter: this methodology has a proven track record, it encompasses the best and most reliable ways of thinking that we have. We’re always better off sticking with it than using any particular alternative. If it has these implications, so be it.
(Still holding the cudgels) Why bother denying there are non-natural properties on grounds of parsimony if we don`t need the denial to underwrite the target moral phenomena? It seems to me that there is no need arising from parsimony to go on and deny that there are such properties. We may be entitled to do so, but why take on additional commitments that aren`t strictly necessary to explain the moral phenomena?
Now, there are independent reasons to do so which lie outside of the domain of parsimony, elegance-in-the-Talbottian-sense, or the like. We may think that the picture of the world which contains, as a part, the denial of the spooky is a more satisfying overall picture of the world. Or, depending on views about explanation, the resulting view might be more explanatorily adequate. But these are different criteria which may inveigh against otherwise less parsimonious views. We`d be better off, I think, firmly separating out parsimony considerations from other good-making properties like explanatory adequacy. Which is what I take at least part of Doug`s point to be
(Jason, I don`t think this is in opposition to anything you say, but just to point out that we should section off different good-making properties of a theory. I take one way of agreeing with you is to think that an committed nominalist theory is preferable to an agnostic one, where both are equally good at explaining the target moral phenomena, on some other good-making grounds. Whether this is so is something I don`t have a committed view on. But is this right as an interpretation of you?)
Hi Jason,
Why is this claim
“We get [the claim that there are no non-natural properties] for free…. In this context, absence of evidence is indeed evidence of absence”
any more plausible than this claim
“We get the claim that there are non-natural properties for free. In this context, absence of evidence for the absence of non-natural properties is indeed evidence for the presence of non-natural properties”?
And what do you mean by “we get it for free”? Do you mean that (a) we are justified in believing it in the absence of any evidence for its truth? Or do you mean that (b) our failure to have any evidence for its being false is itself evidence for its being true — evidence that absent any countervailing evidence justifies our believing it? But why should I accept either of these? Claim (a) seems absurd. For what, then, justifies our believing it is true if there is no evidence for its truth? Non-evidence? Claim (b) seems to lead to my being justified in holding contradictory propositions. For I may fail to have any evidence for P and well as fail to have any evidence for ~P. And, in that case, claim (b) entails that I am both justified in believing P and justified in believing ~P.
And in what sense is the formation of a moral belief a moral phenomenon? It isn’t even normative. That you morally ought to form a certain belief or perform a certain act are moral phenomena, but the formation of a belief and the performance of an action are not moral phenomena. So even if psychology has discovered a lot about the formation and modification of moral beliefs, it has discovered nothing about any moral phenomena such as which beliefs we morally ought to form or modify.
Jack: I do think that is a correct interpretation of me.
Doug: I mean (a) that we are justified in believing it in the absence of any direct evidence of its truth. It is not possible in principle to have direct evidence for the truth of a non-existence claim. It is rationally appropriate to conclude that certain phenomena do not exist when positing their existence has very low explanatory value. This inductive principle is, I think, justified by its reliability.
I take your point that it’s going to be hard to specify the moral phenomena in a theoretically neutral, non-question-begging way.
Those interested in this issue should see Morton and Sampson’s paper “PARSIMONY AND THE ARGUMENT FROM
QUEERNESS” (http://philpapers.org/rec/MORPAT-23). I’m told that what they say is sympathetic to the general thrust of this post, although having just skimmed over the relevant section, it’s clear that what they say is much more sophisticated and much less half-assed than this post. Also, I’m told that Elliot Sober is just finishing a book entitled Occam’s Razors. So we should all be on the lookout for that.
Jack,
Right after I posted I realized that I was probably misreading your previous post, but I didn’t know the etiquette of posting to say that. I think the idea is very interesting but I am definitely not qualified to say much about it.
Doug,
Maybe I am talking about explanatory power. I didn’t mean to be, or if I did mean to be, it’s because I thought it’s what you were talking about. Can I ask a question to try to get what you are saying straight in my head?
It seemed to me initially that:
1) You endorse the view that there’s something good about theories that make fewer claims.
2) Anti-realist naturalist theories make fewer claims.
I still think (1) is roughly true but I’m not sure precisely what it means anymore. Maybe what you are saying is that it’s just good to make fewer claims about what does or does not exist, and that’s it. Anti-naturalist theories make fewer claims about other metaphysical stuff, but the same number of claims as the realist about what does and does not exist. I don’t think you meant that, though, based on “it seems to me that POP-neg is the best of the three. That’s because a theory that has fewer commitments is less open to being falsified than a theory with more commitments.”
My way of reading (1) was that the good making feature was fewer commitments stated explicitly in the theory. This gives us (2).
Here’s another reading of (1), though, that denies (2): the good-making feature is fewer entailments of the theory (or fewer logically atomic entailments, or something like that). In which case realism and anti-realism are on a par. This fits with what you say, because the more entailments, the more ways a theory can be false. Is this what you have in mind?
Hi Jason,
Well, that’s not quite (a), so let’s call it (c). I think that if we plausibly spell out (c) it will say: if I have no direct evidence for there being Fs and I would rightly expect to have some direct evidence of there being Fs if there were Fs, then this constitutes evidence that there are no Fs. Do you agree that this is how to plausibly spell it out?
Now, this is the sort of principle that I was appealing to in my post when I talked about unicorns. Given both that unicorns are supposed to be a kind of vertebrate and that we would expect a vertebrate to leave some fossilized remains, the fact that there are no fossil remains of unicorns is evidence that there were no unicorns (in, say, the period of the dinosaurs). And I think that you’re correct that this principle has been shown to be reliable. (Of course, if you take out the clause about how it must be that we would rightly expect to have some direct evidence of there being Fs if there were Fs, then I would deny that the principle has been shown to be reliable.)
Okay, if all that’s right, then if you’re going to appeal to claim (c), you need to explain what direct evidence you think that we would rightly expect to have if there were causally inefficacious non-natural properties and that, in fact, we don’t have. Obviously, you’re not going to cite our intuition that pain is bad and ought not to be pursued for its own sake as the sort of direct evidence that we would expect to have but don’t have. For we have this evidence (if it is evidence). Instead, you’ll need to appeal to something like the idea that we would expect to have direct evidence of a causal connection between our belief that pain is bad and the causally inefficacious non-natural property of badness, which pain putatively has. But, of course, that won’t work. Because we can’t rightly expect a causally inefficacious property to leave direct evidence of having caused some effect.
So I’m not seeing how an appeal to claim (c) helps you.
Hi Brian,
If I were more careful, I would have said that a theory is more parsimonious (in whatever sense that parsimony is indeed a theoretical virtue) if it has fewer ontological commitments, where an ontological commitment is a commitment to the existence or non-existence of some TYPE of entity, property, or process. So it’s not about claims. It’s about types of things or properties and whether they exist or are instantiated.
And, with regard to (2), if you hold that naturalists deny that there are non-natural properties (and you’ve said that you do), then I don’t think that naturalists have fewer ontological commitments than pluralists do, quite the contrary.
So I would replace your (1) and (2) with:
1*) I endorse the view that there’s something good about theories that have fewer ontological commitments.
2*) Naturalists (those who hold that there are only natural properties) have just as many ontological commitments as pluralists (those who hold that in addition to natural properties, there are non-natural properties).
Hi Doug,
Can I ask where you got your version of parsimony? I thought a theory is more parsimonious when it has fewer kinds of things in its ontology. The SEP seems to agree:
Of course, you can use the term as you like. I take it you are claiming a virtue for your Pop-neg that the more classic form of parsimony lacks. (But it’s not clear to me what virtue that is.)
Hi Jamie,
I just came up with it myself. I was puzzled why some thought that parsimony was a theoretical virtue, and POP-neg was the only interpretation of parsimony that seemed to be a genuine theoretical virtue. I don’t make any claim about whether POP-neg is what other philosophers have meant by parsimony. I think that historically philosophers have meant POP-neu or POP-pos. But I suspect that, as Sampson and Morton point out (having just read their paper), this is because they thought that the world was created by God and that God would make his creation simple. But if we don’t believe in God or don’t think that he has some preference for simplicity, I see no reason for thinking that nature is simple rather than messy, complex, and well-populated with various types of entities and properties.
The virtue of POP-neg is, I thought, obvious: POP-neg avoids the conjunction fallacy. The probability of a conjunction is always less than or equal to the probability of either one of the conjuncts. So if P = natural properties and Q = non-natural properties, then the probability of either (P & Q) or (P & ~Q) is going to be less than or equal to the probability of P. And given that it is rather unclear whether Q or ~Q, I would say that the probabilities of these conjunctions are going to be less than the probability of P. This is the motivation for adopting POP-neg. By contrast, POP-pos holds that the probability of (P & ~Q) is greater than the probability of P. And POP-neu holds that the probability of P is no greater than that of (P & ~Q).
Hi Doug,
You suggest (c): “[I]f I have no direct evidence for there being Fs and I would rightly expect to have some direct evidence of there being Fs if there were Fs, then this constitutes evidence that there are no Fs.”
Several things worry me about this. First, I think we should focus on the positive hypothesis that is favored by the balance of available evidence, and disbelieve things that conflict with it. Relatedly, I want to avoid all mention of evidence of non-existence; maybe there isn’t any such thing. Third, the qualification you introduce (“and I would rightly expect to have some direct evidence of there being Fs if there were Fs”) serves to make the principle more gerrymandered, when meta-induction from science (probably) supports the less-gerrymandered claim, just as it (probably) supports the theoretical preference for simplicity, even in the absence of God. Reasoning in these ways works, even though it is emphatically not deductive reasoning. The alternative, I suspect, is ultimately something like Ancient Skepticism.
So I would say: if the hypothesis that there are Fs has low overall explanatory value, then it is rational to believe that Fs do not exist.
Two notes: First, we should probably think of explanatory value as essentially contrastive. Second, the forms of explanation relevant, here, are the forms of explanation common in the sciences (including the social sciences), as well as something like Schroeder’s constitutive explanation.
I should probably take a look at that Sober book, though …
Hi Doug,
That’s interesting. I mean, if by ‘more plausible’ you mean more probable, I guess the fewer claims a theory makes the more plausible it is. A metaethical theory that says there may or may not be any moral properties is extremely plausible. But it lacks some virtues that we might like in a metaethical theory.
I think we should prefer metaethical theories that explain more of the phenomena. Naturalists (in my experience, anyway) think they can explain the relevant phenomena without appealing to non-natural properties, and indeed that this makes their explanations better. They can then adopt the doxastic attitude toward non-natural properties that most of us take toward the dragons in Richard Yetter Chappell’s garage. I would say it’s quibbling to agonize over exactly what attitude that is.
Hi Jason,
I worry that we’re no longer talking about parsimony. Nevertheless, your proposal is interesting. You write: “if the hypothesis that there are Fs has low overall explanatory value, then it is rational to believe that Fs do not exist.”
What is the overall explanatory value of a logical property such as the property of logically entailing some proposition? You might say that it is low, because we don’t need it to explain any event in space and time. Or you might say that it is high because it helps us explain various logical facts. If the former, the principle is implausible. If the latter, it’s not going to be helpful to us in determining whether there are moral properties.
Hi Jamie,
Note that NAT and PLU are not metaethical views; they’re metaphysical views. And the view that claims that there are natural properties but makes no claim as to whether or not there are non-natural properties is not only more plausible than NAT or PLU, but is also just as virtuous as PLU and NAT if we’re assuming that everything else (such as explanatory power) is equal.
I agree that we should prefer theories that have greater explanatory power. But the question that concerns me here is whether NAT is, other things being equal, more plausible than PLU on the grounds that it is more parsimonious.
Hi Doug,
Interesting post. Here’s my two cents. With regard to your narrow question: I’m inclined to agree with you that NAT is NOT more plausible than PLU simply on the grounds that it is more parsimonious-in-your-sense…because as you say, it isn’t more parsimonious-in-your-sense. It is more parsimonious-in-Jamie’s-sense (which I think is the usual sense of ontological parsimony); I’m unsure whether this by itself makes it more plausible. If so, only marginally.
However, as a metaethical naturalist who has stressed parsimony, I think this is a straw man as a response to metaethical naturalism, for reasons similar to those advanced by others here. Metaethical naturalism as I understand does not necessarily or usually argue from, and is not committed to, the claim that there are no non-natural properties. (Maybe there are, I don’t know. I’m not committed to naturalism about mathematics or God by virtue of my metaethical naturalism). Rather, it argues that moral properties aren’t nonnatural (either because they’re natural, or because they don’t exist). The usual and best line of argument appeals to explanatory power: if (as I believe) we can explain all the data with natural properties alone (which we have independent reason to posit), then we have no reason to posit nonnatural properties. Perhaps in the absence of any reason to posit these properties, parsimony-in-Jamie’s-sense is determinative; at least, I think we should all believe that there are no invisible dragons.
Hi Steve,
Well, it’s pretty uninteresting to observe that NAT is or isn’t more parsimonious in some particular sense, which I take is your point. But I took myself to be arguing for something a bit more interesting: that NAT isn’t more parsimonious in any sense in which parsimony constitutes a virtue. So, if we are to understand parsimony as Jamie is suggesting such that a theory that postulates both that there are Fs and that there are no Gs is more parsimonious than a theory that postulates both that there are Fs and that there are Gs, then I fail to see why parsimony is a virtue. Unless, of course, we’re assuming that the universe is ontologically sparse, which I see no reason to assume absent the assumption that God exists and prefers sparsity. Do you have a reason for thinking that a theory that postulates both that there are Fs and that there are no Gs is better, other things being equal, than a theory that postulates both that there are Fs and that there are Gs. Since, other things are equal, we are to suppose that in terms of any non-parsimony consideration (including explanatory power) there is just as much or little evidence that are no Gs as there is that there are Gs.
Also, I wasn’t giving a response to metaethical naturalism (the view that moral properties are not non-natural) and so my response couldn’t be a staw man response to metaethical naturalism. I was responding to those who argue as follows:
P1) No natural property could be normative.
P2) Moral properties are, by conceptual necessity, normative.
P3) There are only natural properties — that is, the only properties that are instantiated are natural properties.
C) Therefore, nothing instantiates any moral property.
Note that P3 is just NAT. And some argue for NAT on the grounds that it is more parsimonious than either PLU or the theory that is silent on whether there are non-natural properties. And someone who makes this argument needs to give us some reason for thinking that NAT is more plausible than either PLU or the theory that is silent on whether are non-natural properties, for the argument works only if we accept NAT as opposed to either of these other two.
And if this is right, then it all comes down to whether we can explain all the data with just natural properties. But this is where I think that we end up with a bit of stalemate, because the ethical non-naturalist will argue that although we can explain all the natural data without appeal to non-natural properties, it doesn’t follow that we can explain all the data.
Doug,
Apologies for mistaking the breadth of your target! I’m largely in agreement with you that the skeptical argument you identify isn’t very compelling. (In fact, while I tentatively accept P3, I’m more skeptical about P1 and P2.)
Two quibbles:
There’s an important difference between the claim that the universe is ontologically sparse (I agree; what’s the evidence for that?) and the claim that there are no properties of a specific kind (Gs). Perhaps indefinitely many kinds of properties are possible. But it seems to me that we are justified in believing in the nonexistence of any one kind of possible property in the absence of any evidence in its favor. (Perhaps it will be replied that “nonnatural property” is a catch-all category. Then it becomes important what we mean by a “natural property”. On some accounts (e.g. causally efficacious) I guess I’d count as a nonnaturalist.)
The second quibble: I think the stalemate you mention can be broken in the metaethical naturalist’s favor. (That’s one way to describe the project in my book). I argue that all the data that can identified without begging the question against naturalism can be explained by naturalism.
We are justified in believing in the nonexistence of any one kind of possible property in the absence of any evidence in its favor.
On the basis of what are we so justified? Do you think that the universe is ontologically sparse in terms of the number of kinds of properties? If so, then, as you put it, “what the evidence for that?”
On the other point, I still need to read your book. So my bad.
I find the claim strongly intuitive, though I agree it’s difficult to explain why. I’m tempted to say: we know (perhaps from experience?) that many many more things can be imagined than are real. Plausibly, infinitely many kinds of properties are imaginable. So I don’t think this claim implies that “the universe is ontologically sparse in terms of the number of kinds of properties”. How about this: we know that the universe can’t contain infinitely many kinds of properties. So for any kind of property we can imagine, there is a negligible epistemic probability that it exists/is instantiated in the absence of any evidence of it.
Here’s an alternate argument. I’ll present it with some fairly naive assumptions, but I think it can be made more sophisticated. Suppose we’re rationally justified in believing any proposition for which our credence is above .7, in believing it false if our credence is below .3, and in suspending judgment if our credence is in-between. So if we’re justified in suspending judgment in the existence of invisible dragons, and similarly in the existence of invisible fairies, gremlins, and giants, then (assuming these are probabilistically independent) we would be justified in believing the disjunction: either invisible fairies, gremlins, or giants exist. But that’s crazy. (I think you indicated that you think rather that we rationally needn’t have any assignable credence in these propositions, so you won’t be swayed by this argument. But I’m not sure it’s possible to fail to have a credence of any kind in a proposition one contemplates. Still, I’m well outside my expertise here.)
Hi Steve,
How could we know from experience that many many more things can be imagined than are real? Perhaps, we know from experience that many many more things can be imagined than are present in our sphere of experience. Now, some suggest that we live in a multiverse with many (perhaps, infinite) universes. I have no idea whether that’s true. And I have no idea how many galaxies there are in our universe or whether there are galaxies that are hundreds of billions of light years away from of us or whether some (or many) them contain unicorns. And it would be strange to think that you could just intuit whether such contingencies are the case. Can you intuit how many stars there are or how many different kinds of stars there are? Do tell!
You say: “we know that the universe can’t contain infinitely many kinds of properties.” How do we know this? You must have special powers if you’re able to know how many galaxies or universes there are and what entities exists and what properties are instantiated in them.
And you’re right about what I think about your other argument: in the absence of any evidence whatsoever about whether Fs exist, I think that we would be unjustified in having any precise or imprecise credence with respect either to the proposition that Fs exist or to the proposition that Fs do not exist. After all, what would justify that precise or imprecise credence that you think is justified? Not evidence, for this is a case where there is no evidence. And in strange to think that something beside evidence for the truth of proposition could justify a credence in its truth.
Now, you’re not sure that it’s even possible to fail to have a credence of any kind in a proposition one contemplates. What’s the relevance? That just means it’s impossible for you not to have an unjustified credence. I think that’s fine. Someone may not be able to help but believe that the earth is flat even in the face of conclusive evidence for the earth’s not being flat. That just means that this person is doomed to have an unjustified belief. In any case, it certainly doesn’t mean that the person’s belief is justified just because they doomed to have. So the fact that you’re doomed to have a credence in a proposition that you contemplate doesn’t show that you’re justified in having a credence in that proposition.
Hi Doug,
I’ll concede here on the imagination/infinity argument. I was shooting from the hip, and I’m not sure what to say to back it up. To clarify, I was thinking about our experience that (e.g.) we can imagine the world to be many more ways than experience shows it to be. But I don’t know how to extend the point to (e.g.) kinds of properties. So I’ll let that slide.
I’m not satisfied with your response to the credence argument, however. First, are you really okay with saying that it might be impossible not to have an unjustified credence toward some propositions? That seems an unattractive commitment. Second, here’s an argument that there are justified imprecise credences in these cases. For any proposition for which you have no evidence either way, I trust you’ll agree that a credence of 0 or of 1 would be unjustified. Further, I trust you’ll agree that belief in such a proposition would be unjustified, so given the naive assumptions in my argument, there’s an upper bound of .7 on your credence. Furthermore, if you’re right that we’re unjustified in believing the proposition is false, then (on those assumptions) there’s a lower bound of .3. So aren’t you committed to thinking that the widest imprecise credence that would be justified ranges over the interval .3-.7? If so, then my argument gets off the ground: for a relatively small number of things for which we have no evidence whatsoever, we ought to believe the disjunction. I don’t think you’ll embrace that conclusion.
Hi Steve,
You ask, “Are you really okay with saying that it might be impossible not to have an unjustified credence toward some propositions?”
On the assumption that it’s impossible “to fail to have a credence of any kind in a proposition one contemplates,” absolutely.
Don’t you think the same thing? Suppose that P is a proposition “for which you have no evidence either way.” Now suppose that it’s not possible for you to have any credence other than 0 or 1 in P and also impossible for you to fail to have a credence of any kind in P. Isn’t this a case where it’s impossible for you to have a justified credence? For you hold that “a credence of 0 or of 1 would be unjustified.”
You also claim that the view that it’s impossible for such a person to avoid having an unjustified credence is “unattractive.”
How so? It doesn’t violate ought-implies-can or the like. Do you just think that if a person cannot help but have a certain credence in a proposition, then that makes that credence justified? But why would the person’s justification for a level of credence depend on what’s nomologically possible as opposed to her evidence?
Hi Doug,
When I said “I’m not sure it’s possible to fail to have a credence of any kind in a proposition one contemplates”, I didn’t mean that I thought there might be SOME PEOPLE for whom it isn’t (psychologically) possible. I was rather doubting it was possible for anybody. So I took your response to commit you to the view that there may be some propositions for which it is impossible for anybody to have a justified credal state. That’s what I’m suggesting is unattractive.
Now I expect your response will be to deny my impossibility claim. That’s where the rest of my argument comes in. On some commonsense (though certainly controversial) assumptions, if you consider whether p, and neither believe p nor believe that not-p, then ipso facto you have a credal state between belief and rejection (on my assumptions, between .3 and .7). Perhaps that’s as precise as it gets, but that’s still an imprecise credence. And I think this is all my argument needs to get the absurd conclusion.
You could, of course, reject the naive assumptions about the relation between belief and credence, etc. But that requires taking a stand on some issues in formal epistemology. Since I’m not a formal epistemologist, I may well just be embarrassing myself here (I hope not!), but I’m blundering on for the sake of half-assedness…
Hi Steve,
I’m not seeing why it matters whether it’s some people or all people. Imagine a world in which some people cannot help but have a credence of 1 in a proposition P for which they have no evidence. I think that you’ll agree that they are unjustified in their credence of 1. Now, imagine that everyone but those people are killed. It’s now the case that everyone cannot help but have a credence of 1. Do you now deny that those same people are unjustified in their credence of 1?
Now, I think that I also deny your impossibility claim. Let’s simplify things. You seem to arguing in the following way: if you consider whether p, and don’t have a credal state of 0 or 1, then ipso facto you have a credal state between 0 or 1. I don’t see how that follows. That is, I don’t see how it follows from the fact that you don’t have a credal state of either 0 or 1, then you must have some other credal state that is between 0 or 1. Why can’t you just have no credal state? After all, it doesn’t follow logically from your not having a credal state of 0 or 1 that you have some other credal state.
Doug,
Okay.
But what I said about metaethical views also seems to apply to metaphysical views, so I don’t see why this matters.
Here is a view, which we can call ‘(∨)’:
There are natural properties ∨ there are no natural properties.
By your criterion, (∨) is more plausible than any of the other views you consider. Indeed, it is a view than which none more plausible can be conceived. Further, if we assume everything else (e.g. explanatory power) is equal, it is just as virtuous as NAT and PLU.
Still, (∨) seems to be missing some virtue or other.
Hi Jamie,
My view is that, OTHER THINGS BEING EQUAL, a view is more plausible if it is more parsimonious in the sense of having fewer ontological commitments.
On my view, then, the view that there are natural properties but no non-natural properties (viz., NAT) is no more parsimonious (and, thus, is no better on the grounds of parsimony) than the view that there are both natural and non-natural properties (viz., PLU).
But I thought that you were suggesting that a better alternative view is the one that we find in SEP: OTHER THINGS BEING EQUAL, a view is more plausible if it is more parsimonious in the sense of postulating fewer kinds of things or properties.
On this the SEP view, the view that there are natural properties but no non-natural properties (viz., NAT) is more parsimonious (and, thus, better on the grounds of parsimony) than the view that there are both natural and non-natural properties (viz., PLU).
Were you advocating the SEP view over my view, as I thought? Or were you just pointing out something that I’ve admitted all along: that things may not be equal and that a view that is better for being more parsimonious may be worse for having less explanatory power? I interpreted you as saying the former, because I didn’t see the point in saying the latter. After all, I’m not arguing that theories that are more parsimonious are necessarily better, all things considered.
I suspect, though, that I’m probably still missing your point. If so, could you state it more plainly.
Hi Jamie,
One more point: I concede that naturalists think they can explain the relevant phenomena without appealing to non-natural properties. But isn’t that because they think that all the relevant phenomena are natural phenomena? After all, if you think that there are non-natural phenomena, then I don’t see why you would think that you could explain that without appealing to non-natural properties. So the naturalist must have some prior reason for thinking that there is no non-natural phenomena that needs explaining. And why do they think that? I don’t know? Because they think the view that there is no non-natural phenomena is more parsimonious? But it isn’t on my view of parsimony. So I assume that they have to appeal to the SEP view, which I don’t see the appeal of absence some view involving God and his desire to keep things simple for us.
Doug,
I’m not suggesting merely that all actual people might be unable to avoid having a credal state toward a proposition in the absence of evidence (upon wondering whether it is true), but that all possible people are unable to do so–because it’s constitutive of contemplating whether p that one have a credal state toward p. I grant you that the mere contingent fact that some actual person, or even all actual persons, are unable to avoid having some particular credal state would not suffice to make that credence justified. What I’m saying is unattractive is the conclusion that follows (from my constitutivity claim, and the claim that no credence would be justified) that every possible person who contemplated whether p, whenever there is no evidence for or against p, would respond in an unjustified way.
I know you don’t want to say that, so denying the impossibility/constitutivity claim is surely the better option. Here’s my argument for the constitutivity claim. If you consider whether p, and you’re not certain that p or that not-p, then you’re uncertain whether p. But uncertainty whether p just is a (possibly imprecise) credal state toward p, somewhere in the range between 0 and 1 (exclusive). Granted, it doesn’t follow from the fact that you don’t have credence 1 or credence 0 that you have some other credence, but only because you needn’t have any attitude toward p at all. But being uncertain about p is a (credal) attitude toward p.
One other point: your last response to Jamie touches on something I said before, about naturalism being able to explain all the phenomena that can be identified in a non-question-begging way. To claim that naturalism can explain all the data because all the data is naturalistic is, indeed, question-begging. But similarly, to claim that naturalism can’t explain all the data because there is non-naturalistic data is similarly question-begging. A non-question-begging argument for naturalism is that it explains all the phenomena (e.g. moral truths, moral motivation, moral supervenience, moral knowledge) that can be identified in a way that doesn’t start out by assuming that there is/is not anything nonnatural. That’s what I try to offer.
Hi Steve,
Thanks for clarifying. If understand you now, here’s your argument:
P1) If you contemplate P and you don’t have a credence of either 1 or 0 in P, then you’re uncertain whether P. P2) If you’re uncertain whether P, then you have a (possibly imprecise) credal state toward P, somewhere in the range between 0 and 1 (exclusive).
C) Therefore, if you contemplate P and you don’t have a credence of either 1 or 0 in P, then you have a credal state toward P.
You seem to take P2 to be a conceptual truth. That is, you hold that being uncertain whether p just is to adopt an attitude toward it — namely, that of being in a credal state somewhere in the range between 0 and 1 (exclusive). But if that’s right, I reject P1. I don’t see why lacking two credal states (two attitudes) entails having some other credal state (attitude). How is this supposed to follow? Do you think that to contemplate P is just to adopt an attitude toward P? But that doesn’t seem right to me. It seems that just as I can contemplate whether to A without intending to A, intending to not-A, or having some other intention toward A, I can contemplate whether to believe P without believing that P, believing that not-P, or having any other doxastic state toward P.
On the other issue, right: you’ve claimed that naturalism explains all the phenomena (e.g. moral truths, moral motivation, moral supervenience, moral knowledge) that can be identified in a way that doesn’t start out by assuming that there is/is not anything nonnatural. And I need to read your book. But I hope that you don’t expect me to just take your word for it that naturalism can do what you say it can do. It’s not that I think that you’re untrustworthy. It’s just that when it comes to a contentious philosophical claim, I need more than just that someone has made that claim before I’m willing to accept P. And, as things stands now (prior to reading your book), I’m very skeptical that naturalism can explain normative truths, what it’s like to be bat or to feel pain, the fact that, using concepts that are sufficiently comprehensive, I can conceive of experiencing some pain while disembodied, and many other things.
But I take your point that the naturalist needn’t appeal to parsimony in arguing that naturualism can explain all the relevant (non-question begging) data without appealing to non-natural properties.
Doug,
Since you just mean ‘gets higher credence’ by ‘is more plausible’, of course I agree that when A is a logical consequence of B and B is not a logical consequence of A, A is more plausible. Was this your only point?
The reason I keep stressing other virtues is that I think the point of parsimony is that it carries explanatory power with it. That is the main virtue in brings to a theory. A theory can be extremely weak logically, and therefore a consequence of many other theories that are not a consequence of it, but be very bad at explaining things. When theories get very weak logically, they do tend to be bad at explaining things.
Hi Doug,
I think we’re on the same page now. I grant that my argument turns on some controversial (though in my view intuitive) assumptions, so it’s perfectly reasonable for you to reject some of them, and I won’t push my case any further here–except to register that it seems very intuitive, to me, to understand uncertainty in terms of an intermediate credence. Perhaps I should grant that there are ways in which one may contemplate whether p without reaching either certainty or uncertainty–but I don’t see how this is possible in cases where we linger on the question. No need to reply, though.
(In case it wasn’t obvious, the rest of the argument was supposed to show that since believing the disjunction isn’t justified, the justified credence in the proposition with no evidence either way (e.g. there is an invisible dragon in the room) must be extremely low–thereby being/justifying belief that it is false.)
And yes, of course you shouldn’t believe that naturalism can be vindicated merely on my say-so; I was just registering my disagreement with your opinion, and pointing out that there are non-question-begging (I hope!) cases for naturalism that don’t depend on NAT. I expect lots of people will remain unconvinced by my book, too. My challenge to those people is to say what (non-question-begging) data it fails to explain.
Doug,
I don’t think I understand what you say at 10:01.
I don’t know what you have in mind here. The kinds of phenomena I was thinking of were things like: the moral supervenes on the natural; moral claims do not seem to be analyzable in naturalistic terms; moral judgment is practical. These don’t seem like the kinds of things it would be helpful to call ‘natural’ or ‘non-natural’. So it would help if you could say which phenomena you have in mind.
Hi Jamie,
And I’ll let you have the last word if you like. It’s time for me to move on to other things.
My only point was to argue that my view is more plausible than what I called the SEP view.
I think the point of parsimony is that it carries explanatory power with it. That is the main virtue in brings to a theory.
I don’t see it that way. A theory that claims both that there are Fs and that there are no Gs is less parsimonious than a theory that claims only that there are Fs. And this is true even if neither theory is able to explain any phenomena by appealing either to the presence of Fs or absence of Gs. A so the latter theory doesn’t seem to get any extra explanatory power just by being more parsimonious the former.
Regarding 10:01, I had in mind various events in space and time, such as certain neurons firing in the brain (those associated with the judgment that I’m required to X) and this causing the events that constitute my X-ing. Of course, we don’t need to appeal to non-natural properties to explain any events in space and time. But that doesn’t mean that there isn’t any phenomena (perhaps, normativity or what it’s like to be in pain) that we can’t be explained without appeal to non-natural facts.