This is the last installment of the Prosblogion Reading Group. I’ve found reading these posts and comments edifying, and I hope the rest of the readers have as well. I’d like to thank Matthew for setting this up, and for the other participants–both posters and commenters–for their great thoughts.
Below I discuss Tooley’s response to Plantinga’s response to Tooley. Or, put another way, Tooley’s “Yes way!” to Plantinga’s “No way!” To keep my comments at a manageable length I’ve referred back to Trent and Andrew’s posts, rather than presenting the whole dialectic here. But I’ve tried to summarize the dialectic briefly in most places. For more detail on the original argument or Plantinga’s response, be sure to see the discussions of the last two weeks.
1 Plantinga’s Responses to My [that is, Tooley’s] Two Arguments
1.1 Atheism as the default position
We’ve already seen Tooley’s argument that atheism is the default position and Plantinga’s criticism of that argument in last week’s post by Andrew. Very briefly, Tooley argued (to paraphrase Plantinga, pg 165; quoted by Tooley on pg 233):
Step 1: The intrinsic probability of the proposition that there is an omniscient, omnipotent, wholly good being (call that being OOG) is as great as the intrinsic probability of that there is an omniscient, omnipotent, wholly evil being (call that thing OOE), and both of these is as intrinsically probable as the proposition that there is an omniscient, omnipotent, morally indifferent being (OOI) [N.B., they also use the term ‘a priori’ probability to mean the same thing as ‘intrinsic’ probability, see (236)]. So the intrinsic probability for each of these is no more than one third, and hence, the intrinsic probability that the God of theism (OOG) exists is, at most, one third.
Step 2: Given that there is no further evidence for belief in God, and that the intrinsic probability that God exists is one third, atheism should be the default position.
Plantinga questioned both Step 1 and Step 2. Plantinga questioned Step 2 by claiming that, even if Tooley is right that the intrinsic probability of God’s existence is at most one third [or, more strictly, the intrinsic probability of the truth of a proposition that OOG exists is at most one third], why shouldn’t agnosticism be the default position? We see this spelled out in Andrew’s post last week.
Tooley’s response to Plantinga here is to claim that beliefs come in degrees, and that they have to get to a certain degree for us to count as believing them [“once it is recognized that belief admits of degrees, the question arises as to what degree of belief is needed before one can be said to believe something” (234)]. The language here isn’t the best; you need to believe something with a high enough degree of subjective probability for you to count as believing it. The only non-arbitrary degree he thinks we can point to as the high enough degree of subjective probability for belief is one half.
As for agnosticism, Tooley claims “it seems best to use the term ‘agnostic’ to cover cases where one thinks that the existence of God and the non-existence of God are equally likely, or where one has no subjective probability at all concerning the relevant proposition” (234). So agnosticism shouldn’t be the default position, since the probability that OOG exists is, at most, one third, whereas the probability that OOG doesn’t exist is at least twice the probability that OOG does exist, and hence the existence and nonexistence are not equally likely.
I think Tooley’s understanding of ‘agnostic’ is too narrow. Suppose I know that all the humans in the world have been assigned numbers, the men getting odd numbers and the women getting even numbers. A machine just randomly picked a number assigned to a human. I’m asked, do you believe that the number chosen was even? I know that there are slightly more women in the world than men. So I know the probability that an even number was chosen is slightly higher than the probability that an odd number was chosen. So I don’t think that an odd number was chosen is just as likely as that an even number was chosen. So it turns out I’m not agnostic about whether an even number was chosen. And this even if I certainly think both that I’m agnostic about it and that I don’t believe that an even number was chosen.
Plantinga objected to the Step 1 by saying that he thought the only reason one would affirm that the a priori probability of OOE is as great as that of OOG is that “one can’t see a difference in their probabilities” (169; quoted by Tooley, 236). Tooley responds that he has a reason for thinking that the a priori probability of OOE is as great as that of OOG. Tooley asks us to assume a sparse theory of properties, like Armstrong’s (1978). Properties are identified with genuine universals and there are no negative or disjunctive properties. Now consider these two principles:
Principle 1: State descriptions and permutations of individuals. Any two state descriptions that differ only be a permutation of individuals are equally likely.
Principle 2: State descriptions and families of properties. Any two state descriptions that differ only by a permutation of properties belonging to a family of properties are equally likely.
A ‘state description’ is “a certain conjunction of atomic propositions and their negations” (239). and a ‘family of properties’ is “a maximal set of mutually incompatible, non-conjunctive properties” (237).
Now Tooley argues as follows for the claim that the existence of OOE is at least as likely as the existence of OOG. The property of always choosing to do what is right, call it P, is a genuine property, and hence a universal. The property of always choosing to do what is wrong, call it Q, is also a genuine property, and hence a universal. Form the family of properties that contains these two properties, P and Q. Consider a state description that involves an OOG person who has property P. Replace the P with a Q, hence forming a state description with an OOE. By the second principle, these two state descriptions must be equally likely. So it is equally likely that OOG exists as it is that OOE exists. And a similar argument can be run for OOI. So here we have a reason for believing that OOG and OOE are equally likely.
Some comments on this argument. First, I’m not sure that sparse universal theorists want properties like P and Q as their universals. As I understand it, fans of sparse universals don’t want any more universals than those that a complete science would require.
Second, note that the family of properties that contains P and Q either contains just one more property, that which something has when it is morally indifferent, call it R, or many more properties, one for each grade of moral property a thing can have. It seems that it should contain more than just R, since there are many more salient divisions concerning morality to be made than just ‘always does good’, ‘always does evil’, and ‘is morally indifferent.’ In fact, it seems that almost all (if not all) humans lack all three of these moral properties. And we know we have moral properties (at least we know that, if there are moral properties, then we are the sorts of things that have them). So there should be more than just these three, P, Q, R.
The knowledge that there are more properties in this family than P, Q, and R should help Tooley here. In particular, his argument against theism should get stronger, since there aren’t just three possible OO beings that are equally likely (e.g., OOG, OOE, OOI), there are as many as there are incompatible moral properties in the family with P and Q. Surely there are at least two moral properties had by humans. So the family of moral properties contains at least 5 properties (but surely more, given the vast number of moral properties we see in this world). That means that the a priori probability of OOG is no greater than one fifth. And, of course, the more moral properties there are, the worse it gets for theism. So, all that to say, if Tooley can have moral properties as sparse universals (and I don’t see why he can’t, even if it isn’t typical, so far as I know), and can have Principle 2, it seems like he has an argument for why the probability of OOG is no greater than the probability of OOE. It also seems like he has a better argument for atheism being the default position.
I don’t think Tooley can have Principle 2. here’s a potential counterexample. Consider a property that sparse universalists are happy to accept: having a mass of 100 kilograms (call this property U). Consider another property which is incompatible with it: having a mass of 110 kilograms (call this property V). Now form the family of properties which includes these two properties. The family will include all incompatible mass properties. Now consider a state description that includes a human, Bob, with property U. Replace U with V. By Principle 2, the state description with Bob having V is just as a priori probable as the state description with Bob having U. So far so good. But now consider all the other members of the mass family. The state description where Bob has property W, the property of having a mass of 100,000 kilograms, is, by Principle 2, just as a priori probable as the state description where Bob has a mass of 100 kilograms. But that’s not true! It is impossible that a human have that much mass. Likewise, consider the mass of an electron. It is just as a priori probable, given Principle 2, that a human have the mass of 100 kilograms as that he have the mass of an electron. But again, a human can’t have that little mass. So Principle 2 is false.
1.2 The argument from evil
In this section Tooley responds to Plantinga’s critiques of his premises 15 and 16.
(15) No rightmaking properties that we know of are such that we are justified in believing both that an action of choosing not to prevent the Lisbon earthquake would have had those rightmaking properties, and that those properties are sufficiently serious to counterbalance the relevant wrongmaking properties.
As Andrew mentioned last week, Plantinga claims that 15 assumes that belief in God is not justified. But Tooley (and Andrew last week) point out that this isn’t quite right. Belief in God could be completely justified and it still be true that we don’t know of a candidate rightmaking property for the Lisbon earthquake.
(16) For any action whatever, the logical probability that the total wrong-making properties of the action outweigh the total rightmaking properties–including ones of which we have no knowledge–given that the action has a wrongmaking property that we know of, and there are no rightmaking properties that are known to be counterbalancing, is greater than one half.
Plantinga’s argument, as Andrew said, is pretty much the same as Trent’s from two weeks ago: the probabilities in question are inscrutable. So, just as Andrew did, I’ll refer you to Trent’s discussion. Tooley, in his response, offers arguments on behalf of two claims he made in his main selection, claims which he used to support 16:
(1) Judged from a purely a priori view, the mere existence of wrongmaking properties is no less likely than the existence of rightmaking properties.
(2) Judged from a purely a priori point of view, the likelihood that there exists a rightmaking property with a moral weight whose absolute value is equal to M is no greater than the likelihood that there exists a wrongmaking property whose absolute value is equal to M.
Tooley offers arguments on behalf of these two claims, which again depend on his Principle 2, which I’ve argued against above. His argument is that the property, being a rightmaking property, is part of the family which includes these other two properties: being a wrongmaking property and being a morally neutral property. So the state description which includes that a is a rightmaking property will be no more or less a priori likely (by Principle 2) that a state description that includes that a is a wrongmaking property. Thus, (1) is true. And we can give a similar argument for (2). From this, Tooley thinks we have good reasons to affirm (1) and (2), which in turn are reasons for affirming (16).
I don’t have anything to say against the above argument other than to say that I think that Principle 2 is false. Hence the argument, in my estimation, isn’t sound.
One worry I have about claiming Principle 2 to be false, though, is the following. Tooley says of his Principles 1 and 2: “There are certain general principles that serve to capture what is correct in the classical principle of indifference. Those principles, moreover, are needed for inductive logic: if one rejects them, inductive skepticism appears inescapable” (240). Is he right here? Am I stepping into inductive skepticism by rejecting Principle 2? And why think that 1 and 2 are necessary conditions for inductive logic anyway?
2.1 Internalist versus externalist accounts of justification
In this section Tooley considers and rejects the claim which he paraphrases from Plantinga: “a proposition’s “seeming right,” or “an inclination to believe” a proposition, constitutes non-propositional evidence for the proposition, and renders acceptance of the proposition non-inferentially justified in the absence of defeaters” (241).
Instead, Tooley accepts:
“one is non-inferentially justified in believing that p if and only if one is directly acquainted with some state of affairs T that is a truthmaker for p” 241-242)
Some problems with this. If Tooley is following Armstrong here, he will run into some small problems. For, according to Armstrong, there isn’t a state of affairs that makes it true that Armstrong exists. Rather, Armstrong himself, the very man, makes it true. And likewise for all other affirmations of existence. (Sure, some affirmations of existence will require states of affairs, but those will only be the propositions which affirm the existence of some state of affairs.) And this means that one can never be non-inferentially justified in believing any proposition which represents the existence of something (besides, again, states of affairs). Also, Armstrong claims that necessary truths do not require states of affairs, but rather objects, to make them true (Armstrong 2004, pg 98). These small problems are eliminated if Tooley drops “state of affairs” from his definition of non-inferential justification.
2.2 Is there a reliable belief-forming faculty in the case of religious beliefs?
Tooley gives two arguments that there isn’t. First, Tooley argues that in cases where there are reliable, general belief-forming mechanisms, such as perception, memory, and deductive reasoning, we find massive intersubjective agreement. In addition, the intersubjective agreement reached in such cases isn’t dependent on one’s societal background or indoctrination and is universal across human societies. However, we don’t find such agreement concerning religious matters. And there is a very strong correlation between societal background and indoctrination, on the one hand, and religious beliefs, on the other. So there is no reliable belief-forming faculty in the case of religious beliefs.
Second, one might want to limit the belief-forming mechanism that gives rise to reliable beliefs about the nature and existence of God. But even here there are reasons to deny such a limited mechanism. He gives 5 reasons, such as the long duration of human history and vast number of people (even today) who are not monotheists. Or how education or exposure to philosophical thinking correlate with a decline of belief in God. These reasons are meant to show that even a limited mechanism isn’t at work in humanity at large, since if there were one.
But what if, as Andrew Moon suggested in his first comment on Andrew Cullison’s post, the reliable mechanism needs to be pumped with proper action for it to work right. We know the mechanisms for sight work like this, and surely we’ve learned from teaching logic that the mechanisms for logical thought do as well. Oliver Sacks reports of a case study concerning a 60 year-old woman who never had the proper stimulus to learn to use her hands. She had no reliable mechanism for telling what a thing is by feeling it (she couldn’t tell that a hand was a hand, for instance). But after some pumping, she gained a reliable ability to use her hands as normal adults do. Perhaps the SD needs to be properly pumped to get working. And perhaps that proper pumping, or lack thereof, accounts for the discrepancies across groups. But then one wonders, isn’t the SD working in Catholics and Calvinists? If so, is it going haywire in one of these two groups? Or does the SD really say very little? Perhaps it is just the mechanism that gets us the belief that OOG exists, and from there we are on our own?
But do either of Tooley’s arguments show that there isn’t a mechanism that, once properly pumped, produces reliable beliefs about the existence of God?
3 The Argument from Evil Versus Justifications for Believing in the Existence of God
3.1 Non-inferentially justified belief in God?
Here Tooley rehearses what he has done so far.
3.2 Inferentially justified belief in God
Tooley discusses arguments for the existence of God in this section, quickly giving short paragraphs on the arguments from religious experience and miracles, then discussing the ontological argument in a bit more detail.
Tooley appears to claim that the proposition, possibly, p is true if and only if there is no “sequence of propositions that leads from p to some formal contradiction, where each step is related to one or more earlier steps either by formally valid rules of inference, or by substitution in accordance with some definition, or via an incompatibility of universals” (247). Now consider this in relation to the ontological argument. If the ontological argument is sound, then the proposition necessarily, there is an OOG is true. If that proposition is true, then the proposition, possibly, there is no OOG is impossible. If possibly, there is no OOG is impossible, then there must be some derivation of the kind he mentions above showing a formal contradiction from that there is no OOG. But no one has ever produced such a derivation. So we are not justified in believing that there is a derivation of the kind he mentions. And from here we reason backwards to the claim that we aren’t justified in believing that necessarily, there is an OOG is true.
I have a suspicion that there might be a tu quoque here somewhere from the fact that there is no derivation from that there is an OOG to a formal contradiction. If there is such a derivation, why did Tooley waste time with these probabilistic arguments when there’s a valid argument that shows a contradiction from that there is an OOG out there? Anyway, I leave the formation of the tu quoque, of the arguments for why there is no tu quoque, to the comments.
What I would especially like to see discussed is Principle 2. Why do I need it for inductive logic, and why must I be an inductive skeptic if I reject it (which I do). And if I do need it, what’s wrong with the counterexample I give against it?