In a recent article in this journal, Robert Oakes has argued for the intriguing position that theism defeats the ‘doctrine of public-world fallibilism’ [hereafter DPWF].Footnote 1 According to DPWF, it is impossible for there to be infallible knowledge of public-world propositions (such as ‘There is an olive on the floor’). So what Oakes argues is that theism supports an infallibilist epistemology. Given the recent interest in the problem of radical scepticism, this position is of great interest to the contemporary epistemological debate. In this paper, however, I will argue that Oakes has not established his conclusion that theism supports infallibilism.
The argument
In order to get clear on what exactly the argument Oakes presents us with amounts to, let's start with introducing some terminology. The obviously crucial concept in his argument is the concept of infallibilism. But here we immediately encounter a problem. Oakes hesitates between two different interpretations of infallibilism. On the one hand, he seems to accept David Lewis's interpretation of infallibilism which roughly runs as follows (194–195):
- INF
Subject s knows proposition p if and only if s's evidence eliminates all possibilities in which not-p is the case.
But on the other hand, he sometimes defines infallibilism as follows:
- INF2
‘[knowledge] can be had only when the relevant justification precludes the epistemic possibility of error.’ (194)
INF2 is far weaker than INF. The reason is that INF2 speaks of epistemic possibilities of error, whereas INF is usually interpreted as speaking of logical possibilities of error. Here are two problems with choosing to interpret infallibilism as INF2 rather than as INF. In the first place, since infallibilism in epistemology is usually interpreted as a thesis about logical possibilities of error, and not as a thesis about epistemic possibilities of error, then if Oakes argues against INF2, he is not, in effect, arguing against infallibilism as it is usually interpreted in epistemology. In the second place, if Oakes would argue against INF2 his argument is not very startling. For it seems obvious that INF2 does not lead to radical scepticism at all. If the only possibilities one has to eliminate in order to know a proposition are epistemic possibilities, then sceptical possibilities are clearly out from the start. For given what we know, one might argue, sceptical scenarios (like the brain-in-a-vat, or BIV, scenario) are not (epistemically) possible. What this shows is that, contra Oakes, we don't need to accept theism if we want to be infallibilists. Thus I submit that Oakes should interpret infallibilism as INF in order to present as strong an argument as possible.
Two more principles play a crucial role in Oakes's argument. He first subscribes to the principle of God's impeccable veracity:
- PGIV
God does not deceive us.
He then continues to propose that an important part of the PGIV is the ‘correctability requirement’:
- COR
God could not possibly permit us to have a false belief that we could not correct.
With these principles in hand, Oakes argues that ‘theism requires the falsity of the DPWF’ (194). That is, according to Oakes, theism will allow that there are cases in which we have infallible knowledge of public-world propositions. The argument Oakes presents might be parsed as follows.
Theism-infallibilism argument (TIA):
(1) PGIV
(2) COR
(3) If (1) and (2), then [for all false beliefs I hold, these false beliefs can in principle be corrected].
(4) Hence, for all false beliefs one holds, these false beliefs can in principle be corrected.
(4) is equivalent to:
(5) One cannot have false beliefs that cannot be corrected.
And (5) leads to:
(6) If one cannot have false beliefs that cannot be corrected, then one can have infallible knowledge of public-world propositions.
(7) Hence, one can have infallible knowledge of public-world propositions.
Comment on the argument: If I were a brain-in-a-vat (or BIV) that was ‘fed’ the perception of an olive on the floor, my belief that there is an olive on the floor would be false, but, crucially, I would have no means to correct this false belief. The reason for this is that there is no perceptual or phenomenological difference between perceiving an olive on the floor and having a BIV-perception of an olive on the floor. Now it seems that what Oakes's argument says is that we can eliminate the BIV-possibility (as well as all other sceptical possibilities) on the basis of theism and PGIV/COR: we cannot have false beliefs that cannot be corrected. In this way, theism supports infallibilism.
Against the argument
There are various problems with this argument. To begin with, we would need positive reasons that show that COR is indeed true. For one obvious question – that Oakes does not answer – is: How does God ensure that we will not believe falsehoods we cannot correct? In what follows, however, I will just presuppose PGIV as well as COR. I will now argue that even if we accept these two principles, the argument remains problematic.
Atheists
In the first place, the argument implies that atheists cannot have infallible knowledge of public-world propositions. For consider the following example: John is an atheist. He does not believe in God and he certainly does not believe that theism is true. One day, John comes to believe that there is an olive on the kitchen floor. He wonders whether he really knows that this is the case – after all, he cannot eliminate that he is an envatted brain. He thinks of the following way out: If God existed, He would not deceive me and He would not permit me to have any false beliefs I could not correct. And so I can eliminate that I am an envatted brain and I do know that there is an olive on the kitchen floor!
It seems unlikely that John is indeed able to eliminate the envatted brain scenario. For the mere presupposition that theism is true cannot enable John (or anyone else) to eliminate the envatted brain possibility. And thus, the mere presupposition that theism is true cannot give John (or anyone else) infallible knowledge of the public world.
Irrational theists
In the second place, not even all theists can have infallible knowledge of public-world propositions. Consider: Jill is a theist. She believes that God exists. She also believes that He is not a deceiver and that He will not permit us to have false beliefs we cannot correct. Her beliefs are unjustified and irrational, however, because they are the product of wishful thinking. One day, Jill comes to believe that there is an olive on the kitchen floor. She wonders whether she knows that this is the case – after all, she cannot eliminate that she is an envatted brain. She thinks of the following way out: I believe that God exists. If God exists, He would not deceive me and He would not permit me to have any false beliefs I could not correct. And so I can eliminate that I am an envatted brain and I do know that there is an olive on the kitchen floor!
It seems unlikely that Jill is indeed able to eliminate the envatted brain scenario. One can believe that some error-possibilities do not obtain, and one can believe this on the basis of some of the other (in this case, religious) beliefs one holds, but this does not mean that one is thus able to eliminate all possibilities in which not-p is the case. If that would be possible, elimination would be all too easy. I would thus say that Jill does not infallibly know that there is an olive on the kitchen floor.
Rational theists
In the third place, not even a theist who has rational and justified theistic beliefs can have infallible knowledge of public-world propositions. Consider: Jill is a theist. She believes that God exists. What is more, her belief that God exists is a rational belief. She also rationally believes that God is not a deceiver and that He will not permit us to have false beliefs we cannot correct. One day, Jill comes to believe that there is an olive on the kitchen floor. She wonders whether she really knows that this is the case – after all, she cannot eliminate that she is an envatted brain. She thinks of the following way out: I rationally believe that God exists. If God exists, He would not deceive me and He would not permit me to have any false beliefs I could not correct. And so I can eliminate that I am an envatted brain and I do know that there is an olive on the kitchen floor!
What follows from Jill's rational theistic beliefs is that she will also rationally believe that she is not an envatted brain. And thus she will also rationally believe that there is an olive on the kitchen floor. But, crucially, from all this it does not follow that she will now infallibly know that there is an olive on the kitchen floor. For in the first place, it seems plausible to say that theism could only give us infallible knowledge of the public world if it were known infallibly that theism is true. Thus, in the second place, if it would follow that Jill will now infallibly know that there is an olive on the kitchen floor, then she should also infallibly know that theism is true. Presumably, if ‘rationality’ is sufficient to give infallible knowledge in the public-world case, why should it not be sufficient to give infallible knowledge in the theistic case? Now I doubt that even many theists will want to argue that they have infallible knowledge of the truth of theism. And thus I would say that Jill does not infallibly know that there is an olive on the kitchen floor.
Conclusion
Putting the preceding points together, the key problem with the argument is that the conclusion of the argument only follows if we infallibly know that theism is true, and, in particular, if we infallibly know that premises (1) and (2) are the case. But even though theism might be rational, justified and warranted, no-one will presumably want to argue that we can have infallible knowledge of the truth of theism. Hence, the conclusion (7) does not follow from the premises.
At this point, Oakes could object that even though his argument does not work in its strongest form, it might work in a weaker form. That is to say, Oakes could argue that even though theism does not imply infallibilism, it does imply that one can be justified or entitled to believe that one has infallible knowledge of public-world propositions. If (1) one rationally believes that God exists, (2) one rationally believes that He does not deceive us, and (3) one rationally believes that God will not permit us to believe falsehoods we could not correct, then (4) it would also be rational to believe that one can infallibly know public-world propositions.
This argument is invalid, however. What follows from (1)–(3) is that it would be rational to maintain that there is no epistemic possibility of p's being false. (Indeed, Oakes says as much on 198). However, as said at the beginning of this section, infallibilism is usually not seen as the thesis that the subject's evidence should eliminate all epistemic possibilities in which not-p is the case, but that the subject's evidence should eliminate all logical possibilities in which not-p is the case. So although it might be rational to think – on the basis of (1)–(3) – that brain-in-a-vat possibilities are epistemically impossible, it would be irrational to think – on the basis of (1)–(3) – that brain-in-a-vat possibilities are logically impossible.
Divorcing theism from infallibilism
What my discussion of Oakes's argument shows is that theism can only provide us with infallible knowledge of public-world propositions if theism itself is (or can be) known infallibly. But even though philosophers might have shown that theistic belief is rational, it has not yet been shown that we (can) have infallible knowledge of the truth of theism.Footnote 2 Hence, I submit that Oakes has not established the conclusion that theism supports infallibilism about public-world beliefs. The marriage between theism and infallibilism – if there even is one – has certainly not been made in heaven.Footnote 3
