Richard Swinburne’s argument from religious experience (AFR) as given in The Existence of God (2nd ed.- hereafter: EOG) is based on three key epistemological  principles:

EXPERIENCE

…(in the absence of special considerations), if it seems (epistemically) to a subject that x is present (and has some characteristic), then probably x is present (and has that characteristic)… (EOG, p. 303)

MEMORY

If it seems to a subject that in the past he perceived something  or did something, then (in the absence of special considerations), probably he did. (EOG, p.303)

TESTIMONY

…(in the absence of special considerations) the experiences of others are (probably) as they report them. (EOG, p.322) 

There are some interesting issues and complexities involving probability calculations that I have run into recently in thinking about this argument.  Let’s start simple, and then work towards more complicated and realistic scenarios. The simple scenario  I have in mind is this:

Just one person has just one religious experience of a generic theistic sort (i.e. this person has an experience which seems (epistemicallly) to him or her to be an experience of the presence of God).  

What is the evidential force of this experience for that person who has the experience, given Swinburne’s principles?

If the person in question is having this religious exprience right now, then he or she does not need to make any assumptions about the reliability of his or her memory, nor is there a need to make use of testimony about the religious experiences of others, since we are assuming that there is just one religious experience on just this one occasion.  The reasoning of this person would go like this, based on Swinburne’s principle concerning experiences:

1. I am now having an experience in which it seems (epistemically) to me that God is present here and now.

2. There are no special considerations that cast doubt on the veridicality or reliability of this experience.

3. In the absence of special considerations that cast doubt on the veridicality or reliability of this experience, if it seems (epistemically) to a subject that x is present (and has some characteristic), then probably x is present (and has that characteristic).

Therefore:

4. It is probably the case that God is present here and now.

One obvious “special consideration” against the veridicality of this religious experience is evidence against the existence of God.  Swinburne recognizes that this is relevant, and he has saved the argument from religious experience for the end of his case for God.  So, he thinks that he has already dealt with various reasons and arguments against the existence of God, including the problem of evil, and thinks he has shown that there is at least a significant probability that God exists, even taking negative evidence into account.   I interpret him to claim that the probability for the existence of God is between about .4 and .5 prior to consideration of AFR.

But Swinburne thinks that he only needs to show that the probability of God’s existence is something greater than “very low” prior to consideration of religious experience.  I interpret that to mean that he only needs to show that there is a probability of at least .2 (two chances in ten) that God exists, prior to consideration of AFR.

I’m not going to directly challenge the above reasoning that is based on a single instance of a theistic religious experience.  I’m more interested in looking at the issues that arise in more complicated scenarios.

One obvious complication is that religious experiences usually only last for a few seconds or a few minutes.  This means that the above reasoning will only be of temporary relevance to the person who had the religious experience.  Once the experience is gone, the person who had the experience must rely on a MEMORY of the experience to justify his or her current belief in God:

5. It seems (epistemically) to me that last Friday night, I had an experience which (at that time) seemed (epistemically) to me to be an experience of the presence of God.

 6. There are no special considerations that cast doubt on the  veridicality or reliability of my apparent memory of having had this experience last Friday night.

7. If it seems (epistemically) to a subject that he or she had a certain experience at a particular time in the past, then (in the absence of special considerations that cast doubt on the veridicality or reliability of that apparent memory) he or she probably did have that experience at that particular time in the past.

Therefore:

8.  It is probably the case that last Friday night I had an experience which (at that time) seemed (epistemically) to me to be an experience of the presence of God.

The apparent memory does NOT absolutely guarantee that the experience really happened as one thinks it happened.  An apparent memory can only make it very probable that the experience happened and was of a certain character.  Furthermore, even at the very moment that the religious experience was occurring, the experience did not absolutely guarantee that God was in fact present; it only made the presence of God probable.  In remembering a religious experience, one makes two probable inferences.  The first probable inference is from the apparent memory to the occurrence of the religious experience, and the second probable inference is from the occurrence of the religious experience to the existence of God.  Each probable inference in a chain of inferences lowers the probability of the conclusion.

At the time I was having the religious experience, I could be very confident that I was having an experience which seemed (epistemically) to me to be an experience of the presence of God.  So, at that time we could say that I was justifiably certain that I was having such an experience.  The probability that I was having such an experience could be said to be 1.0 for me at that time.  But that time has come and gone, and I can no longer be certain that I had that religious experience and that it was of the described character.

Suppose that given the apparent memory of having had a religious experience (of the sort described above), and given the absence of special considerations that cast doubt on the reliability of the apparent memory, the probable inference to the conclusion that the religious experience really occurred gives that conclusion a probability of .8.   That I am right now having an apparent memory of this event is something I can know with a very high degree of certainty, so let’s just say that the occurrence of the apparent memory is certain, that it has a probability of 1.0.  In this case, the conclusion that I had the religious experience (as described) last Friday night would be .8, based on the apparent memory of having had that experience.

If it were certain that I had an experience that seemed (epistemically) to me to be an experience of the presence of God, this would NOT make it certain that God exists, but if there are no special considerations that cast doubt on the veridicality or reliability of that expereince, then, according to Swinburne, I can justifiably infer that it is probable that the experience was veridical and thus that God probably exists.  Let’s suppose that given that it was certain that I had a religious experience of the sort described, this would make the probability of the  existence of God .8.  It is tempting at this point to reason along the lines of a hypothetical syllogism:

9. If someone has an apparent memory of a religious experience of the presence of God (and there are no special considerations casting doubt on the memory), then that person probably did have a religious experience of the presence of God.

10.  If someone had a religious experience of the presence of God (and there are no special considerations casting doubt on that experience), then that person  probably was in fact in the presence of God and God probably does exist.

Therefore:

11.  If someone has an apparent memory of a religious experience of the presence of God (and there are no special considerations casting doubt on that memory, and there are no special considerations casting doubt on the experience), then that person probably was in the presence of God and God probably does exist.

From (11) we can form an argument for the probability that God exists, by adding a few premises:

12.  I have an apparent memory of a religious experience of the presence of God.

13.  There are no special considerations casting doubt on that apparent memory.

14. There are no special considerations casting doubt on that religious experience.

Therefore:

15.  I probably was in the presence of God and God probably does exist.

However, there are a couple of problems with the logic of the argument for (11).  First of all, the following is NOT a valid deductive argument:

16. If P, then probably Q.

17. If Q, then probably R.

Therefore:

18. If P, then probably R.

The concusion does not follow logically, because in a chain of probable inferences, the probability is reduced at each step.  Suppose that the truth of P made the probability of Q  equal to .6.  In that case, premise (16) would be true (if we interpret “probably” to mean having a probability greater than .5).  Suppose that the truth of Q makes the probability of R equal to .6.  In that case, premise (17) would be true. The truth of P would thus only make Q somewhat probable (.6), so we would not be certain that Q was true, and thus we would not be certain that premise (17) applies. There is only a probability of .6 that Q is the case, so only a probability of .6 that premise (17) applies.  Only if Q turns out to be true will the logic of premise (17) be activated.  Thus, we must multiply the probabilities of the two probable inferences:  .6  x .6 =  .36.    So, if P is the case, then this argument only supports the conclusion that the probability of R would be .36 , or rounding to one digit: .4.  But a probability of .4 is too low to justify the conclusion that R is “probably” true.  In order to conclude that R is “probably” true, one would need to show that the probability of R was greater than .5 (at the least).

Another way to put this point, is to note that this relationship (If X, then probably Y) is NOT transitive, as opposed to the similar sounding relationship If X, then Y, which is transitive.  In a chain of many implications or entailments, the strength of the logical connection does not weaken:

19. If P, then Q.

20. If Q, then R.

21. If R, then S.

22. If S, then T.

Therefore:

23. If P, then T.

The above reasoning is deductively valid.  The logical connection between P and T in the conclusion is just as strong as the logical connection between P and Q in premise (19).   Swinburne is very much aware of this basic logical point that distinguishes probable inferences from implications or entailments.

There is another problem or complexity involved in this argument form:

16. If P, then probably Q.

17. If Q, then probably R.

Therefore:

18. If P, then probably R.

With inductive reasoning, the probability of a claim or belief can change with new or additional evidence.  Thus, although P might well make Q probale in most circumstances, there are possible circumstances in which although P is the case, Q would definitely be false.  For example, suppose that you see that I frequently drive a late-model Mercedes-Benz sedan.  You might reasonably infer that I am probably NOT poor. But if you learn that I have a part-time job as a driver for a wealthy business man, then your previous inference is cast into doubt.  I might well be poor, even though I frequently drive a late-model Mercedes-Benz sedan.  That is how inductive reasoning works.  New information can alter the probability of a claim or belief.

This means that in order for premises like (16) or (17) to be true, we must understand them to involve an unstated qualification: other things being equal.

16a. If P, then probably Q (other things being equal).

17a. If Q,then probably R (other things being equal).

Therefore:

18a. If P, then probably R (other things being equal).

In other words, probable inferences and inductive reasoning are always to be thought of as contextual, as referring to a certain collection of information or assumptions, and so there is always the possibility that new or additional information could alter the probabilities.

To be continued…