Swinburne’s Argument from Religious Experience – Part 4

Although I have been considering the implications of the idea that the veridicality of a Theistic Religious Experience (TRE) is independent of the veridicality of other TREs, this is NOT the view of Swinburne.  In fact, Swinburne clearly holds the opposite view, the view that the veridicality of a TRE is dependent on the veridicality of other TREs.  I will get into the details of this shortly.

First, let me back up for a moment and provide a key definition.  Swinburne defines “religious experience” in Chapter 13 of The Existence of God (2nd edition, hereafter: EOG, where he presents his Argument from Religious Experience, hereafter: AFR):

For our present purposes it will be useful to define it [a ‘religious experience’] as an experience of God (either of his just being there, or of his saying or bringing about something) or of some other supernatural thing. (EOG, p.295) 

Note the emphasis on TREs: “an experience of God”.  Swinburne does not limit religious experiences to experiences of God, since the definition also includes experiences “of some other supernatural thing”.  However, Swinburne immediately points out that his focus is on TREs, especially on one specific kind of TRE:

For most of the discussion I shall be concerned with experiences that seem to be simply of the presence of God and not with his seeming to tell the subject something specific or to do something specific. (EOG, p.296) 

So not only is Swinburne’s argument focused on TREs, but it is focused on a specific subset of TREs, what I have referred to as “generic” TREs.

Statements of key points in his argument also focus on TREs:

…a religious experience apparently of God ought to be taken as veridical unless it can be shown on other grounds significantly more probable than not that God does not exist. (EOG, p.321)

One who has had a religious experience apparently of God has, by the Principle of Credulity, good reason for believing that there is a God… (EOG, p.325)

Swinburne’s definition of ‘religious experience’ has a flaw, if taken as it stands.  It is not clear that one can have “an experience of God” if there is no God.  Swinburne does not intend to beg the question about the existence of God, and in the context of  the opening of Chapter 13, it is fairly clear that what he had in mind is an experience in which is seems (epistemically) to the subject that God is present (or that God is communicating a message to the subject, or that God is performing some action).

Swinburne leaves open the possibiilty that it might seem (epistemically) to a person that God is present when in fact there is no God, and thus God was NOT present to that person.  In other words, one can have a TRE that is non-veridical.  Having a theistic religious experience does NOT imply or entail that God was present or that God exists.  It might be the case that all TREs are non-veridical, that all TREs are misleading experiences.  Therefore, the occurrence of TREs does not in and of itself logically imply that God exists.

Back to the issue of dependency between the veridicality of generic TREs.  One obvious point is that if just one single generic TRE is veridical, then that means that God was present at least on that particular occasion.  But since God is omniscient and omnipotent and eternal (by definition), if God was present on one occaision, then it follows logically that God is present at any and every place at any and every time.  If God exists at one moment, then God exists in all moments, for any person who exists for only a finite duration of time cannot be ‘God’.  Any person who can only influence events in one particular part of the universe cannot be ‘God’.    Any person who is only aware of events in a particular place or at a particular time cannot be ‘God’.  In short, if God was present at one moment of time in one particular location, then God exists.  If God exists, then God is present at all times and at all places.

Recall that Swinburne saves his presentation of AFR until after all other major considerations for and and against the existence of God have been covered (in his view).  He believes that other relevant evidence shows that the existence of God is somewhat probable, that theism has a probability somewhere between .4 and .5:

g: God exists

.4 < P(g) < .5 

But Swinburne is clearly talking about a conditional probability, a probability that is based on the evidence in the premises of his previous arguments for and against God.  Let’s use a letter to represent this background evidence that was considered prior to examination of AFR:

k: [the background evidence of the premises of the inductive arguments for and against God previously presented by Swinburne]

Now we can represent the probability range more accurately:

.4 < P(g|k) < .5

Swinburne believes he has a bit of wiggle room here, because all that is required for the success of AFR, in his view, is that the prior probability of the existence of God be more than just ‘very improbable’.  I would interpret that to mean the following assumption is required for the success of AFR:

P(g|k) > .2

If AFR is as good as Swinburne thinks, then the evidence in the premises of this argument should bump up the probability significantly, to make the existence of God “more probable than not”:

e: Many people have had generic TREs which are not subject to special considerations that cast doubt on the veridicality of those TREs. 

P(g| e & k) > .5

If we have before us a collection of clean (i.e. no special considerations apply) generic TREs, and if we could somehow determine that one TRE in this collection was in fact veridical, that would, by itself, make it certain that God exists.  From that point forward any further instances of TREs would need to be evaluated on the basis of a NEW prior probability of the existence of God.  This new information, that at least one TRE was veridical, would shift the prior probabililty of the existence of God from somwhere between .4 and .5, all the way up to the maximum probabilty: 1.0.  In other words, as soon as one single TRE has been determined to be veridical, we have good reason to be much less skeptical about the veridicality of other TREs.

This is kind of like the idea of a miracle.  As soon as one single miracle has been determined to be valid, that establishes both the existence of God and the fact that God is, at least on some occasions, willing to intervene in nature for the sake of some human (or some animal) and to cause a violation of a law of nature.  Once one single miracle has been determined to be valid, then we would have good reason to be much less skeptical about other miracles.

A similar sort of relationship appears to hold in the case that we determine a particular TRE to be non-veridical.  If someone claims to have had an experience of the presence of God, but we determine that God was NOT present on that occasion, then we have also determined that God does NOT exist.  For if God DID exist, then God would have been present in the time and place that the person who claims to have experienced God had this experience that seemed to him/her to be an experience of the presence of God.  If God exists, then God exists at all times and at all places.

Furthermore, according to Swinburne, if God exists, then God is involved in the causation of any religious experience that seems (to the subject) to be an experience of God:

But, if there is a God, he is omnipresent and all causal processes operate only because he sustains them.  Hence any causal processes at all that bring about my experience will have God among their causes; and any experience of him will be of him as present at a place where he is.  And so, if there is a God, any experience that seems to be of God, will be genuine–will be of God. (EOG, p.320)

It appears that if there is just one single TRE that we determine was non-veridical, then we have determined that God does NOT exist, and that all other TREs are also non-veridical.  If God exists, then all TREs are veridical.  Therefore, if just one TRE is non-veridical, then God does NOT exist.

So, at least at first blush, it appears that if one single generic TRE is determined to be veridical, that shows that God exists, and that other generic TREs are also veridical, and it appears that if one single generic TRE is determined to be non-veridical, that shows God does NOT exist, and that other generic TREs must also be non-veridical.  Given these two sorts of logical dependencies, the probability tree diagram for generic TREs would look like this:

3 TREs with Dependency

As soon as the status of the first TRE is determined, so is the status for any other TREs.  If the first TRE was veridical, then God exists, and all other TREs must then also be veridical, based on Swinburne’s views about the implications of the veridicality of a TRE.  If the first TRE is non-veridical, then all other TREs must then also be non-veridical.  Assuming that the prior probability of God’s existence is .4, we must either determine that the first TRE is veridical and raise that probabilty to 1.0, or determine that the first TRE is non-veridical and lower that probability to 0.

I don’t think Swinburne was aware of this implication of his view of the implications of determining a TRE to be veridical:

There are large numbers of people both today and in the past who have had religious experiences apparently of the presence of God and that must make it significantly more probable that any one person’s experience is veridical. (EOG, p.323-324)

It seems to me that the occurrence of large numbers of “religious experiences apparently of the presence of God” does NOT help the case for God.  The probability of the veridicality  of the first TRE that we consider will depend on the prior probability of the existence of God, but once the veridicality of that TRE is determined, the question of the existence of God will be answered, and no further TREs need be considered, because the veridicality of the remaining TREs will be determined by whether the first TRE was veridical or not, given Swinburne’s assumption that IF God exists, then ALL  generic TREs must be veridical.