The limitations of Transcendental Arguments

0. Introduction

In the very first post on this blog, back in November 2015, I looked at a paper by Michael Butler, which is a first-rate examination of transcendental arguments (TAs) and their relationship to the presuppositional TAG of Van Til, Bahnsen, etc. I recommend reading Butler’s paper closely.

Part of what is interesting about that paper is the breadth of reading that Butler has done on the subject. It comes with a comprehensive literature review of the contemporary literature on TAs in analytic philosophy, and a satisfying look at the Kantian origins of this tradition. It also provides a look at Van Til’s approach, and how that was taken forwards by Greg Bahnsen. Butler, though a presuppositionalist himself, (and I think trained by Bahnsen) raises four interesting criticisms of TAG. He explains Bahnsen’s attempts to rebut these criticisms, and then criticises these rebuttals in turn. The four objections are:

(1) the nature of TAG; (2) the uniqueness proof for the conclusion of TAG; (3) the mere sufficiency of the Christian worldview; (4) the move from the conceptual necessity of God’s existence to the actual existence of the Christian God.

The last objection is probably the most interesting, and the one I paid least attention to in my previous post. As Butler says:

I consider this stricture to be the most powerful argument against TAG and the most difficult to answer.

The explanation of the objection runs as follows:

This objection revolves around the consideration that proving the conceptual necessity a worldview does not establish its ontological reality. Kant, for example, argued that the notion of causation is transcendentally necessary for thought (or at least human thought).  Without the concept of causality there could be no thought.  But just because causality is necessary for thought does not mean, so Kant argued, that the things in themselves (ding an sich) which exist independently of our conception of them, undergo causal relations.  Conceptual necessity does not guarantee ontological necessity.  In the same way, assuming that TAG is sound, all that is proved, so this objection goes, is that we must, in order to be rational, believe that God exists.

What this objection is highlighting is that transcendental arguments, such as those employed by Kant, even if thought to be successful, do not establish the existence of something in reality. Rather, they establish synthetic a priori truths, which are about the forms our experience must take. They establish, at best, only conceptual necessities, not ontological necessities.

For example, we may be forced to experience the flow of time as one event happening after another, in the past, present and future. We may be unable to conceive of how to make sense of the contrary – what would it even mean to have all your life’s experiences ‘at the same time’? Yet, even if this were so, we would still be left wondering whether this is because time is ‘in itself’ something which really does flow, or whether it is just a “stubbornly persistent illusion” as Einstein once said. It may be required for us to make sense of the world, yet not indicative of the way the world is. To put the point another way, it may be a necessary precondition for intelligibility (a phrase we hear a lot with internet presuppositionalists), yet not true.

  1. Bahnsen’s rebuttal

Butler quotes Bahnsen’s rebuttal to this objection:

…because this is an apologetical dialogue (giving reasons, expecting argument, etc.), both parties have assumed that the true viewpoint must affirm rationality.  Van Til argues that if the unbeliever’s worldview were true, rationality would be repudiated, whereas if Christianity were true, rationality would be affirmed and required.  So while the whole argument may be stated in hypothetical terms, the conclusion is actually established as true, since the hypothetical conditions was granted from the outset by both parties.  (If the unbeliever realizes this and now refuses to grant the legitimacy, demand, or necessity of rationality, he has stepped outside the boundaries of apologetics.  Furthermore, he forfeits the right to assert or believe that he has repudiated rationality, since without rationality assertion and belief and unintelligible.)

Butler is impressed with this response, though ultimately concedes that it does not address the point in hand. His positive comments are the following:

Bahnsen’s answer is that the issue is one of rationality.  If TAG establishes that Christianity is the necessary conceptual precondition of human experience (including rationality) it follows that we must hold to the Christian worldview in order to be rational. And if somebody refuses to accept the Christian worldview or God’s existence, he has no foundation for rationality and, without such a foundation, has no rational basis to object that the conclusion of TAG.

This defense carries a great deal of force.  It effectively undermines the unbelievers ability to rationally reject the Christian faith.

2. My objection

Before we go on to look at Butler’s criticism of this, and how he tries to improve on Bahnsen, I want to point out that I think we can bring up an interesting objection at this stage.

In a previous post on Stephan Molyneux, I pointed out an objection to one of his arguments. He made the following claim:

If I tell you that I like chocolate ice cream, and you tell me that you like vanilla, it is impossible to “prove” that vanilla is objectively better than chocolate. The moment that you correct me with reference to objective facts, you are accepting that objective facts exist, and that objective truth is universally preferable to subjective error.

However, this seemed to boil down to an implausible inference, namely: if you argue that p is true (“with reference to objective facts”), you are showing a preference for truth over falsity. The reason this seems implausible is that we can easily come up with examples of people engaging in arguments (even “with reference to objective facts”) who do not necessarily have any such preference for truth over falsity. My examples were:

  • Putting forwards an argument arguendo
  • Participating in debate competitions
  • A lawyer arguing because she gets paid to do so
  • A politician arguing to win votes
  • An internet troll arguing to cause irritation
  • An undergraduate arguing to impress others

What these examples show is that there are contexts in which someone can put forward an argument yet not be primarily directed towards truth, or be of the opinion that ‘truth is preferable to error’. I described these as ‘non-standard’ contexts, in contrast to what Molyneux seemed to have in mind, which should be called the ‘standard context’.

The standard context is something like where both parties are solemn truth-seekers, and ‘play by the rules’ of correct logical behaviour (or whatever – its not even clear what this is supposed to be). But the point is that Molyneux is effectively just presuming that all argument goes on in standard contexts, and that therefore if you make an argument you have the attitudes and beliefs of those in the standard context, i.e. you are earnestly seeking truth, and prefer truth to falsity.

However, the non-standard contexts I describe above are examples where people simply have different attitudes, and in which they are not earnestly seeking truth at all (they are doing something else). They may prefer a useful falsehood (if it wins the case, or gets them laid). This shows that you cannot make the inference that Molyneux describes; just because someone is making an argument does not establish that they prefer truth over falsity.

I might even go so far as to say that the ‘standard’ context is a kind of fantasy. Does it ever actually exist? What even are its necessary conditions? I certainly don’t know. It seems at least possible that the ‘non-standard’ contexts are in fact ubiquitous (although this is a dangerous ‘post-modern’ thought…).

Anyway, let’s tie this back in to Bahnsen’s reply. Recall, he said:

…because this is an apologetical dialogue (giving reasons, expecting argument, etc.), both parties have assumed that the true viewpoint must affirm rationality.

Here, Bahnsen is invoking the notion of the standard context, just like Molyneux, which requires that the parties “assumed that the true viewpoint must affirm rationality”.

But this seems too much of an assumption already. One could go through the motions of a debate with an interlocutor without being convinced of their rationality. I might doubt your rationality at the start, but be prepared to give you the benefit of the doubt. Maybe I took you for a rational fellow at the start, but increasingly come to doubt it as you engage with me. Am I not engaging in apologetical debate if I do that?

One could even engage in the process while being unsure of one’s own rationality. Maybe I am open to being persuaded about that question, and in the mean time I do what seems right to me in the moment of the debate. Does Bahnsen mean to say that I would not actually be participating in the debate if I was of that mindset while I did it?

And I mean, to some extent that description feels right as a description of my own mental state much of the time. My own phenomenology of rationality involves both the feeling that I am rational enough to engage in things, like conversation (and even in completing a PhD in philosophy), yet plagued with a sense of my own fallibility, seeing as I make mistakes in reasoning every day (just ask my girlfriend!). I feel aware of both the presence and the absence of my own rationality, and thus its precise extent remains somewhat unclear to me. So while I will engage in a conversation with someone like Bahnsen, he cannot assume that we both share the assumption that the truth of the matter involves my rationality, if that means to presuppose that my rational faculties are perfect. Far from it. I have a recognition that they are sometimes right and often wrong. Am I disqualified from the conversation as a result? If not, what does it even mean to say that we must both assume that the truth affirms rationality as a precondition for the debate?

But maybe all he means by the shared assumption that the truth will involve ‘affirming rationality’ is that he can assume that we both affirm enough rationality to get by in a debate. Yet, even that seems like it admits of counterexamples. Consider a variation on familiar Chinese Room thought experiments, where your interlocutor has no rationality at all.

Imagine Bahnsen is in a room and he is fed slips of paper with comments seemingly from a curious atheist about the Christian worldview, but which are in fact produced by some sophisticated mechanical algorithm. Bahnsen writes back a reply to the comments and posts them back through the slot in the door from which they came, and in this fashion enters into what he takes to be a conversation.

Imagine that Bahnsen and the algorithm debate TAG, and got into this particular bit of debate. Let’s say that the algorithm had raised Butler’s fourth objection, i.e. it had pointed out that even if TAG is sound, that only establishes that believing in God is necessary for rationality rather than establishing that God exists. At this point, imagine that Bahnsen wrote a reply on his slip of paper, which read as follows:

…this is an apologetical dialogue … [and so] both parties have assumed that the true viewpoint must affirm rationality

In this context, it is unclear whether this is even true any more. Is it still an example of ‘apologetical debate’? It is one person debating with an algorithm. Does that count? Whether we are allowed to assume all the trappings of the standard context in this setting seems rather doubtful, to say the least. What seems certain is that there isn’t really another party in the debate at all, let alone one who has assumed that ‘the true viewpoint must affirm rationality’.

This is important because the objection that the mechanical algorithm came up with is as relevant to Bahnsen as it would have been had a rational person came up with it. What difference does it make to the salience of the point itself (that TAG only establishes conceptual necessity, and not ontological necessity) if the point is made by an unthinking algorithm? It seems like Bahnsen should be just as worried about it either way.

If he left the room at this point, and then found out that there was nobody on the other side of the door, Bahnsen’s rebuttal to the fourth objection certainly seems to have been disarmed. But does he have any reason to think that the objection itself has been dealt with? I think not. What the algorithm said to Bahnsen remains true, even though it was not made by an ‘agent’ as such.

What this shows, I think, is that the problem is really his problem, not his interlocutor’s. Even if we grant that no rational agent can make the objection, it remains an objection anyway. And what this shows, I think, is that Bahnsen is really just doing a complex burden shifting exercise. Declaring victory because the interlocutor is not allowed to make the objection (while still claiming to be rational) just does not address the objection.

3. Back to Butler

Butler sees this issue in a similar way to me, although he does not put it like I do. He explains the problem like this:

The challenge is, thus, to bridge the gap between having to believe the Christian worldview because it provides the necessary preconditions of experience and showing that the Christian worldview is true.

The problem with Bahnsen’s reply to this, according to Butler, is as follows:

The problem, however, is that while TAG … demonstrates that the Christian worldview is necessary precondition for experience, it does not prove that the Christian worldview is true.  For it may be that our experience of rationality, morality, science, etc. are illusory.  Bahnsen’s reply to Montgomery, that we must make the “gratuitous assumption” that at least one worldview must be right, is without foundation. Surely, we can, for argument’s sake, conceive of the world being ultimately irrational and amoral. And if can do this, it follows that TAG, on this interpretation, fails to prove that Christianity is true.

How does Butler propose to bridge the gap between conceptual and ontological necessity? His answer involves distinguishing a conceptual scheme from the Christian worldview. The types of transcendental arguments which get us only to conceptual necessity involve conceptual schemes, whereas ones involving the Christian worldview go beyond that to ontological necessity (about how the world actually is in itself):

Stated another way, the necessity of a conceptual scheme cannot guarantee anything about the way the world must be.  For while such a scheme may organize our experience, it itself is dumb and mute and cannot, definitionally, tell us anything about the world itself.  But the Christian worldview is not a mere conceptual scheme.  It claims to do more than simply provide us with the necessary preconditions of experience. The Christian worldview posits a sovereign, creator God who is both personal and absolute in his nature. This God is, moreover, a speaking God who reveals truths to us about himself and the world. In his revelation to us he declares that he has made a world and that this world exists independently from himself and us. On the basis of his revelation, therefore, which is itself the necessary precondition of experience, we can know truths about the world and God.

The answer seems to be that Christianity involves a God who tells us what the world is like, whereas a conceptual scheme cannot. It is not clear to me how this is supposed to bridge the gap though. It seems to me though that we have simply moved from one conceptual necessity to another.

Before, we had the conceptual necessity being that ‘God exists’, i.e. the conclusion of TAG could only be that it is necessary (for rationality, etc) to believe that God exists (rather than it being true that God exists). Now though, as an attempt to bridge the gap between those, Butler has expanded the proposition in question to include the explicit statement that God tells us about the world. So now we should change the conclusion to being: it is necessary (for rationality, etc) to believe that a God exists who tells us what the world is like. Notice the scope of the belief here though. It is necessary to believe that: [a God exists who tells us what the world is like]. The content of what God tells us matching what reality is like is something that is under the scope of the belief in question; the veracity of the revelation is itself an article of belief. Unless the content of the messages that God gives us were somehow known to go beyond the conceptual necessity, to ontological necessity, we have not advanced one step from where we were before. This is because Butler’s proposal is still compatible with it being false that that God exists. It is still just a conceptual necessity. What he needs to do now is bridge the gap between believing what God tells him about the world is true, and it actually being true.

But, of course, being able to bridge the gap between something being believed to be true, and it actually being true is the very thing Butler’s suggestion was supposed to be clearing up. So unless we beg the question here, and assume we have bridged the gap somehow, his proposal is ineffective. Ultimately, the belief that what God has revealed is truthful is just another synthetic a priori truth.

4. Conclusion

Butler’s response here is utterly disappointing. I genuinely enjoyed his paper, and think it is worthy of being much more widely read. The shame of it is that the ending is so flaccid.