Can a transcendental argument reach the world, or only our beliefs?

Some arguments try to prove God from evidence: the universe had a beginning, life looks designed, people report miracles. The transcendental argument does something different and more ambitious. It says: never mind the evidence. Look at what you are doing when you argue at all. Reasoning presupposes things — that logic holds, that the world is regular enough to learn from, that your mind and reality are made for each other. Those presuppositions need an explanation, and only one explanation works. So anyone who argues has already conceded the conclusion.

It is a genuinely clever strategy, and it has a genuinely hard problem, which is the subject of this page. Nearly everything in the argument turns on a single step, and its own best exponents can be found stating the objection to that step more sharply than most of its critics do — and then, later in the same recording, walking straight into it.

The question, stated exactly

Strip the argument to its skeleton and it has three lines.

1. Human thought and knowledge happen.
2. Necessarily, they can only happen if some state of affairs X obtains.
3. Therefore X obtains.

As logic, this is fine; if the second line is true, the third follows. Everything depends on what the second line says.

Here is the difficulty. There are two versions of it, and they are very easy to confuse.

Only the strong version reaches the world. The weak version tells you something about minds. And every argument of this shape faces the question of which one it has actually established — because the evidence for the two looks identical from the inside. Whatever it feels like to be unable to think without assuming something, it feels exactly the same whether that something is true or merely unavoidable.

The objection, named

The canonical statement of the problem is Barry Stroud’s, from a 1968 paper that has shaped the literature ever since. Stroud’s move is called the substitution objection, and it is simple to the point of being annoying.

Take any proposition the proponent claims must be true for thought to be possible. Substitute the weaker claim: what must be true is only that thinkers believe it, or that the world appears that way. The proposition itself may still be false. Unless the proponent can rule that substitution out, all he has shown is that the belief is indispensable — and indispensability is a fact about us.

There are two ways to answer this. One is to strengthen the claim: show that denying X is not merely unwarrantable but self-falsifying — that the act of denying it makes it true. That works for a small privileged class of propositions. The other is to attack Stroud’s underlying assumption that unavoidable belief and truth can always come apart, which requires a form of idealism or verificationism — and if you have that, the transcendental argument has become redundant, because the idealism is doing the work.

That is the fork. Everything below is about whether anyone has taken either branch successfully.

The proponent states the objection himself

The most striking thing in the source material is that the objection does not have to be imported. In a two-and-a-half-hour discussion on transcendental arguments, Jay Dyer reads out, and endorses, a passage from Russell Manion that puts the problem precisely:

does the mere fact that we are psychologically disposed such that we cannot think of the world in any other way than logical categories justify us in concluding that the world therefore is bound by those categories¹

He immediately calls that transition a “leap” and a “non sequitur,” and offers the obvious defeater: human beings may simply be built to read reality that way. Twenty minutes later he puts it in his own words:

why should we think that our psychological limitations are descriptive of the entire universe […] a psychological inability to modify the way that we think about the universality of logic does not prove it is in fact universal²

This is Stroud’s objection, stated by a proponent, without prompting, and applied against rival (Thomist and Aristotelian) versions of the same manoeuvre. It is the right objection and it is well put. It also means that no critic can be accused of inventing a problem the tradition does not recognise.

What is proposed instead

Dyer’s own answer, later in the same discussion, is not “we cannot think otherwise.” It is a substantive metaphysical proposal about what X actually is, and it is worth stating at full strength because the popular caricature of presuppositionalism misses it entirely.

my answer is god created the world the world operates in a regular fashion through divine providence and that gives me a basis for belief in induction […] i have a belief in logic because i believe in abstract principles which are grounded ultimately in the mind of god³

So X has three parts: a creator who makes minds and world fit one another; providence, which makes the world regular enough for induction to be a reasonable bet; and divine conceptualism — the view that logical and mathematical truths are thoughts in an eternal mind rather than free-floating abstractions or human conventions. And the starting point is revelation: “my jump start my starting point is divine revelation so that’s what grounds and gives me an account for the world the past the future the self abstract concepts logic.”⁴

Assembled charitably, the argument is this.

1. Genuine human knowledge, reference, predication and inductive reasoning occur.
2. Knowledge is factive and non-accidental; reference and predication require objective structure; induction requires a regular and knowable order.
3. Those conditions obtain only if reality and human minds are intentionally coordinated by an omniscient creator, abstract objects are grounded in the divine mind, and regularity is secured by providence.
4. That relation is not merely believed. It is the actual metaphysical condition of the practices in line 1.
5. Therefore such a creator exists.

Line 4 is the strong version of the conditional, stated explicitly. That is a real advance on vaguer formulations: it names what has to be true and says plainly that belief is not enough. The remaining question is whether anything supports it.

Where the argument stops

The clearest test comes near the end of the same discussion, when an interlocutor asks whether the Orthodox worldview might itself be something “cooked up and conceived” — that is, whether the whole account might be an indispensable representation rather than the truth. The reply:

the problem of the external world isn’t a problem for a christian because he believes god created the world because divine revelation says there is an external world⁵

Read carefully, that answers a different question. It says what the Christian believes and why the belief is internally coherent. It does not show that the belief is true rather than indispensable — which is exactly the substitution the objection proposed, and exactly the objection Dyer himself had stated an hour and a half earlier against other people’s arguments.

The general shape of what is missing can be stated exactly. Here is a countermodel — a coherent story on which everything the argument observes is true and its conclusion is false.

1. Finite minds are constituted so that they necessarily apply logical, causal and object concepts.
2. Their practices are coherent and locally reliable enough for thought, language and action to work.
3. They necessarily believe those concepts track universal, mind-independent structure.
4. Those beliefs might nonetheless misdescribe the ultimate ontology. Abstract truths might not be divine thoughts; regularity might not be providential; the apparent fit between mind and world might hold only within the range those minds can access.
5. So the practices can occur, with the belief intact, while the full metaphysical claim is false.

Nothing in the sources rules that out. And note that Dyer’s own earlier critique positively licenses it: the argument that psychological necessity does not entail worldly necessity works just as well against his conclusion as against the Thomist ones he aimed it at.

There is a further problem, raised by Bálint Békefi, that is worth stating because it is structural rather than rhetorical. If the argument’s authority rests on revelation, and revelation is accessed historically and perceptually — through texts, testimony, and the senses — then the argument is leaning on precisely the faculties whose reliability it was supposed to underwrite. And if biblical authority may simply be granted at the outset, the transcendental argument becomes superfluous: you already have what you were trying to prove.⁶

What actually survives

Two results survive intact, and both are worth having.

Factivity is a real, if narrow, bridge. If someone genuinely knows that p, then p is true — that is what knowledge means. So an argument that establishes knowledge of some particular proposition does thereby establish something about the world. The limitation is severe: it delivers the truth of the thing known, not the truth of whatever theory you propose as the precondition of knowing it. Factivity gets you across the gap for one proposition at a time; it does not get you to divine conceptualism.

Stroud’s objection is not a general refutation. This matters, and it comes from an independent direction. Simon Schüz argues that to run the substitution objection across the board, the sceptic needs a realist conception of objectivity — a standard on which beliefs about the world concern facts independent of being believed. But applying “this too might be false” to that constitutive standard itself generates an antinomy. So the sceptic must distinguish ordinary empirical beliefs, where substitution stays available, from transcendental beliefs that are constitutive of applying the standard of objectivity at all. Against the second class, substitution cannot simply be presumed.⁷

That is a genuine rehabilitation, and it means the blanket verdict “all such arguments fail” is unavailable. It is also carefully bounded: Schüz shows a restricted class of world-directed transcendental arguments is possible in principle. He does not show that Christian theism, providence, divine conceptualism or revelation belongs to that class, and a candidate would still have to survive self-reflective doubt by way of a very strong necessity claim.

A rival that does not need a divine mind

An argument of this shape depends on there being no alternative. So the honest test is whether an alternative exists — a worked-out account of objective logical reality that neither reduces logic to human convention nor grounds it in God. One does, and it is not a sketch.

Gila Sher’s formal-structural realism runs roughly as follows. Logic’s job is to supply an especially powerful method of inference that expands knowledge by reliable shortcuts. It is not a game: logic’s rules “are constrained by the world.” Reality itself has formal features — identity, cardinality, union, intersection, symmetry, transitivity — that are patterns of ordinary objects and properties rather than residents of a separate abstract realm. What makes an operator formal is a precise mathematical condition: it “is formal iff it is invariant under all isomorphisms of its argument-structures.” Because formal laws are invariant in that strong way, they have wider counterfactual reach than physical laws, and that is what gives logical inference its distinctive necessity. Logical norms then follow from truth: “the source of the normativity of logic, on the present account, is its truth.” Our knowledge of all this is holistic, fallible and revisable, with reason dominant because formal features are too abstract for direct observation.⁸

Whatever one finally thinks of it, this is a real countermodel. It motivates objective logical structure independently — from correspondence with the world and from invariance — rather than deriving it from what thinkers cannot help believing. It therefore blocks the inference from “objective logical facts obtain” to “a divine mind grounds them,” because here is an account on which the first is true and the second is not asserted.

Blocking an inference is not winning an argument, and the rival’s own costs are real. Three in particular. It explains why logic is normative for inquiry aimed at truth, which leaves the person who declines that aim untouched — in Sher’s later work the normativity of truth is located in a chosen, if objective, human value. It gives an excellent account of how error is corrected, through anomaly detection and reciprocal testing, without fully explaining why finite minds should be able to select the uniquely correct formal structure in the first place. And it leaves intentionality — the fit between thought and world that the theistic account was built to explain — as something described rather than grounded.

So the position is not “atheism wins.” It is that the exclusivity premise, on which the whole strategy depends, has not been established against a live, published, technically developed alternative.

Where this leaves the question

Open, with the burden clearly located.

The argument is not refuted. Its logic is valid, its strong conditional is now stated explicitly rather than left vague, and the general objection against it has itself been shown not to be a general objection. Anyone who says these arguments are obviously worthless has not read the recent literature.

What has not been shown is why thought and knowledge require the truth of the full metaphysical account rather than the unavoidable belief in, representation of, or reliable access to some order. “Revelation says there is an external world” is a description of the proposed account. It is not yet a reason to think the account is true rather than indispensable.

Six things would move this. A proponent would have to pick one exact capacity and one exact worldly condition; defend the strong conditional rather than the weak one; show either that denying the condition is self-falsifying or that the condition is uniquely constitutive of objectivity; defeat an explicit belief-only countermodel and at least one structural rival such as Sher’s; avoid using the truth of the disputed revelation as an unargued premise; and state the scope without quietly smuggling in uniqueness, Trinity or a particular church.

That is a demanding list. It is not an impossible one, and the Schüz result is reason to think the space is not empty. But no source examined here has done it, and until someone does, what the argument establishes is that certain beliefs are unavoidable — which is a fact about thinkers.

References

References

1. Jay Dyer, “Transcendental Arguments, Epistemic Justification, Foundationalism & Circularity,” 1 April 2022, 2h36m, at 00:32:32–00:32:53, reading and endorsing R. M. Manion, “The Contingency of Knowledge and Revelatory Theism.” Quotations are from the automatic English captions. ↩

2. Same source, 00:35:46–00:36:12. ↩

3. Same source, 02:23:57–02:24:26. ↩

4. Same source, 02:25:24–02:25:43. ↩

5. Same source, 02:34:02–02:34:32. ↩

6. Bálint Békefi, “Van Til versus Stroud: Is the Transcendental Argument for Christian Theism Viable?”, TheoLogica 2, no. 1 (2018), 136–160. Békefi distinguishes a denial that is merely self-stultifying — one you could not rationally affirm — from one that is performatively self-falsifying, where the act of affirming it makes it false, and argues that only the second would carry the required weight. Barry Stroud’s original objection is “Transcendental Arguments,” The Journal of Philosophy 65, no. 9 (1968), 241–256, especially 253–256; the publisher’s full text was not accessible during this research, and the passages were checked against two full-length scholarly papers and the Stanford Encyclopedia of Philosophy rather than read in the original. ↩

7. Simon Schüz, “Rehabilitating Transcendental Arguments: A Dialectical Dilemma for Stroud’s Meta-Epistemological Skepticism,” Ergo 10, no. 22 (2023), 612–648, especially 645–646. On the question of what kind of necessity such an argument could claim, see also James N. Anderson, “No Dilemma for the Proponent of the Transcendental Argument,” Philosophia Christi 13, no. 1 (2011), 189–198, who clarifies the modal structure while conceding that defending the crucial premise remains a “formidable” task. ↩

8. Gila Sher, “The Foundational Problem of Logic,” Bulletin of Symbolic Logic 19, no. 2 (2013), 145–198, and “On the Explanatory Power of Truth in Logic,” Philosophical Issues 28, no. 1 (2018), 348–373. The quoted phrases are, in order, Sher 2018, 352–353; Sher 2018, 355; Sher 2013, 176; and Sher 2013, 191. The independent map of the position, and the access and structure-identification costs noted here, are from Tuomas E. Tahko, “A Survey of Logical Realism,” Synthese 198, no. 5 (2021), 4775–4790, especially 4780–4783. The location of truth’s normativity in a chosen human value is Sher’s own later statement: “The ‘Post-truth’ Crisis, the Value of Truth, and the Substantivist-Deflationist Debate,” Australasian Philosophical Review 9, no. 1 (2025), 7–31, at 14–15, where she says explicitly that the value of truth is not “forced upon us (by nature, God, the property of truth, the principles of rationality, or what have you).” Sher’s own religious views are not known and are not at issue; the countermodel is an analyst’s test of what her account entails, not a claim about what she believes. ↩