Aquinas, the a priori/a posterioridistinction, and the Kantian dependency thesis
Department of Philosophy, Fordham University, Bronx, New York
Abstract: This article re-examines the applicability of Kant’s dependency thesis to Aquinas’ cosmological proofs for the existence of God. The first part of the paper provides a summary of Kant's dependency thesis, followed by a review of a defence of Aquinas by J. William Forgie. The second part of the paper explains some of the logical apparatus upon which Aquinas' argument hinges - specifically his understanding of the a priori/a posterioridistinction. I conclude by calling attention to certain distinct metaphysical assumptions within Kant and Aquinas' respective logical apparatus that would have to be addressed prior to the more specific question of whether Kant's critique of the cosmological argument is applicable to Aquinas.
This article re-examines the applicability of Kant’s dependency thesis to Aquinas’ cosmological proofs for the existence of God. The first part of the paper provides a summary of Kant's dependency thesis, followed by an account of why, given that much of the literature in natural theology has moved past this question, I think it worthwhile to bring it up again. After this I concentrate specifically on a defence of Aquinas by J. William Forgie. The second part of the paper explains some of the logical apparatus upon which Aquinas' argument hinges - specifically his understanding of the a priori/a posterioridistinction. I conclude by calling attention to certain distinct metaphysical assumptions within Kant and Aquinas' respective logical apparatus that would have to be addressed prior to the more specific question of whether Kant's critique of the cosmological argument is applicable to Aquinas.
Kant's Dependency Thesis
Kant divides the cosmological argument into two parts. In the first part, one reasons from the existence of any being at all to the existence of a necessary being. In the second, one reasons from the existence of a necessary being to that of a highest being.1 I shall follow Forgie in holding that ‘it is in the second stage [of the proof] that one commits oneself to the ontological argument’ (Forgie (1995), 89).
Kant himself details the problem with the second part of the cosmological proof as follows:
Consider the proposition … that any absolutely necessary being is also the maximally real being. If this proposition is correct, then it must, like all affirmative judgements, be convertible at least per accidens - i.e., at least to this proposition: some maximally real beings are also absolutely necessary beings. However, one ens realissimum is in no respect different from another; and hence what holds for some beings contained under this concept holds also for all. Hence I shall (in this case) be able to convert the proposition also simply - i.e. to thisone: any maximally real being is a necessary being. Now because this proposition is determined a priori and merely from its concepts, the mere concept of the maximally real being must carry with it also this Being’s absolute necessity. But this is precisely what the ontological proof asserted, and what the cosmological proof did not want to acknowledge but nonetheless laid at the basis of its inferences, although in a covert manner. (B 636-37)
The relevant points of Kant's reasoning can be summarized as follows: Given the meaning of ens realissimum,2 it is impossible for anything that fits this description to have any qualities that another being fitting this description does not have - that is, every ens realissimum is identical to every other in every formal respect. Therefore, ‘any absolutely necessary being is an ens realissimum’ entails ‘any ens realissimum is an absolutely necessary being’. But since the truth of this statement is determined a prioriby the meaningof the term ‘ens realissimum,’ for the cosmological argument to implicitly rely on it is for it to rely on the ontological argument; for that argument, in its most basic structure, is this generation of the claim ‘God exists’ from the concept of an ens realissimum. 6) Therefore, the cosmological arguer can only deny the probative force3 of the ontological proof at her own expense.
Why reopen the question?
Aquinas is commonly extricated from Kant's critique by pointing out certain differences between the version of the critique that Kant attacks and that which Aquinas advances. On Kant's side, it has been pointed out since Russell that the version of the cosmological argument that Kant critiques is not Aquinas', but is closer to those advanced by Leibniz and Wolff.4
Concomitantly, many commentators have pointed out that the notion of necessity used by Aquinas in his five ways is different from that used by Kant. This point has been established by Peter Geach,5Patterson Brown,6 Anthony Kenny,7 and William Rowe, among others, the last of whom writes the following: ‘Aquinas does not mean [by ‘necessary being’] a being whose non-existence is a logical impossibility. Instead, he means a being in which there is no capacity for generation and no capacity for corruption.’ (Rowe (1975), 40)
This point is presumably meant to block Aquinas from Kant's critique in the following way: Kant's proof of DT relies crucially on the simple convertibility of ‘every absolutely necessary being is an ens realissimum’. But not only is this proposition not convertible simply on Aquinas' account: it is not even true, since other beings (e.g. angels and human souls) also necessarily exist. Therefore, though other arguments are susceptible to the DT, Aquinas' five ways do not fall within its scope.
But this is not a sufficient defence of Aquinas. To make this clear, let us leave aside arguments in Aquinas’ five ways that are perhaps more covertly similar to Kant’s description and focus on the argument made by Aquinas that most explicitly resembles Kant’s reconstruction:8 Aquinas’ third way, the second half of which reads as follows:
Every necessary being does or does not have the cause of its necessity from elsewhere. But it is not possible to proceed in infinitum among things that have a cause of their necessity, just as is the case with efficient causes, as was proven. Therefore, it is necessary to posit something that is per se necessary, not having the cause of its necessity from elsewhere, but itself being the cause of the necessity of other [necessary beings]: this all call God. (ST I. ii. 3, co)9
Here, ‘necessary being’ is not simply convertible with ‘highest being’. But this is largely irrelevant to Kant's critique: Kant's conversion is not one between ‘necessary being’ and ‘highest being,’ but one between ‘absolutely necessary being’ and ‘highest being’. And Aquinas has, if not exactly this, a very close analogue in his notion of a per se necessary being, which is convertible with ‘highest being’ - and this seems to situate Aquinas' third way directly in the scope of Kant's dependency thesis. As Forgie rightly points out, ‘We cannot free Aquinas from the DT by pointing out that he does not use a vulnerable notion of necessary being if in the end he uses the notion of uncaused necessary being in a vulnerable way.’ (Forgie (1995), 93).
I take the above to establish two points: first, that one cannot assume that Aquinas' cosmological proofs escape the scope of Kant's dependency thesis on the grounds that they are in many respects different from the Leibniz-Wolff token of the argument that Kant takes as his paradigm, since they have not been shown to be relevantly dissimilar; and second, that therefore, modern adaptations of Aquinas' natural theology that have thought themselves justified in leaving behind Kant on these grounds are in fact not so justified. Thus, the applicability of Kant's dependency thesis to Aquinas' five ways remains, at this point in our discussion, an open question.
In sum, the literature thus far attempts to extricate Aquinas from the DT by means of an evasion thesis: while Kant's dependency thesis remains a valid critique of certain cosmological arguments, Aquinas' happen to fall outside of its scope.
Forgie himself has advanced an alternative defence of Aquinas against the DT, which aims not at evading, but rejecting it. Forgie takes Aquinas to hold that ‘God exists’ is not an a prioritruth, which he takes from Aquinas' rejection of the ontological argument at ST I. ii. 1.10 Forgie then combines this with Aquinas’ statement that ‘God exists’ is per se notumsecundum se - which he interprets as ‘necessarily true’ - to argue that ‘“God exists” is a necessary, but a posteriori truth. It is self-evident in itself, i.e., necessarily true, but not self-evident to us, i.e., not something whose truth can be seen simply by understanding it.’ (ibid., 99).
Forgie's adopts this reading as a modification of a position taken up by Alvin Plantinga, who interprets Aquinas' distinction between per se nota quoad nos and in se propositions as one between a) propositions the truth of which is immediately knowable upon understanding, and b) necessary truths.11 But Plantinga also holds that ‘[t]he name ‘God’ in the sentence ‘God exists’ is not a proper name of God and does not serve to express the essence of God. Rather, that word is short for some such definite description as ‘the Creator and sustainer of all there is’ or perhaps ‘the first cause’ or 'the unmoved mover’’ (Plantinga (1975), 136). Consequently, the proposition uttered by us when we say ‘God exists’ is a contingent truth - that is, it is contingent on His being the thing predicated of Him under the disguised description ‘God’, which He in every case could not have been had He chosen not to create. On the other hand, there is a different proposition ‘God exists,’ in which the name ‘God’ serves as a proper name. This proposition is necessarily true, but unknowable to us.12
Forgie's critique of Plantinga's account,13 which I take to be sound, can be sketched as follows: 1) Aquinas holds that ‘God exists’ is a) self-evident in itself b) not self-evident to us, and c) demonstrable a posteriori - that is, on Plantinga's reading, it is a’) necessarily true, b’) not immediately graspable by the intellect, and c) demonstrable a posteriori. 2) But on this account, any form of the proposition ‘God exists’ that we can understand fails to satisfy the first criterion. Therefore, there is no proposition that we can understand that meets Aquinas' requirement of being self-evident in itself but not to us. 3) On the other hand, the proposition ‘God exists’ that is self-evident in itself - i.e. necessarily true - is unknowable. Therefore it is indemonstrable, and consequently not demonstrable a posteriori. Therefore, this proposition also fails to satisfy the three requirements that Aquinas thinks hold of some one proposition ‘God exists’. 4) Therefore, on Plantinga's account, there is, in fact, no one proposition that satisfies Aquinas description of the proposition ‘God exists’. 5) But on a straightforward reading of Aquinas,14 there is some such proposition. C) Therefore, Plantinga's reading of Aquinas on this point is likely mistaken.
After refuting Plantinga's reading of Aquinas, Forgie puts forward an alternative interpretation of the proposition ‘God exists’ as Aquinas understands it:
Suppose we think of proper names not as having 'senses' or as 'expressing' the essence, or any property of their bearers, and think of them instead more as Kripkean 'rigid designators', or in the way Mill thought of proper names - as terms which simply denote but lack connotation. Then we can regard the subject term 'God,' in 'God exists', as a proper name, not a disguised description, without requiring a grasp of the essence of God for the proposition to be intelligible. (Forgie (1995), 98-99)
For some time, two things puzzled me about Forgie's argument up to this point. First, I did not understand the pertinence of the critique of Plantinga's exegesis of Aquinas to the question of how Aquinas is to be extricated from Kant's critique of the cosmological argument. Second, though Forgie does refute certain alternative positions attempting to defend Aquinas from Kant, he does not provide any direct support to the thesis in the paragraph quoted directly above: he neither a) puts forth reasons that show that the thesis successfully diffuses the DT, and does not merely disagree with it; nor b) brings forth additional texts to confirm that the thesis is in fact held by Aquinas.15 Consequently, his thesis is expressed, towards the end of the argument, in the following conditional: ‘If there is no inconsistency in Aquinas' views then there must be something wrong with the DT.’ (ibid, 99). This, however, is just the contrapositive of a thesis that any prospective opponent of Forgie's would assent to, i.e. that if there is not something wrong with the DT, then Aquinas' position is inconsistent.
Apology for Forgie
Retrospectively, I would like to explain what I now take to be the unity and weight of Forgie’s argument. Though what I am about to say is not explicit in Forgie's paper, I take it to supply the implicit motive behind the critique of Plantinga, and shall allude to evidence in support of this position where necessary.
Even apart from the question of whether Plantinga provides an adequate account of Aquinas' position, Plantinga's analysis of the proposition(s) ‘God exists’, thinks Forgie, is essentially incompatible with the thesis that ‘God exists’ is an a posterioritruth. This is because, in holding the name ‘God’ to be a disguised definite description, Plantinga has in advance set up the task of proving God's existence as being, to a large extent, engaged in discerning the meaning(s) of the name ‘God’. But meaning analysis is precisely the provenance of the a priori, and is only indirectly relevant to the a posterioritask of discerning matters of fact about the world. To put this another way, Forgie thinks that Plantinga's analysis relegates any possible theistic proof to the realm of sense when it should be engaged in fixing reference. Hence, he writes that Plantinga's descriptions ‘[only] provide the … meaning for 'God' in any intelligible proposition 'God exists’’ whereas these meanings should themselves ‘serve, in Kripke's idiom, to fix the reference of the name 'God'‘ (Forgie (1995), 98-99) - a service into which they cannot be pressed in Plantinga's own system on account of the depth of his apophaticism.16 This also explains why he mentions and criticizes Plantinga's comparing the truth of the proposition ‘God exists’ to that of Goldbach's conjecture, since this latter is, if true, notan a posterioritruth, but rather a synthetic a prioritruth. Forgie's point here seems to be that on Plantinga's account, ‘God exists’ likely would be a synthetic a priori truth, whereas ‘for Aquinas, 'God exists' is not an a priori truth at all’ (Forgie (1995), 99).
Now, thinks Forgie, since 1) Aquinas says that ‘God exists’ is known a posteriori; and 2) an a posterioritruth is one about objects that can, at least in principle, be named, quantified over, put into a set, etc., it follows that 3) Aquinas' proofs must - like any a posteriori proof for the existence of God - be engaged in the task of fixing the reference of the name ‘God’. 4) Thus, though there are likely some differences, Aquinas' proofs must have been engaged in the same reference-fixing kind of work that today is best aided by recourse to the apparatus of Kripke's Naming and Necessity.
Read this way, Forgie's conclusion is not meant to tie every loose end so much as to be a kind of instauration for research in natural theology. Given this, the arguments against Plantinga would not be construed as definitively entailing a Kripkean approach; but a step in that direction would likely be viewed as both the most natural and the most logical. Forgie writes: ‘We need an account in which the proposition 'God exists' satisfies all the conditions (a), (b) and (c).’17 The implication, if I am reading Forgie correctly, is that Kripke's account of names as rigid designators can do this, whereas no other well-known account can.
The argument of Forgie's paper, then, can be summarized as follows:
If there is valid a posteriori proof for the existence of God (e.g. Aquinas’ third way), then it either merely evades the DT or rejects it. (Proofemp) (ET v ~DT)
~ET (established in section I of Forgie's paper)
Furthermore, if there is a valid a posteriori proof for the existence of God , then the term ‘God’ in the proposition ‘God exists’ must be either treated as a disguised definite description or as a rigid designator. (Proofemp) (Plantinga v Kripke)
If it is treated exclusively as a disguised description, then the dependency thesis remains valid (established in section II). (Plantinga DT)
Con. Therefore, if any of Aquinas' a posterioritheistic proofs are valid - that is, if he can establish that ‘God exists’ is necessary a posteriori - then the term ‘God’ must be treated as a rigid designator, and the dependency thesis must be rejected (Proofemp) (Kripke & ~DT)18
Forgie's conclusion is ambiguous insofar as it attempts to both a) lay the groundwork for further systematic work in natural theology, and b) establish an interpretation of Aquinas. The following section shall consist in a critique of third of the above premises. Specifically, I will show that the dichotomy in the consequent it fails to encompass Aquinas’ actual position, thereby depriving the conditional of its pragmatic force.19 In other words, I shall undermine Forgie's philosophical hypothesis by showing his interpretive hypothesis to be mistaken.
Critique of Forgie's account of per se nota propositions.
Forgie's dilemma between Kripkean and Plantingan analyses of the term ‘God’ in the proposition ‘God exists’ arises from a) his gloss of Aquinas' per se notum secundum se/quoad nos distinction; and b) his understanding of the a priori/a posteriori distinction: ‘God exists’ must be necessary, but not a priori; if a proof is conducted in Plantinga's manner then it is a priori; this leaves a Kripkean analysis as the only, best, or most obvious means by which to ensure that ‘God exists’ be an a posteriori truth.
A minor objection to Forgie's gloss is that Aquinas alreadyhas an a priori/a posteriori distinction, as well as a distinction between necessary and contingent truths. Given this, the lexical items ‘per se notum quoad nos’ and ‘per se notum secundum se’ would seem to serve no unique purpose in Aquinas' language. This superfluity would add unnecessary confusion to the expositions of an otherwise generally lucid and clear writer: in other words, Aquinas could have simply said that ‘God exists’ is necessary and a posteriori instead of developing additional vocabulary that could potentially mislead his reader.
But more decisively, Forgie's interpretation of Aquinas schema of per se nota truths is inconsistent with Aquinas’ own remarks on the subject. To show this, let us compare Forgie's account with Aquinas' own. Aquinas writes:
Something's being self-evident occurs in two ways: in one way, in itself and not to us; in the other way, in itself and to us. From this some proposition is per se nota: that its predicate is included in the meaningof the subject ... If, then, the quiddity of the predicate and subject is understood by all, the proposition [that combines them] will be per se nota to everyone … But if the quiddity of the subject and predicate is not known to some, then that proposition will be per se nota in itself, but not among those who do not understand the subject and predicate of the proposition. And therefore it happens, as Boethius states in his book de hebdomadibus, that certain conceptions of the soul are common and per se nota only among the wise. (ST I. ii. 1, co)20
Now in the first sentence of the above, Aquinas makes it clear that every per se nota proposition is per se nota secundum se. From here, this group of all per se notae propositions is divided into those that a) are and b) are not per se notae quoad nos. To put this into Aquinas' own parlance, we might say that the relationship between secundum se and quoad nos per se notae propositions is one of genus to difference. This means, in principle, that Aquinas' larger category of per se notae secundum se propositions is divided by one easily distinguishable factor into two distinct groups. Call this the genus-difference criterion (GDC).
Furthermore, Aquinas delimits per se notae secundum se propositions as those having the following mark: the predicate is included in the meaningof the subject. Let us call this the subject-predicate containment criterion, or the containment criterion (CC) for short.
Forgie's Aquinas has problems with meeting both the GDC and the CC.
Concerning the former, Forgie construes per se nota secundum se as ‘necessary truth’. But he is ambivalent regarding the demarcation the sphere of per se notum quoad nos propositions. Recall that he writes that ‘'God exists' is a necessary, but a posteriori truth. It is self-evident in itself, i.e., necessarily true, but not self-evident to us, i.e., not something whose truth can be seen simply by understanding it.’ (Forgie (1995), 99). Here, the parallelism in the first half of the two sentences can be interpreted as implying or not implying parallelism in the latter half. If the implication is there, ‘not self-evident to us’ would be coextensive with ‘a posteriori’. But this cannot be the case, because Forgie presumably does not think that all a posterioritruths are necessary, as would be the case if ‘necessary’ were a genus of ‘a posteriori’. If, on the other hand, ‘not self-evident to us’ is not coextensive with ‘a posteriori,’ but perhaps includes, for instance, certain synthetic a prioritruths as well, then the unity of the category as a whole would be left vague. Depending on how this unity is construed, Forgie's Aquinas might still be unable to adequately account for the existence of both necessary and contingent a posterioritruths.
Furthermore, by virtue of the CC, the necessarily true propositions must be those wherein the predicate is included in the meaning of the subject. Therefore, Forgie cannot both hold that ‘God exists’ is a necessary truth, and that ‘God’ in the a posteriori, necessarily true proposition ‘God exists’ is a rigid designator - that is, it is a term ‘which simply denote[s], but lack[s] connotation’ (ibid., 98). On account of the former, it follows that the predicate ‘exists’ must be contained in the meaning of ‘God’. But on account of the latter thesis, the name ‘God,’ must be literally meaning-less: it does not contain within it any sense or property, and consequently, does not contain within it the sense or property ‘exists’. Therefore, ‘God exists’ both would and would not be a necessary truth.
Perhaps Forgie can avoid this problem by holding that it is not ‘God’ that must contain within itself the predicate ‘exists,’21 but God, who contains the predicate ‘existence’ by actually existing. More generally, a subject (b) would be said to contain its predicate (F) simply by being F. But this account cannot work either, since the propositions that meet the containment criterion must be coextensive with the necessary truths. If, then, for some subject (b) to contain the predicate (F) is interpreted simply as (Fb), then it follows that all truths are necessary truths. But Aquinas clearly intends that the propositions that are per se notae secundum se are only a subset of all true propositions. Therefore, Forgie's reading on the above points must be mistaken.