The ontological argument from Anselm to Kant

The OA first appears in Anselm's Proslogion, in which Anselm argues on the basis of his definition of God as id quo maius cogitari nequit, ‘that than which nothing greater can be thought’ (IQM). The demonstration, reconstructed according to the conceptual (or hyperintensional) formulation of the ontological argument (Oppy (Reference Oppy1995) ), proceeds as follows:

Anselm's argument was later revived by Descartes, who was to give a set of interconnected arguments, rather than a unique proof. For our purposes, however, it is sufficient to consider two. In the first, Descartes starts from two assumptions: that we all have the idea of God; and that God can be defined as an infinite substance. As I do not possess any of the perfections that are represented in this idea – the perfections of being eternal, omniscient, omnipotent, etc. – I cannot have created it, since the cause of an idea requires at least the same amount of perfection as the represented idea. Therefore, the cause of an idea of an infinite substance can only be an infinite substance. In the second argument, Descartes draws on the analogy of a geometrical demonstration. If I have a genuine concept of a triangle, I can infer that a triangle's internal angles add up to 180 degrees, such an inference holding whether or not any actual triangle exists in the world. Similarly, I can clearly and distinctly conceive God as a supremely perfect being, and so as having every perfection. From this idea I can infer various facts about God, just as I can about triangles – facts that hold whether or not God actually exists. I can infer, for example, that God is omniscient, and that he is omnipotent, for I would contradict myself if I thought that a being with every perfection lacked a perfection. But existence itself is a perfection; therefore, unlike the case of the triangle, I can infer that God exists.

In turn, Leibniz offered a rectification of Descartes's proof, arguing that it was first necessary to show that the supremely perfect being of Descartes's proof is possible – that is, that the idea of such a perfect being is coherent. This, he thought, could be done, because the concept of perfection cannot be analysed, and therefore it cannot be demonstrated that perfections are contradictory or incompatible. Once it is shown that there can be an entity having all the perfections, Descartes's argument, according to Leibniz, becomes valid (Leibniz (Reference Leibniz, Remnant and Bennett1996), 437–439).

In the eighteenth century, however, Kant offered a refutation of the OA that was destined to have an impact on all further philosophical discussions, and specifically, an impact on Hegel. Existence, he argued, should not be understood as a property among others, something represented by a feature that forms part of the concept of the object in question (Kant (Reference Kant and Smith2007), A592–603/B620–631); rather, the existing object is to be found outside thought and therefore must be able to be intuited in space and time. That existence can be ascertained only empirically, not deduced intellectually, is shown in the example of the hundred thalers: ‘A hundred real thalers do not contain the least coin more than a hundred possible thalers. … My financial position is, however, affected very differently by a hundred real thalers than it is by the mere concept of them (that is, of their possibility).’ It follows that the demonstration of the existence of something (God) by predicating a property of that object is fallacious.

We now might consider Hegel's position in relation to this history of the OA. Hegel was clearly attracted to the Anselmian formulation of the OA, and this can be explained, we suggest, by the presence of something characteristically ancient that coexisted with something peculiarly modern in Anselm's argument. Anselm wanted to capture the unity of thought and being (the IQM cannot exist only in the mind, so it must exist in reality), and this aspect of the proof harks back to that which can be considered as the default position in ancient thought. For example, Aristotle's idea of how the mind comes to have thoughts of things effectively presupposes the reality of the things – effectively, corporealized forms – that allow our thoughts to be thoughts. This realist complexion of Aristotle's logical and epistemological thought, which relies on a fundamentally extra-mental conception of forms, is in contrast to the general direction of modern thought, in which such forms become essentially subjective entities – 'ideas’ in the modern sense.

This unity of being and thought as presupposed by the extra-mental existence of forms for Aristotle is not presupposed in the modern OA; the starting point of the argument is a purportedly determinate concept of God understood as a possible existent, and only then does the argument proceed to the actuality of God on the basis of what can be derived from features of that initial concept. Hegel points out that such a type of proof which starts with a concept and tries to infer to being was unknown to classical thought, and that it starts off from a separation of thought and being, which was destined to become most obvious later in the early modern period. This separation implies an advanced stage of thought (Hegel (Reference Hegel, Hodgson, Brown, Hodgson, Stewart, Fitzer and Harris1996), 434): the idea that we have ‘thinkables’ that are independent of the question of whether what is thought in them exists or not is an important achievement in contrast to the ancient-Aristotelian ‘realist’ approach. This stage of thought is necessary and problematic at the same time.

Let us consider why it is necessary first. Hegel claims that the unity between being and thought needs to be demonstrated (Hegel (Reference Hegel, Geraets, Suchting and Harris1991), § 193 remark), and not simply presupposed. Now, a demonstration implies judging (nothing can be ‘demonstrated’ if no judgments are used), and judging implies a separation. And the separation is precisely what this stage of thought provides. In fact, this stage of thought is what would be nowadays called a form of ‘representationalism’ – that is, the view that the immediate object of knowledge is an idea in the mind distinct or separated from the external object which is the occasion of perception. From this angle, the OA, in its Anselmian form, is made up of a mixture of the ancient ‘unity’ conception and the modern ‘separation’ conception of the relation of thought and being. The separation (the distinction between the conceptual realm and the realm of being that the OA subsequently attempts to bridge) is there, but only as presupposed; and, as such, it cannot, in Hegel's view, work. To demonstrate the unity between being and thought, a more advanced stage of thought is necessary, one that develops the idea that concepts can have determinate natures independently of any questions concerning the existence of that which the concepts represent.

Nevertheless, this stage of thought, albeit necessary, is also problematic. Descartes, Hegel says, needs to be praised for his ‘sublimest thought, that God is that whose concept [Begriff] includes within itself its being’; however, ‘the defect in Anselm's argumentation’ (the presupposition of the distinction between being and thought, conceived as being independent from thought, rather than posited by it) is also present in Descartes and other modern philosophers who employ the ontological argument (ibid., § 193 remark). The core of Descartes's ‘geometrical’ version of the ontological argument is that it is possible to infer various proprieties of God, whether or not God actually exists. While this idea was already implicit in the Anselmian version of the argument, it was Descartes who made explicit this type of ‘representationalist’ account of thought, in which the mind is conceived of knowing itself (that is, its own subjective states or processes conceived of as ‘ideas’ or ‘thoughts’) first, thus generating the familiar sceptical questions about the reality of the apparent world ‘external’ to the mind. The conception of thought as the realm of possibilities had become more radical with Leibniz. For Leibniz, possibility is simply the absence of contradiction. Thus the existence of God becomes, in Hegel's words, ‘only an inference from eternal truths’, and thus we can again have the ‘wearisome [langweilige] proof of His existence’ consisting in the inference that ‘He has the prerogative of existing immediately in His potentiality’ (Hegel (Reference Hegel, Haldane and Simson1995), 339).

Against this background, we can now understand why Hegel's position in relation to Kant's refutation of the OA was ambivalent. Kant was right to the extent that his refutation captured the aspect of the proof that represents the most significant flaw of the OA, namely, the distinctly modern separation between thought and being. It is in its ‘faulty form’, Hegel says, that the OA ‘finally succumbed to the Critique of Reason and to the thought that existence cannot be extracted from the notion’ (Hegel (Reference Hegel and Miller1969a), 705).Footnote ² As already stressed, the distinction between thought and being is regarded by Hegel himself as an achievement. However, Kant was wrong, in Hegel's view, to conclude that ‘being’ can only be ascertained empirically. Here Hegel is critical of the conception of concept presupposed in Kant's critique of the OA. ‘In ordinary life’, he notes, ‘we do indeed call a representation [Vorstellung] of a hundred dollars a concept [Begriff]. It is no concept, however, but only a content-determination of my consciousness; an abstract simple representation [Vorstellung] …, or a determinacy of the understanding that is within my head, can of course lack being’ (Hegel (Reference Hegel, Hodgson, Brown, Hodgson, Stewart, Fitzer and Harris1996), 436). In other words, Kant's critique applies only to what we might call ‘concept-representations’ – mental representations that may be applied to external objects – not to proper concepts (Begriffe) as Hegel understands them. As a consequence of this, we argue, Hegel is critical of the conception of the relation of thought to being that follows from this inadequate conception of concept.

Hegel's most extensive engagement with this issue is found in his comments on the Critical Philosophy in the Encyclopaedia Logic. Here it is remarked that the elevation ‘out of the empirical worldview to God … has the thinking consideration of the world as its only foundation, not the merely sensory one that we have in common with the animals’ (Hegel (Reference Hegel, Geraets, Suchting and Harris1991), § 50 remark). This means, Hegel explains, that when we speak of God, we are dealing with an object ‘of quite another kind than one hundred dollars, or any other particular concept [Begriff], notion [Vorstellung], or whatever other name you want to give it’. Then Hegel concludes: ‘God has to be expressly that which can only be “thought as existing”, where the concept includes being within itself. It is this unity of the concept and being that constitutes the concept of God’ (ibid., § 50 remark).Footnote ³ But is Hegel's line of reasoning plausible? Was not the unity of thought and being a naïve assumption of ancient thought, finally overcome with the modern separation of the two realms? Does not Hegel's argument, as with the more traditional proofs, still take us no further than the concept of God?

This objection is legitimate and, in fact, acknowledged by Hegel himself. Let us go back to Anselm's argument. The outcome of the proof is the unity of thought and being, but the starting point is a concept of God understood as a possible existent. Kant's refutation is effectively a consequence of the modern viewpoint, namely, that which starts from the separation between thought and being: he shows that implicit presupposition, and then claims that the proof fails. Once again, Kant is right in showing that presupposition, which is clearly there in Anselm's proof and in its modern variations. ‘Since Anselm's day’, Hegel comments, ‘we have come no further in any respect’. As such, the OA does not take us further than the concept of God, as ‘to presuppose means to accept something immediately as primary and unproved’ (Hegel (Reference Hegel, Hodgson, Brown, Hodgson, Stewart, Fitzer and Harris1996), 440). However, Hegel continues, ‘The defective feature is the fact that this is a presupposition and therefore something immediate, and so one does not recognize the necessity of this unity’ (ibid.). In other words, the argument could work, according to Hegel – provided that one were able to demonstrate the unity of being and thought, namely, the transition of the absolute concept into existence.

This may look like a desperate move, as Kant's understanding of conceptuality is generally recognized to be a sophisticated one. For him concepts are no longer thought of as the pale remnants of perceptions, as in much seventeenth- and eighteenth-century philosophy. Rather, they are understood in terms of the role they play in a normatively conceived cognitive life. In theoretical contexts, they are general representations applicable to objects in judgements capable of truth or falsity, judgements made on the basis of those objects being made present in the empirical ‘intuitions’ of experience (Kant (Reference Kant and Smith2007), A320/B376–377). From his perspective, what right do we have to speak freely of objects about which we could have no sense as to what could differentiate speaking truly of them from speaking falsely? In his critique of Kant's critique of the OA, Hegel appeals to a different understanding of the nature of the concept, but one for which he purports to find the essentials within Kant's own philosophy.

Conceptions of concept

Kant had taken the application of particular empirical concepts in experience to be dependent on the default application of what he called ‘pure’ concepts, the possession of which was included among the necessary conditions of experience. These were pure concepts of the understanding (categories) and of reason (ideas). It is within the latter (which ‘transcend the possibility of experience’ (ibid., A320/B377) ) that Hegel finds a conception, that of the ‘I’ of the ‘I think’, the ‘transcendental unity of apperception’, that he will take as a model for what he refers to as ‘the concept’ (der Begriff) and that will provide the model for the concept of God. Thus, Hegel states that ‘it is one of the profoundest and truest insights to be found in the Critique of Pure Reason that the unity which constitutes the nature of the concept [Begriff] is recognized as the original synthetic unity of apperception, as unity of the I think, or self-consciousness’ (Hegel (Reference Hegel, Moldenhauer and Michel1969b), 584). To avoid confusion we will henceforth leave Hegel's term Begriff untranslated, and use ‘concept-representation’ to refer to those empirical concepts, such as the concept ‘thaler’.Footnote ⁴ We ask: might not a version of the OA using something closer to the transcendental ‘I’ as a model for the Begriff ‘God’ prove more satisfactory, according to Hegel's thought, than one presupposing a concept-representation? Kant himself in his criticism of the OA presupposes that the concept of God functions as a concept-representation, but if Kant's concept of ‘I’ is itself not able to be understood along the lines of a concept-representation, then he has implicitly provided an alternative way to conceive of God to that presupposed in his own critique of the OA. Let us work through this more carefully.

Kant had introduced the idea that a finite rational being must have a certain conception of him or herself as a rational being. I need to grasp myself in terms of a representation (Vorstellung) – 'I’ – that ‘must be capable of accompanying all other representations, and which in all consciousness is one and the same’ (Kant (Reference Kant and Smith2007), B 132, see also B 407–408). Where there is reasoning there must be a presupposed unity of consciousness. If I believe ‘p’ and you believe ‘not p’ we have disagreement, not irrationality, as we both might have good reasons to believe what we do, even though one of us is wrong. But if I simultaneously believe both ‘p’ and ‘not p’ it is an altogether different matter. To grasp myself as having the requisite unity requires a concept that I can apply to myself. This is the concept ‘I’. We can now ask after the general features of this concept, and how it relates to empirically applicable ‘concept representations’ such as ‘thaler’.

According to Kant, when I apply the concept ‘thaler’ in a judgement such as ‘there is a thaler in the numismatist's window’ I must also, in some way, be applying the concept ‘I’ to myself as the unified subject making that judgement. We can, of course, in a descriptive sense, hold incompatible beliefs, but it is a requirement of rationality that we do not consciously hold them. There indeed may be individuals with the psychological condition once known as ‘multiple personality disorder’ who seem to harbour different ‘minds’ within the one body. Speaking as one personality or ‘alter’, such a person might assert that p, and, speaking as another, assert that not p. But, as in many such cases, the pathology here points to an important feature of the normal. In the pathological case there is, seemingly, no grasp of the self as a unity – a grasp that is otherwise a feature of the normal. This grasp, Kant thinks, is a necessarily conceptual one, and the ‘I’ is the concept in question.Footnote ⁵ Without it there is nothing to ensure that my spatio-temporally unified body harbours multiple cognitive subjects with incommensurable beliefs. And in judging I must apply this concept to myself. The question will now be whether one can understand one's application of ‘I’ to oneself on the model of the application of the empirical concept-representation to some object. Is ‘I’ just another concept-representation, as Kant's reference to it as a Vorstellung might seem to suggest?

At first sight there seems to be an obvious problem looming for the affirmative answer. My application of a concept-representation to an object, referring to the thing in the numismatist's window as a thaler, for example, in the context of making a judgement about it, is said to presuppose my application of ‘I’ to myself. But if ‘I’ is a concept-representation, then will not this act already presuppose the application of ‘I’ to the thinker? And does not a type of regress threaten?

Some clarification might come from thinking of ‘judging’ here as analogous to asserting, and from examining the role of the pronoun ‘I’ in this latter context. Thus Wittgenstein pointed to two different uses of the ‘I’ – we might call them object-determining and subject-expressing uses (Wittgenstein (Reference Wittgenstein1958), 66–67). On the one hand, ‘I’ can be used like a proper name to which empirical concepts (concept-representations) are predicated to state some empirical fact. We can think of someone – Arthur – saying of himself, for example, ‘I have a bump on my forehead’, and others able to pass on this information by saying ‘Arthur has a bump on his forehead’. But when a speaker applies ‘I’ to him or herself as the maker of an assertion to another, the ‘I’ does not work in this way. When I inform you, say, that the value of the Euro weakened overnight, I am not primarily informing you of my belief or conviction that it is the case that the Euro weakened overnight. I am purporting to tell you something about the world – in this case, the value of the Euro – not about my current beliefs about it. Furthermore, unlike a proper name, the ‘I’ is not fixed in its designation, being swapped between speakers in conversation. I address you, but when you answer, you become the ‘I’ addressing me, that is, your ‘you’. For the subject-expressing use, the idea of the ‘I’ as like a proper name is clearly misleading (Anscombe (Reference Anscombe1981) ),Footnote ⁶ and yet the object-specifying use has clear parallels with the use of a proper name, as the example of Arthur shows.

One way of capturing the difference between these uses of ‘I’ is to say that in the subject-expressing use, the ‘I’, rather than functioning to designate the bearer of some empirical property, signals a claim to some normative status.Footnote ⁷ From the perspective of the hearer of an assertion, what is important is whether or not the world is as the speaker asserts it to be – whether what he is told is true or false. Moreover, to evaluate another's acts is to construe that other as responsible for them – to construe her as an autonomous agent. But at the same time, in its object-designating function, the speaker's use of ‘I’ also allows hearers to learn empirical facts about the speaker – the assertion allows the hearer to learn what it is that the speaker believes despite the fact that it is not primarily about conveying that information.Footnote ⁸ Perhaps these aspects of the linguistic use of ‘I’ might now help to shed light on some of the features implicit, but unacknowledged, in Kant's conception of the transcendental I, and in turn help us to differentiate features of the Hegelian Begriff from the ‘concept-representation’.

The first thing to note is that with the subject-expressing use of ‘I’ there can be no clear separation of issues of existence from those of conceptual content – the separation that is found in empirical concept-representations such as ‘thaler’. Empirical concepts can be misapplied, giving false judgements. I can mistake an old French franc for a German thaler and mislead you by telling you I saw a thaler in the numismatist's window. But the ‘I’ in its performative subject-expressing role doesn't function in the way that fallible predications are being made with it.Footnote ⁹ When identifying myself with the ‘I think’ of an assertion, I am purporting to hold myself to the norms governing the use of the concepts I employ in making the assertion – I am construing myself as a free agent who can be taken to be responsible for them. In this sense, I can only be an ‘I’ by implicitly applying and holding myself to the concept ‘I’, and, at least within certain limits, I cannot fail to be an ‘I’ when using the concept of myself in this way. A relevant analogy here is the Austinian ‘performative’ such as the speech act of the marriage celebrant via which the couple come to instantiate the concept ‘married couple’ simply in virtue of the celebrant pronouncing the couple to be a married couple in the appropriate circumstance (Austin (Reference Austin1962) ). But in the case of the ‘I’, the speaker is at the same time the object of the pronouncement.

Kant seems to allude to something like this peculiar functioning of ‘I’ when he refers to it as ‘an act of spontaneity’ (Kant (Reference Kant and Smith2007), B132). I may, of course, fail properly to live up to the norms associated with I-hood, but when I so fail I am normally held accountable for not living up to them, not excused on the basis that I wasn't really an ‘I’. But there is a further sense in which the application of ‘I’ differs from that of a name used with an empirical concept-representation in that the misapplication of the ‘I’ is impossible. Suffering from amnesia or some delusion, say, a person might not be able to identify herself by name correctly. But if the use of the personal pronoun is intact, one can't go wrong: one can't mean someone else when one uses it because the rules of its use are simply that it refers to whoever utters it.

With these considerations in mind we might approach Kant's idea of the thinking of the concept ‘I’ as an ‘act of spontaneity’ by describing such an act as self-actualizing. That is, it is a concept whose expression establishes the normative status of that which it is meant to express, and does so in a way such that there can be no question of who it is that bears that status. As in the Austinian performative, there is a coincidence of concept (the use of the concept ‘I’ on the one hand, ‘married’ on the other) and being (the speaker's being an I, the couple's being married). In both cases, the coincidence of thought (concept) and being cannot be thought of as it is in knowledge, where the relationship is that of the truth of the concept to the object. In so far as this coincidence of concept and being is brought about by the exercise of the concept, the nature of such a concept is not independent of questions concerning the existence of that which it is a concept of. For some of the functions of the ‘I’, we might say that the distinction between being and thought is not independent of thought, but posited by thought. One can appreciate the advantage of starting with this model of concept when thinking of the ontological argument.

We might understand, then, how even in Kant's system the concept ‘I’ cannot simply be understood on the model of an empirical concept-representation, and that a possible way seems to have opened for thinking of this concept as providing a model for a way of conceiving God that will not fall prey to Kant's own critique of the OA. We may say that as long as the concept-representation is taken to be his default model of the concept, Kant's conception of thought remains broadly ‘representational’ despite the fact that Kant is critical of the idea that thoughts ‘represent’ the world ‘in itself’: rather than represent things as they truly are, thought strives to give objective representation to ‘appearances’.Footnote ¹⁰ In the Science of Logic, however, Hegel points to the peculiarity of this position.Footnote ¹¹ Kant seems to be a sceptic about metaphysics on the basis that metaphysical claims cannot be justified from experience in the way that empirical claims can. But why evaluate metaphysical claims on the measure of an ‘objectivity’ that is not true?

Very general meta-philosophical issues arise here concerning the relation between Hegel's and Kant's philosophies; we can do little more than locate ourselves within this contested field. On our understanding, Hegel is to be seen as generally ‘post-Kantian’ in that he accepts much of Kant's ‘Copernican revolution’ and tries to extend it by criticizing remnants of pre-Copernican thinking in Kant himself.Footnote ¹² One way of seeing this is to regard Hegel as extending the anti-representationalist dimensions of Kant's philosophy alluded to above. And this must, of course, affect the way in which the OA is to be understood since, as we have suggested above, it had been discussed from Anselm to Descartes mainly from within metaphysical frameworks that were fundamentally representationalist. But it is not simply a matter of the OA needing to be understood from within an anti-representationalist account of metaphysics; rather, it is the OA, we suggest, that foregrounds Hegel's strongly anti-representationalist general account of concepts as it makes explicit Hegel's new conception of ‘concept’.

The transcendental unity of apperception

We have suggested that with his idea of the transcendental unity of apperception Kant had shown the way beyond a representationalist account of concepts but had not built on this insight. Indeed, his criticism of the OA was on the basis of a more conventional representationalist idea of concepts. Generalizing Kant's novel approach to concepts found in the case of the ‘I’, Hegel believed he could thus undercut Kant's own objections to the OA. In the interval between Kant and Hegel it had been Fichte who had explicitly developed the concept of the ‘I’ which was left largely undeveloped in Kant; it was also Fichte who led the idealist purging of residual representationalist elements in Kant. We therefore briefly turn to elements of his account in order to bring into focus the sorts of consequences that such a development of Kant might have for the OA, once it is constructed within the new framework. We will subject this ‘Fichtean’ gloss to the internal critique that was part of Hegel's appropriation of Fichte, with the purpose of reconstructing how an Hegelian OA would look freed from the constraints of pre-Kantian representationalism.

In his influential 1794–1795 version of the ‘Doctrine of Science’ (Wissenschafteslehre) Fichte (Reference Fichte, Heath and Lachs1982) had sketched an account of what might now be discussed as the ‘intentional’ structure of consciousness based on Kant's notion of the ‘I think’. Fichte's account follows from Kant's comments about the spontaneity of the ‘I’ reviewed above, but it does so in ways that attempt to free the Kantian picture from Kant's fundamental distinction between ‘concepts’ and ‘intuitions’ – a distinction central to the model of the concept-representation. Fichte's first principle, the ‘Principle of Identity’, captures the necessary unity of the ‘I’ in Kant. To be rationally conscious one has to strive to be a rationally self-identical I=I. But Fichte gives this a significance beyond that found in Kant in that the I=I must be entirely self-positing or self-determining. Kant had thought of some aspect of the I's theoretical cognition – the content of empirical intuitions – as ‘given’ rather than determined by the subject, and as necessitated in a natural way. But with the idea that the I is rationally self-determining, Fichte had suggested that the I cannot be relieved of even this responsibility. Effectively Fichte's epistemological approach here represented a fallibilist approach to empirical knowledge: reflection on the particular conditions of experience can lead one to reinterpret what had initially been accepted with certainty as given. Thus, like Hegel after him, he rejected what is now commonly called the ‘Myth of the Given’: the idea that naturally conceived givens of experience could play a foundational role in the justification of knowledge claims.Footnote ¹³

Fichte's transformation of Kant may have been the launching pad for the German idealist movement, but a common criticism was that Fichte's account of the ‘I’, with its clear similarities to Descartes's conception of the ego, was still one-sidedly ‘subjective’ and that it failed to capture the objectivity belonging to a subject in virtue of that subject's embodiment and locatedness in the objective world – in the locution of the time, their belonging to ‘being’ (Sein). There is one way in which Fichte's ‘I’ was like that of Descartes that might be valuable for us, however, in evaluating the consequences of this ‘Fichtean’ turn in thought about the ‘I’ for the OA. Both Fichtean and Cartesian conceptions of the self include a necessary internal reference to a type of perfect version of itself. Recall that in Descartes's version of the OA there is reference to a conception of a God whose existence is required to explain how an imperfect self can have such an idea of a perfect one. In Fichte's case, the reference is more abstract. It is to the idea of the self-identical I=I, which any finite self must apply to itself in order to possess the cognitive capacities that it has – a self-identity that provides a goal for its own striving. This would appear to make Fichte a natural successor to Descartes with respect to the OA. However, with Fichte it is explicit that were the self-positing I=I to be regarded as somehow analogous to God, it would still in no real sense be separate from the finite I's own self-positing. For Fichte, on this basis, there seems to be no question of any separate existence of God at all, but this should not be surprising, as Fichte had inherited something of Kant's own conception of the idea of God as a necessary ‘postulate’ made by the self and required by it so that it better hold itself to the norms of its own reasoning – in Kant's case, specifically the norms of moral reason (Kant (Reference Kant and Gregor1997), bk II, ch. 1, sect. V). If a version of the OA were to be constructed from the perspective of Fichte's transformation of Kant, clearly its God would need to be more than a subjective posit.Footnote ¹⁴

Here we might relate Hegel's stance towards Fichte to those of his contemporaries at the Tübingen Stift – Hölderlin and Schelling – for despite his clear differences from the ‘romantic’ direction of their thought, he shared elements of their critique of Fichte. For Hölderlin and Schelling, Fichte's mistake had been to picture the finite I as understanding itself and its opposed objects as emerging from the ‘division’ of something that could itself be thought of as an I, the ‘I=I’, and at the same time, to think of that which is, ‘being’, as simply that which can be ideally presented to such an ‘I’ as determinately knowable by it. For Hölderlin, however, that from which the finite I and its knowable content (‘not I’) emerged was simply ‘being’ itself. To be an I capable of judgement (Urteil) presupposes a primal separation (Ur-Teilung) within ‘being’ (Hölderlin (Reference Hölderlin and Pfau1988), 37–38). A similar thought is found in the early Schelling (for example, Schelling (Reference Schelling, Harris and Heath1988), 10), and the general picture that was emerging was one in which the I was necessarily embodied and located in a much strong sense than that found in Fichte, complemented by a Spinozist non-anthropomorphized conception of God (that is, not treating God as an I) as the ground of human existence. Hegel was no friend of romanticism, but there is a sense in which he was as opposed to the problematic subjectivistic features of Fichte's I as the romantics had been themselves. And yet, we suggest, neither Hegel nor the romantics wanted to regress to any pre-Kantian conception of these issues. From Hegel's perspective, the problem was to rectify the problems found in both Kant and Fichte from within. Hegel's OA must, then, be able to be seen as utilizing the Kant–Fichte model of Begriff as based upon the transcendental unity of apperception or I=I, while subjecting this to a critique of its one-sidedly subjective character. Here we turn to the logical context in which these issues are played out in Hegel.

Concept, judgement, and inference in Hegel's logic

Peter Hodgson has written that Hegel's OA encompasses the entire Science of Logic, but here we will have to limit ourselves to some aspects of Hegel's logic that might bear on the more specific question we have formulated: that of what might become of any Fichte-inflected reading of the OA in Hegel's hands. Hegel accepted certain insights from the ‘romantic’ criticism of Fichte's development of Kant's notion of self-consciousness into the I=I, but he also rejected anything like the primordial concept of ‘being’ which Hölderlin had used to replace Fichte's I=I as designating that from which ‘I’ and ‘not I’ both emerged. We will briefly appeal to two sections of the Science of Logic relevant to these two critiques, starting with Hegel's examination of the ‘principle of identity’ and then turning to his earlier treatment of the ‘thought determination’, ‘Being’, from which the logic starts. We then move to Hegel's explicit treatment of der Begriff in book 3 of the Science of Logic.

Hegel's two-volume work starts with ‘Objective Logic’, which runs over the two books making up volume I. It is towards the beginning of book 2 of the ‘Objective Logic’, ‘The Doctrine of Essence’, that we find Hegel's critique of the ‘principle of identity’. Hegel considers the ‘law of identity’ ‘in its positive expression A=A’ as, ‘in the first instance, nothing more than the expression of an empty tautology’ that ‘has no content and leads no further’ (Hegel (Reference Hegel and Miller1969a), 413). Hegel notes that a person who set out to say something with the sentence beginning ‘The plant is –’ and who concluded with the predicate ‘a plant’ would say nothing, and thereby ‘contradict’ herself. A tautology is clearly not a formal contradiction, but it may be considered as contradictory at a pragmatic level when the normative act of asserting is considered: an act that purports to be an assertion but says nothing might be thought not to be an assertion at all. Significantly, Hegel adds that the concept ‘God’ should not be thought of as having this formal identity: ‘If anyone opens his mouth and promises to state what God is, namely God is – God, expectation is cheated, for what was expected was a different determination; and if this statement is absolute truth, such absolute verbiage is very lightly esteemed’ (ibid., 415). And so Hegel's criticism of the Fichtean analogue of the concept of God – Fichte's I=I – must surely also fall prey to this criticism. But if Hegel shares this type of criticism with his romantic contemporaries, he is equally critical of their own key assumptions.

For Hölderlin, human judgement (Urteil) presupposed a primal separation (Ur-Teilung) within ‘being’ such that the judging finite subject separates itself from being, which only then becomes for it a domain of objects for judgement. For Hegel, the pantheistic connotations of the romantic transformation of Fichte might have solved what could be considered the existence problem facing any Fichtean God. As the ground of the I's existence, and relieved of Fichte's anthropomorphism, the romantics’ pantheistic God would not have the status that the concept ‘God’ had for Kant and Fichte – that of a projected ideal of the finite self. And it seems hardly plausible to doubt the existence of that which is referred to as ‘being’, as it simply means that which is! But it is evident from the very beginning of Hegel's Science of Logic that there will be a problem looming for any attempt to equate God with ‘being’.

‘Being’ is the first thought determination considered within the first book of ‘Objective Logic’, ‘The Doctrine of Being’. The determinations are here meant to be understood primarily as structures of being itself, and only secondarily as structures of thought about being, an attitude found in Aristotle's Categories. One might think that the category ‘being’ provides a secure ontological starting point for capturing what is, as this makes no claims about the nature of what is other than that it simply is. But any apparent epistemological virtue here turns out to be a semantic vice, as in saying no more about what is other than the fact that it is, we fail to say anything at all. Such an indeterminate ‘being’ can have no features that separate it from its intuitive opposite, ‘nothing’, and so the thought of being cannot be held separate from, and so becomes, the thought of nothing. The Doctrine of Being, it would seem, is not the appropriate place in which to look for a thought determination adequate to God.

While ‘Objective Logic’ had expressed the categorial realist stance of ancient thought, ‘Subjective Logic’ starts with the expression of the contrastingly subjective orientation of modern thought. As we have seen, Hegel takes Kant's ‘spontaneous’ concept ‘I’ as a true model for concepts in general, and the details of this are developed in his treatment of concepts, judgements, and inferences in the Subjective Logic. There we see a relentless critique of the idea that judgement can ultimately be thought of as made up of a ‘subject’ term naming something in the objective world and a ‘predicate’ conceived of as a general concept-representation. As with Hölderlin, a judgement (Urteil) involves an original separation (Ur-Teilung) within a unity, but not the unity of ‘being’ but rather of der Begriff itself. This means that a judgement is not to be thought of as the application of a concept-representation to some given existent: a judgement will be thought of as a relation between two determinations of the concept, the three determinations of conceptuality being universality, particularity, and singularity (analogous to the structure of the Trinity (Bubbio (Reference Bubbio2014b) ). Aristotle in his syllogistic had only allowed the quantities of universality and particularity, while singular judgements had come to be added in the Middle Ages, raising the question of exactly how to incorporate them, the favoured solution being to class them with universal judgements.Footnote ¹⁵ Kant had in many ways returned to the Aristotelian position: concepts are necessarily general while objects qua singulars are represented by non-conceptual ‘intuitions’ (Anschauungen). Thus, for Kant, intuitions function in a way like singular terms, picking out worldly individuals somewhat in the way that names do (see, for example, Hintikka (Reference Hintikka, Penelhum and MacIntosh1969) ), but in the wake of Fichte's critique of the concept–intuition distinction, singularity re-emerges in Hegel's logic as a conceptual determination rather than as something non-conceptual and name-like. It is only in the judgement's initial and most immediate form and when taken as a representation that the subject term of the judgement operates, Hegel says, as a ‘kind of name’.Footnote ¹⁶

As has been suggested by Robert Brandom (see, for example, Brandom (Reference Brandom2002) ), Hegel rejects a ‘representationalist’ account of the content of judgements in which the judgement is thought to represent independent worldly objects or states of affairs. There is a holistic, ‘inferentialist’ dimension to Hegel's approach such that standing in inferential relations to other judgements is a necessary condition for any judgement to have a meaningful content. As Hegel puts it, the syllogism is the ‘truth of the judgment’ (Hegel (Reference Hegel and Miller1969a), 669). But in contrast to Brandom, we understand Hegel as a ‘weak’ rather than a ‘strong’ inferentialist, in that while necessary, the fact of a judgement's standing in such inferential relations is not sufficient for it to have content (Brandom (Reference Brandom2009), 8).Footnote ¹⁷ Hegel makes it clear that in everyday life we cannot avoid taking our concepts as functioning in this representational way: we assume that our subject terms name concrete existents and that our predicate terms express what we think about such existents. This sets up a tension between everyday life and philosophical reflection:

To ‘recognize the concept in … mere representation’ is to grasp that the judgement that was immediately taken to be about or unproblematically picture some self-sufficient concrete object or state of affairs can be reasoned about – for example, that it can function as an antecedent from which other judgements can be deduced, or can itself be seen as a deducible consequence from other judgements. This in turn can lead to a new understanding of what is presented in the judgement, as the judgement will no longer be simply thought to ‘picture’ isolated things or states of affairs. A judgement's connections to other judgements will be taken to reveal the connections among those things and states of affairs the judgements are purportedly about.Footnote ¹⁸ Nevertheless, as a ‘weak inferentialist’, Hegel relies on some initial purportedly ‘representational’ input from the world as is given in perception, with his ‘anti-representationalism’ taking the form of the demand that such seemingly atomic content be liable to some type of translation or ‘redetermination’ so as to give it a form adequate to its status as antecedent or consequent of an inference.Footnote ¹⁹

Thus Hegel's representations are akin to Kant's appearances in that they should not be taken as ‘representing’ the way the world is in itself, but unlike Kant's appearances they do not cover or hide from us a ‘real world’ of things in themselves as presupposed in the Kantian account. That is, remnants of the representationalist picture are retained in Kant in that Kant takes our cognitions to represent appearances rather than things in themselves. Recognizing conceptual structure in the representation means something quite different for Hegel. It allows the representation to be reinterpreted, freeing it from the fixity of those elements of thought that had hitherto been thought of as ‘given’ and as beyond the scope of thought's normative capacities.

The self-actualizing ‘I’

We have drawn on this ‘inferentialist’ dimension of Hegel's account of der Begriff but must issue here a warning. Logic standardly deals with the formal relations between thoughts, abstracting from their ‘matter’, but this cannot be the case for Hegel.Footnote ²⁰ Formal logic in this sense presupposes a representationalist understanding of the concepts with which it deals, and we have already seen how for Hegel the paradigm of concept is one which ‘posits’ its own content. It should not come as a surprise then that Hegel's quasi-formal taxonomizing of judgements and syllogisms in the Subjective Logic leads to the discussion of a type of inference described as the syllogism of necessity, in which ‘the formalism of the syllogistic process, and with it the subjectivity of the syllogism and of the Begriff in general, has sublated itself’. Here ‘the Begriff as such has been realized; more exactly, it has obtained a reality that is objectivity’ (Hegel (Reference Hegel and Miller1969a), 702–703). In the Encyclopaedia Logic, this syllogism of necessity is explicitly the locus of the ontological argument (Hegel (Reference Hegel, Geraets, Suchting and Harris1991), § 193).

We have earlier stressed the ‘pragmatic’ issues on which many of Hegel's ‘logical’ claims turn. In this respect, we follow Brandom's pragmatic-inferentialist view of logic, in which formal inferential relations between judgements are dependent upon actual practices of non-formal or ‘material’ inference in the ‘language game’ of the asking for and giving of reasons. But while Brandom's pragmatic inferentialism is open to the charge that it is insufficient to account for the content of empirical judgements, and the experience upon which empirical judgments draw (see, for example, Kremer (Reference Kremer, Weiss and Wanderer2010), McDowell (Reference McDowell, Weiss and Wanderer2010) ), Hegel's ‘weak’ version, as we have argued, is not subject to this critique as it does not deem inferential relations as sufficient for the content of judgements. Thus while Brandom's version does not have an input from conscious experience into judgement contents, Hegel's does so, and, moreover, does so in a way that leaves open the question of the type of experience involved, leaving a place for the content of religious forms of experience in which the subject of experience grasps herself in relation to God. By making inferences necessary for the rational content of thought, however, Hegel rejects any foundationalist approach to experience, including religious experience. But that expressions of experience can be put into material inferential relations amounts to the redetermination of that very content in ways that make experience relevant to reason, and reason responsive to experience.

Elsewhere, Redding (Reference Redding1996, 156–158) has argued that the ‘concrete’ syllogism of necessity with which Hegel's treatment of the syllogism ends should be regarded, not as any formal articulation among abstract judgement contents, but rather as articulating human acts of judging and inferring – effectively, intersubjective communicative acts of asserting and giving reasons to be understood in terms of Hegel's broader notion of recognition. That such recognitive acts are to be thought of as the constituents of ‘spirit’, the proper successor to the Fichtean I=I, is what Hegel had attempted to show in chapter 4 of his Phenomenology of Spirit. Human activity is governed by conceptualized social roles or statuses which are acknowledged, actualized, and reproduced in the acts themselves via the use of the concepts involved. It is by belonging to such forms of life that humans become self-conscious, free, and rational beings – it is in this sense that the concepts articulating them are, like the I, self-actualizing. The question, however, concerns how the concept of God on this model can be saved from collapsing into the status of a collective human posit.Footnote ²¹

We might start thinking about the ontology being invoked here by an appeal to Anscombean ‘institutional facts’ rather than ‘brute facts’ (Anscombe (Reference Anscombe1958) ). Institutional facts populate the realm of Hegel's objektiver Geist, a realm within which the holding of some fact and the fact of its social recognition cannot be easily prized apart. The married couple are married inasmuch as they were pronounced to be married by the appropriate person, that they recognize each other as the person to whom they are married and are so recognized by the community at large. But such considerations commit neither Anscombe nor Hegel to the counter-intuitive idea that all things are what the relevant community takes them to be, as each can appeal to an important distinction: that between institutional and brute facts (Anscombe), and between Geist and Natur (Hegel). Unlike the married couple, what we call water would still be water were there no ‘us’ to call it by that or any other name. But as absolute spirit, God clearly could not be dependent on his creatures in just the way that, say, the existence of a married couple is dependent on the recognitive practices of its community. And so if Hegel wants to think of God as being actualized by the use of some analogous self-actualizing concepts, he will need to have an account of how this picture could be meaningfully applied to something infinite, and not simply to the sorts of finite beings that the model paradigmatically fits. God would need to speak to his creatures in self-actualizing acts.

Once more we might see Hegel as capitalizing on and developing an approach already found in Kant. The other side of Kant's critique of the traditional arguments for the existence of God had been his treatment of God as a moral ‘postulate’ (Kant (Reference Kant and Gregor1997), bk II, ch. 1, sect. V). Because we are subjected to the forces of our material natures, to apply the moral imperative we need the help of belief in the ‘objectivity’ of ideas of God and the immortal soul. Furthermore, Kant had also insisted that a moral agent must ‘hear’ the moral law as addressed to him as if it were the voice of God, since qua finite being, no human agent could think of himself as the source of the law (Kant (Reference Kant, Förster and Rosen1995) 22:51; Redding (Reference Redding, Diego Bubbio and Redding2012) ). But from the point of view of theoretical reason, such claims look to be just about a certain psychological necessity, and once more we do not advance beyond the necessity of the mere idea of God. Hegel, we suggest, needs the idea of a form of human speech in which it is God who is recognized as the one ‘doing the speaking’. For Hegel, this can then be the way that God self-actualizes in the context of human interaction. We can gain a clue as to how this might be thought of from the section of the Phenomenology of Spirit where Hegel reflects on the limits of the Kantian idea of the internal voice of morality as the voice of God – the scenario in which a ‘hard-hearted’ judge hears the confession of a ‘beautiful soul’ (Hegel (Reference Hegel and Miller1977), §§ 632–671).

The beautiful soul is brought as a confessor before a judge, but this judge is initially understood in a quasi-Kantian way, as the bearer of a conscience that represents the pure moral law. As in the Kantian system, there is a fundamental and unbridgeable conflict between reason and nature here. But the judge, as an embodied human being, must also be a sinner. And so despite his implicit claim to be representing the universality of moral law, the judge must have his actions and intentions, and hence judgements, marked by the same finitude (cf. Bubbio (Reference Bubbio2012), 812–813).

The contradiction of such an arrangement can only be resolved by mutual confession and forgiveness that brings about reconciliation. But how can these finite individuals taken separately be capable of this selfless act of forgiving? Hegel here invokes the voice of God, not as the voice of the commanding moral law, but as the forgiving voice (Hegel (Reference Hegel and Miller1977), § 670). The speech act of mutual forgiveness, says Hegel, ‘is God manifested in the midst of those who know themselves in the form of pure knowledge’ (ibid., § 671). In the human act of forgiveness, God is fleetingly given corporeal existence, and what look like human acts provide the occasion for the self-actualization of the divine. Indeed, Hegel seems to conceive of the apparently human act of proving the existence of God in the same way. It is an act in which God passes from ‘mere’ concept into existing ‘idea’. And it is fitting that this happens within the medium proper to Hegel's God – thought.Footnote ²²

This might still look as if Hegel's OA only concerns, as in Kant and Fichte, the necessity of the ‘idea’, and not the existence, of God, but this is to subject Hegel's views to a criticism couched in terms of the representational framework that he has taken himself to have criticized. That is, this criticism still relies on treating the concept ‘God’ on the model of a concept-representation such as that of ‘thaler’, but as naming a distinct object hidden behind the veil of appearances. In the realm of spirit, any spiritual object's existence is dependent on its being posited in recognitive acts: none of us have the form of existence that could be there ‘anyway’, independently of our being recognized. We do not have the form of existence of natural elements that would still be there independently of whether they were recognized or not. Hegel can now turn the tables on Kant's criticism: to demand that God have this type of existence modelled on the independence of nature is to degrade his status to something unfit for the concept of God. Moreover, Hegel's recognitive account of Geist has already shown the error of thinking of God as some mere projection of the ‘human subject’, as the existence of human subjects conceived in terms of what makes them distinctively human – self-consciousness – is as conditional upon the existence of God as God is on human subjects.

It is not surprising then that Hegel's version of the OA cannot be evaluated independently of his idealism. Many, of course, would agree with Harrelson that Hegel has not advanced the ‘proof’ here beyond the idea that what is proved is merely the necessity of the ‘idea’ and not the ‘existence’ of God (Harrelson (Reference Harrelson2009), ch. 7), but such a position is an external one, presupposing the representationalist conception of knowing that Hegel seeks to undermine. What we deny is that one could accept Hegel's idealism and find his version of the OA wanting in this way. Here we have offered only the briefest sketch of what we take his idealist metaphysics to be, and have tried to sketch what we take Hegel's OA to be when considered in terms of such a reading of this form of idealism. It is true that it is difficult to find in Hegel's writings a succinct presentation of the OA, but this does not mean that his writings do not contain the outlines of a reasoned argument from the concept of God to the being of God. If we start from Hegel's own concept of concept – Begriff – rather than presuppose some non-Hegelian understanding of concept (as a concept-representation), and if we are aware of what, from Hegel's perspective, are the limits of any notion of ‘proof’ that presupposes any such non-Hegelian concepts of concept, then features of his distinctive version of the ontological argument can start to come into view.Footnote ²³