1. Introduction
Luis de Molina (1535–1600), the early modern Scholastic philosopher and theologian who caused such a stir in post-Reformation Europe two centuries before Kant launched his critical project,Footnote 1 is certainly not the first one who comes to mind when thinking of Kant’s intellectual ancestors. And yet, perhaps at second view, Molina’s role in the establishment of a powerful libertarian account of freedom, which – in Kant’s immediate vicinity – found its most vocal representative in Christian Crusius, suggests that such a connection may not look entirely far-fetched after all. Footnote 2
In previous contributions (Ertl Reference Ertl, Kaufmann and Aichele2014, Reference Ertl2020), I have tried to show that the core element of the ‘machinery’ of human free agency, as Kant sees it, namely the intelligible character, can indeed be understood in terms of Molina’s celebrated (or infamous) counterfactuals of freedom.Footnote 3 Counterfactuals of freedom are concerned with hypothetical scenarios and state how a finite free agent would act in a fully specified possible situation. My suggestion implies that, paradoxically, this Molinist legacy is pertinent, first and foremost, for a position Molina himself did not share, namely the claim that natural causal determinism and libertarian freedom of human agents can coexist.
Hence, the thesis of Kant as a Molinist (minimally defined, as we shall see) obviously involves a good deal of conceptual work, since the similarities do not lie on the surface. In this article, I am going to explore whether those similarities also emerge on Molina’s own turf, as it were, with regard to the question as to whether human freedom can be upheld in the face of God’s foreknowledge and his role as a creator.Footnote 4 A third, closely related issue, namely freedom in the face of divine concursus, i.e., God’s causal activity with regard to human agency beyond his role as creator, of which grace is a prime example, will have to be treated at a later occasion.Footnote 5
Obviously, these questions have far less prominence in Kant’s œuvre, but one might expect that the similarities (or differences for that matter) are perhaps more obvious in this regard. As it will turn out, this is in one sense indeed the case, but in another sense it is not. The Molinist dimension when it comes to the defence of freedom in the face of creation is undeniable, although not straightforward. Ironically enough, though, Kant’s take on free will in the face of foreknowledge is perhaps the least Molinist element in his overall account.
Strikingly, once again Kant’s Molinism gains contours against the background of views sometimes strongly diverging from the historical Molina. Molina is convinced of compatibility when it comes to the creation facet (because, in his opinion, creation does not involve determinism), while Kant’s position is more complex. Kant distinguishes what we can understand through the insight of our own reason from how things may be beyond that. With regard to how far our own insight reaches, a form of theological libertarianism and hence incompatibilism is the position to take in this respect. This incompatibilism, however, is meant to guarantee the self-accrued status of the intelligible character, and this in turn is the cornerstone of Kant’s Molinism.
The article is structured as follows: I shall first give, in section 2, a very brief summary of my claim that Kant – in the context of his take on free will and natural causal determinism – is committed to Molinism minimally defined (MMD). I then turn in section 3 to Kant’s discussion of foreknowledge in the theology lectures. As we shall see, in Kant’s opinion, if a problem arises at all for the human free will, it does not arise by virtue of divine knowledge strictly speaking but by virtue of the involvement of the divinity in the generation of the respective truths. In section 4, this issue will be explored further, and I shall try to show that such a threat indeed arises by drawing on the idea that God himself, through his will, brings about the truthmakers of what is known.
This idea – ironically enough – was at the heart of a Thomist conception of divine knowledge and diametrically opposed to what Molina had in mind. Strikingly, as we shall see, while Kant rejects the compatibilism inherent in a Thomist approach, he first reaches an unsatisfactory position in that he denies the need for transcendental freedom when we try to uphold the demands of morality in the face of creation. In the final section, I shall look at Kant’s later attempts to rectify this shortcoming. In these attempts, Kant’s MMD – albeit in an unusual disguise – is in full display.
Examining Kant’s Molinist leanings in this regard is hopefully not just an exercise of a somewhat esoteric historiography of ideas. Rather, such an approach is meant to help draw some of the contours of Kant’s practically grounded metaphysics. It is important to realize that what re-enters into the critical system on practical grounds, after being expelled from its theoretical realm, is in large parts not a fully-fledged Leibniz-Wolffian, and hence ultimately Thomist, style of metaphysics. Such an examination also promises to chart new paths for discussing Kant’s position on compatibility altogether, so that this vast topic that easily eludes a concise account can be explored on safer grounds. Finally, such an investigation can also highlight tensions within Kant’s attempted solution to the compatibility problem.
2. Kant’s MMD
In Ertl (Reference Ertl, Kaufmann and Aichele2014 passim; 2020: 34–47), I tried to show that Kant – in his account of free will and causal determinism in nature – is committed to a form of Molinism. More precisely, Kant is committed to MMD. This is to say, his doctrine of a self-accrued intelligible character (see, e.g., CpR, A539/B567, A556/B584; CprR, 5: 98; Rel, 6: 27 and 31f)Footnote 6 can be read in terms of (i) the claim that there are pre-volitional counterfactuals of freedom, which are even beyond God’s control. To repeat, a counterfactual of freedom indicates what a finite free agent would do in a fully specified possible situation, and ‘prevolitional’ refers to God’s will. In the light of his doctrine of God as the scrutinizer of the heart, Kant at least implicitly (ii) accepts that there is something like genuine so-called ‘middle knowledge’ (scientia media) by virtue of which God knows how free human agents would behave in all possible situations without, however, using this term in the pertinent context (see CprR, 5: 140; MM, 6: 392f). By virtue of his ascription of transcendental freedom to human agents (see CpR, A547/B575; CprR, 5: 3-5), he (iii) endorses a version of libertarian free will.
Kant’s MMD can be obfuscated by the very fact that it is pertinent for what we can call ‘the natural causal facet’ of the compatibility question of free will and determinism. (i)–(iii) play major roles in his attempt to establish this compatibility, and – in my reading – Kant’s position amounts to a theologically inflected version of altered-laws-compatibilism: the dependency of the laws of nature on libertarian human freedom can be elucidated through God’s special kind of knowledge about it, knowledge that informs the divine creative activity that in turn accounts for the (systematicity of) the laws of nature. To be sure, the assumption of the existence of such a God is a matter of the regulative use of reason and the postulates of pure practical reason. As I said, the historical Molina denied compatibility of freedom and natural necessity and, unlike Kant, rejected determinism in nature (see, e.g., Conc 4, 47, 2). In contrast, Molina’s natural causal indeterminism resurfaces on the level of things in themselves in Kant, so that – in this regard at least – it seems appropriate to speak of an ‘adaptation’ of Molina’s position.
Crucially then, one needs to distinguish various different facets of the compatibility problem. In my opinion, these are the creation, concurrence, natural causal determinism, divine knowledge, and logico-semantic facets of this problem. This is to say, it is not obvious that a created being, a being whose actions God concurs with, a being operating in a causally determined environment, a being whose future actions God foresees, and a being about whose future actions true propositions exist, can be free.Footnote 7
Before I can turn to the discussion of the foreknowledge facet in Kant, let me make the following four remarks of caution about what not to expect in what follows.
(1) The overall contours of Kant’s view on free will (see CpR, A532-58/B560-86; GMM, 4: 456–7; CprR, 5: 94–106) are well known while almost every element of it is contentious. His account draws on the distinction between things in themselves and appearances. Human agents are free insofar as they are things in themselves; causal determinism holds in the temporal realm of appearances with time being a form of human intuition. There is no correlate of time and no natural causal determinism on the level of things in themselves. Free actions, while originating in the non-temporal realm of things in themselves, also have an appearance-dimension or appearance-aspect. It is notoriously difficult to spell out Kant’s contention that, by way of their transcendental freedom, human agents (as things in themselves) act, or make a difference, in space and time. Kant’s discussion clearly presupposes that this can be done in a satisfactory manner and presupposing this is something I also need to do in what follows.
(2) Kant grants God (whose existence is assumed on practical grounds) cognitive access to the sphere of things in themselves (see, e.g., L-PhDR/Pölitz, 28: 1052f) while maintaining that God’s way of representing entities in space and time is not itself spatio-temporal. The details of Kant’s account of God’s representation of spatio-temporal entities merit a detailed examination in their own right. In particular, it is worthwhile in the light of Kant’s main strategy in transcendental philosophy to identify the conditions of the possibility of cognition of objects with the conditions of the possibility of objects. This then allows that in the case of objects in space and time the pertinent cognition must be human and not divine cognition. There is a tension here insofar as God’s representation of spatio-temporal entities is in some sense inadequate, but he is still considered to be omniscient.Footnote 8 When it comes to free human actions, the difficult part to cognize for humans is the noumenal ground of these actions, while for God it is their temporal manifestation. I need to emphasize, however, that Kant’s views on the divine way of representing spatio-temporal entities altogether will not be analyzed in detail in what follows.
(3) Strikingly, when discussing what is commonly known as the problem of freedom and foreknowledge, Kant hardly mentions the transcendental idealist doctrine of time, which in the case of his discussion of the natural causal facet of compatibility is of such importance. At any rate, as Desmond Hogan (Reference Hogan2014: 50) has indicated, Kant clearly does not think that this account of time renders the solution to the problem any easier. As we shall see, the real issue for Kant is how God can know about the free actions of his creatures, whether they are temporal or not, given that according to the theist conception of God Kant is committed to, God is a cognitively autarkic subject.
(4) Kant, moreover, in the critical period at least, does not discuss questions often arising in connection with the foreknowledge facet, namely the ontological status of the future, i.e., whether it is real or not real.Footnote 9 Kant’s subscription to causal determinism on the phenomenal level does not pre-empt this question. Even if there is only one future and what that future will be is not open, the status of the future may still be real or unreal. This issue points to the question as to whether Kant’s idealist theory of time has (per McTaggart’s well-known distinction) A-theoretical or B-theoretical leanings. To simplify things a bit, it is in B-theoretical accounts that the future standardly counts as real.
It has sometimes been suggested (see the discussion in Zagzebski Reference Zagzebski1991: 47–52) that an eternity view of God, to which Kant also subscribes, entails a B-theoretical reading of time. If this is correct, it would be true of Kant’s account in terms of the form of inner intuition as well. Again, I will not discuss this issue in what follows.
With all these points hopefully clarified, I hope I can avoid the impression that at times I must be talking about somebody other than Kant.
3. Free will and divine foreknowledge
When it comes to the critical period, Kant’s treatment of foreknowledge compatibility is to be found exclusively in the lectures (and reflections), and here the lectures on metaphysics and the lectures on ‘natural theology’, as it is called in Baumgarten (M §800), are pertinent. Standardly, in Kant’s lecture arrangement, natural theology was part of the metaphysics lecture course (corresponding to §§800–1000 of the Metaphysica), but on various occasions (i.e., in 1774, 1783/4, 1785/6 and 1787) Kant gave a separate course on natural or philosophical theology.Footnote 10 In both cases, Kant’s courses were based mainly on Baumgarten’s textbook, while in the separate theology courses Eberhardt’s textbook and Meiners’ concise history of theological doctrines also played some role. I shall focus on Kant’s engagement with Baumgarten here, although it is in Eberhardt that Molina and the Concordia are explicitly mentioned (see 28: 596; Eberhard and Kant Reference Eberhard, Kant, Fugate and Hymers2016: 62). Baumgarten touches upon the problem of divine foreknowledge in M §§875 and 878 of the section De intellectu Dei.
I shall now describe the main outlines of what Baumgarten has to say. First and foremost, there is an important contextual point. Baumgarten is arguing against what he calls ‘philosophical Socinism’ (M §875). Socinism, named after the Italian theologians Laelius (1525–1562) and Faustus (1539–1604) Socinus, is a position, which denies that God has foreknowledge of what will happen in the future. Baumgarten is adamant in pointing out that this position is erroneous and that God’s knowledge encompasses all three modi of time. Strikingly though, Baumgarten does not mention the reason why Socinism denies divine foreknowledge, and at least part of the reason is the conviction that human freedom could not be upheld in the light of God’s foreknowledge. Similar to most of the so-called ‘Open Theists’ of our time, the Socinists were foreknowledge incompatibilists and what one could call ‘foreknowledge libertarians’. In their opinion, the determinism generated through the foreknowledge relation is not benign. Neither does Baumgarten even mention the issue of a possible threat to human freedom arising here. Since, however, he supports the claim that the human will is free (see M §§700–32 as well as Allison Reference Allison, Fugate and Hymers2018: 171–6 and Schwaiger Reference Schwaiger, Fugate and Hymers2018), he must be taken to at least implicitly subscribe to something like foreknowledge compatibilism. Secondly, Baumgarten uses the tripartite distinction of different kinds of divine knowledge prominent not only in Molina (Conc 4, 52) but also in Molina’s early modern Scholastic opponents, such as the Thomists (see Freddoso Reference Freddoso and Freddoso1988: 11–13, 47). As it happens, though, this is a rather superficial similarity and, by itself, does not get us very far for establishing or assessing a link between Molina and Kant. However, taking Baumgarten’s handling of this distinction into consideration is an important step in an attempt at doing this.
Baumgarten distinguishes three main types of divine knowledge, scientia naturalis, scientia libera (visionis), and scientia media. His understanding of at least two of these kinds of knowledge is significantly different from early modern sources. This concerns natural knowledge, which for Baumgarten covers what is metaphysically possible, and middle knowledge, which concerns the counterfactuals of freedom. For reason of constraints imposed by the present venue, I shall focus on middle knowledge. Baumgarten subscribes to what one can call an ‘unfulfilled counterfactual reading’ (UCR) of middle knowledge. In Molina’s original conception, God’s scientia media is present even prior to creation and covers those counterfactuals the antecedent of which will be fulfilled in the actual world. For Baumgarten, by contrast, scientia media only covers cases, which are purely hypothetical and indeed counterfactual from the point of view of the actual world. As we shall see in more detail, in Baumgarten – unlike in Molina and his Thomist opponents – scientia media plays no role in explaining how God’s knowledge of future free human actions comes to pass.
Finally, there is scientia libera (visionis) in terms of which God knows the determinations of the actual world. It is, however, not exclusively about free human actions. Libera here rather refers to God’s free will as the ground, or at least part of the ground, of the contingent truths of the actual world and hence ultimately to God’s freedom with regard to creating the entities, which make up the core of the actual world (see Insole Reference Insole2013: 29–57 and Kain Reference Kain and Look2021). This is of particular importance for the following reason. While scientia libera (visionis) is not restricted to human free actions, it is a fortiori concerned with them since they are, or at any rate are supposed to be, prototypically contingent entities among the actual entities. To be sure, Baumgarten’s usage of the term scientia libera (visionis) should not be read as by itself committing him to the claim that all the contingent truths of the actual world are dependent only on God’s will. After all, Molina, whilst denying this, used this term as well. No doubt, there is a certain tension in the combination of the expressions libera and visionis.Footnote 11 While visionis (at least in a non-technical, in particular non-Kantian sense) suggests something like mere truth tracking, libera seems to indicate that the truth in question is somehow generated by the knowing subject, in this case God.
Let me now return to one of the major differences between Baumgarten and Molina mentioned above and explain a little further.Footnote 12 As I said, in Baumgarten, scientia media is not used to explain how God can have foreknowledge of future free human actions. In general, while in Molina the three types of divine knowledge are more like different elements or sources sometimes playing together to account for an instance of divine knowledge (see Freddoso Reference Freddoso and Freddoso1988: 23f) in Baumgarten, each instance of divine knowledge belongs in exactly one category. In Molina, the role of scientia media in explaining foreknowledge is to contribute an important part of the ground of this knowledge. By knowing how an agent would behave freely in all possible situations, God ‘only’ needs to know which of these situations will actually arise to reach knowledge of how they will act freely.Footnote 13
The Thomists acknowledge the key role of scientia media in this regard (see Conc 4, 53, 10–18; Freddoso Reference Freddoso and Freddoso1988: 36–42). According to the Thomists, however, it is God who fixes the truth value of counterfactuals of freedom by bringing about their truthmakers through an act of his will, while according to the Molinists it is the finite agents themselves who contribute to doing that (and the difficulty is that these truths are supposed to hold prior to any world being actual). For the Molinists, this human factor is crucial for foreknowledge compatibility while the Thomists are creation compatibilists in the sense that, in their opinion, determination through God’s will does not undercut human freedom.Footnote 14 The Thomists can claim that on their account it becomes clear how God can generate this knowledge all by himself, without depending on external input. For example, natural knowledge, which in their conception is about causal necessities, and knowledge about which of the natural agents and the free agents he has made actual is all that is required to this end apart from scientia media. And natural knowledge ultimately depends on the essences of things, which are nothing but restrictions of the divine essence. Molina, by contrast, needs to claim that human free agents – even prior to their actualization – have sufficient ontological independence to provide for the truths of the counterfactuals of freedom. This puts pressure on the idea that divine knowledge needs to be generated in an autarkic manner.Footnote 15
With this in mind, let us now turn to Kant (L-PhDR/Pölitz, 28: 1053–6; L-RTh/Danzig, 28: 1271–3). Perhaps most strikingly, he dismisses scientia media as a ‘useless distinction’ (L-PhDR/Pölitz, 28: 1055) and claims that the object of scientia media is covered by scientia naturalis. To be sure, this first and foremost concerns Baumgarten’s ‘distorted reading’ of scientia media in terms of UCR. In line with this dismissal, he follows Baumgarten in not attributing any role to scientia media in a foreknowledge account. As a matter of fact, this is also true of the genuine Molinist scientia media Kant is committed to (without however using this term) in his critical writings. Recall that in his doctrine of God as the scrutinizer of the heart (which in turn is backed up by the ‘God-postulate’) Kant maintains that God knows the Gesinnung or fundamental attitude of each human agent, and this involves knowing what they would do in all possible cases (see CprR, 5: 140).
We are hence facing a rather murky situation, and it may look somewhat desperate to try to identify possible allegiances or differences to Molina or the Thomists here. And yet, in this thicket of shifts of meaning and conceptual reconfigurations there are clear traces of a distinct conception of scientia media detectable. Strikingly, though, the variant at issue is the Thomist variant. I will now try to identify these traces and then discuss what this means for my claim that Kant is committed to MMD. In any event, the emphasis on MMD plainly does not require agreement on the part of Kant with everything the historical Molina said. We need to focus on the defining criteria of MMD, and here (i) in the list in section 2 above is of particular importance, but it will take a bit of time to get to how the question of foreknowledge connects to this. Clearly, if Kant were to endorse this Thomist doctrine, my claim that he subscribes to MMD could not be upheld. Ultimately, we shall realize that it is outside the context of considerations regarding divine cognition that this issue becomes pertinent. In order to see how to get there, let us now look at what Kant has to say about foreknowledge:
How God would foresee the future free actions of human beings is no more difficult than how he foresees present human actions; for he does not see as we do; otherwise he would be passible. Since he knows purely through the extent to which he is conscious of himself and of his nature as the cause of their [i.e., the actions’] causality, all difficulty falls equally away. Now, here one difficulty is indeed avoided, but a new one arises from this, namely: God is then regarded as the cause of the free actions of human beings. However, this difficulty always remains, whether one supposes that God knows all things in advance through knowledge of his own nature or not. (L-RTh/Danzig, 28: 1271)
In this passage, Kant makes three striking claims: (i) Divine knowledge as such is not the most pressing problem for human freedom; the point is that this knowledge needs to be generated only through God’s nature and his self-awareness. No external factors may be involved, such as in perception realistically construed. (ii) The future status of free human actions is of no particular concern; if a problem arises for them, it arises with regard to all modi of time. (iii) If divine self-awareness is a problem for human freedom, this problem remains even if knowledge turns out not to be ‘parasitic’ on God’s self-consciousness.
What Kant says here and elsewhere (see L-PhDR/Pölitz, 28: 1055) is indicative of his overall foreknowledge compatibilism, as has been noted in the literature (see Hogan Reference Hogan2014): nothing in knowledge itself, even if the knowing subject is located in an eternal realm, is a threat to human freedom. The determinism involved in knowledge, simply insofar as it is knowledge, seems to be benign for Kant.Footnote 16 Kant’s position on the one hand amounts to what one could call an ‘altered-eternity compatibilism’ or, alternatively and following Vicens and Kittle (Reference Vicens and Kittle2019: 25),Footnote 17 a ‘multiple atemporal realms response’. According to such a view, the atemporal realm is not fixed: if the human agent had acted differently, God’s timeless knowledge would have been different. (Note that this holds regardless of whether human actions themselves are timeless or not). On the other hand, however, one of the key moves of those approaches is blocked. Due to God’s cognitive autarky, one cannot maintain that the human action causes God’s belief. Even though God’s belief may not cause the human action, it would appear that something in God must do so if he is supposed to have knowledge.
With this in mind, I will now consider the potential threat to human freedom ensuing from God in more detail. To simplify things a bit, I shall focus on the divine will. Divine nature is of course more comprehensive and includes more than, or is often even contrasted with, the divine will. Divine nature, more precisely God’s status as ens realissimum, still plays some role for the critical Kant in accounting for the real possibility of things in themselves and therefore also of a putatively free human agent, even though claims like these are beyond the realms of theoretical cognition. Clearly, though, the nature and status of this account of what is realiter possible in itself are rather unclear.
Also, I shall look at a particularly obvious case of the involvement of the divine will, namely creation. That there is a causal role on the part of God is perhaps particularly clear in this context. There are other types of operationes Dei in Baumgarten’s classification (see, e.g., M, synopsis, pp. 50f), and he summarizes the other main type under the title ‘providence’ (with a number of subtypes such as governance and concursus).Footnote 18 In general, there is of course an interplay of divine intellect and will when it comes to divine actions, insofar as he is supposed to be the perfect rational being. In this respect as well, I will have to simplify things a bit.
4. Free will and creation
Let us stay with the theology lectures for a moment. While Kant does not draw on any of the rival early modern conceptions of scientia media when discussing foreknowledge and – from the point of view of the early modern perspective – has a distorted ‘official conception’ of scientia media in the first place, clear traces of the Thomist account of scientia media, albeit not under this name, are detectable in Kant’s discussion of the creation facet of compatibility. As we shall see, crucially, for Kant the threat to freedom arises out of these traces, and initially at least, he does not have a satisfactory answer as to how to defuse it. This is to say, he struggles to show how the freedom and the creation thesis can be upheld together, because of the very connection of the concept of creation to these traces of a Thomist account of scientia media.
According to Kant in the theology lectures, in creation, in particular when creation involves bringing about entities which are themselves causes, what occurs is something like a configuration or arrangement of these created causes by the creating cause. Kant uses the term Einrichtung (L-RTh/Danzig, 28: 1280), which can indeed be translated literally as ‘arrangement’. Kant is not very explicit as to what this ‘arrangement’ or ‘configuration’ amounts to, but it is fair to say that in his opinion the creating cause somehow determines how the created cause ‘reacts’ to external input. In other words, the idea is that the agential profile of the created entity is getting fixed through creation.
Strikingly, it is the notorious example of the Bratenwender, i.e., turnspit or roasting jack, which can make this clear. Kant discusses this example in a slightly more extended fashion in LMet L1/Pölitz (which admittedly is of the mid-1770s) while the more famous passage from the CprR (5: 97) is not only disarmingly short but may even refer to a different kind of such a device altogether.Footnote 19 In any event, the Bratenwender of L1 indeed seems a far cry from the comprehensive theocentric metaphysics the Thomists invoke in their account of scientia media, but appearances are deceptive here. These two issues have more in common than their shared early modern origin. Let me first describe the type of Bratenwender Kant obviously had in mind. It is not the simple device on sale today, but a so-called Gewichtsbrater, a weight-driven kind of clock jack, i.e., a rather sturdy mechanical device often fixed close to an open fire place or an oven and whose main function is to rotate a roast fixed to a skewer. Its engine functions like a clockwork installed on a tower; these clockworks are typically driven by a weight suspended on a rope, and this weight, via an elaborate mechanism of gear wheels, keeps the skewer turning at a reasonable speed. There is a crank or winder fixed to a reel by means of which the weight can be lifted. This is what Kant says in L-Met L1/Pölitz:
Spontaneity in some respect (spontaneitas secundum quid) is when something acts spontaneously under a condition. So, e.g., a body which is shot off moves spontaneously, but in some respect (secundum quid). This spontaneity (spontaneitas) is also called automatic spontaneity (spontaneitas automatica), namely when a machine moves itself according to an inner principle, e.g. a watch, a turnspit. But the spontaneity is not without qualification (simpliciter talis) because here the inner principle (principium) was determined by an external principle (principium externum). The internal principle (principium internum) with the watch is the spring, with the turnspit the weight, but the external principle is the artist who determines the internal principle (principium internum). The spontaneity which is without qualification (spontaneitas simpliciter talis) is an absolute spontaneity. (L-Met L1/Pölitz 28: 267f)
When it comes to the turnspit, Kant considers the weight as the ‘inner principle’ of its agency, the craftsman determining the mechanism as its ‘external principle’. This is obviously still a somewhat abbreviated account. Determining must include constructing here, since normally it is not the craftsman who winds up such a device. Nonetheless, we need to assume that the device has indeed been wound up so that the weight can exert its driving force through gravitation, and that it has been fixed in an appropriate position close to the fire. I will therefore speak of the craftsman ‘installing’ the mechanism, and this is meant to include all this.
Kant focuses on the merely comparative internality of the inner principle that moves such a device. But the example furthermore excellently illustrates in what sense it is never up to an individual Bratenwender how it behaves. In this respect, it is important to see that the mechanism, installed by the craftsman, underwrites conditional statements, counterfactuals about the ‘behavior’ of the device. Of course, with regard to a Bratenwender, counterfactuals underwritten in this manner concern only a rather limited range of its behavior. Importantly, though, some of those at least concern the individual token of the Bratenwender, not just the type. An example could be:
If a roast of a certain size and weight were to be fixed on the skewer of this particular machine, the device would move at a certain speed and in a certain manner (perhaps evenly, perhaps smoothly, slowly, haltingly etc.)
This narrowness of scope should not distract us, however. The point is the comparison or association of the respective inner principles moving an agent in the case of the Bratenwender on the one hand, and a putatively human free agent on the other. As Kant maintains in CpR, A538-541/B566-9, in the case of a free agent or at any rate putatively free agent, the inner principle of agency, i.e., the particular capacity of transcendental freedom, is accounted for by the intelligible character. This would suggest that – in Kant’s view – what happens through creation, when creation involves configuration, is nothing other than fixing the intelligible character by the creator.
This is all the more poignant, since in my interpretation the intelligible character of a human agent consists of all the counterfactuals of freedom true of him. And this is why I take Kant’s thought of configuration through creation to be reminiscent of the Thomist conception of scientia media. To be sure, here the issue is completely disconnected from questions of knowledge, in particular divine knowledge, or ‘cognition’ if we wish to follow Kant. We are on the constitutive level. It does not matter whether anybody knows the counterfactuals true of a particular Bratenwender. The point is that the mechanism installed by the agent is that, or at least a major part of it, by virtue of which these counterfactuals are true. And this is precisely the same thought developed by the Thomists: it is God who has brought about the truthmakers of these counterfactuals. Of course, this very much helps to explain how he can know about them, but bringing about truthmakers and knowing the truth are different issues.
With regard to the theology lectures, the transcripts present configuration through creation as the fact of the matter, purely descriptive, the way things are, although there are tensions already detectable. Of course, this could be an example of the students not understanding Kant’s Rollenprosa, as it would be called in German, i.e., his speaking on behalf of somebody in favor of such a doctrine without himself supporting it. Kant concedes:
In the world, everything has qualified spontaneity (spontaneitas secundum quid); for the arrangement (Einrichtung) of creatures still comes from God. An original being (ens originarium) has absolute spontaneity; for there is nothing external to it that would be the external cause of its means. (L-RTh/Danzig, 28: 1280)
On the face of it, this is a rather puzzling statement: Kant’s mature moral theory precisely requires absolute spontaneity on the part of the human agent, and absolute spontaneity – in its practical variant – amounts to transcendental freedom (even though it is beyond theoretical cognition). Here, however, the Einrichtung or ‘arrangement’ of the creatures through an act of the divine will is said to undercut this form of spontaneity.
Of course, it is contentious when precisely Kant starts maintaining that morality ultimately requires transcendental freedom; it is also true that a structurally similar conflict between creation and absolute spontaneity by virtue of configuration is detectable in L-Met L1/Pölitz, 28: 267f, and this conflict concerns the theoretical variant of spontaneity so that it may predate the conflict between creation and transcendental freedom with regard to morality. In any event, though, as soon as he requires transcendental freedom for morality, we have such a conflict. Without arguing for this further, I shall assume here that it is already in the first Critique that the main contours of the doctrine emerge in which he indeed takes morality to require transcendental freedom, even though at least until the Groundwork he still thought that there are cognitive routes that connect us to transcendental freedom other than merely the consciousness of the moral law: for example, the experience of practical freedom (in the Canon of the CpR, A802f/B830f) or something akin to an inference from insight into the spontaneity of reason related to generating ideas (e.g., in GMM III, 4: 452).
This makes it all the more surprising that, as I said, in the theology lectures Kant presents the connection of creation and configuration as a matter of fact. What is his strategy to defuse this tension? When exploring the difference between the divine and human agent further, Kant says:
Freedom is practical when I act independently of all sensuous impulses, according to the prescriptions of reason alone. I must presuppose this practical freedom in humans, and indeed in the moral and in all the practical sciences, if their laws are to be valid. But this freedom is only a mere idea, and we cannot prove its actuality. However, whoever acts according to this idea and who believes that he must so act is actually free – indeed, not theoretically, but practically. (L-RTh/Danzig, 28: 1280, emphasis added)
What he is recorded to have said here amounts to a remarkable position. Kant, when comparing divine and human agency, explicitly contrasts transcendental and practical freedom. He holds something like hard determinism when it comes to creation and transcendental freedom with regard to human agents, but maintains that practical freedom, short of being grounded in the idea of transcendental freedom, is sufficient for morality. To be sure, even practical freedom, so understood, cannot be proven theoretically according to Kant (In L-DPhR/Pölitz, 28: 1067, Kant also ascribes practical freedom to God). If it is indeed his position since the first Critique that morality requires transcendental freedom (or a form of practical freedom, which is grounded on the idea of it) when it comes to the threat of determinism in nature, it is striking to suggest that, when it comes to the context of creation, morality can somehow survive without it. This is inhomogeneous at best.Footnote 20 We have hence discovered an extra layer of difficulties in Kant’s critical thinking up to around 1784.Footnote 21 Although his efforts at showing how transcendental freedom and practical freedom grounded on it can coexist with determinism in nature were deemed successful by him, he did not have the means yet to solve a further difficulty engulfing these key notions from a different direction along similar lines.Footnote 22
If my reading of Kant’s treatment of the natural causal facet of the compatibility problem is correct, this issue is particularly pressing, since the creation thesis is an integral part of the solution of the natural causal facet of the compatibility problem. As I mentioned above, according to this reading, drawing on God’s creative act in the light of his knowledge can elucidate how (at least some of) the laws of nature can be dependent on or be sensitive to human freedom. But now it turns out that major ingredients of the solution to one of the facets of the compatibility problem generate challenges in their own right with regard to another facet (which is not entirely surprising given the history of the problem).
We could of course trace possible developments of his take on ‘configuration’ in later transcripts of metaphysics lectures, but instead of doing this, I will move on directly to Kant’s published writings. There are two famous passages in his critical works in which Kant treats the creation facet of compatibility, namely CprR, 5: 100–3, and Rel, 6: 142f. Brewer and Watkins (Reference Brewer and Watkins2012) have provided a comprehensive reading of the one in the second Critique, a reading moreover that is not only fully consistent with my claim that Kant subscribed to MMD but moves Kant closer to the historical Molina in this matter, insofar as creation – at least when properly understood – is not seen as a threat to human freedom any longer. I can therefore be very brief here.
Brewer and Watkins draw on and improve earlier attempts by Beck (Reference Beck1960: 206–7) and Franks (Reference Franks2005: 115–6) and claim that by clarifying that creation concerns things in themselves, not appearances, and by making creation itself an act of freedom on the part of God guided by moral principles, Kant can substantiate the claim that the causality involved in creation does not license the assumption of transitivity. This is to say, while God is the cause of the agent as a thing in itself and the agent is the cause of his free action, God is thereby not eo ipso the cause of this action. Rather, what God creates is an agent equipped with a certain (universal) nature, in this case a nature, which includes the capacity of transcendental freedom, so that it can indeed be up to the agent to use this capacity in one way or another and hence acquire at least large parts of their agential profile by themselves. Since it is also in the CprR (5: 98) that self-acquisition of the intelligible character is identified as crucial for transcendental freedom to hold, Brewer and Watkins thereby also show why the idea that creation involves configuration would indeed undercut transcendental freedom in the first place. Configuration through creation entails that the overall agential profile is provided by God.
Strikingly, when assessing Kant’s position here, it is fair to say that Kant clearly holds that creation and transcendental freedom can coexist and are in this sense compatible. This, however, should not distract us from the fact that Kant, in an important sense, remains a creation incompatibilist.Footnote 23 He still holds that if creation amounted to determinism, it would undercut transcendental freedom, because – unlike in the case of knowledge insofar as it is knowledge – this determinism would be causal and would violate the causal-sources condition of freedom. What he now suggests, however, and to repeat, is that creation does not involve determinism.Footnote 24
Moreover, with Kant now holding that the assumption of creation entailing configuration (and hence determinism) is erroneous, for him, this assumption is in some ways ultimately a correlate of mistaking time as a feature of things in themselves. What precisely Kant has in mind here is certainly not obvious and getting to the bottom of all this would require an article in its own right. Therefore, I should rather like to proceed to the Religion passage; it provides perhaps even more interesting material for my original question regarding Kant’s overall Molinist leanings.
It is here that Kant more explicitly resumes the thoughts that we have encountered in the lectures on (metaphysics and) theology. And in this passage, he sounds far less convinced that creation compatibility can be established by drawing on key theses of transcendental idealism about time. It now rather sounds as though the assumption of configuration through creation is unavoidable, and in this respect one could argue that it might be something like a case of transcendental illusion. In any event, the creation of free agents is involved in one of the mysteries of pure rational faith. It is here too that he makes the extraordinary suggestion of exempting free human agents from the creation thesis, as long as we intend to remain within what we can figure out according to our own ‘rational insight’. This lengthy passage deserves to be quoted in full:
We can form a concept of the universal and unconditional subjection of human beings to the divine legislation only insofar as we also consider ourselves his creatures; just so can God be considered the ultimate source of all natural laws only because he is the creator of natural things. It is, however, totally incomprehensible to our reason how beings can be created to use their powers freely, for according to the principle of causality we cannot attribute any other inner ground of action to a being, which we assume to have been produced, except that which the producing cause has placed in it. And, since through this ground (hence through an external cause) the being’s every action is determined as well, the being itself cannot be free. So through our rational insight we cannot reconcile the divine and holy legislation, which only applies to free beings, with the concept of the creation of these beings, but must simply presuppose the latter as already existing free beings who are determined to citizenship in the divine state, not through their nature dependence, in virtue of their creation but because of a purely moral necessitation, only possible according to the laws of freedom, i.e., through a call. So the call to this end is morally quite clear; for speculation, however, the possibility of beings who are thus called is an impenetrable mystery. (Rel, 6: 142f; trans. Kant Reference Kant and Wood2018: 168) Footnote 25
In this passage, Kant implicitly concedes the shortcoming of our own rational insight with regard to the fact of the matter regarding creation compatibility, and I shall come back to this below. Nonetheless, he suggests what one could call ‘metaphysical existentialism’ as the position to take when it comes to our possibly limited perspective.
Let us examine this position in a bit more detail. The main idea is to take the existence of the human rational agents of the actual world as a basic fact, not further accountable in terms of their causal origin. (I shall ignore the question about other possible dependencies here, e.g., those concerning human agents merely as possible beings). Obviously then, the creation thesis has a restricted as opposed to a comprehensive reading. To be sure, even in a comprehensive reading of it, not everything in the world of things in themselves should be and needs to be regarded as created. As Kant himself remarked in the Magath version of the Pölitz lectures, even though he disparaged those points as ‘subtilities of the schools’ (Schulsubtilitäten, L-RTh/Magath, 28: 1092), essences and real negations (absence of realities) are not to be regarded as created anyway (in our rational insight, presumably). The point at issue is rather that in the restricted reading of the creation thesis, some causal agents (which are also natural agents) are exempted from creation.
What Kant is aiming at here is similar to the traditional doctrine of the so-called pre-existence of the soul, a topic Aquinas discusses in his highly influential treatise on creation, as one position to be dismissed after all (see STh, Ia qu90 art1 obj1). But whereas in this approach human souls are treated as being of the divine substance, metaphysical existentialism rather suggests that human agents (insofar as they are things in themselves) exist independently of God and hence are agents in their own right, as it were, similar to the idea constitutive of Manicheism, according to which there are two independent principles that account for the world.
Kant’s suggestions of a kind of metaphysical existentialism here are of course highly unorthodox, and they certainly are not something the historical Molina endorsed.Footnote 26 There is therefore an undeniable point of disagreement between Kant and Molina. This disagreement as to the reading of the creation thesis, however, should not distract us from the fact that, arguably and perhaps even paradoxically, metaphysical existentialism is one suggestive way of underwriting one of the defining criteria of MMD, the ‘existence’ of true pre-volitional counterfactuals of freedom. Clearly, if the existence of a rational agent is beyond the control of God, it is perhaps easier to see how the way these possible agents freely react to certain situations is beyond his control as well. Expressing the previous point in Kant’s parlance, metaphysical existentialism facilitates the assumption that the intelligible character is self-accrued by the agent whose existence is a bedrock fact.Footnote 27
Of course, how one should conceive of such character acquisition is no easy question to answer, but I take it that one can here resort to conceptions developed in the context of the discussion of Kant’s take on the natural causal facet of compatibility. My own approach (Ertl Reference Ertl2020: 40) is to extend the account developed by Willaschek and by McCarty, who maintain that the intelligible character is acquired by virtue of actual free actions (such as telling a lie) an agent is carrying out. I would suggest that it is not only the actual but the hypothetical free actions the agent would carry out that matter in this regard.
While Kant certainly has a point here, as it is obviously easier to see how an intelligible character can be self-accrued once we take the existence of the agent in question as bedrock, this should not make us overlook the fact that – in Kant’s opinion, given the limitations of our insight of reason – such a self-accrual may in fact be compatible with the comprehensive creation thesis. This is to say, it may be possible after all that a free agent is both created and has a self-accrued intelligible character. Presumably, this would have to count as a ‘mystery’ in its own right. In any event, even though the idea of creation ‘only’ belongs to the regulative employment of reason, this idea may even be construable in a realist manner without undercutting the assumption of freedom.
The last question I wish to consider is whether such a compatibility could in fact be established along the lines of Thomism and hence along creation compatibilist lines. If this were possible, Kant could not be considered as subscribing to MMD with regard to the realm beyond our insight of reason (because Thomism rejects (i) in the definition of MMD). Here, I disagree with Hogan (Reference Hogan2014: 63), who suggests, or at least seems to suggest, that Kant intends to leave open the possibility that configuration along with creation and human transcendental freedom can indeed be the fact of the matter. In this case, we would have to be wrong in assuming that configuration undercuts transcendental freedom, but I cannot see any indication that Kant granted this option. Kant confidently puts forward self-accrual of the intelligible character as the constituent feature of transcendental freedom (without insinuating that this may be only how we understand this matter), and it very much looks like a conceptual truth that the configuration of an agent through God undercuts the requirement that the ultimate causal source of the action in question must be internal. Rather, for Kant the core of the difficulty is how it is possible that creation does not entail configuration. In other words, the difficulty concerns the historical Molina’s conviction that counterfactuals of freedom can be true prior to and in combination with creation. But conceding this concerns an issue beyond Kant’s minimal Molinist credentials.
The historical Molina even thinks he has an answer to the question as to how this is possible, but this concerns his doctrine of concursus,Footnote 28 and as I said above, this will be the topic for another occasion. Suffice it to say that even in this regard there may be surprising similarities to Kant (when it comes to the realm of things in themselves). There may of course be other reasons why freedom falters in the face of creation, so that the absence of configuration may only be one of a number of conditions that need to be met. Moreover, I take it that Kant has perhaps exaggerated the difficulties here. It is not obvious to me why we necessarily think that creation entails configuration, unless perhaps it is part of an explanation how a cognitively autarkic mind has knowledge of the free actions of other persons. In this case, we would have moved full circle. Kant, at any rate, has not provided a comprehensive, let alone convincing foreknowledge account, which goes beyond the minimalist solution of appealing to divine omniscience. And this, while consistent with MMD, is indeed the perhaps least Molinist element of Kant’s Molinism.
5. Concluding remarks
It is, unfortunately, time to draw these investigations to a close. My primary, albeit limited, aim has been to show that my idea that Kant is a Molinist, minimally defined, can be upheld against the background of what Kant has to say about human free will in the light of divine knowledge and creation, Molina’s own domain. As we have hopefully seen, the real issue for Kant is creation, and in this regard he comes up with the extraordinary suggestion of a metaphysical existentialism in terms of which his key idea of a self-acquisition of the intelligible character, or in Molina’s language of the pre-volitionality of the counterfactuals of freedom, can be secured.
Molina has often been charged with an undue ‘externalism’, which would exclude an important aspect of reality from the all-encompassing cover of God. Kant’s limitation of the scope of the creation thesis can be seen as a paradigmatic example of such a move, but the irony is that for Kant such a move becomes necessary precisely because, in his opinion, we are trapped in the Thomistic idea that creation involves configuration.
Hopefully, all this is indeed not simply a piece of idiosyncratic intellectual history, although I think linking Molina and Kant amounts to an important and overdue correction of a widespread assumption of Kant’s intellectual pedigree.
In addition to this, looking at Kant’s treatment of the free will problem through the lens of Molinism could perhaps shed light on so-far still rather neglected aspects of his thought on this matter. This not only concerns the multifaceted nature of the compatibility problem as such but also what I have elsewhere called ‘Kant’s practically grounded metaphysics’ (Ertl Reference Ertl2020: 2). The God-postulate is anything but a simple deus-ex-machina-move but generates tensions in its own right, in particular in connection with the idea of creation in its regulative use by reason. It seems as though it is ultimately the quest for systematicity of the laws of nature on the one hand and the ‘freedom-sensitivity’ of some of those laws on the other, which – from our perspective at least – pull in different directions and require us to assume a comprehensive creation thesis in one context and a limited variant in another, thus opening up a similar rift to that of freedom and natural causal determinism in the first place. While we have also seen that in Kant’s opinion the fact of the matter may allow for a comprehensive account of creation to coexist with freedom, to understand how this is possible is – in his opinion at least – something that eludes human insight. In order to understand freedom in the face of creation, at any rate, we need, according to Kant, to go one step further than Molina himself and cannot restrict human emancipation to the realm of the possible.
Acknowledgments
This is the slightly extended and emended text of the Mary Gregor Lecture that I delivered at a session of the North American Kant Society at the APA Central meeting in Denver on 24 February 2023. I am deeply grateful to NAKS for their most generous invitation. Moreover, I should like to thank Andrew Chignell, who chaired the session, Dai Heide, who acted as a commentator, and members of the audience for the ensuing discussion. I also thank Eric Watkins and Karl Ameriks for their comments and suggestions and Ian Platt. Work on this paper has been supported by a kakenhi grant from the Japan Society for the Promotion of Science.