1Schelling’s middle period works have always been a source of fascination. They mark a break with the idealism (in both senses of the word) of his early works and hence with the Fichtean and then Hegelian tradition while not being weighed down by the reactionary burden of his late lectures on theology and mythology. But they have been equally a source of perplexity. The central work of this period, the Essay on Human Freedom (1809) or Freiheitsschrift takes as its topic the moral problem of freedom, but spends much of its time telling a mystical-metaphysical story about the creation of the world that attempts to paint a picture of a kind of irreducible metaphysical contingency in nature. What is the relation between the moral and the metaphysical elements of the Freedom essay?

2It has never been obvious. But some recent Anglophone scholarship suggests an answer. In her Freedom and Reason in Kant, Schelling and Kierkegaard Michelle Kosch¹ presents a strong case for the view that Schelling’s Freiheitsschrift is motivated by a critique of Kant’s conception of autonomy in ethics. The idea is that Kant’s problems start not if he fails to show that it is rational to be moral, but if he succeeds. For if he does succeed, then it is irrational to be immoral, and hence it is impossible for a rational agent to act immorally: an act not in conformity with duty is evidence of irrationality (perhaps one’s rational agency has been thwarted by pathological inclinations). This is broadly the Socratic position: evil, as a knowing immoral act, is not possible. Officially Kant rejects the classical view, most explicitly in his 1793 Religion with the Bounds of Mere Reason, for a Christian position in which freedom is understood as the freedom to choose between good and evil without thereby abandoning one’s agency. But he has no obvious means for doing so.

3Kosch attempts to show that Schelling comes, by 1809, to identify his early work with something like this Kantian dilemma: when it comes to our freedom, being pushed around by reasons is just as bad as being pushed around by causes: the Hegelian ruse of reason is as inimical to our agency as phenomenal determinism. The Freiheitsschrift is the result of Schelling’s attempt to defend a muscular, Christian conception of the possibility of evil.

4Nevertheless, for Kosch, Schelling’s move to the metaphysical level is still problematic: just where one would expect an account of norms that does not infer them from our rational autonomy, we get, puzzlingly, a cosmology. I argue that Kosch’s matrix for interpreting Schelling in fact does give us a way of understanding Schelling’s metaphysics of contingency. In broad outline, Schelling’s argument is that our choice of metaphysical schema is constrained by a correct understanding of agency: the world cannot be causally determined, because that would be inconsistent with our understanding of ourselves as agents; but equally the world cannot be rationally determined, and for the same reason. Schelling’s metaphysics of contingency is an unpacking of the consequences of this inference, and not a botched attempt to ground values outside of rational autonomy.

1. Intelligible Fatalism

5The basic contours of Kant’s account of morality are well-known. In one influential reconstruction,² Kant distinguishes between motivating reasons and justifying reasons. This explains the otherwise apparently circular endeavor of Groundwork I,³ which promises an analysis of the content of morality, but proceeds by supposing that we know it and then giving an account of moral motivation. The account of motivation shows that an action can be formally in accord with what morality demands, but incorrectly motivated e.g. by long-term self-interest or immediate inclination. This is to say that the motivating reason for a formally moral action can be different from its justifying reason, i.e. from the fact that it is the right thing to do.⁴ A properly motivated moral action is, Kant demonstrates, motivated precisely by the fact that it is actually the right thing to do. This identity between justifying and motivating reasons is what grounds Kant’s derivation of the content of morality. The account of motivation establishes that in a properly motivated moral action we must set aside all empirical motives, both interests and inclinations. The identity of motivation with justification then permits the inference that it is the setting aside of all empirical motives that comprises the justification of the content of morality. But what is the result of setting aside everything empirical for the content of the moral law? It entails that the moral law does not have any content: it is purely the form of law. And this is the premise that yields, in Groundwork II,⁵ the famous first formulation of the categorical imperative: the maxim of an action is moral iff it is logically consistent to will it as a universal law.

6Since morality (at least of the kind that Kant is interested in) uncontroversially makes demands on us, requiring some actions and forbidding others, it is not surprising that Kant thinks of morality as a kind of law. What is more surprising is his account of freedom. In a brief argument at the outset of Groundwork section III⁶ Kant tries to show that freedom of the will also entails submission to law. The argument goes like this: freedom is a kind of causality; all causality is law-governed; therefore freedom is law-governed. But freedom cannot involve external constraint. So the law governing it cannot be a natural causal law. In fact there is only one way freedom can be reconciled with submission to law (Kant argues), and that is if freedom gives itself the law. This is the thesis of autonomy. Kant argues that the law that freedom gives itself in its autonomy is the same as the moral law so that ‘freedom’ and the ‘self-legislation of the will’ are ‘reciprocal concepts’,⁷ comprising what Henry Allison⁸ calls the ‘Reciprocity Thesis.’

7Most of this argumentation comes from the early Groundwork text. And it is well-known that Kant changed his mind quite radically before writing the Critique of Practical Reason in 1788, essentially putting forward two quite distinct accounts of the ground of morality. In the Groundwork account, he attempts first to show the hypothetical claim that if we are free, then we are bound by the moral law (sections I and II), and then goes on to show that we are in fact free (section III). By contrast in the Critique of Practical Reason, it is the moral law that we have direct epistemic access to, via the notorious ‘fact of reason’ (AA V, 31).⁹ Combined with the principle that ‘“ought” implies “can,”’¹⁰ this shows that we are free: we know directly what we ought to do under the moral law; since we ought to do it, we can; and it’s only possible for us to act morally if we are in fact free. As Kant himself points out, this reversal is less drastic than it seems; freedom remains the ratio essendi of the moral law. What the second argument emphasizes is that the moral law is how we come to know that we’re free, it is the ratio cognoscendi of freedom.¹¹ And for the purposes of this essay, Kant’s introduction of the mysterious ‘fact of reason’ and change in the direction of our epistemic inferences does not affect the Reciprocity Thesis: in both cases freedom and the moral law are still ‘reciprocal concepts.’ What changes is only which side of reciprocal turns out to be easier to establish.

8If true, the Reciprocity Thesis has a clear if awkward consequence. Simply framing the thesis formally makes this obvious: (x)(x is a freely chosen action ↔ x is a moral action). From this it follows trivially that (x)(x is not a moral action → x is not freely chosen). In other words, it is not possible to choose freely to act immorally; in yet other words, the fact that an action is not moral comprises sufficient evidence for the claim that it was not freely chosen. As Kosch¹² points out, such reasoning is not completely beyond the pale. Indeed, with an uncontroversial extra premise, it is characteristic of Greek moral thinking. If one cannot will immoral actions, and yet they occur, there must be some other explanation for their occurrence. The best alternative is that the actor was not aware of what she was doing, i.e. she lacked knowledge. And this is Socrates’s position: vice is a form of ignorance, or equivalently, virtue is knowledge. Nevertheless, what Kosch does not emphasize is that such a position is antithetical to traditional Christianity because it makes it impossible to freely choose to do wrong¹³.

9It is worth noting that part of Kant’s argument for the Reciprocity Thesis goes by way of an analogy between the law-likeness [Gesetzmäßigkeit] of the noumenal causality of freedom (what would today be termed ‘agent causality’) and the phenomenal causality of natural laws. One aspect of this is particularly problematic, almost paradoxical, for law-likeness always entails necessity for Kant. So the free will is autonomous because it gives itself a necessary law. It is not always obvious what Kant means by law-likeness: in particular, you would expect there to be a big difference between the descriptive necessity with which effects follow from causes under natural laws and the prescriptive necessity of actions in conformity with the normative moral law. However the analogy between the law-likeness of free agency causation and natural (phenomenal) causation only makes sense if the former is understood as describing the actual process of noumenal volition. And Kant says just this in other places. In the Groundwork for instance he claims that a will unencumbered by sensible inclinations would not stand under the categorical imperative (a normative construct) as we do because the ‘volition’ of a holy will ‘is itself necessarily in accord with the law’.¹⁴ And this suggests that Kant thinks of the moral law as a descriptively correct law describing the volitional processes of holy wills in the same way that the laws of natural causality are correct descriptions of phenomenal processes. This view probably gains what plausibility it has from the fact that the volitional processes of a free agent are the processes of practical reason, i.e. reason applied to volitions. And rational inferences are necessary.

10This idea of rational necessitation leads directly to the same conclusion as the Reciprocity Thesis. If I am free iff I give the law (rationally, necessarily) to myself, then I cannot have been free (or rational) in acting immorally. But rational necessitation is clearly stronger, since it suggests that I am really not free even when I am acting rationally (because now reason is necessitating my behavior) whereas the Reciprocity Thesis itself leaves open the possibility of free choices between moral actions, e.g. in the cases of imperfect duties, where the time and manner of fulfillment are not stipulated.

11Kosch¹⁵ points out that this was a serious sticking point for Kant’s contemporaries. In particular Carl Christian Erhard Schmid, in his 1790 Versuch einer Moralphilosophie, accuses Kant of saving our freedom from the frying pan of phenomenal determinism only to cast it into the fire of what he picturesquely terms ‘intelligible fatalism.’ In a nutshell: Kant admits that we are not free if all that is happening to us is that we’re being pushed around by (necessary) natural laws; but Schmid argues in parallel, that we’re not free if all that’s happening to us is that we’re being pushed around by (necessarily acting) reasons.¹⁶ Interestingly, Schmid argues further that the only way for immoral actions to occur (given the rational necessitation of such actions at the noumenal level) is for there to be another intelligible causality distinct from that of freedom and reason¹⁷. Schelling, who owned Schmid’s Versuch, will make use of this view.

2. Radical Evil

12Quite possibly in response to Schmid, or at least to similar objections concerning the implications of the Reciprocity Thesis, Kant developed, in his 1793 Religion innerhalb der Grenzen der bloßen Vernunft, a more properly Christian conception of evil, emphasizing the claim that I cannot be held morally responsible for evil unless I have positively chosen it: ‘But only our own act is something morally evil, i.e. something that we can be responsible for’.¹⁸ This claim is what Allison terms the Incorporation Thesis¹⁹ and is connected with (although not obviously identical to) what Kant himself terms radical evil, das radical Böse.²⁰

13Kant is trying to avoid the model of evil action in which my will is striving for the good, but is unfortunately overcome by my inclinations. This model is a tempting interpretation of the consequences of the Reciprocity Thesis under the condition that human beings have both a rational aspect (committed to freedom and hence the moral law) and a sensible one (the source of interest and inclination). Thus when I fail to act in accordance with the law, if it is because I did not in fact choose freely but was overcome by my sensible inclinations, then everyone who does something wrong would be able to plead crime passionnelle not just in mitigation but in exculpation.²¹ In the Religion text Kant explicitly denies this: the ‘ground of evil cannot be placed … in man’s Sinnlichkeit’ because we are not responsible for it.²² And this is why Allison terms this view the Incorporation Thesis: sensible (pathological) motives are not themselves evil or the source of evil; they only become so when incorporated into my personality by means of an active choice.

14But the Incorporation Thesis is not in fact what Kant means by describing evil as ‘radical.’ Rather, Kant uses this term in relation to his rational reconstruction of original sin  ‒ not for nothing is this a religion text. Kant rejects the strictly theological conception of original sin as something inherited, Erbsünde, as this would clearly violate the Incorporation Thesis, that we are only responsible for what we have chosen.²³ But Erbsünde appears phenomenologically as something innate, and this is the view Kant defends, using the term angeboren, inborn, to describe original sin, more in keeping with English usage: evil is something ‘radicales, angeborenes’ and ‘woven into human nature [Menschheit] itself; it has, as it were, taken root there’, while at the same time being ‘incurred [zugezogen] by us ourselves’.²⁴

15Clearly Kant has a conceptual problem here: how can there be anything that is both angeboren and chosen? The problem contains two puzzles. First, what can this thing be? And second, how can it be both angeboren and chosen? In answer to the first question, Kant argues that the maxims we form for action (that are tested for their conformity to the moral law by the universal law formulation of the categorical imperative) are nested into a hierarchy: my adoption of a local maxim in my current situation is itself the result of my adoption of a more global maxim. On pain of infinite regress, the hierarchy must terminate at some point in a most basic maxim that has the form of prioritizing either the moral law or my interests and inclinations. This is the choice of what Kant describes as Gesinnung [disposition].

16It should be noted that Kant precludes the possibility of doing wrong because it is wrong. He describes this as teuflisch [devilish]²⁵ and something human beings are incapable of. It seems clear that the Reciprocity Thesis would exclude this possibility. But in the Religion book Kant also refuses to think of evil as merely the result of inclination. So what is evil? It has to do with the lexical ordering of two principles: one principle is the moral law, and the other principle is the principle that incorporates (in line with the Incorporation Thesis) as the basic guides for my willing. So Kant doesn’t even think it is possible for me (as a free agent) to choose to guide my behavior by my inclinations. Rather the worst I can do is to choose to give priority to my inclinations. This is what Kant describes as Verkehrtheit [perversity].²⁶ In other words, at the most basic level I determine (as in the counterfactuals in Section I of the Groundwork) what I will do in case acting in conformity with the moral law is inconsistent with my interests and inclinations: either I will do the right thing (prioritizing the law) or not (in which case I have prioritized my interests and inclinations).

17This answer to this first puzzle in fact directly raises the second, for we can only act on the basis of maxims, and the adoption of a lower level maxim presupposes the adoption of higher level ones. So any action in fact presupposes the adoption of a most basic maxim. But how can we choose our basic maxim prior to any action; and yet the choice be of our own doing so that we can be responsible for it? Kant’s solution is to trade on the account of transcendental freedom in the Critique of Pure Reason:²⁷ choice of basic maxim takes place in the noumenal realm and hence independently of temporality; this independence from temporality makes it appear as something innate  ‒ but it is still a choice, and hence something for which we can be responsible.²⁸

18Kant represents this conception of evil as a ‘positive’ one, and contrasts it explicitly with the classical equation of virtue with knowledge.²⁹ But has he really made conceptual room for the free choice of evil? Kant himself admits that the choice of evil, although empirically universal, is rationally inexplicable [unerforschlich].³⁰ In a sense, this stems simply from the fact that to explain my adoption of maxim a is to mention the higher order maxim b warrants it. If there is a highest order maxim, then by definition it cannot be adopted because of some higher order maxim and therefore lacks an explanation. If this were all there was to it, then the inexplicability would not have anything specific to do with the question of evil. But it does, for the adoption of a good maxim is motivated, precisely by its conformity to the moral law = the consistency of practical reason. It is the adoption of an evil maxim that is inexplicable.³¹

19Some commentators³² see this in itself as an insuperable objection. But Kant introduces another innovation in the Religion book designed to help. This distinction is within the human will itself between Wille (which legislates the moral law) and Willkür (which is the bare power of choice). It is easy to see how this could in principle help to explain how evil is possible, i.e. how we could actively choose evil, act evil: Wille shows us what the law is, but Willkür doesn’t have to go along with it. On its own though this distinction also doesn’t help, for if the Reciprocity Thesis is true, then the moral law (now conceived as given by Wille) still implies and is implied by freedom (now located in Willkür). Either the Wille/Willkür distinction is consistent with the Reciprocity Thesis, in which case nothing has changed; or it is not, in which case Kant has made a much deeper change to his moral theory than he admits, and one that doesn’t have anything to do with the Wille/Willkür distinction: he has denied the Reciprocity Thesis. If that goes, then there’s no particular reason to divide the human will along the lines of the Wille/Willkür distinction. The rationale for such a distinction was to preserve the biconditional linking freedom with morality. But if there is no link, then morality is separate from the question of the freedom of the will and cannot be defined as freedom giving the law to itself i.e. as autonomy. And there is no evidence Kant wants to give that up.

20Ultimately the two cornerstones of Kant’s moral theory, the Reciprocity Thesis and the Incorporation Thesis are inconsistent. The reciprocity thesis entails that immoral (evil) actions are unfree, and hence not really even actions, but mere ‘happenings,’ due, canonically, to the baleful influence of pathological inclinations on the will; but the Incorporation Thesis argues that such inclinations can only be a source of evil if they are freely chosen. Kant wants to have it both ways, but can’t: he is ‘at war with himself’³³ or ‘ends up straddling two worlds’³⁴.

3. ‘Freedom is the capacity for good and evil’

21‘Freedom is the capacity for good and evil.’ This is Schelling’s famous definition of freedom from the Freiheitsschrift. In this section I will try follow Michelle Kosch in arguing that Schelling chooses this formulation of freedom  in terms of evil  because he has become skeptical not only about whether Kant had a real theory of freedom, but also about whether the idealist movement in general, his own former self included, had one either. But first it is important to realize that the split in Kant’s thought mentioned above is not one of the splits that the idealist movement took itself to be healing. Rather, the idealist movement (including the early Schelling) lines up almost completely on the side of rejecting radical evil. And it is this rejection that Schelling addresses in the Freiheitsschrift. Along the way I also want to make some comments about the relation of the metaphysical preoccupations of the idealist tradition to the normative problematic of freedom. These comments will pave the way for the next section in which I address the extent to which Schelling’s insistence, against the idealists, on the fundamental importance of evil also has metaphysical consequences, but ones quite different from the idealists’.

22The diagnosis that Kant is a conflicted thinker is a commonplace in post-Kantian German idealism: the motivation of almost all the big philosophical players in the early nineteenth century was to heal the divides that Kant had so carefully constructed: concept/intuition, phenomenon/noumenon, rational/pathological. But the way in which Kant is in conflict with himself in this case does not align with any of the conflicts that the idealist tradition (early Schelling, prior to the Freiheitsschrift of 1809, included) sought to thematize and overcome. It corresponds rather to one where there was something approaching unanimity in favor of one side and not the other: Kant’s own qualms about radical evil were set to one side and the characteristic form of early German idealism accepted, in one form or another, a conception of freedom as something akin to rational autonomy.

23Indeed, it is possible to see this whole tendency of post-Kantian German idealism as giving rise to a speculative metaphysics on the basis precisely of a fundamentally normative problematic. The next three paragraphs will summarize this argument, which implies that Schelling’s rejection of the idealist conception of freedom will at the same time involves a rejection of the idealist metaphysics of nature. Just as moral action, for Kant, involves the possibility of suppressing empirical inclinations, it also involves the possibility that one’s action will not turn out empirically as intended, and the possibility that one will not be rewarded for one’s virtue in the empirical world. The catalogue of viciousness, worthy of one of Sade’s novels, that Kant provides as evidence for ‘a radical innate evil in human nature’,³⁵ shows just how hostile the empirical world is to the realization of the kingdom of ends. But Kant argues that it is a condition of moral action that the agent have reasonable grounds for hope that the empirical world will not turn out to be radically hostile to this realization. Thus such hope is a practical postulate of reason.

24It is easy to see why such a delicately balanced system might turn out to be intellectually unstable. By the time of the Critique of Judgment, Kant himself is beginning to have doubts. But these doubts are immediately formulated not in the normative terms of the hope that the world will turn out to be consistent with our normative requirements, but in terms of a theoretical question and the metaphysical constitution of nature. The possibility that nature will not turn out right is tabled by Kant in terms of the possibility that empirical laws might demonstrate ‘so infinite a manifoldness, and so great a heterogeneity of natural forms…that the concept of a system according to these (empirical) laws must be completely alien to the understanding,’³⁶ indeed this manifoldness and heterogeneity ‘might be infinitely great, and present to us a raw chaotic aggregate without the slightest trace of a system’.³⁷

25As is well-known, the Critique of Judgment attempts to provide a rationale for thinking that the messiness of the empirical is consistent with human rationality, to bridge the ‘incalculable gulf fixed between the domain of the concept of nature, as sensible, and the domain of the concept of freedom, as the supersensible’.³⁸ The dominant trend of post-Kantian idealism resolves the tension inherent in the rational hope of the Critique of Practical Reason and the purely reflective judgments of the Critique of Judgment in favor of the full-blown rational constitution of nature in its entirety. Thus the developments of the speculative metaphysics of the post-Kantian period can be understood as a response to a specifically normative question about freedom.

26But rationalization of nature makes explicit and large scale what had been in Kant implicit and small scale: that reason is an impersonal force just as implacable and inimical to a libertarian conception of human freedom as causal determinism is. For Kant, the universality of reason is internalized in each of us as a rational being; but in the rational history of Hegel, or the Schelling of the System of Transcendental Idealism, nature and history are both directly the working out of a rational system. In Kant, the worry about evil was that if we choose wrong, we are being irrational and hence failing to be rational agents and hence not responsible for going wrong; but if nature is itself speculatively identified with the rational, then it is simply not possible to go wrong, since the irrational (pathological) impulses that for Kant make us deviate from the moral law are simply reason that has not recognized itself as such yet. Thus the gulf between the rational and the pathological is only healed at the expense of ignoring the split between Reciprocity thesis and the Incorporation thesis. This gulf is resolved unambiguously in favor of freedom as submission to rational necessity and hence to the elimination of any possibility of evil.

27It is here that Schelling’s intervention is most original, I think. At least some of the speculative metaphysics that Schelling develops is also motivated by normative concerns; but by 1809 Schelling rejects, or at least seriously problematizes, the idea that nature is rationally constituted; and he does so again for normative reasons. This time however, the reason is to make room for a notion of freedom that permits the logical possibility of evil, for freedom is ‘the capacity for good and evil’³⁹.

28It is characteristic of course of Schelling that he never accepted the Fichtean and then Hegelian principle that philosophy consisted only in developing nature out of reason, or real out of the ideal. As Iain Grant⁴⁰ has shown, the idea of a Naturphilosophie is crucial to the whole of Schelling’s philosophy in all its phases. But prior to 1809 the two projects, of deriving the real from the ideal, and the naturphilosophische one of deriving the ideal from the real, appear to be mutually supportive and counterbalancing approaches to the same problem. Thus the counter-movement from real to ideal does nothing to disrupt either the (speculative) identification of reason and nature or its corollary, the identification of freedom with obedience to reason. In the most arresting description he gives of such freedom, one that can serve as a summary of idealism as a whole (until he himself punctures it in 1809), Schelling writes ‘if we think of history as a drama, in which everyone who takes part plays his role entirely freely and according to his pleasure, then we can conceive of a rational development of this confused play only if we think that a single spirit speaks in everyone … [so that] we are the co-authors of the whole, and creators of the particular roles that we play. … The absolute acts through each individual intelligence’⁴¹.

29It is a familiar critique of Hegel ‒ and applies in only a slightly modified form to the Schelling of the System of Transcendental Idealism  ‒ to say that his conception of freedom is totalitarian, since I am free only insofar as I successfully occupy an appropriate role with Sittlichkeit; and the role is itself legitimated as part of the development of spirit towards the necessary over-coming of its own self-alienation. But two aspects of this critique should be distinguished in order to understand Schelling’s 1809 position, not least his position in relation to his own previous works. First the necessity with which reason works itself out can be emphasized; this necessity is prima facie inconsistent with individual freedom of choice. Second, the problematic that I have been describing can be emphasized. The problem with rational self-determination as a conception of freedom is only tangentially related to the necessity with which reason works (either in Kant or in Hegel). More basic is the exclusion of the possibility of choosing wrong, of choosing evil. And this exclusion is consistent with some level of choice i.e. with non-necessity (e.g. as above in the question of imperfect duties for Kant or which aspects of one’s role to perform in Hegel). Thus the freedom/necessity dichotomy, though important for Schelling, is not ultimately the most crucial issue. The Freiheitsschrift is about the possibility of the free choice of evil, not about mere non-necessitation, even if the latter were in some sense a necessary condition of the former.⁴²

30In any event, whether one emphasizes the necessity of reason’s working out of nature or the fact that in constituting nature as reason, freedom is the freedom to choose only the rational, Schelling introduces his characteristic definition of freedom as ‘a capacity for good and evil’⁴³ in order to resist this idealist line of thought.

31The passage is not easy to interpret in part because Schelling immediately becomes interested in what seem to be specifically theological issues surrounding the problem of evil. I shall argue that Schelling’s concerns are not in fact mainly theological, but stem rather from an attempt to defend the possibility of freedom by appealing to the possibility of evil. Immediately after giving this definition, Schelling initiates a discussion of the question of privative or merely negative conceptions of evil⁴⁴. The discussion appears fundamentally theological, concerned with the problem of evil as the ‘profoundest difficulty’⁴⁵ that emerges from the definition of freedom as the capacity for good and evil: the definition appears to entail that evil must be attributed to god. And it is the theological impropriety of doing so that motivates merely negative or privative definitions of evil, which Schelling dismisses as attempts simply to deny ‘the reality of evil’⁴⁶. Schelling elaborates a number of positions as part of a more general background argument attempting to show that these conceptual problems are not specific to pantheism.⁴⁷ Some commentators see Schelling’s whole view here as essentially theological: he is providing a new theodicy on the model of Leibniz.⁴⁸ But the discussion is not essentially theological. The clue for its interpretation lies in the identification Schelling makes of negative conceptions of evil with denial of the reality of evil: negative conceptions of evil are not really conceptions of evil at all; it is these negative conceptions that are theologically motivated, precisely by the need not to attribute evil to god. But the reality of evil is attested in two important ways, both independent of any theology. First, as an empirical matter, evil is not identified with imperfection but is associated with relatively more perfect entities, like human beings (SW I/7, 368-9).⁴⁹ Second, since the ‘living’ definition of freedom entails the possibility of evil, the denial of evil entails a denial of freedom. And the assumption of the existence of human freedom is both basic to the Freiheitsschrift and again not necessarily theological in origin.

32In the philosophical tradition, the belief that evil is a merely privative concept is associated with Leibniz’s notion of ‘metaphysical evil,’ which, according to the definition at Theodicy § 21, ‘consists of simple imperfection’.⁵⁰ But Schelling’s discussion continues by adverting to ‘the Platonic view’⁵¹, a reference that Kosch⁵² argues is comprehensible only as identifying the views Schelling has been discussing about privative evil with the classical equation virtue and knowledge. As I have already argued, this tradition entails a conception of vice as mere ignorance and hence amounts to a denial of the possibility of free choice of evil, i.e. a denial of the reality of freedom.

33Schelling’s subsequent elaboration is unmistakably Kantian. The same doctrine of a merely privative conception of evil is also manifest in the Kantian view that ‘freedom consists in the mere mastery of the intelligence principle over the desires and inclinations of the senses, and the good is derived from pure reason’⁵³. Schelling immediately concludes that in this case ‘evil can have no freedom,’ a formulation he corrects to ‘evil is completely lost sight of’ or denied. His argument is as follows:

the weakness or ineffectualness of the principle of understanding can indeed be a ground for the lack of good and virtuous actions, yet it cannot be a ground of positively evil ones and those adverse to virtue. But, on the supposition that sensuality or a passive attitude to external impressions may bring forth evil actions with a sort of necessity, then man himself would surely only be passive in these actions; that is, evil viewed in relation to his own actions, thus subjectively, would have no meaning; and since that which follows from a determination of nature also cannot be objectively evil, evil would have no meaning at all. That it is said, however, that the rational principle is inactive in evil, is in itself also no argument [Grund]. For why does the rational principle then not exercise its power? If it wants to be inactive, the ground of evil lies in this volition and not in sensuality. Or if it cannot overcome the resisting power of sensuality in any way, then here is merely weakness and inadequacy but nowhere evil.⁵⁴

34This is exactly the sort of argument I introduced in section 1 to explain Kant’s desire in bring in a notion of radical evil, lining the Incorporation Thesis up against the Reciprocity Thesis. The ‘evil’ that would result merely from the rational will being overcome by inclinations (senses, passions etc.) cannot be ‘positive’ for in that conception of evil the subject is merely ‘passive,’ and the ‘evil’ would be a ‘mere determination of nature’ lacking ‘meaning’ for the subject as well as objective significance. It would, in other words, be no conception of evil at all. Taking Schelling’s use of the term ‘passivity’ seriously, it can be seen that such events are not even actions, in the sense that their notional agent is not actually responsible for them: nature is. The mere inactivity of reason is not explanatory here, because there’s no rationale for such inactivity (‘why does the rational principle then not exercise its power?’) except if the will has itself chosen to be inactive, or in Kantian vocabulary, to prefer inclination to the moral law. But in that case it is not ‘sensuality’ that is responsible, but the will: and this is the Incorporation Thesis. Thus Schelling is effectively arguing for the Incorporation Thesis as against the Reciprocity Thesis. He is doing the exact opposite of the trend of German idealism, going strongly against its grain.

4. Metaphysics

35Kosch⁵⁵ argues that the Schelling of the Freiheitsschrift is confused. On the one hand, Schelling sees the incompatibility of the Incorporation Thesis with the Reciprocity Thesis, and rejects the latter: freedom entails the possible free choice of evil. But this involves the rejection not just of the equation of free agency with moral norms, but also a rejection of the Kantian understanding of those norms themselves, for the Reciprocity Thesis (as the name implies) works both ways: freedom implies the moral law; but the moral law also implies freedom, i.e. autonomy, the giving of the law to oneself. Schelling needs therefore to give an account of the constitution of moral norms that separates it from agency. Such an account, Kosch argues, cannot equate morality with autonomy, for that equation is precisely what enabled Kant to identify freedom with morality. If morality is not autonomy, it must be heteronomy.⁵⁶ And this is what Kosch goes on to argue in the rest of her book, devoted to the later works of Schelling and above all to Kierkegaard: to freely choose is (possibly) to freely choose evil; and good and evil cannot be definitionally tied to one’s (free) agency through autonomy  ‒ they must appear from elsewhere. The moral revelations of the Christian god, for instance, would fit the bill.⁵⁷

36But Schelling tells us almost nothing about the content of the moral law or its justification and, confusingly, ‘[a]t the point in the essay at which he seems obliged to turn to ethics, however, Schelling turns instead to cosmology’.⁵⁸ Kosch regards this as quite unsatisfactory, a kind of displacement of the real (Kierkegaardian) problem that free agency ultimately requires non-human i.e. divine revelation of valuesonto an inappropriately metaphysical or naturalistic register. Ultimately this naturalistic register is ‘far from fully comprehensible’⁵⁹ in its own terms; and even if it were, it would not be able to answer the question required of it. This commits her to the view that the Freiheitsschrift as a transitional work, and that its intellectual labor is incomplete without the later philosophy of revelation.

37I think this appraisal is incorrect: it is true that the metaphysical understanding of nature that Schelling turns to in the Freiheitsschrift does not provide content or justification for moral norms; but this is not its intention. Rather it represents a working out of the metaphysical features that nature must possess as a result of Schelling’s new conception of freedom.⁶⁰

38In a way my representation of Schelling as inferring naturalistic metaphysical structures from normative commitments is continuous with Kant and the development of post-Kantian idealism that culminates in Hegel’s Phenomenology. The Groundwork III argument for the reality of freedom (which, in Kant’s sense of autonomy, already entails the moral law) proceeds by making two observations: (1) it is a necessary condition of practical action in general that we behave as if we were free (which, for Kant, entails that we behave as if obliged by the moral law); (2) we have some independent reason to believe we actually possess transcendental spontaneity because it is a condition of possibility of (theoretical) experience. That Kant presents the second observation suggests that he does not think the first suffices. And yet the first corresponds to his official story: the Second Analogy in the Critique of Pure Reason shows that experience is exhaustively determined by natural causal law; but the Third Antinomy shows that this does not exclude the ‘possibility’ of a transcendental freedom or spontaneity at the noumenal level. Belief in transcendental freedom is a condition of rational agency. And we are entitled to act on the belief for practical purposes because it is not impossible that we are (in fact, noumenally) free. If Kant thought this argument convincing, he would not need (2), which looks like it steps over the bounds of what we theoretically affirm (and equally he would not need to abandon this line of argument in the Critique of Practical Reason and reverse the epistemic order of freedom and moral law). Thus Kant is already clearly tempted to do what the idealist tradition embraces: make a transcendental inference from the conditions of free action to metaphysical claims about the constitution of nature. Schelling follows this line of argument, but in a different direction from that of the idealists, as a result of his new conception of freedom.

39The Freiheitsschrift opens with a long discussion of pantheism, evidently designed to position the new work in relation to the Natur- and Identitätsphilosophie. The burden of the argument is to refute those who claim that pantheism as such entails either ‘fatalism’⁶¹ or ‘does away with all individuality’ (SW I/7, 343), and to establish that ‘pantheism does not make formal freedom, at least, impossible’ (349). Spinozism, Schelling admits, is a form of fatalism; but the fatalism is not related logically to Spinoza’s pantheism, but to another aspect of his philosophy. Here Schelling introduces what is to be a vocabulary register that pervades the text, arguing: ‘the error of his system is by no means due to the fact that he posits all things in God, but to the fact that they are things … he treats the will, too, as a thing’ and it is this that least to the ‘lifelessness of his system, the harshness of its form, the bareness of its concepts and expressions, the relentless austerity of its definitions’ (349).

40Although Schelling’s refutations are quite intricate, he makes one major conceptual point. He argues that those who claim that pantheism is inconsistent with freedom and individuality are mistaken because they misunderstand what identity means in the signature pantheist claim that god is nature (SW I/7, 341f). Schelling anticipates the clarity of logical developments in the twentieth century by distinguishing clearly between the ‘is’ of identity and the ‘is’ of predication (341). And he also notes correctly that many identity claims are actually reductionist claims (342): the identity theory of mind for instance does not just identify mind and body in some way, but tacitly accepts the reality of the body and reduces the mind to the body. As far as an identity claim goes however, it could just as well work the other way round. But his central claim here is that identity statements are predications, and in predications, subject and predicate are related as antecedent and consequent or ‘unexpressed and expressed’ (342).This has the effect, as he points out, of equating the principle of sufficient reason (that everything has a ground) with the principle of identity (345-346). It also has the corollary that to say god = nature is to say that god is the ground of nature; and this kind of dependency is neutral with respect to the constitution of natural beings, leaving open the possibility that some of them could be free.

41At a gross level, it is obvious that Schelling’s positioning here reflects a basic argumentative structure along the following lines: if a metaphysical conception of nature in the broadest sense is inconsistent with freedom, then that conception must be rejected. In other words, this positioning makes sense only on the assumption that he is arguing from freedom to a metaphysics of nature. Pantheism in general is not one of the class of metaphysical positions that must be rejected on these grounds; but Spinoza’s ‘dead’ pantheism is. Schelling argues that (SW I/7, 417) the pantheistic relation between god and nature is independent of whether the relata are conceived realistically (as substance and individual things) or idealistically (as primal will and individual wills). And he therefore clears the conceptual space for himself to defend a non-Spinozistic pantheism. This form of pantheism substitutes ‘will’ for substance, leading to Schelling’s powerful declaration that ‘in the final and highest instance there is no other being [Seyn] than willing [Wollen]. Willing is primordial being [Wollen ist Urseyn]’ (SW I/7, 350).

42It is only at this point, which Schelling identifies with ‘Idealism’ (351) in general and Kant (351-352) in particular, that the problem of freedom can be posed at all; but it is posed inadequately, in a merely ‘negative’ (351) way or, as he says, ‘merely formally’ (352). What Kant got right is the equation of freedom with extra-temporality (351-2); but he got three things wrong. First, he failed to make will a basic (positive) ontological category.⁶² Secondly, even if Kant had properly ontologized the will, making it a ‘positive conception of the in-itself in general’ (352), he would not have said anything specific about human freedom, since the ‘intelligibility on which freedom was based’ would be common to the in-itself of all entities and the specific difference characterizing human freedom would be lacking. It is probably this point that makes Schelling describe Kant’s theory of freedom as negative and merely formal. Lastly, of course, Kant’s conception of freedom is not ‘living’ precisely because it is not that of the ‘capacity for good and evil’ (352). But it is worth noting that even the merely formally correct conception of freedom has serious metaphysical implications: it implies the transcendental/empirical distinction, locates freedom in the transcendental, identifies extra-temporality as at least a necessary (if not a sufficient) condition of freedom and demands that will be the most basic ontological category.

43As argued above (and in Kosch), the tacit contrast between Schelling’s ‘living’ definition of freedom and that of Kant suggests that Schelling thinks Kant (and those neo-Kantians, including perhaps his former intellectual selves, who followed Kant in this) did not conceive freedom in such a way as to make evil possible. What difference does Schelling’s definition make? It is here (see above) that Schelling deals with the problem of evil in a purely theological sense, and the rest of the introductory material (357) is taken up with defending the claim that the problem of evil is something that every system has to deal with, and therefore Schelling’s pantheism shouldn’t be particularly taxed with it (356). The treatise proper therefore opens with an ontological discussion, one concerning the distinction between ‘Wesen insofar as it exists and Wesen insofar as it is merely the ground of existence’ (357). As Schelling points out, the distinction stems from his 1801 Presentation of my System, published in his Journal for Speculative Physics. And one question has to be whether Schelling’s argument proceeds from already established metaphysical premises to the ‘deduction’ of freedom (or evil, insofar as this is a condition of freedom). I do not believe that this is the case: although Schelling might want to present his intellectual development as continuous, his discovery that Spinoza and even Kant had not in fact been talking about freedom effectively modifies his ontological commitments.

44A second problem arises from the fact that many of the most famous formulations of the distinction from the Freiheitsschrift are apparently theological in character. For instance, Schelling argues that everything has a ground, so god has a ground; but god is everything; so god’s ground is ‘that within god which is not god himself’ (SW I/7, 359). This kind of argument can make it seem like Schelling’s premises here are theistic, and that his argument starts from a theological premise and derives freedom from this premise.⁶³ Again, if Schelling has a basic metaphysical commitment to god, then his account of freedom is either derivative from this commitment or merely negative, seeking simply to show that his existing commitments are at least consistent with his new understanding of freedom. But I claim that it is this new conception of freedom that is driving the commitments. Two facts militate against seeing Schelling here as arguing from prior commitment to a theological premise. One is that the same ground/existence distinction can be derived from ‘things’ as well as god (358); and the other is that the question of god’s freedom impressed itself on Schelling in the later drafts of the Weltalter; and in those texts god increasing suffers from the same dilemmas as human beings do in the Freiheitsschrift. So it looks as though those dilemmas, i.e. Schelling’s conception of freedom, are the primary datum for his argument.

45Schelling elaborates the distinction between ground and existence in a series of metaphors (or examples): gravity versus light (359-360), darkness versus light (360), contraction versus expansion, particular will versus universal will (363), order versus das Regellos [the unruly or anarchic] (359), selfhood versus love(365) and, perhaps most significantly, ‘the incomprehensible basis of the reality of things’ versus ‘reason’ (360).

46Doubtless there are at least some intrinsically ontological motivations for this distinction: one plausible one is that the dark ground is a condition of individuality or particularity tout court. But Schelling uses the distinction between these principles to explain the possibility of evil. In well-ordered cases, the rational principle subordinates the chaotic one, transforming it into ‘its instrument.’ (SW I/7, 361). Thinking of the chaotic principle as that of individuality, this well-ordered outcome is precisely that in which the Kantian individual dismisses her particularity in free moral action; and the instrumentalization of the individual with respect to the cunning of reason corresponds to a Hegelian understanding of the relation of universal to particular. Neither of course satisfies Schelling as a conception of freedom because neither is consistent with the possibility of evil. Evil, Schelling claims, ‘perverts [verkehrt]’ the appropriate subordination of particular to universal (371) and attempts to ‘overturn [umzukehren] the relation of principles’ and ‘to raise the ground above the cause’ (365). So, while the principles themselves may be motivated directly by ontological considerations like the condition of personality or individuality in general, the possibility of the ‘dissolution’ of their relations is an ontological feature explained by Schelling’s understanding of freedom (365).

47Strictly the possibility of this ontological perversion belongs only to human beings: neither god (365) nor animals (372-3) can undergo it. But it has clear correlates in the existence of disease, which represents the same ‘independence’ of the particular in relation to the universal⁶⁴ and the creation of a ‘false unity’ (371). Indeed Schelling’s definition of evil as something ‘formal’ (370) i.e. something that bears on the relation between principles suggests that it is a possible ontological correlate of any organization of the principles of particularity and universality.⁶⁵

48Still, the notion of a reordering of principles of self-love and universality is surprisingly close to Kant’s late view in the Religion book: both Kant and Schelling reject the view that human sensibility or finitude is itself evil; they both agree that human beings are conflicted at the level of principles; they both agree that the conflict of principles is between willing particularity (self-interest and inclination) and willing universality; and they both agree that evil is not only not mere sensibility, it is also not even the choice of sensible motives: evil is only a reversal in the ordering of the principles of particularity and universality.

49So has Schelling made a real advance on Kant here? I think the answer to this question is ‘yes’ precisely because of the metaphysical implications of Schelling’s view. But even these are similar. The most obvious inference from a robust (libertarian) notion of freedom to metaphysics would be that freedom entails the falsity of determinism. Schelling rejects this move, a rejection itself entailing that any metaphysics of ‘contingency’ emerging from Schelling cannot be understood as inconsistent with the exhaustive (necessary) determination of the empirical world. He appears to argue that both empirical determinism and empirical indeterminism are inconsistent with freedom: determinism because it entails that actions are determined by ‘motives or other causes … which are no longer in our control at the time of the action’⁶⁶ so that we cannot be responsible for them; indeterminism because, as the famous case of Buriden’s Ass makes clear, it makes ‘individual actions completely accidental’ (SW I/7, 383) and suffused with ‘arbitrariness [Wilkür]’ (382).⁶⁷

50Schelling does not of course want to leave things like this: he resolves the dilemma by appealing to ‘the Kantian conception’ (384) of freedom. Here he is referring primarily to the argument, drawn from the solution to the Third Antinomy in the Critique of Pure Reason that transcendental freedom is at least logically consistent empirical determinism. This discussion is useful to Schelling because Kant’s notion of transcendental freedom is more prior (both logically and chronologically) to the equation of freedom with the moral law. Nevertheless, it can seem as if Schelling is endorsing some version of freedom as autonomy, i.e. giving the law to oneself, in particular if one focuses on the necessity of law. For instance, Schelling refers to his notion of freedom as a ‘higher necessity’ or ‘an inner necessity that springs from the essence of the active agent’ (383) or ‘an inner necessity of the free being.’ Do these formulations not suggest that we are back at an ‘intelligible fatalism’ again? No. This would only be so, Schelling argues, if the being in question were ‘dead:’ in that case indeed ‘its activity would only ensue from necessity’ and ‘imputability and all freedom would be vitiated’ (385). But for ‘living beings,’ the case is quite different and this ‘inner necessity is itself freedom’ for, specifically in our case, ‘das Wesen des Menschen ist wesentlich seine eigene That’ (385).

51In other words, although Schelling is hard to interpret, the ‘higher necessity’ is not a necessity that applies higher up i.e. at the transcendental level. Rather, the necessity is down here at the empirical/phenomenal level. But its quality has changed because it is not longer a necessity that stems from the empirical level, for instance from some arbitrary essence that we happen to be endowed with. Instead, my actions certainly flow with necessity from my essence; but my essence is itself something that ‘I’ (in some sense) have freely chosen. This enables Schelling to reconcile Kant with Spinoza, whom he refers to in free indirect speech, saying ‘only that is free which acts according to the laws of its own inner Wesen and is not determined by anything else either within it or outside it’.⁶⁸

52So we are now in a position to understand how it is precisely Schelling’s account of the metaphysical implications of his conception of freedom that distinguish his position from Kant’s (and the rest of the idealist tradition). The distinction between Kant’s view that my noumenal self extra-temporally chooses an ‘intelligible character’ whose highest level maxim prioritizes private inclination over moral law and Schelling’s view that I choose my Wesen can seem rather subtle. But in fact, given Schelling normative critique of Kant, there is a lot packed into it. For Kant, despite the attempts of the Religion text, the self that chooses intelligible character is already constituted by autonomy and practical rationality, since Kant equates freedom to choose with autonomy and practical rationality. Thus the choice can really only go one way, for if I choose evil, then this outcome undermines my ability to say that I was a chooser i.e. a free agent. Schelling on the other hand takes the argument of the Religion text to its conclusion: that my constitution as an agent is independent of any notion of autonomy or practical rationality and hence of the moral law.⁶⁹

53As Kosch points out, Schelling says almost nothing about the content or justification of moral goodness in the Freiheitsschrift. He seems tacitly to assume that it is extensionally equivalent to the Kantian moral law. But he may also think it is justified in the same way i.e. in terms of practical rationality. For Kosch, this a serious problem because she thinks this must entail that the free agent is also constituted in the same way, and hence Schelling would fall back into Kant’s dilemma. But this is not necessarily so. Indeed it seems plausible that Schelling, although perhaps conflicted about the issue, does indeed identify morality, both in content and justification, with Kant’s view, while at the same time maintaining that the choosing agent should not be identified with practical rationality.

54Of course this would entail that agency is in part irrational. But Schelling suggests something very like this when he argues that ‘we do not … believe … that anyone can become virtuous or a hero or any kind of great man through pure reason’ (SW I/7, 413). More to my point, just this is the central lesson of Schelling’s ontologization of the two principles,⁷⁰ and especially of the dark ground: the dark ground comprises, in the phrase that animates Žižek’s reading of Schelling, the ‘indivisible remainder that cannot be resolved into reason’ (SW I/7, 360). At the level of the transcendental decision, this opacity of ourselves to our own rational scrutiny manifests itself in the fact that the ‘free act which becomes necessity cannot occur in consciousness, insofar as it is mere self-awareness and only ideal consciousness, since the act precedes it as it precedes all Wesen and indeed produces it’ (SW I/7, 386). Although only human beings face this free act, the two possible choices are constituted by principles with ontological depth. Agency requires the possibility of non-rational action; and non-rational actions are only possible if we are constituted by fundamental ontological principles that include something like a dark ground. Thus Schelling’s deepest objection to Kant is the latter’s identification of in-itself of things with something intelligible: the in-themselves, things must be inscrutable at a certain level, for otherwise it would be impossible for us to conceive of ourselves as agents. What for Kosch is the most problematic aspect of Schelling’s middle period works  ‒ his apparent acceptance of Kant’s understanding of morality through reason  ‒ is in fact the most fruitful and productive.

Conclusion

55Following Michelle Kosch’s analysis, I have argued that Schelling’s middle period works are responding to a deep worry about the nature of agency. The worry is that Kant and Hegel (and Schelling himself, earlier) are unable to account for a genuine notion of freedom, for freedom is linked definitionally to morality in various ways; and this link makes immoral actions ipso facto unfree, hence not actions. Thus we can (in a certain sense) only perform moral actions. And this is inconsistent with the idea that we can choose. This is why Schelling becomes interested in the notion of evil.

56Schelling can, I think, be understood in the general spirit of Naturphilosophie, as elaborating in the Freiheitsschrift the consequences of a robust libertarian notion of agency for our metaphysical conception of reality. There are several aspects to this elaboration, but the most significant is the idea that the dark ground can rise to the surface. The distinction of principles may have independent metaphysical motivation; but the protrusion of the dark ground into the empirical world is directly connected with the possibility of our own agency, as manifest in the possibility of choosing evil.