Ought implies can, asymmetrical freedom, and the practical irrelevance of transcendental freedom

In this paper, I demonstrate that Kant's commitment to an asymmetry between the control conditions for praise and blame is explained by his endorsement of the principle Ought Implies Can (OIC). I argue that Kant accepts only a relatively weak version of OIC and that he is hence committed only to a relatively weak requirement of alternate possibilities for moral blame. This suggests that whether we are transcendentally free is irrelevant to questions about moral permissibility and moral blameworthiness


| INTRODUCTION
In his landmark article on Kant's compatibilism, Allen Wood observes that for Kant freedom is fundamentally asymmetrical: For Kant, freedom consists in a one-way difference, so to speak.[…] If an action is performed from sensuous motives, then it is a free action only if the agent could have acted instead on a priori motives.If a being acts on a priori motives, however, it may be free even if there is no possibility that it could have failed to act as it does (1984, pp. 82-83).
The notion of asymmetrical freedom is familiar from contemporary discussions on moral responsibility.Based on a number of so-called "Luther cases," Susan Wolf (1980Wolf ( , 1990) ) argues against a generalized version of the Principle of Alternate Possibilities (PAP).Whereas PAP requires alternate possibilities for both moral praiseworthiness and moral blameworthiness, Wolf maintains that while agents are morally blameworthy only if they could have done otherwise, they can be morally praiseworthy even if (or precisely because) they could not have done otherwise. 1 Timmermann explores Wood's observation further and contends on the basis of compelling textual evidence that Kant rejects PAP and is committed only to a requirement of alternate possibilities for moral blameworthiness (Timmermann, 2003, pp. 45-65).I will follow Timmermann and assume that Kant is committed to an asymmetry in the control conditions for moral praise and blame.Rather than defending this asymmetry, my aim is to explain it.The key to the explanation, I shall argue, lies in the principle Ought Implies Can (OIC).The upshot of my argument is that the concept of transcendental freedom is irrelevant to questions about moral permissibility and moral blameworthiness.This paper is structured as follows.In the following section, I demonstrate that one can derive from OIC a requirement of alternate possibilities for moral blameworthiness, but not one for moral praiseworthiness, and cite some textual evidence showing that Kant endorses the premises of this derivation.I show that this derivation underlies Kant's analysis of his key examples in his discussion of the third antinomy in Section 3. In Section 4, I demonstrate that the derivation potentially yields wrongness and blame incompatibilism and suggest that Kant employs the doctrine of transcendental idealism to avert this threat.In Section 5, I distinguish between various senses of "can" and show that whether the derivation yields wrongness or blame incompatibilism-and consequently whether Kantians needs transcendental idealism-depends on the type of "can" implied by "ought."Section 6 examines which version of OIC Kant accepts.I argue that he accepts only a relatively weak version of OIC and hence is committed only to a weak compatibilist requirement of alternate possibilities for blame.This implies that Kantians can reject wrongness and blame incompatibilism even if transcendental idealism is assumed to be false.

| OIC AND THE DERIVATION OF PAP-B
In one of his articles, Harry Frankfurt mentions almost in passing that "the appeal of PAP may owe something to a presumption that it is a corollary of the Kantian thesis that 'ought' implies 'can'" (Frankfurt, 1988, p. 95).Yet, he immediately proceeds to reject the idea that PAP and OIC are corollaries.In contrast to PAP, he explains, "the Kantian view leaves open the possibility that a person for whom only one course of action is available fulfills an obligation when he pursues that course of action and is morally praiseworthy for doing so" (1988, p. 96).Let us expound this insight in detail.
We can start by dividing the general version of PAP into separate requirements for praise and blame: PAP: S is morally responsible for A only if S could have done otherwise than A.
PAP-P: S is morally praiseworthy for A only if S could have done otherwise than A.
PAP-B: S is morally blameworthy for A only if S could have done otherwise than A.
PAP entails both PAP-P and PAP-B.To reject PAP-P or PAP-B (or both, for that matter) is thus to reject PAP.I will now demonstrate that PAP-B can be derived from OIC in combination with two other premises.OIC runs as follows: Kant is widely regarded as bringing OIC into prominence.The question, as Stern puts it, "is not whether Kant adopted the principle, but which version of the principle" (2004, p. 56).I postpone the discussion of this complex question until Section 6 and here merely stipulate the meaning of the various elements of the principle.
Since Kant endorses OIC only in relation to moral obligations (Kohl, 2015, pp. 701-703), I will assume that the "ought" is the "ought" of moral obligation. 2 This notion of "ought" is broader than Kant's narrow use of the notion in the passage in the Groundwork, where he argues that the "ought" is "out of place" in relation to the divine and holy will (GMS 4:414). 3Since Kant claims that the divine and holy will, too, "stand under objective laws (of the good)" and are hence subject to moral obligations, the proposed broader notion also applies to the divine and holy will.The sense of "can" featuring in OIC is crucial to my purposes, and for that reason, I defer the discussion of various notions of ability to Section 5, leaving the notion of "can" unspecified for now.Various accounts of the nature of the relation between "ought" and "can" have been proposed, but it is widely held that Kant understands the relation between obligation and ability as one of logical entailment. 4I will proceed on this assumption.
Kant's preferred mode of inference for OIC is modus ponens.He understands the "ought" of the moral law as a "fact of reason" and then, applying OIC, he infers that we can comply with the moral law.With Martin (2009, p. 110), we can call this the "Capacity-Expanding Application" of OIC: The derivation is valid.An example will be helpful to get the gist of it.Assume that Jane is morally blameworthy for making a false promise.If Jane is morally blameworthy, making a false promise was morally wrong.And if making a false promise was morally wrong, Jane morally ought to have acted otherwise.But Jane morally ought to have acted otherwise only if she could have acted otherwise.Therefore, Jane is morally blameworthy for making a false promise only if she could have acted otherwise.
Importantly, PAP-P cannot be derived from OIC.To be able to derive PAP-P from OIC, we would have to revise either (P2) or (P3).The easiest option is to revise (P3) as follows: (P3') S is morally praiseworthy for A only if it is morally wrong for S to A Although this strategy would enable us to validly derive PAP-P from OIC, (P3') is obviously false: we often praise agents for doing what is morally right.A more plausible version of (P3) that would enable us to derive PAP-P is as follows: (P3") S is morally praiseworthy for A only if it is morally permissible for S to A To be able to derive PAP-P from OIC by substituting (P3'') for (P3), however, we would also have to revise the second premise as follows: (P2') S morally ought to omit A if and only if it is morally permissible for S to A Like (P3'), (P2') is obviously false.That means that both strategies for deriving PAP-P from OIC fail.Since PAP-B can and PAP-P cannot be derived in a sound way from OIC, the derivation provides us with a powerful explanation of Kant's commitment to asymmetrical freedom.
Although it is not my aim in this paper to demonstrate for each premise that Kant accepts it, it is nonetheless helpful to begin by citing some textual evidence in support of each premise.(P1) simply states OIC and it seems beyond doubt that Kant accepts it.To cite but a few examples, Kant claims that an agent "must judge that he can do what the law tells him unconditionally that he ought to do" (MS 6:380), that "if the moral law commands that we ought to be better human beings now, it inescapably follows that we must be capable of being better human beings" (RGV 6:50), and finally that "duty commands nothing but what we can do" (RGV 6:47). 6The central textual evidence will be discussed in Section 6.
(P2) is an uncontroversial axiom of deontic logic.Besides the fact that Kant needs it to be able to derive moral duties from the categorical imperative, there is compelling textual evidence that he accepts the premise.In the Metaphysics of Morals, for example, he defines the deontic notions "obligatory," "permitted" and "forbidden" in terms of each other, taking the notion of duty or obligation as a starting point: "Duty is that action to which someone is bound [verbunden]."He subsequently defines permissibility in terms of obligation: "That action is permitted [erlaubt] (licitum) which is not contrary to obligation [Verbindlichkeit]."Finally he adds, "hence it is obvious what is meant by forbidden [unerlaubt] (illicitum)" (MS 6:222).Allow me to state the obvious: an action is morally forbidden (i.e., wrong or impermissible) if and only if it is "contrary to moral obligation," that is, if and only if we morally ought to omit it.That is what (P2) says.
(P3) can be called "no blameworthiness without fault" (Wallace, 1994, p. 135) or No Fault, for short (Haji, 2002, p. 189).There is little Kantian literature on this principle, but one clear confirmation that Kant accepts No Fault can be found in his discussion of praise and blame for attempts and intentions in the Lectures on Ethics.There, Kant claims that a morally objectionable attempt can be morally blameworthy, whereas a morally objectionable intention, defined as "a mere entertaining of the action" (VE 27:565), cannot.He justifies this claim as follows: "Where there is propositum [intention], no action can be imputed, since it is not yet an action, but where there is conatus [attempt], there can be such imputation, since that is already an action" (VE 27:292).This reasoning presupposes No Fault.
Although there is certainly more to be said about Kant's endorsement of each of the premises, a close reading of all the relevant passages is beyond the scope of this paper.The textual evidence cited above suffices, I take it, if it can be shown that the derivation plays a pivotal role in Kant's discussion of the third antinomy.

| THE DERIVATION AND THE THIRD ANTINOMY
The best confirmation of Kant's commitment  Turn to MALICIOUS LIE first.Determinism is the thesis that every event is necessary given a preceding state of the world in conjunction with the laws of nature. 7Now, assume that determinism is true and suppose someone tells a malicious lie.From the truth of determinism, Kant notes, it follows that the lie was necessary given the laws of nature and preceding conditions such as "a bad upbringing, bad company, […] the wickedness of a natural temper insensitive to shame, […] carelessness and thoughtlessness [and] the occasioning causes" (A554/B582).Now Kant makes an interesting observation: "Even if one believes the action to be determined by these causes, one nonetheless blames the agent" (A554-555/B583-584).
Why do we blame the liar despite the fact that his lie is fully determined by preceding causes?Let us briefly consider a position commonly referred to as "semi-compatibilism" (Fischer & Ravizza, 1998, p. 53).Semi-compatibilists grant that causal determinism is incompatible with genuine freedom to do otherwise, but they hold that moral responsibility is nonetheless compatible with causal determinism because agents can be morally responsible for their actions even if they could not have done otherwise.They would thus argue that we blame the liar because moral blameworthiness does not require alternate possibilities.

Kant gives a different explanation:
This blame is grounded on the law of reason, which regards reason as a cause that, regardless of all the empirical conditions just named, could have and ought to have determined the conduct of the person to be other than it is (A555/B583, emphasis added).This explanation takes for granted that if the liar were unable to do otherwise, our blame would be unjustified.It thus presupposes PAP-B.Kant thinks that our blame is justified and hence that PAP-B is satisfied.Indeed, he claims that despite the truth of determinism, the liar "could still have refrained from the lie" (A556/B584).The passage suggests that Kant thinks the liar could have done otherwise because he ought to have done otherwise, which in turn indicates that he relies on the capacity-expanding application of OIC.
Admittedly, in the quoted passage Kant merely juxtaposes "could have" and "ought to have" and does not explicitly state that the "can" is derived from the "ought."In THEFT, however, he makes the justificatory relation between "can" and "ought" explicit.He introduces the example as follows: If I say of a human being who commits a theft that this deed is, in accordance with the natural law of causality, a necessary result of determining grounds in preceding time, then it was impossible that it could have been left undone (KpV 5:95).This is a clear statement of what follows from determinism.Kant then raises the following question: How, then, can appraisal in accordance with the moral law make any change in it and suppose that it could have been omitted because the law says that it ought to have been omitted?(KpV 5:95, emphasis added) Whereas in MALICIOUS LIE, the two notions were merely juxtaposed, the word "because" in this passage clearly indicates that the "can" derives from the "ought."The passage thus suggests that Kant accepts PAP-B and thinks that PAP-B is satisfied based on the capacity-expanding application of OIC.Indeed, Kant thinks that the thief could have done otherwise despite the truth of determinism: So considered [i.e., as a thing in itself], a rational being can now rightly say of every unlawful action he performed that he could have omitted it even though as appearance it is sufficiently determined in the past and, so far, is inevitably necessary (KpV 5:98).
The careful reader might have noticed that Kant's reasoning here relies on the doctrine of transcendental idealism.This strategy will be discussed in the following section.Here, we should note that MALICIOUS LIE and THEFT confirm, first, that Kant insists on a requirement of alternate possibilities for blame based on his endorsement of OIC and, second, that he takes this requirement to be satisfied based on the capacity-expanding application of OIC. 8   To conclude this section, I want to call attention to the remarkable fact that Kant's examples in his discussion of the third antinomy are without exception examples of morally wrong actions.Besides MALICIOUS LIE and THEFT, Kant discusses a case of "unlawful conduct" (KpV 5:98-99), the actions of certain "born villains" (KpV 5:99-100), "a man who strikes another dead" (VE 27:502-503), again a thief (VE 27:540), and some unspecified "criminal cases" (VM 29:1019).In all of his discussions of the third antinomy, Kant mentions not a single example of a morally right action.This can hardly be a coincidence.And if it is not a coincidence, there must be an explanation for it.By showing that PAP-B can be derived from OIC, while PAP-P cannot, the derivation provides us with a powerful explanation of Kant's choice of examples in his discussion of the third antinomy.

Kant does cite examples of morally right actions in other contexts, but these examples differ markedly from
those in the third antinomy: when discussing examples of morally right actions, Kant stresses precisely that agents can freely do the right thing even if they could not have done otherwise.For example, he claims that a dutiful son who simply cannot help fulfilling his duty to his sick father "proves his freedom in the highest degree by being unable to resist the call of duty" (MS 6:382n.).Similarly, he relates of "an honourable man [who] cannot lie," and tells us that this man nevertheless "refrains [from lying] of his own will" (VE 27:267).This confirms the explanatory power of the derivation once more.

| THE DUTY-RESTRICTING APPLICATION OF OIC AND THE DOCTRINE OF TRANSCENDENTAL IDEALISM
Kant relies on the capacity-expanding application of OIC in MALICIOUS LIE and THEFT.Recall, however, that one person's modus ponens is another person's modus tollens.By accepting OIC, Kant thus also allows for the duty-restricting application.I will first show how determinism can undermine moral wrongness and moral blameworthiness on the assumption of OIC and then suggest that Kant employs the doctrine of transcendental idealism to prevent the dutyrestricting application of OIC.
Strikingly, Van Inwagen illustrates how determinism can undermine blameworthiness on the assumption of OIC by means of an example of a lie: If someone charges you with, say, lying, and if you can convince him that it was simply not within your power not to lie, then it would seem that you have done all that is necessary to absolve yourself of responsibility for lying.Your accuser cannot say, "I concede it was not within your power not to lie; none the less you ought not to have lied."Ought, as the saying goes, implies can (1983, p. 161).
To understand the force of this example, let us return to the derivation and focus on the intermediary conclusion, namely: (C1) if it is morally wrong for S to A, then S can omit A. The intermediary conclusion expresses a requirement of alternate possibilities for moral wrongness.This becomes more apparent when we reformulate it as the following materially equivalent claim: PAP-W: It is morally wrong for S to A only if S can omit A Van Inwagen assumes that you cannot omit the lie and infers from this that it was not morally wrong for you to lie (i.e., that it was not the case that you morally ought not to have lied).This follows on the duty-restricting application of OIC.But Inwagen concludes that you are absolved from moral responsibility for lying, thus assuming implicitly that if it was not morally wrong for you to lie, then you are absolved from moral responsibility for lying.He assumes, in other words, that PAP-W entails PAP-B.The derivation shows that this is not so.PAP-B follows only on the assumption of No Fault.
If, for the sake of the argument, we stipulate that determinism rules out the kind of alternate possibilities featuring in PAP-W and PAP-B, then anyone who accepts the first two premises of the derivation is committed to wrongness incompatibilism, and anyone who also accepts the third premise is committed to both wrongness and blame incompatibilism. 9I have argued that Kant accepts all three premises of the derivation.To be able to draw the conclusion that "nature […] does not conflict with causality through freedom" and that "this antinomy rests on a mere illusion" (A558/B586), he must thus show that determinism does not rule out alternate possibilities of the relevant kind.
Kant finds the "key" to doing so in transcendental idealism (A490-497/B518-525).Since it is not my aim to defend this strategy, the following rough sketch should suffice.According to transcendental idealism, things considered as appearance are subject to the forms of intuition (i.e., space and time) and the categories of the understanding (e.g., substance and causality), whereas things considered as things in themselves are not subject to these conditions.Agents, too, can be considered as having an "empirical character" subject to time, space, substance and causality and as having an "intelligible character" not subject to these conditions (A538-541/B566-569).
Based on the doctrine of transcendental idealism, Kant can claim that the liar could have omitted the lie in the sense that (a) had he chosen a different intelligible character, he would have had a different empirical character; and (b) had he had a different empirical character, he would have omitted the lie (Wood, 1984). 10Notably, these two counterfactual statements can be true even if considered as appearance the lie was necessary given the liar's empirical character, the circumstances of action and the laws of nature.Transcendental idealism thus enables Kant to claim that determinism does not rule out alternate possibilities of the relevant kind, thereby preventing the duty-restricting application of OIC.
Note, however, that this speculative tour de force seemed called for in the first place only because it was implicitly assumed that, if transcendental idealism were false, the kind of alternate possibilities featuring in the versions of PAP-W and PAP-B that can be derived from OIC would be incompatible with determinism.In the following two sections, I will argue that this assumption is false.

| DIFFERENT ABILITIES, DIFFERENT PRINCIPLES
Thus far, I have left the sense of "can" in the various principles unspecified.If we specify the relevant notions of ability, we will be able to see that accepting PAP-B need not amount to accepting a genuine requirement of alternate possibilities for moral blameworthiness.The reason is that there are stronger and weaker versions of PAP-B, depending on the type of ability at stake.
The distinction between general and specific abilities is helpful here (Honoré, 1964;Mele, 2003). 11Consider Janet.Janet has had an excellent philosophical education at a prestigious university, in the course of which she acquired the ability to give splendid conference talks.Three months ago, her paper was accepted for a conference and now she stands in front of the audience, ready to shine.Now consider Paul.Paul has had an excellent philosophical education at a minor college, in the course of which he acquired the ability to give splendid conference talks.
Three months ago, he received his PhD but now he is unemployed and, having no affiliation, he is not invited to conferences.Whereas Janet will give a splendid conference talk if she tries to do so, this does not hold for Paul.We can express this by saying that while Janet has the general and the specific ability to give a splendid conference talk, Paul only has the general ability to do so.
For our purposes, it is necessary to distinguish a third type of ability.Consider Kant's definition of transcendental freedom as "the ability [Vermögen] of beginning a state from itself, the causality of which does not in turn stand under another cause determining it in time in accordance with the law of nature" (A533/B561).Where Janet's general and specific ability to give a splendid conference talk depended on preceding conditions (e.g., her being invited to the conference), transcendental freedom is an ability to do something regardless of the preceding conditions.As Kant puts it in his discussion of MALICIOUS LIE, it is the ability to do or omit something "regardless of all the empirical conditions just named" (A555/B583).
We have thus distinguished between the following three types of ability: 1. General ability: S is competent to A 2. General ability and specific ability: S is competent to A; and if S tries to A, then S will A 3. Transcendental ability: S is competent to A; and S can do A regardless of the preceding conditions Similarly, we can distinguish between three versions of PAP-B: PAP-B 1 : S is morally blameworthy for A only if S had the general ability to do otherwise PAP-B 2 : S is morally blameworthy for A only if S had the general and specific ability to do otherwise PAP-B 3 : S is morally blameworthy for A only if S had the transcendental ability to do otherwise Note that the versions of the principle become stronger as the numbers in subscript go up: PAP-B 3 entails PAP-B 2 and PAP-B 2 entails PAP-B 1 , but not the other way around.
A brief explanation of each version of the principle is in order.PAP-B 1 is so weak that it is not genuinely a requirement of alternate possibilities for blame.This is because the notion of general ability is not under discussion in the free will debate: to the best of my knowledge, no incompatibilist in the contemporary debate has ever argued that if determinism is true, then anyone who at a given point of time does not give a conference talk must lack the general ability to give conference talks.Since PAP-B 1 is very weak, even semi-compatibilists can accept it; and indeed, many semi-compatibilists hold that agents are morally blameworthy only if they have the general ability to comply with moral principles.Wallace, for instance, holds that agents are morally blameworthy only if they possess "the powers of reflective self-control" (1994, p. 157) and Fischer and Ravizza consider "moderate reasons-responsiveness" to be a necessary condition for moral blameworthiness (1998, pp. 44-45).
PAP-B 2 is stronger than PAP-B 1 .The principle operationalizes the sense of "could have done otherwise" originally formulated by the classical compatibilists and further developed by proponents of the conditional analysis.In addition to requiring a general ability to comply with moral principles, PAP-B 2 says that agents are morally blameworthy only if they would have acted as they morally ought to do had they tried to do so.Determinism is a proposition about what must necessarily happen given the actual past and the laws of nature, but this is perfectly compatible with other things happening given a different past. 12Consequently, if determinism is true and someone told a lie, it can still be true that the person would have told the truth had she tried to do so.That means that PAP-B 2 does not yield blame incompatibilism. 13  PAP-B 3 is the strongest requirement.Whereas determinism entails that "among the causes in appearance there can surely be nothing that could begin a series absolutely and from itself" (A543/B571), transcendental freedom is "the ability of beginning a state from itself" (A533/B561).The transcendental ability to do otherwise is thus incompatible with determinism-unless transcendental idealism is true.Indeed, Kant claims that if transcendental idealism is false, "the necessity in the causal relation can in no way be united with [transcendental] freedom; instead they are opposed to each other as contradictory" (KpV 5:94).But if transcendental idealism is true, then "nature […] does not conflict with causality through [transcendental] freedom" and "this antinomy rests on a mere illusion" (A558/B586).
PAP-B 3 thus yields blame incompatibilism unless transcendental idealism is true.Now turn to OIC.Again, we can distinguish between three versions of the principle: OIC 1 : If S morally ought to A, then S has the general ability to A OIC 2 : If S morally ought to A, then S has the general and specific ability to A OIC 3 : If S morally ought to A, then S has the transcendental ability to A As with the various versions of PAP-B, the various versions of OIC become stronger as the numbers in subscript go up: OIC 3 entails OIC 2 and OIC 2 entails OIC 1 , but not the other way around.This has important implications for the derivation.Since the conclusion of an argument cannot be stronger than its weakest premise, PAP-B 1 can be derived from all versions of OIC; PAP-B 2 can be derived from both OIC 2 and OIC 3 but not from OIC 1 ; and PAP-B 3 can be derived only from OIC 3 .The crucial question is thus which version of OIC Kant accepts.

| KANT'S VERSION OF OIC
Kantians commonly take it that Kant is committed to OIC 3 .Vilhauer, for one, claims that Kant "appeals to the 'OIC' principle to argue from the claim that we know we ought to act in certain ways to the claim that know we can act in those ways, which he claims to imply that we know we are transcendentally free" (in press, p. 4).Assuming that knowledge is justified true belief, Vilhauer claims that Kant endorses OIC 3 .I will argue, by contrast, that Kant endorses a significantly weaker version of the principle.

| The gallows examples
Kant's central analysis of OIC is found in §6 of the second Critique, where the aim is to show that we become aware of our freedom through our consciousness of the moral law.Kant illustrates his analysis by means of two examples, known as the GALLOWS EXAMPLES.Concretely, the question under discussion is whether we have the ability to resist our sensuous desires and how we become aware of this ability.Many courts use the so-called "policeman at the elbow" test to determine whether a defendant could have resisted her desires.The test asks whether the defendant would have committed the crime if a police officer had stood at her elbow.If the answer is negative, the defendant could have resisted the desire; if it is affirmative, the defendant acted on an irresistible impulse.
Kant develops a similar test.Consider his first example BROTHEL: Suppose someone asserts of his lustful inclination that, when the desired object and the opportunity are present, it is quite irresistible to him; ask him whether, if a gallows were erected in front of the house where he finds this opportunity and he would be hanged on it immediately after gratifying his lust, he would not then control his inclination.One need not conjecture very long what he would reply (KpV 5:30).
The answer that is understood is that the man would know for sure that he would not act on his sexual desire in those circumstances.The counterfactual scenario thus makes the man aware of his ability to resist this desire.
Note, however, that the most plausible explanation of why the man believes he would resist his sexual desire in the counterfactual circumstances is that he believes that in those circumstances he would be unable to resist another sensuous desire, namely the desire to preserve his life.Consequently, even if BROTHEL makes the man aware of his freedom from his sexual desire, it does not yet make him aware of his freedom from sensuous desire as such.To this end, Kant develops a second counterfactual test.Consider FALSE TESTIMONY: But ask him whether, if his prince demanded, on pain of the same immediate execution, that he give false testimony against an honorable man whom the prince would like to destroy under a plausible pretext, he would consider it possible to overcome his love of life, however great it may be.He would perhaps not venture to assert whether he would do it or not, but he must admit without hesitation that it would be possible for him.He judges, therefore, that he can do something because he is aware that he ought to do it and cognizes freedom within him, which, without the moral law, would have remained unknown to him (KpV 5:30).
Note that the two examples have a different outcome.In BROTHEL, the man knows that he would resist his sexual desire due to his inability to resist his desire to preserve his life; in FALSE TESTIMONY, on the other hand, he is unsure about whether he would refuse to give false testimony and swing from the gallows or submit in face of the threat of execution.The capacity-expanding application of OIC explains why.The decisive difference between BROTHEL and FALSE TESTIMONY is that only the latter example involves a moral "ought."Kant stipulates that the man morally ought to refuse to give testimony even if he is under threat of immediate execution.One could of course deny this, but that is beside the point.The point is that if the man believes that he morally ought to refrain from giving false testimony, then he must believe that he can do so.That explains why he is in doubt about how he would act in FALSE TESTIMONY while having no doubts about how he would act in BROTHEL.Kohl (2015) provides us with the most elaborate analysis of Kant's endorsement of OIC to date.My analysis of the GALLOWS EXAMPLES diverges from his in at least one important respect.I have stressed that if the protagonist of FALSE TESTIMONY morally ought to refrain from giving false testimony, then he must believe that he can refrain from doing so.Call this principle Ought Implies Epistemic Possibility.By contrast, Kohl claims that for Kant the "ought" entails not merely "a mere open deliberative alternative" but rather "a genuine power to perform acts of choice on the basis of moral motives" (2015, pp. 698-699).On this analysis, the kind of possibility that is entailed by the moral "ought" is thus ontological rather than epistemic.Let us refer to this principle as Ought Implies Ontological Possibility.
In defense of his claim, Kohl adduces the following argument: "If oughts could not govern rational efforts, they would be pointless and false as principles of practical reason" (2015, p. 694).Call this the argument from action-guidance.Copp (2003Copp ( , 2008) ) developed this argument outside of the Kantian context.Assessing it will helpful in determining whether the GALLOWS EXAMPLES support an ontological or merely an epistemic version of OIC.This is how Copp summarizes the argument from action-guidance: If an agent is morally required to do A in a particular situation, then all other options she faces are morally ruled out.If the agent cannot do A, then doing A is not among her options.Hence, if an agent is morally required to do A but cannot do A, then all of her options are morally ruled out (Copp, 2003, p. 274; see also Copp, 2008, p. 71).
Copp claims that if all options were ruled out, the moral requirement would fail to guide action.Since guiding action is precisely the point of moral requirements, however, the moral requirement would in those circumstances be pointless and hence invalid.
To evaluate this argument, note that the options under discussion are deliberative options.According to Copp, deliberative options can be ruled out morally (when doing A would be morally wrong) or ontologically (when the agent cannot do A).In constructing the argument, Copp assumes that if the agent cannot A, then doing A is not among the deliberative options of the agent.This assumption is false."From the mere fact that an agent in fact cannot do A," Fischer correctly observes, "it does not follow that she knows that she cannot do A" (Fischer, 2006a, p. 349).In other words, a lack of ontological options does not entail a lack of deliberative options.
To make sense of this, consider FRANKFURT-STYLE FALSE TESTIMONY: Prince Black demands, on pain of immediate execution, that Jones give false testimony against an honorable man White, whom Black would like to destroy under a plausible pretext.Unbeknownst to Jones, Black implanted a device in Jones' brain, which ensures that if Jones were about to refrain from giving false testimony, he would still choose to act as Black desires.Jones eventually submits in face of the threat of execution, leaving the device inactive.
The moral requirement clearly guides Jones' deliberation under the assumed conditions.After all, if giving false testimony against honorable men were morally permitted, Jones would probably not have given it a thought at all.It would thus seem that Jones morally ought to refrain from giving false testimony against White even if, given the setup of the example, he is unable to do so.The argument from action-guidance thus fails to support Ought Implies Ontological Possibility.Indeed, FRANKFURT-STYLE FALSE TESTIMONY reveals that a moral requirement can guide an agent's deliberation as long as long the agent believes that she can comply with the moral requirement.This implies that the argument from action-guidance supports an epistemic rather than an ontological version of OIC.If the argument Epistemic-OIC 3 : If S morally ought to A, then, for all S knows, S has the transcendental ability to A Allison, for example, elucidates Kant's idea as follows: "We cannot conceive of ourselves as rational agents without attributing to our agency an 'intelligible character,' capable of determining itself to act on the basis of rational principles, 'independently of the conditions of time '" (1990, p. 41).Similarly, Korsgaard (1996, p. 163) understands the idea as entailing that "we must regard ourselves as having free will," where free will is understood to be compatible with determinism (p.162) only if transcendental idealism is true (p.187, footnote 29). 15  The insight underlying Kant's idea is that when deliberating about whether to do A or B, we exhibit the belief that we can do A and the belief that we can do B. 16 The crucial question, however, is which sense of "can" is referred to here.To determine the relevant notion of ability, Bok (1998, p. 109) cites Sartre's example of a young man who is torn between joining the Free French and caring for his mother.She imagines a determinist telling the young man that he has only one genuine alternative, even though the determinist has to admit that she does not know which of the young man's alternatives it is.The young man retorts: The truth of determinism gives me no reason to abandon either my mother or my country; and as for my "alternatives," it implies only that whichever choice I make will have been the only choice that I could possibly have made.I don't know which choice that is, and your arguments imply that I cannot possibly know which of my alternatives you would call my only real possibility until I have already chosen one of them.So I am left with exactly the same decision, to be made among exactly the same alternatives, whether or not determinism is true.And now, if you will excuse me, I'll try to make it (Bok, 1998, p. 109).
Bok's example illustrates that it is perfectly compatible to believe in determinism and to rationally deliberate about a choice between two options as long as we do not know which option we will choose.We can imagine the protagonist of FALSE TESTIMONY engaging in a similar discussion with the determinist.This reveals that the protagonist of FALSE TESTIMONY can rationally deliberate about whether he should give false testimony or swing from the gallows without manifesting the belief that he has the transcendental ability to choose either option.It reveals, in short, that epistemic-OIC 3 is false. 17  A Kantian objector might reply as follows: "I agree that rational deliberation is compatible with a belief in determinism, but I deny that one can rationally deliberate about a choice between two options without manifesting the belief that one has the transcendental ability to choose either option.Rational deliberation is compatible with determinism because transcendental idealism is true." Let me make the nature of this dispute clear.We can explain our intuitions about the example either by saying that (a) epistemic-OIC 3 is false, or by saying that (b) epistemic-OIC 3 is true and transcendental idealism is true.Which explanation is preferable?Applying Ockham's razor, we can say that if there are two explanations of the same phenomenon, then the simpler explanation is to be preferred.And in this case the first explanation is no doubt the simpler one.Moreover, it would seem that the Kantian objector introduces transcendental idealism as an ad hoc hypothesis merely to prevent epistemic-OIC 3 from being falsified.But the biggest problem for the Kantian objector is that if transcendental idealism is true, the young man's belief in transcendental freedom will not be ruled out in any possible scenario.Yet there will definitely be scenarios in which rational deliberation is ruled out.
Let us slightly amend Bok's example to illustrate this.Suppose that the young man's mother lives in England and that due to the war the only way he can reach her is by swimming across the Strait of Dover.Even if the young man knows how to swim and how to care for other people, he is convinced that he cannot make it to his mother by swimming across the Strait of Dover.This puts an end to the young man's deliberation.The young man may certainly regret not being able care for his mother, but there is nothing to deliberate about in the assumed circumstances: the young man should simply join the Free French.
How can this be explained?We may assume that the young man is aware of his general ability to care for other people and his general ability to swim, but quite plausibly the young man lacks the belief that he also has the specific ability to care for his mother under the stipulated conditions; that is, he does not believe that if he tries to reach England by swimming across the Strait of Dover, then he will reach England.
The two versions of the example thus support the following principle of rational deliberation: Rational deliberation: If S rationally deliberates about whether to do A or B, then S manifests the belief that he has the general and specific ability to do A and the belief that he has the general and specific ability to do B Assuming that a moral "ought" is valid only if it can guide rational deliberation, this yields the following version of OIC: Epistemic-OIC 2 : If S ought to A, then, for all S knows, S has the general and specific ability to A The proposed principle of rational deliberation provides us with a simple explanation of our intuitions about the two versions of the example.The young man can rationally deliberate about whether to care for his mother or join the Free French in Bok's version of the example, but not in the amended version of the example, because the consequent of the principle is true in the former and false in the latter version of the example.

| "Possible under natural conditions"
Thus far, I have focused on passages in which Kant defends an epistemic version of OIC, but it cannot be denied that in some passages Kant asserts a version of Ought Implies Ontological Possibility.In the third antinomy, for example, he writes that "the action must be possible under natural conditions if the ought is directed to it" (A547/B575).Similarly, in On the Common Saying he writes that "it would not be a duty to aim at a certain effect of our will if this effect were not also possible in experience" .Kohl (2015) rightly notes that the notions of "possible under natural conditions" and "possible in experience" go beyond both logical possibility (pp.692-693) and epistemic possibility (pp.691-692).The kind of possibility at stake here must be of some ontological kind.
Let us examine what Kant means by "possible under natural conditions" and "possible in experience."Here my analysis concurs with that of Kohl.Kohl reads the modality at stake in these passages in light of the notion of "real possibility," a notion that Kant commonly contrasts with logical possibility.Whereas logical possibility requires merely that a concept be free of contradiction, Kant claims that "the postulate of the [real] possibility of things […] requires that their concept agree with the formal conditions of an experience in general" (A220/B267).The formal conditions of experience are, of course, the forms of intuition (i.e., time and space) and the categories of the understanding (e.g., causality).
Based on this interpretation, we see that in the passages quoted above Kant endorses a principle that might be called Ought Implies Nomological Possibility: Ought Implies Nomological Possibility: If S morally ought to A, then A is consistent with the forms of intuition and the categories of the understanding.This principle rules out that we morally ought to do things that are incompatible with linear time, Euclidean space or the laws of nature.Notable examples are seeing tomorrow's sunrise already today, making past actions undone, being in Berlin and New York at the same time, and traveling faster than the speed of light. 18 This principle is plausible enough, but it is weaker than OIC 1 .To see why, note that it is compatible with Ought Implies Nomological Possibility that we morally ought to do things that we do not have the general ability to do.For example, it would be compatible with Ought Implies Nomological Possibility that we morally ought to fly like a bird or live under water like a fish.We lack the general ability to do these things, but birds and fish demonstrate that these things are nomologically possible.
Possibly for this reason, Kohl revises the Ought Implies Nomological Possibility by including facts about the general abilities of the addressee of a moral "ought.""A valid prescription that an agent should aim at a certain effect," he asserts, "implies that the addressee has the general capacities to produce this effect" (2015, p. 693). 19Although it may not be easy to find explicit textual evidence in support of this revision, the revision makes sense from a systematic perspective and seems coherent with the broader Kantian framework.I will thus adopt it.Importantly, however, the revised version of Ought Implies Nomological Possibility is not stronger than OIC 1 .

| Kant's versions of OIC, PAP-W and PAP-B
Let us take stock.A detailed analysis of the GALLOWS EXAMPLES and the idea that we must act under the idea of freedom revealed that Kant is committed to epistemic-OIC 2 , which says that if S ought to A, then, for all S knows, S has the general and specific ability to A. Other passages suggest that Kant also understood OIC ontologically such that if S morally ought to A, then A is nomologically possible and S has the general ability to A. None of the central textual passages supports either OIC 3 or epistemic-OIC 3 .That means that the moral "ought" implies neither the actuality of nor the belief in transcendental freedom.
Let us combine these findings into a single principle.Since general ability entails nomological possibility (as we do not have the general ability to break the laws of nature), we can leave out nomological possibilities for the sake of simplicity and formulate Kant's combined version of OIC as follows: OIC Kant : If S ought to A, then S has the general ability to A and, for all S knows, S has the general and specific ability to A This principle ties in nicely with the widely shared assumption that, in contrast to objectivist accounts of moral obligation, the notion of moral obligation is fundamentally tied to the deliberative framework for Kant.Since the conclusion of an argument cannot be stronger than its weakest premise, the strongest versions of PAP-W and PAP-B that can be derived from Kant's version of OIC are the following: PAP-W OIC-Kant : It is morally wrong for S to A only if S has the general ability to omit A and, for all S knows, S has the general and specific ability to omit A PAP-B OIC-Kant : S is morally blameworthy for A only if S had the general ability to omit A and, for all S knew, S had the general and specific ability to omit A I have argued that these requirements of alternate possibilities for moral wrongness and moral blame are compatible with determinism even if transcendental idealism is assumed to be false.

| CONCLUSION
In this paper, I have demonstrated that the Kant's commitment to an asymmetry between the control conditions for moral praise and blame can be explained by reference to his endorsement of OIC.A close analysis of the key 1.If S ought to A, then S can A 2. S ought to A ∴ S can A But one person's modus ponens is another person's modus tollens.Martin (2009, p. 110) refers to this mode of inference as the "Duty-Restricting Application" of OIC: 1.If S ought to A, then S can A 2. S cannot A ∴ It is not the case that S ought to A We can now examine how PAP-B can be derived from OIC.Let us refer to this argument as the derivation of PAP-B or simply as the derivation 5 : (P1) If S morally ought to omit A, then S can omit A (OIC) (P2) S morally ought to omit A if and only if it is morally wrong for S to A (axiom of deontic logic) (C1) If it is morally wrong for S to A, then S can omit A (P3) S is morally blameworthy for A only if it is morally wrong for S to A (No Fault) (C2) S is morally blameworthy for A only if S can omit A (PAP-B) to the premises of the derivation is found in his examples in his discussion of the antinomy between freedom and determinism.The central examples are MALICIOUS LIE in the dialectic of the Critique of Pure Reason (A554-556/B582-584) and THEFT in the critical elucidation of the Critique of Practical Reason (KpV 5:95-98).