Kant's System of Practical Perspectives

by Stephen Palmquist (stevepq@hkbu.edu.hk)

Morality, by itself, constitutes a system. [Kt1:839]

1. The Shift from the Theoretical to the Practical Standpoint

Having proved their usefulness in the arduous task of interpreting the first and most widely studied of Kant's three Critiques, the interpretive tools established in Parts One and Two can now be employed in interpreting his second Critique, which develops a system based not on the theoretical, but on the practical standpoint. By interpreting 'system_p' in terms of the formal structure specified in Figure III.6, we shall find once again twelve essential steps in Kant's argument. But, whereas system_t adopts the logical perspective as its standpoint, the system examined here in Chapter VIII adopts the hypothetical perspective as its standpoint. (As we saw in II.4, the transcendental perspective serves as the Perspective for Kant's entire System.) Once this basic difference in their standpoints is discerned, numerous fundamental differences emerge between system_p and system_t. As Weber says in W14:251: 'In this new domain, the problems raised by the Critique of Pure Reason change in aspect; doubts are dissipated, and uncertainties give way to practical certainty.' So before delving into the details of system_p, we need to discuss some of the general similarities and differences between it and system_t, which arise out of their formal relationship.

Whereas system_t analyzes how the faculty of representation (i.e., theoretical reason, or the understanding [see VII.1]) functions in the process of gaining knowledge, system_p analyzes how the 'faculty of desire' (i.e., practical reason, or the 'will') functions in the process of a person's moral experience.[1] Kant defines the faculty of desire in a living being as 'the faculty such a being has of causing, through its ideas, the reality of the objects of these ideas' [Kt4:9n; s.a. 48; Kt7:178n]. Accordingly, the will itself functions as 'a kind of causality of living beings so far as they are rational' [Kt5:445-6]. Thus, system_p carries on within the limits set by system_t, picking up from stage four of system_t its concern for practical belief in ideas; yet it goes beyond system_t in the sense of forging a new path of perspectives. The crucial difference in this new path is that its 'object' is no longer 'a thing but only the manner of acting' [Kt4:60]. Therefore, 'the objects of these ideas' should no longer be regarded as objects of knowledge (i.e., representations), but as objects of action (i.e., ends).

Kant points out in Kt4:3 that the titles of the first two Critiques are not exactly parallel with each other: Kt1 is a critique of pure speculative reason, while Kt4 is a critique of impure practical reason. The difference between the two titles, then, is that the title of Kt4 leaves out the word 'pure' and inserts 'practical'. The reason for these changes is rooted in the difference between the improper use of reason from these two standpoints, against which Kant is warning. Whereas from the theoretical standpoint reason runs into illusion when it tries to reach conclusions without considering its object empirically (so that in response, the transcendental limits of sensibility must be put on its pure employment), from the practical standpoint reason runs into illusions when it tries to reach conclusions by considering its object empirically (so that in response, the transcendental limit of freedom from sensibility must be put on its impure employment) [see D2:36; cf. note VIII.10]. For this same reason, Kant adds '... of Pure Practical Reason' to the heading of each main section in Kt4, whereas the headings of the Aesthetic, Analytic and Dialectic in Kt1 do not specify '... of Pure Theoretical Reason': in Kt1 Kant argued against those who believe knowledge is possible through pure reason by showing how it must be impure; in Kt4 he now argues against those who believe morality has an impure source by showing how it must be pure.

Unfortunately, Kant devotes even less effort to explaining the general architectonic structure of his argument in system_p than he does in system_t.[2] But in Kt4:16 he seems to imply that the architectonic relationship between the stages of system_p is different from that between the stages of system_t. He states that the main change in the outline of the second Critique is that

the order in the subdivision of the Analytic will be the reverse of that in the critique of speculative reason. For in the present work we begin with principles and proceed to concepts, and only then, if possible, go on to the senses, while in the study of speculative reason we had to start with the senses and end with principles.

Whereas Kt1 starts with sensibility (stage one) and proceeds through concepts (stage two) to pure principles (stage three), Kt4 starts with the pure principle of the free will (stage one) and proceeds through concepts (stage two) to actual choices presented by sensibly determined inclinations (stage three).

This might tempt us to infer that the order of the first three stages in system_p is no longer 'matter, form, synthesis', but 'synthesis, form, matter'. However, Kant is not referring here to a change in the architectonic structure of the system [cf. Tables VII.1 and VIII.1]; rather, the way the elements manifest themselves in the system changes as a direct result of the change in standpoint. Thus, although the first stage of system_t is sensible, whereas that of system_p is nonsensible, both perform the same, material function. Likewise the stage effecting the synthesis of form and matter is that of the pure (nonsensible) principles in system_t; but in system_p it will be that of the (sensible) choice to follow duty rather than inclination. (The formal stage is unproblematic because it supplies the conceptual elements in both systems.) Although Kant's use of terms such as 'formal' and 'material' is not always easy to follow, an awareness of the standpoint assumed will usually reveal such references to be consistent with his architectonic logic. For the perspectival function of each stage relative to its own standpoint is always the same in both systems, even though its function relative to another standpoint might look different. In both systems, the first stage provides the necessary limiting condition which must be assumed by any proper use of reason from that standpoint; the second stage defines the logical form for conceiving the object; and the third stage explains how the real judgment (or moral action) comes about. Thus, both systems follow the same logical form: taking account of their respective Dialectics, both outlines are divided into four stages.

From this and other hints as to the procedure he is following [see Kt4: 89-91; Kt5:392], we can see that Kant wants the structure of the outline of Kt4 to follow as much as possible the same pattern as in Kt1: both books are divided into a Doctrine of Elements and a Methodology; both contain an Analytic and a Dialectic as major divisions of the Doctrine of Elements; and both develop systems with four main stages. There is, however, an important difference: instead of having an Aesthetic and two 'books' of Analytic, as in Kt1, Kt4 simply has an Analytic divided into three chapters. The reason for substituting an Analytic in place of the Aesthetic of system_t is that, as we have seen, system_p begins not with the faculty which limits reason (viz., sensibility), but with reason's own limitation of itself (free will).[3]

In his own account of this difference Kant makes a claim at one point which, as has often been pointed out, is 'objectively false' [see e.g., F5:294]. He says: 'The Analytic of theoretical pure reason [i.e., in Kt1] was divided into Transcendental Aesthetic and Transcendental Logic', with the latter divided into concepts and principles, whereas 'that of the practical reason is divided ... into logic and aesthetic', with the former divided into principles and concepts [Kt4:90]. What is usually ignored by commentators, however, is that Kant admits his 'analogy' is 'not entirely suitable' [90]; surely this means he has not forgotten that the Aesthetic in Kt1 precedes the Logic, and that the Logic is divided into Analytic and Dialectic [see Table III.1]. Rather than regarding this simply as a lapse of memory or a careless mistake, we should regard it as evidence that Kant was not entirely satisfied with the divisions made in Kt1, so he was still toying with other, more suitable arrangements of his architectonic pattern.[4] This revised organization of Kt1 is actually more suitable, because it correctly places the Aesthetic content of sensibility within the bounds of the Analysis of true knowledge without making it a part of Logic, and because it distinguishes more clearly between this Analytic task and the task of dispelling the illusions of false logic (Dialectic).

One of the chief dangers in interpreting system_p is to neglect the fact that, when Kant uses the same technical terms in both systems, terms such as 'principle', 'concept' or 'sense' [see e.g., Kt4:16], their meanings are bound to be different in the two systems, because their standpoints differ. Even Kant's key distinctions, such as a priori-a posteriori, analytic-synthetic and transcendental-empirical, take on a rather different meaning in this new system [see G14:liv]. In system_t the hypothetical perspective gives rise to certain analytic a posteriori beliefs [see IV.3]. When system_p adopts this perspective as its standpoint, such beliefs are no longer merely 'regulative' (as in system_t [see VII.3.B]); rather they gain 'objective reality' [Kt1:836]. This is possible only because the hypothetical perspective generates 'a special kind of systematic unity, namely the moral' [855]. Within the context of the resulting system, pure reason now formulates principles which themselves can be regarded as objectively valid for moral experience, and thus as synthetic a priori, even though their status as viewed from system_t would be analytic a posteriori.[5] This crucial difference stems from the fact that the four main stages of system_t adopt perspectives on knowledge (scientific experience), whereas those of system_p adopt perspectives on intentional action (moral experience) [Kt35:(1-2); Kt10:110(116)]. In system_p, as Kant asserts, 'reason is given to us as a practical faculty, i.e., one which is meant to have an influence on the will' [Kt5:396]; and 'will' is 'the capacity of acting according to the conception of laws, i.e., according to principles' [412e.a.]; indeed, 'will is nothing else than practical reason.' As Paton suggests, Kant's 'fundamental assumption is that the will is as rational in action as intelligence [i.e., the understanding] is in thinking.'[6]

The implications of this essential perspectival difference for the content of system_p are brought out by Kant in a concise description of the relationships between its first three stages:

... practical reason is concerned not with objects in order to know them but with its own capacity to make them real ... Consequently, it does not have to furnish an object of intuition, but as practical reason it has only to give a law of intuition [i.e., freedom] ... [It] must begin from the possibility of practical fundamental principles a priori [stage one]. Only from these can it proceed to concepts of the absolutely good and evil in order first to assign them in accordance with those principles [stage two] ... Only then could [it deal] with the relation of pure practical reason to sensibility and with its necessary influence on it, i.e., the moral feeling [stage three] ...

... the division of the Analytic of Pure Practical Reason must turn out to be similar to that of a syllogism, i.e., proceeding from the universal in the major premise (the moral principle [stage one]), through a minor premise containing a subsumption of possible actions (as good or bad) under the major [stage two], to the conclusion, viz. the subjective determination of the will (an interest in the practically possible good and the maxim based on it) [stage three]. [Kt4:89-91; cf. G14:li-liii]

If we assume that the Dialectic [Kt4:107-48] is the fourth stage, this summary of the threefold division of the Analytic reveals the following progression of stages in system_p: (1) the synthetic a priori perspective on action applies the principle of freedom to the undetermined will; (2) the analytic a priori perspective formally defines morality in terms of the moral law and its categories; (3) the synthetic a posteriori perspective describes the real actualization of duty through moral feeling; and (4) the analytic a posteriori perspective posits the ideal ends or implications of morality.

These four stages correspond directly to the four stages of system_t, as depicted in Figure VII.3 and in Table VII.1. Hence, we can construct a parallel table for system_p. All the aspects of system_p shown in Table VIII.1 will be discussed more thoroughly in the following two sections, which, like VII.2-3, will deal first with the abstract stages (free will and the moral law) and then with the concrete stages (moral judgment and the final end of morality).[7]

Table VIII.1: Basic Perspectival Relations in System_p

2. The Abstract Conditions of Moral Action (-)

A. Free Will (--)

The starting-point of system_p is the 'disposition' of an individual agent.[8] As 'the ground of moral responsibility' [S11:cxviii], the disposition guides 'the adoption of the particular maxims on which individual decisions are based' [cxv]; in so doing it 'gives rise to morality' [Kt35:(22)]. The disposition is 'the spirit of moral laws' [(52)]; since 'dispositions are the cardinal principles of action ... and their grounds of impulse', ethics itself can be called the 'philosophy of disposition' [(71)]. Yet Kant laments [(71)]: 'It is not easy to give an explanation of the exact meaning of disposition.'

The problem is that this 'ultimate subjective ground of the adoption of maxims' [Kt8:25(20)] is not, as it were, out in the open for all to view. On the contrary, we must 'confess', insists Kant, 'that we cannot cite a single sure example of the [good] disposition' [Kt5:406]. This is because an action's 'moral worth', viewed from the empirical perspective, is 'always doubtful' [406]: 'for when moral worth is in question it is not a matter of actions which one sees but of their inner principles which one does not see' [407]. Thus the disposition is to system_p what the thing in itself is to system_t: the transcendent presupposition which is required for entry into the system. Because it is transcendent, it is both impossible and unnecessary to judge as to the nature even of our own disposition: 'Only God can see that our dispositions are moral and pure' [Kt35:(80); s.a. Kt8:49-50(44-5); Kt6:392-3]. What is necessary is to adopt a moral faith, which might simply be called a desire to act. As such it is directly parallel to the theoretical faith required for system_t [see V.1], which is, as it were, a desire to know, and which is implicit in the original act of representation.

A morally good act ultimately requires a good disposition. But, because the disposition transcends both our knowledge and our action, we must regard the exact nature of its 'influence' on the elements in system_p as a mystery. Nevertheless, here in the first step of Kant's presentation of the conditions for moral activity, the agent must base its desire to act on the assumption of a good disposition, thus positing for itself a good will. A good will is 'the supreme condition of all good [actions]' [Kt4:62; s.a. Kt5:396 and P3:43], for only it can be 'good without qualification': it 'constitutes the indispensable condition even of worthiness to be happy' [Kt5:393]. Indeed, 'a good will is that whereby man's existence can alone possess an absolute worth, and in relation to which the existence of the world can have a final end' [Kt7:443]. It is, however, 'undetermined with reference to any objects' [Kt5:444]. Just as the first step in system_t establishes the most general, but as yet undetermined objective condition for knowledge (the transcendental object, with its assumed relation to the thing in itself), so also the first step here in system_p establishes the most general, but as yet undetermined subjective condition for moral action. This step can now be summarized as follows:

The second step also follows the pattern of system_t (where the object is intuited in space and time), by determining the formal condition of stage one to be 'the rational concept of transcendental freedom' [Kt7:211], viewed now from the practical standpoint [Kt4:3]. Although 'transcendental freedom' is a transcendent idea in system_t [Kt7:468], it becomes immanent in the form of the idea of 'practical freedom' here in stage one of system_p.[9] That practical freedom is the formal condition of the transcendental perspective of system_p is suggested by the fact that, like pure time and space in system_t, such freedom is a kind of brute fact which must be presupposed, even though it can never be 'an object of experience' [Kt7i:195]; 'how this presupposition itself is possible can never be discerned by any human reason' [Kt5:461]. This parallelism is, no doubt, what leads Kant to treat the second step of these two systems as the 'two pivots' around which Criticism itself turns: 'first, the doctrine of the ideality of space and time, which ... merely points towards what is supersensible but unknowable by us ...; second, the doctrine of the reality of the concept of freedom as a concept of the knowable supersensible' [Kt69:311]. This 'mere idea' of freedom 'holds only as the necessary presupposition of reason in a being that believes itself conscious of a will' [Kt5: 459]. Thus, when Kant declares that 'practical reason ... provides reality to a supersensible object of the category of causality, i.e., to freedom' [Kt4:6], he immediately adds: 'This is a practical concept and as such is subject only to practical use ...' By this he means that, as Paton rather boldly puts it, 'unless we can act on this presupposition there is no such thing as action, and there is no such thing as will' [P3:219].

In the third stage, as we shall see in VIII.3.A, 'the freedom of the will' manifests itself as 'autonomy, i.e., the property of the will to be a law to itself ... Therefore a free will and a will under moral laws are identical' [Kt5:446-7]. Since 'the Kantian definition of freedom (or autonomy) in general' is 'self-limitation' [Y2:975] or 'internal determination' [W10:9, cf. 53], the realization of the freedom of choice in moral activity (i.e., in stage three) depends on the prior assumption of practical freedom here in step two, as a limiting condition of the will 'which must be assumed, presupposed and believed of our own will if moral volition in general is to be conceived as a possibility for us' [W24:36; s.a. W5:213]. Just as the limiting condition of intuition in step two of system_t gives rise to a manifold of appearances, so also the limiting condition of freedom here in step two of system_p enables the will to produce what Kant calls a 'manifold of desires' [e.g., Kt4:65]. We can therefore summarize this step accordingly:

As with intuition in the second step of system_t, practical freedom can be divided into 'inner freedom' and 'outer freedom' [Kt6:406]. But this empirical distinction can be ignored, since it plays no part in Kant's Critique [Kt4], but only in his Metaphysics [Kt6], of morality.

Adopting freedom as the transcendental condition of reason's practical standpoint releases us from a strict confinement to sensible intuition (step two of system_t) as a means of interpreting our experience [Kt4:46; cf. Kt8: 50-1(46) and W24:74]; for 'freedom ... transfers us into an intelligible order of things' [Kt4:42], but only in the sense that it gives us a purely rational perspective on our experience.[10] Such freedom does not require 'another kind of intuition than the sensuous', i.e., knowledge of 'a causa noumenon' [Kt4:49], as some interpreters mistakenly allege [e.g., G14:11; see Ap. VIII]. For as Kant explains,

practical reason ... does not concern itself with this demand ... But the concept which reason makes of its own causality as noumenon is significant even though it cannot be defined theoretically for purposes of knowing its supersensuous existence [Kt4:49-50].

Now the concept of a being which has a free will is that of a causa noumenon ... But because no intuition, which could be only sensuous, can support [the] application [of this concept to transcendent reality], causa noumenon is, for reason's theoretical standpoint, an empty concept, although a possible and thinkable one [55-6].

But despite its inability to function positively in system_t, freedom does have a positive function in system_p: 'although an intelligible world ... is for us a transcendent conception [in the context of system_t]-as is also freedom itself, the formal condition of that world-yet it has its proper function' as an immanent conception in system_p [Kt7:404].

From the standpoint of system_p, then, freedom 'makes necessary the concept of an intelligible world' [Kt5:458e.a.] as a nonsensible perspective on experience. Because the fact of human freedom functions positively as 'a causality of reason' for system_p [458], Kant virtually identifies it with the
noumenal substrate of phenomena, enabling this practical concept to serve as 'the keystone of the whole architecture of [the Critical philosophy]' [Kt4:3]. In system_p this 'unconditioned causality, and its faculty, freedom, ... are determinately and assertorically known; thus is the reality of the intelligible world definitely established from a practical standpoint, and this determinateness, which would be transcendent (extravagant) for theoretical purposes, is for practical purposes immanent' [105]. But aside from this single positive role, practical reason does not generally enable us to 'think theoretically and positively' about the idea of freedom [133], for system_p does not convert the 'might be' of the fourth stage of system_t into an 'is' [104], but only into an 'ought to be'.

The third step in system_t was never clearly expounded by Kant, so we had to conjecture, based on the requirements of his architectonic, what he must have intended the reader to take for granted. But Kant expounds the third step in system_p much more clearly: it is that in order for a person to act freely there must be an 'end' presented as a possible 'object' of choice. Since its object also participates in system_t, the causality of freedom, though itself nonsensuous, is nevertheless always 'a cause that is sensuously determined in respect of its effects' [Kt7:465]. In system_p this object is supplied in the form of a 'maxim', a term Kant discusses quite thoroughly.

'A maxim is the subjective principle of volition' [Kt5:400n; s.a. 421n]; its function in stage one is to specify 'the will to act in a certain kind of way in a certain kind of situation' [P3:83]-for instance in a form such as 'I will do X as a means to Y' [84]. Because it is based on a good or evil will, a maxim is 'absolutely good or evil in all respects and without qualification' [Kt4:60]. In Kt4:33 Kant relates maxims directly to step two: 'freedom is itself the formal condition of all maxims, under which alone they can all agree with the supreme practical law [stage two]'. This clearly requires the formulation of the maxim to be located in step three. However, 'there is a great gulf between the maxim and the deed' [Kt8:46(42)], for maxims are not acted upon until stage three, where they are described as having three aspects, each of which corresponds to one of the other three stages: (1) 'A form, which consists in universality', through the influence of stage two; (2) 'A material, i.e., an end', through the influence of stage one; and (3) 'A complete determination ... [enabling them] to harmonize with a possible realm of ends [stage four of system_p] as with a realm of nature [system_t]' [Kt5:463]. Kant alludes to the maxim's dual function when he says: 'The term "act" can apply in general to that exercise of freedom whereby the supreme maxim ... is adopted by the will [in step three], but also to the exercise of freedom whereby the actions themselves [in step nine] are performed in accordance with that maxim.'[11] Wood describes the former as the act in which 'the autonomous agent himself rationally determines the matter of his maxim, and gives himself ends, rather than merely receiving his ends from nature' [W24:61].

According to Kant, an action's 'moral value' depends 'merely on the principle of volition [i.e., the maxim] by which an action is done, without any regard to the objects of the faculty of desire' [Kt5:399-400e.a.; s.a. 425]. As Paton explains, 'the setting of ends before oneself is the essential mark of freedom' [P3:181], because in this way the manifold of an agent's desires, which as such has no moral value, is distilled into a discrete maxim, which does. Therefore, any maxim through which an end (i.e., a possible effect) is chosen can serve to constitute the material of a moral action, provided it conforms to the further conditions set out in stage two. These considerations give us ample ground for summarizing step three as:

The activity of the free will provides the foundation for synthetic a priori propositions in a moral system, just as sensibility does so for a theoretical system [cf. IV.3 and VII.2.A]. Such propositions themselves are not fully revealed until stage three. So at this point in system_p we still know little more than the a priori 'possibility' of freedom: 'We do not understand [freedom], but we know it as the condition of the moral law which we do know' [Kt4:4]. For, although freedom is a necessary condition for the moral law [4n], nevertheless, looking back from the perspective of moral experience (stage three), 'it is the moral law which leads directly to the concept of freedom' [30]. Indeed, this practical freedom 'lies at the foundation of all moral laws' [96]; without it 'no moral law and no accountability to it are possible' [97]. The details of this close relationship between freedom and the moral law become more clear in stage two, to which we will now turn in order to discover how the free will can take on a truly moral character.

B. The Moral Law (+-)

The second stage in system_p corresponds in several respects to the second stage in system_t. Just as the heart of system_t is the Transcendental Deduction of the Categories, so also the former requires 'the deduction of freedom as a causality of pure reason' [Kt4:47-8], or more simply 'the deduction of the moral law'.[12] Probably Kant's most quoted description of how this moral law relates to the freedom of stage one is that, 'though freedom is certainly the ratio essendi of the moral law, the latter is the ratio cognoscendi of freedom' [4n]-a statement closely paralleling his famous account of the reciprocity of intuitions and concepts in Kt1:75. For just as intuitions would be blind without concepts, freedom would be an unjustified assumption without the moral law to explicate its cognitive form; and just as concepts would be empty without intuitions, so also the moral law 'would never have been encountered' without freedom as its essential content. In stage two, as Greene puts it rather loosely, practical reason 'organizes blind moral intuition into a rational moral apprehension' [G14:lii].

Unlike its counterpart in system_t [see Kt1:129-69], however, the deduction of the moral law, as Kant tells us in Kt4:46,

does not concern knowledge of the properties of objects, which may be given to reason from some other source; rather, it concerns knowledge in so far as it can itself become the ground of the existence of objects, and in so far as reason, by virtue of this same knowledge, has causality in a rational being.

This difference allows Kant's account of this deduction to be far simpler than that of the categories of system_t. After giving some clues in Kt5:453-63 as to how to put forward an a priori argument for its validity, Kant appeals in Kt4:42-50 to what he believes is the universal awareness of the moral law, as a 'fact' of our immediate experience, which is 'absolutely inexplicable' from the standpoint of system_t [43]. This law 'is given, as an apodictically certain fact, as it were, of pure reason, a fact of which we are a priori conscious, even if it be granted that no [empirical] example could be found in which it has been followed exactly.'[13] In this sense the moral law is for system_p what the categories are for system_t: viz., given forms which can be proved neither by logical nor by empirical proofs, but which must be utilized by any rational being who wishes to act morally (for system_p) or know empirically (for system_t). In both cases 'we impose [the law] on ourselves and yet recognize [it] as necessary in itself' [Kt5:401n]. But because reason gives itself the moral law, rather than drawing it from objects, Kant concludes that 'the objective reality of the moral law can be proved through no deduction' [Kt4:47]. Or, stated more accurately, its only deduction is the simple fact that it imposes itself upon the moral experience of all rational beings as the only possibility for the formal stage of system_p. As such, it 'is the formal rational condition of the employment of our freedom' [Kt7:450e.a.]: without this logical perspective in system_p, practical freedom would be, so to speak, 'stranded' in the realm of transcendent ideas.

As a result of this way of satisfying the need for a deduction, Kant never clearly and unambiguously explains how 'the law, as the formal aspect of the will' [Kt30:282n(67n)]-i.e., as the formal (+-) stage of system_p-develops according to a three step argument. Nevertheless, he does cite various ways of formulating the moral law in terms of principles (which he calls the 'categorical imperative'), apparently hoping that one or another of these will strike morally attuned readers as a 'fact' of their own moral experience. As it turns out, the basic moral fact of an inner 'law-abidingness' [P3:71; cf. 180] itself enjoys a status parallel to the categories in system_t (i.e., step five), and his other formulations of the categorical imperative give clues as to how we can reconstruct steps four and six as well [see note VIII.25].

Each of Kant's formulations of the categorical imperative includes one of two elements that fit into system_p at this point. The first is the requirement that moral agents be willing for their maxims to 'serve at the same time as the universal law (of all rational beings)' [Kt5:438; s.a. 402,421,434]. In Kt4:30 Kant expresses this formula of 'universality' as: 'So act that the maxim of your will [step three] could always hold at the same time as a principle establishing universal law.' And in Kt6:225 he describes the process more fully:

You must ... begin by looking at the subjective principle of your action. But to know whether this principle is also objectively valid, your reason must subject it to the test of conceiving yourself as giving universal law