1. Introduction
The Transcendental Dialectic of the Critique of Pure Reason is largely given over to Kant’s famous critique of reason’s metaphysical misuse. Indeed, so lengthily does Kant dwell on reason’s self-thwarting dialectical tendencies, it almost comes as a surprise when he insists in the Appendix that reason can play a legitimate role in advancing empirical knowledge. Kant’s account of this legitimate role hinges on his claim that reason operates according to a supreme principle of systematic unity. By regulating the activity of the understanding via this principle, reason helps unify and systematize the empirical judgements generated by the understanding. Thus, although reason’s ideas cannot provide knowledge of objects beyond the bounds of sense, as classical metaphysics wrongly assumes they can, reason’s chief idea – that of systematic unity – nevertheless plays an extremely useful, if not indispensable, role in enabling our acquisition of knowledge within those bounds.
While the broad outlines of Kant’s account are clear enough, the details are not. The chief problem is that Kant presents theoretical reason’s principle in two seemingly contradictory ways, all the while claiming that the principle has both, apparently contradictory, sets of features. On the one hand, he claims that the principle is transcendentally necessary for any coherent use of the understanding at all, while elsewhere suggesting that it is no more than a conditionally necessary heuristic aid.Footnote 1 Equally, Kant claims that the principle is ‘subjective’, and a mere ‘maxim’, while also stating that it is ‘objectively valid’ and, indeed, ‘objectively necessary’ (A648/B676, A651/B679, A666/B694, A680/B708).
Most interpreters concerned to clarify the status of reason’s principle are interested in clarifying the relationship of reason to the faculty of the understanding, and therewith the relationship of the Transcendental Dialectic to the Transcendental Analytic.Footnote 2 The motivating worry is that in the Analytic Kant strongly implies that the understanding can apply its concepts to experience without reference to the systematic unity ideal given by reason. If, therefore, reason’s principle turns out to have transcendental status, as Kant at least occasionally claims in the Appendix, this would imply a major revision of the conditions of the possibility of experience laid out in the Analytic. Whether the Analytic should indeed be reread through the lens of this putative revision is thus the focus of debate, with the very coherence and meaning of Kant’s first Critique account of cognition hanging in the balance.
While the above concerns are undoubtedly weighty, my own interest in reason’s regulative principle comes from a different direction. Kant’s critical enterprise concerns the status, meaning and value of reason as a whole, or to say the same thing, reason taken as a ‘unity’.Footnote 3 The character of reason’s supreme theoretical principle as presented in the Appendix must therefore not only be made to cohere with his first Critique account of cognition, but also with his Critical reconceptualization of reason as a whole. While the methodological reading of theoretical reason I examine in this article aims to solve coherence problems internal to the first Critique, it does so – I shall argue – in a way that is at odds with Kant’s larger Critical reconceptualization of reason. My aim is to suggest a revision of the methodological reading that brings it into harmony with Kant’s larger Critical project.
In section 2 I examine Kant’s presentation of reason’s regulative principle in the Appendix. In section 3 I turn to examining the methodological interpretation of this principle, defended in different iterations by Patricia Kitcher, Paul Guyer, Marcus Willaschek and others.Footnote 4 I argue that, on this interpretation, the principle is taken – implicitly if not explicitly – to be normative in character. I go on to claim that if the principle is indeed normative, then it may usefully be interpreted as a species of practical principle, the likes of which Kant discusses in the Groundwork of the Metaphysics of Morals. While this suggestion may strike some as too unorthodox, I argue that it makes sense both exegetically and philosophically, and that through it we may clarify the full implications of the methodological view. In section 4 I argue, on both conceptual and exegetical grounds, that adherents of this view are wrong to hold that reason’s principle binds hypothetically, and that if the principle is indeed normative, then it binds agents categorically instead. In section 5 I argue that the hypothetical construal of reason’s principle also conflicts with Kant’s overarching attempt to vindicate reason, in all of its employments, as a self-determining power. This last shortcoming is particularly important since it suggests that the methodological view misconstrues the nature of Kant’s Critical enterprise as a whole.
2. The Regulative Principle of Systematic Unity
In the Appendix to the Dialectic Kant seeks to analyse and justify the regulative use of the ideas of theoretical reason, and in particular reason’s fundamental idea, that of the systematic unity of nature. Kant’s analysis relies on a distinction he draws between subjective and objective principles of systematic unity, each of which has a distinct status and appears to bind with a distinct kind of necessity. Subjective principles bring with them no valid claim about objects and, Kant implies, bind with mere methodological force (A652/B680, A663–4/B691–2, A666/B694, A671/B699, A680/B708). On this picture, the use of such principles may be a necessary means of advancing the frontiers of science, say, but this goal is by no means taken as obligatory for all to pursue. Kant suggests that objective principles of systematic unity, by contrast, do make valid a priori claims about objects, and indeed about the character of empirical nature as such (to wit: that it is a systematic unity) (A648/B676, A651/B679). Moreover, Kant suggests that objective principles bind with a stronger, transcendental form of necessity, since they represent an a priori condition on the cognitive representation of nature in general (A644/B672, A650/B678, A663/B691). On this picture, linking bits of cognitive information in systematically unified ways (as though nature were itself a unity) just is what all cognitive agents necessarily do whenever they apply empirical concepts and/or form empirical judgements.
Kant seeks to justify the regulative use of reason in the Appendix by arguing that the subjective-methodological principles of systematic unity are grounded in an underlying set of objective principles, which themselves bind with transcendental necessity (A658/B686, A662/B690). Kant writes:
In fact it cannot even be seen how there could be a logical principle of rational unity among rules unless a transcendental principle is presupposed, through which such a systematic unity, as pertaining to the object itself, is assumed a priori as necessary. (A650/B678)
Reason, however, like all mental faculties on Kant’s view, has ultimately only one set of principles, or to be more precise, only one supreme principle according to which it operates (CPrR, 5: 120). The upshot of his justificatory argument, then, is that there can be no coherent choice between subjective-methodological and objective-transcendental versions of reason’s principle. The supreme principle of systematic unity must itself have both these aspects, dimensions or sets of features.
Despite the clear conclusion Kant wishes to draw here, many commentators struggle to make sense of the single principle he has in view. This is because Kant’s discussion leaves it far from clear how there could be a merely regulative (or, if you will, subjective-methodological) principle that is nonetheless transcendentally grounded. To many it seems that transcendental principles are objective in a manner that makes them necessarily non-regulative, and that regulative principles are subjective-methodological in a manner that rules out the possibility of their having transcendental force. Hence many suppose that the principles for which Kant seeks transcendental status cannot be genuinely regulative; and that the principles for which he claims mere subjective-methodological status cannot be genuinely transcendental, or at least not in any usual sense.
The above problem makes interpretation difficult. One way of coping with this difficulty is to suppose that Kant himself would ultimately agree that reason’s principle cannot in fact have both sets of features, so that what is ultimately conveyed in his justificatory argument is, finally, his inability to make up his mind concerning the full transcendental status of reason’s principle. This reading is most famously expressed by Norman Kemp Smith, who accuses Kant of ‘wavering between [mutually incompatible] subjective and objective interpretations of the Ideas of Reason’ (Kemp Smith 1964: 547).Footnote 5 In keeping with the spirit of Kemp Smith’s reading, many approach the Appendix by privileging either a subjective-methodological or an objective-transcendental version of reason’s principle, while seeking to discount those passages in which Kant suggests that the principle has the other set of features.Footnote 6 The broadly shared assumption is that the principle of systematic unity cannot in fact be both subjective-methodological and objective-transcendental at the same time.
3. Reason’s Regulative Principle Methodologically Construed
Despite their differences, those who read reason’s principle of systematic unity methodologically tend to agree that it is not a transcendental condition on the possibility of experience, affirming instead that it is a heuristic device or methodological aid, necessary for facilitating the expansion of empirical knowledge in particular ways.Footnote 7 On this reading, it is only by inquiring into nature as though it were a systematic unity that we are able to achieve knowledge of a certain kind.Footnote 8 This as-if supposition about the character of nature is understood as a merely useful aid to our practices of inquiry. It is not thought to be a transcendental claim with any kind of objective import. So, for example, Patricia Kitcher writes that we are justified in using reason’s principle methodologically even though the principle’s validity ‘does not rest on the correctness of Nature as systematically unified’ (Kitcher Reference Kitcher1986: 211).
This interpretation draws upon a wealth of textual evidence in which Kant emphasizes the non-objective character of reason’s principle and, indeed, those passages where he explicitly claims a merely subjective and/or methodological status for it, as for instance when he writes: ‘[t]he unity of reason is the unity of a system, and this systematic unity does not serve reason objectively as a principle, extending it over objects, but subjectively as a maxim’ (A680/B708; cf. A652/B680, A663–4/B691–2, A666/B694, A671/B699). Although much of Kant’s characterization of the principle proceeds in this vein, proponents of the methodological reading must also make sense of those passages in which Kant suggests that it has transcendental and/or objective status instead. Indeed, they must make sense of the justificatory argument in which Kant argues that the subject-methodological principles of systematic unity do in fact have transcendental grounding. Different authors handle these passages differently. One strategy is to argue that Kant cannot consistently be committed to the transcendental claim, since it contradicts his arguments for the merely regulative character of the principle, and that these arguments are the ones that reflect his more settled view.Footnote 9 Others insist that the sort of transcendental grounding Kant wishes to attribute to reason’s principle cannot possibly be on a par with the transcendental status enjoyed by the principles of the understanding, since this would contradict its ‘merely subjective’ character.Footnote 10
The interpretative difficulties here are formidable. My aim, however, is not to evaluate the merits of the methodological reading on textual grounds, but rather to draw our attention to one of its central implication. For present purposes what I wish to highlight is that the methodological reading is committed to a normative construal of reason’s principle. What does this mean? Proponents of the reading see the principle as establishing normative standards for guiding cognitive activity, rather than as causally determining or descriptively characterizing the nature of that activity. This comes out most sharply in the way the methodological reading contrasts with the transcendental one, its main foil. The broadly shared methodological idea is that seeking systematic unity is not something all agents necessarily do – as though automatically – whenever they use their understanding in an attempt to cognize nature. Rather the idea is that seeking such unity is something agents rationally ought to do under certain circumstances, but may in fact fail to do (through exercises of cognitive agency that are less than fully or perfectly rational). Unlike the principles of the understanding, therefore, which agents cannot help but use in representing objects, reason’s principle, methodologically construed, neither determines anything given in the world nor anything given in our way of understanding it.Footnote 11 Rather, it normatively specifies how agents ought to try to understand it (namely, as a systematic whole), on the condition that they aim to advance, extend or perfect their empirical knowledge in certain ways. This is what is implied when Guyer, for instance, writes that ‘there are a variety of different ways in which ideals of reason can guide our conduct of scientific inquiry’, and when he goes on to explain how reason’s ideas indicate we ‘should’ proceed in these matters (Guyer Reference Guyer2006: 170).Footnote 12 The notion that we can choose to be guided by rational ‘oughts’ or ‘shoulds’ in our manner of inquiring into nature is of course entirely normative. Guyer further drives home the point in stating that there is ‘a right way as well as a wrong way to use the ideas of reason’ (Guyer Reference Guyer2006: 172). This type of normative language is warranted in the context of the methodological reading since, as I have argued, it takes reason’s principle to be a norm for guiding inquiry toward specific cognitive ends.
There are, to be sure, a wealth of passages in which Kant himself uses explicitly normative language to characterize reason’s principle and its derivative requirements, suggesting that on this point adherents of the methodological interpretation enjoy firm exegetical support (A509/B537, A548/B576, A570/B598, A646/B674, A671/B699, A750/B778). As mentioned, however, it is not my purpose to endorse the methodological reading or any of its features, but rather to explore its implications. The critical issue I wish to raise is that Kant himself has an ambitious and well worked-out theory of normative principles, canonically expressed in his practical philosophy, which is largely ignored by adherents of the methodological reading.Footnote 13 As a result, these adherents leave it unclear whether the implicit account of normativity on which they draw in conceptualizing reason’s principle is in fact Kant’s own. While this reluctance to characterize reason’s principle as a species of practical principle is understandable for reasons I’ll explore in a moment, it leads to problems, which are the focus of my discussion below.
Is there good reason to restrict the scope of Kant’s account of practical principles found in his ethical writings, as adherents of the methodological reading are implicitly inclined to do? Does it make sense to discuss reason’s regulative principle qua normative principle as though Kant himself does not have a well-worked out conception of such principles? One can imagine that defenders of the methodological reading may wish to answer ‘yes’ to these questions. To them it might seem that the account of practical principles Kant offers in Groundwork cannot possibly refer to the overarching class of normative principles as such, including normative principles directed to guiding our theoretical activity. In support of this view they could cite the many passages in which Kant writes as though practical and theoretical reason are two distinct faculties, operating according to principles of fundamentally different kinds. For instance, in the second Critique Kant famously discusses the potential conflicts between practical and theoretical reason in the context of arguing for the priority of the former over the latter.Footnote 14 Many take this discussion – amongst others – to suggest a fundamental difference between our two uses of reason and their principles. And if such a fundamental difference can be substantiated, then surely adherents of methodological reading would be correct to resist reading reason’s regulative principle as practical, however normative it is taken to be.
I wish to argue that this objection misses the mark, and does so by failing to distinguish two different ways in which Kant uses the term ‘practical’: one narrow, which contrasts with his use of the term ‘theoretical’, and one broad, which does not. If we allow that reason’s regulative principle is normative in character, that is, directed to guiding the activity of cognizing agents, then, I shall argue, it qualifies as practical in Kant’s broad sense of that term.
It is plainly the case that Kant often contrasts practical and theoretical reason, as was pointed out above. What is important to note, however, is that in these sorts of passages Kant is almost always distinguishing classes of rational activity with reference to their proximate aims, with theoretical activity taken to aim at systematic knowledge of objects, and practical activity taken to aim at the highest good (cf. CPrR, 5: 120ff). Yet Kant writes that the very idea of an aim is practical in a broad sense (WOT, 8: 131). This, I believe, is the same general point he has in mind when writing that ‘all interest is ultimately practical’, even the interest of theoretical reason, which is ‘complete in practical use alone’ (CPrR, 5: 121). It is, moreover, the same idea that surfaces repeatedly in the third Critique, as when Kant claims that ‘in the end all the effort of our faculties is directed to what is practical and must be united in it as their goal’ (CJ, 5: 206). What is clear from these sorts of passages is that Kant conceives of the overarching class of rational activity – aimed at ends, guided by norms and undertaken by free rational agents – to be practical in a broad sense.Footnote 15 According to this sense, even rational activity aimed at theoretical ends counts as a form of practical activity. That the term ‘practical’ has both this wide and narrow meaning for Kant is well recognized, for example by those who take his conception of the broad practical to ground his unity of reason thesis.Footnote 16 Whatever one’s views on the unity of reason, however, the point here is just that if Kant’s conception of the broad practical refers to the overarching class of purposive, rational activity, as I have argued it does, then adherents of the methodological reading do in fact construe reason’s regulative principle as practical in this sense.Footnote 17 For recall that on their reading reason’s principle is a norm directed to guiding the activity of agents in pursuit of cognitive ends.Footnote 18
But does this entitle us to consider reason’s principle a ‘practical principle’ in the sense in which Kant uses that term in his ethical writings? Are practical principles as Kant discusses them in Groundwork, say, also to be construed in this broad sense? The answer here is an unequivocal ‘yes’. It is well known that in Groundwork II Kant begins with an entirely general conception of the will and proceeds by laying out an exhaustive typology of the various kinds of imperatives that may be directed to guiding it (G, 4: 412–20). That the generic idea of an imperative – which represents an action as in some sense good or practically necessary – is broadly captured by what, in contemporary parlance, we would call a ‘norm’ or ‘normative principle’ is uncontroversial. It is for this reason that all normative principles, even those that guide rational activity in pursuit of theoretical ends, may be considered practical imperatives, which concept relies on Kant’s broad conception of the practical, mentioned above. Thus the conceptual framework for analysing practical imperatives found in Groundwork is in fact perfectly suited to analysing reason’s regulative demand for unity. Unsurprisingly, adherents of the methodological reading tend to borrow from Kant’s practical vocabulary in analysing the principle of systematic unity in any case, however reluctant they may be to countenance the larger philosophical and exegetical consequences of this move.
In this section I have argued that adherents of the methodological reading interpret reason’s principle of systematic unity as normative, which is to say, directed to guiding the activity of rational agents in pursuit of theoretical ends. I have also argued that if the principle is indeed normative in this way, then it counts as a practical imperative in Kant’s broad sense of that term. Moreover, I have shown that we are therefore entitled to analyse this principle in reference to Kant’s doctrine of imperatives as presented most clearly in the Groundwork of the Metaphysics of Morals. Unless adherents of the methodological reading can present good grounds for restricting Kant’s account of practical principles, they have every reason to view the principle of systematic unity as a principle of just this kind.
4. Is the Regulative Principle a Hypothetical Imperative?
Now if the methodological interpretation is correct in taking reason’s principle to be normative, and if we accept that Kant’s conception of practical principles offers us a comprehensive framework for analysing such principles, then we have a new conceptual vocabulary with which to interrogate the status of the principle. This allows us to reframe the question of its necessity as a question concerning its categorical or hypothetical status.Footnote 19 At least part of what appears unclear in Kant’s discussion in the Appendix, from this perspective, is whether the principle binds hypothetically, on the condition that agents adopt particular cognitive ends, or categorically, under all conditions.
Before turning to examine what adherents of the methodological reading have to say on this point, let us take a moment to consider what is at stake. If the command to seek systematic unity is categorical, it follows that cognitive unity would be an objective end obligatory for all to pursue, whatever agents’ particular cognitive goals and interests may be.Footnote 20 If reason’s principle is hypothetical, by contrast, then systematic unity would be a subjective, non-obligatory end agents are conditionally required to seek as a means to a further subjective goal that has been assumed. On this latter scenario, only those who, say, decide to do science would be required to pursue systematic cognitive unity as a means to that further end. Moreover, on this latter scenario, there would be nothing inherently irrational or wrong with, for example, claiming dogmatic insight into the character of reality as such, or in other ways flouting reason’s principle. For recall that hypothetical imperatives tell us only that a type of action is necessary as a means to an end, whilst remaining silent on the goodness and necessity of the end in question. If the command to seek cognitive unity is a hypothetical imperative, therefore, this leaves open the possibility that there is nothing good or necessary per se about the unity and coherence of our thought, and thus nothing necessarily good about the exercise of reason in its theoretical employment. Rather, whatever goodness or necessity systematic unity might have would depend on whatever further goal the pursuit of this unity might instrumentally serve. And whatever further goal that might be – science, or knowledge of nature, say – its goodness would not be something given as necessary through reason itself. This is clearly a vastly different situation from one in which the pursuit of unity is categorically required. For here reason would tell us that cognitive unity is indeed something unconditionally good and necessary for everyone to pursue, not something of mere instrumental worth to some. On this scenario, even the activity of theoretical reason would be bound by a supreme telos.
Now although they do not always employ Kant’s explicitly practical vocabulary in articulating their position, I submit that proponents of the methodological reading are nonetheless committed to the claim that reason’s principle is hypothetically binding. For on their view, the principle offers normative guidance to those who would guide their cognitive activity towards specific cognitive ends, ends which agents may but need not adopt. According to Guyer, for example, the principle is not a transcendental condition on the possibility of experience, nor a normative requirement all cognitive agents must meet, but rather a methodological device that is instrumentally necessary for facilitating the expansion of scientific knowledge in fairly specialized ways. Moreover, Guyer nowhere states or implies that all agents are normatively required to seek or expand scientific knowledge. It thus follows that the principle is merely hypothetical on his view.Footnote 21
Marcus Willaschek’s reading broadly follows Guyer’s, and his remarks on it are particularly useful for present purposes. He writes:
Only pure practical reason issues categorical imperatives – unconditionally binding principles. It may be in the interest of speculative reason to cognize ‘the object up to the highest a priori principles’. But still, it is up to us, as rational beings, how far to indulge our reason in this respect. Speculative reason can issue only hypothetical imperatives: ‘If you want to satisfy your speculative interest, don’t stop inquiring until you have gained knowledge of the highest a priori principles.’ But obviously it is not irrational to terminate inquiry (or even never to start it) … (Willaschek Reference Willaschek2010: 185)
For Willaschek, as for adherents of the methodological reading more generally, following the command to seek unity is only required for those who first decide to satisfy their speculative interest. And this satisfaction is taken to be a subjective, non-obligatory end. (If we were normatively obligated to satisfy our speculative interest, then terminating inquiry or never even starting it would be irrational in the sense of being forbidden by reason, contrary to Willaschek’s claim.) Since, however, Willaschek thinks it is ‘up to us’ to decide how far to satisfy our speculative interest, and since the necessity of seeking unity depends on this decision, reason’s principle is hypothetical not categorical on his account. Accordingly, Willaschek states outright that speculative reason, which is to say reason in its theoretical use, ‘can issue only hypothetical imperatives’. It follows that on his reading reason’s principle can at most be justified on instrumental grounds, as a conditionally necessary means to the end of satisfying one’s speculative interest. Crucially, this end – however proponents of the methodological view construe it – is not supplied by reason itself, since in its theoretical employment reason prescribes no other end apart from the systematic unity of cognition.
Now the hypothetical construal of reason’s principle certainly has its advantages, as well as its textual support. For instance, it gains credibility from those passages in the Appendix where Kant seems to claim a merely subjective and/or methodological status for it. It achieves this advantage, however, at the price of having to discount those passages in which Kant claims objective and/or transcendental status for the principle instead. Putting this difficulty to one side, it should also be conceded that the methodological view does a good job of explaining how reason’s principle could be necessary for scientific knowledge without being transcendentally constitutive of the objects of experience. It thus preserves a sharp distinction between the constitutive role of the understanding and the merely regulative role of reason.
Despite these advantages, however, I believe the hypothetical construal of reason’s principle suffers three serious faults. First, it is internally incoherent; reason’s principle cannot be a hypothetical imperative, since this would contradict Kant’s conception of what hypothetical imperatives are. Second, it fails to make the best sense of the text. And third, it pays insufficient heed to Kant’s overarching attempt to vindicate reason – in all of its employments – as a fundamentally self-determining power.
5. The Regulative Principle as a Categorical Imperative
In this section I advance the aforementioned conceptual and exegetical arguments against the claim that reason’s principle is merely hypothetically binding, while in the following section I explore my larger criticism, to wit that the reading fails to cohere with Kant’s Critical enterprise as a whole. It should be noted, however, that all three of my negative arguments against the hypothetical character of the principle function simultaneously as positive arguments for its categorical status, since the distinction between these forms of necessity is binary and exhaustive. The larger claim I wish to support in these sections, therefore, is that, if the principle of systematic unity is construed normatively, then it ought to be read as binding all cognitive agents categorically, under all conditions.
Before getting to my arguments, however, it may be useful to head off two anticipatory objections: first, that any argument that attributes categorical necessity to reason’s principle can be nothing but a reductio ad absurdum, since it is quite simply implausible that reason’s chief theoretical principle should bind in this way; second, that a categorical demand for systematic unity is implausible for the specific reason that it would be too demanding, placing an unreasonably large burden on our every cognitive act, and so failing to pass Kant’s ought-implies-can test. Because these two prospects may lead one to dismiss the direction of my arguments from the outset, let me briefly say something about them.
To be sure Kant argues that there can only be one valid categorical imperative, and that this is the moral law. Thus it may be objected that the existence of a valid categorical imperative directed to guiding our theoretical activity would either have to be derivable from the moral law, or else inconsistent with a major tenet of Kant’s practical philosophy. Neither option may look promising. While I shall have more to say about this later on, let me briefly indicate the shape of my response. I agree with the objector that these are indeed the alternatives, and embrace the first of them. Namely, I accept that if there is indeed a valid categorical command to pursue the systematic unity of our thought then this imperative must be derivable from the Categorical Imperative, and for reasons I shall later indicate find the prospects of this derivation more promising than may at first appear. I reject, however, the further claim that deriving the regulative principle of systematic unity from the Categorical Imperative would necessarily confer on it a moral status. This is because, in agreement with the well-known interpretation advanced by Onora O’Neill, among others, I conceive the Categorical Imperative as a supreme meta-norm governing all rational activity, only a subset of which ought to be considered explicitly moral (see O’Neill Reference O’Neill1989: 3–27).
The second objection grows out of the thought that a categorical requirement to seek systematic cognitive unity would be implausibly demanding, requiring that everyone pursue a grand unified theory of everything whenever they make a cognitive judgement. According to the ought-implies-can principle, such a requirement could not possibly be valid. Happily, I do not think a categorically binding principle of systematic unity would demand anything of the kind. Although I may refer by way of short-hand to reason’s principle as a command to seek systematic cognitive unity, I believe that the normative task associated with the principle, when precisely expressed, is in fact much more minimal. What the principle requires is just that we think in a manner that is consistent with the possibility of pursuing maximal systematic cognitive unity (A689–95/B717–23). That is to say, it requires that we avoid thinking in ways that block or thwart this possibility. Only by avoiding such ways of thinking do we use our reason in a manner that is consistent with reason’s fundamental ideal. But this requirement is not at all a demand that individuals systematically advance the frontiers of science every time they make a cognitive judgement or anything similar. Of course more can be said on this point, but I trust that the general direction of my response has dispelled some of the initial worries countenanced above, opening the way to a consideration of the case I shall develop below.
I turn now to the conceptual argument from Kant’s theory of imperatives to the categorical necessity of reason’s principle, which turns out to be surprisingly straightforward. To begin, let us recall that Kant thinks all practical ends are either subjective, because grounded in contingent conditions, or objective, because projected a priori by reason. Let us also recall that the adoption of objective ends is prescribed as unconditionally necessary by categorical imperatives, while the adoption of subjective ends may either be presupposed by hypothetical imperatives, or else commanded as conditionally necessary by them (as means to further subjective ends). It follows that no categorical imperative ever commands the pursuit of a merely subjective end. (It cannot be unconditionally necessary that everyone adopt an end only some may be capable of adopting because of contingent conditions.) Equally, no hypothetical principle ever requires that we seek an objective end. (It cannot be merely conditionally necessary to pursue an end that, by definition, all agents unconditionally ought to seek.)
At issue therefore in determining the necessity of reason’s regulative principle is the status of the particular end it requires us to seek. If it can be shown that this end is projected by pure reason and no other faculty then we shall know that it is an objective end, and hence that the command to seek it is a categorical imperative. If, by contrast, systematic unity refers to a state of affairs that could only be desired by agents contingently, then we know that the command to seek this end cannot be categorical, but must be hypothetical instead.
Now as Kant makes clear throughout the Dialectic, the ideal of systematic unity is the fundamental expression of pure reason itself. That is to say, systematic unity is not only an idea of pure reason, but the quintessential idea of which all others are but specifications. Moreover, it is clear that the ideal of cognitive unity could only be a projection of pure reason and no other faculty, since pure reason alone is capable of projecting ideal ends beyond the bounds of possible experience. As Kant puts it, only pure reason ‘does not follow the order of things as they are presented in intuition but with complete spontaneity makes its own order according to ideas’ (A548/B576). Hence Kant writes that the idea of systematic unity is ‘inseparably bound up with the essence of our reason’ (A695/B723).
If, however, cognitive unity is indeed a projection of pure reason and no other faculty, then it follows that the command to seek it must be a categorical imperative, holding unconditionally for all beings endowed with reason. For, as we have seen, ends of action that are given a priori through reason are obligatory for everyone to adopt, never non-obligatory. Their pursuit is commanded categorically, never hypothetically. Leaving aside the substantive question of how any categorical imperative can be shown to hold for human beings (i.e. the question of our freedom), and the additional question of what it would mean for reason to issue a categorical imperative in the theoretical domain, it is clear that the principle must have the structure of a categorical command.
Not only is the categorical necessity of reason’s principle entailed on conceptual grounds, however, I believe it also makes better sense of the text. Although in the Appendix Kant does not use the overtly practical vocabulary of imperatives, I shall argue that he nonetheless implicitly defends the principle’s categorical necessity. He does this negatively, by arguing against its conditional necessity, and positively, by showing that it binds all agents on a priori grounds. I examine passages in which Kant deploys both strategies.
Kant appears particularly eager to ward off hypothetical misinterpretations of reason’s principle. Thus, for instance, his central argument for the principle’s necessity functions as a highly compressed reductio of the hypothetical reading. He writes: ‘For the law of reason to seek unity is necessary, since without it we would have no reason and without that, no coherent use of the understanding, and, lacking that, no sufficient mark of empirical truth’ (A651/B679). Let us assume, with Kant, that reason’s essential theoretical function is to guide our use of the understanding through the principle of systematic unity. In light of this premise, claiming that we are only conditionally required to follow this principle leads to absurdity and contradiction, as I believe the above passage illustrates. For, if seeking unity were only hypothetically required as a means to the subjective goal of gaining knowledge, say, then making the rationally permitted choice to abandon this goal would mean annulling the command to seek unity. But this would be tantamount to willing the annulment of reason’s essential function, according to our above premise. In Kant’s own words: ‘without [the law of seeking unity] we would have no reason’. Moreover, as Kant points out, choosing to annul reason’s principle by abandoning the goal of gaining knowledge would also mean choosing to employ our understanding incoherently, and thus forfeiting ‘any sufficient mark of empirical truth’. The above passage sets up a reductio precisely because Kant assumes that reason does not permit its own annulment, and with it the forfeiture of ‘any standard of empirical truth’. This does not mean that free, imperfectly rational beings cannot choose to annul their reason, it only means that doing so would violate reason’s own categorical requirement. Such a choice would be forbidden by reason, not permitted by it. It follows that the notion of necessity Kant employs in stating that ‘the law of reason to seek unity is necessary’ is categorical, not hypothetical. By implicitly ruling out a hypothetical construal of the principle, as he does in the above passage, he negatively argues for its unconditional status.
But not all Kant’s arguments for the principle’s categorical necessity are negative. Kant also uses a variety of metaphors in an attempt to characterize the principle’s unconditional necessity directly. And although these arguments are less developed than one might wish, they nonetheless reveal Kant’s commitment to the categorical necessity of reason’s command. Thus, for example, he frequently describes the imperative to seek unity as an instance of pure reason’s own ‘legislation’. He writes:
The greatest systematic unity, consequently also purposive unity, is the school and even the ground of the possibility of the greatest use of human reason. Hence the idea of it is inseparably bound up with the essence of our reason. The very same idea, therefore, is legislative for us … (A695/B723)
Several pages later Kant returns to the same theme:
For the regulative law of systematic unity would have us study nature as if systematic and purposive unity together with the greatest possible manifoldness were to be encountered everywhere to infinity. For although we may light on or reach only a little of this perfection in the world, yet it belongs to the legislation of our reason to seek for it and presume it everywhere. (A701/B729)
By claiming that the pure rational idea of systematic unity is itself ‘legislative for us’, Kant implies that the command legislated by this idea is grounded in pure reason. Accordingly, he states that the principle of seeking unity ‘must therefore rest on pure transcendental and not empirical grounds’ (A660/B688). As is widely acknowledged, however, only categorical imperatives abstract from empirical grounds in this way. Hypothetical imperatives, by contrast, gain their necessity in reference to them, since they assume the adoption of contingent ends. By using the metaphor of pure reason’s own legislation to characterize reason’s command, Kant strongly suggests that it is a categorical and not a hypothetical imperative.
Earlier in the Appendix Kant evokes the categorical necessity of reason’s command through a different metaphor. In arguing for the principle’s transcendental status, he refers to it as ‘an inner law of reason’s nature’ (A650/B678). This phrase suggests that the principle’s validity holds on the basis of reason’s nature considered in itself, that is to say, in abstraction from all possible conditions and relations. If this is correct, however, then ‘inner’ and ‘absolute’ may be used interchangeably in this context to characterize the necessity of an imperative whose validity rests on pure reason alone. But, as we have seen, only categorical imperatives have validity of this kind. Kant again reinforces the point by claiming that reason’s regulative principles ‘carry their recommendation directly in themselves and not merely as methodological devices’ (A661/B689). Presumably, principles that carry their validity (or their ‘recommendation’) ‘directly in themselves’ do not carry it elsewhere, i.e. in reference to assumed ends or other contingent conditions. Kant therefore again suggests that the principle of seeking unity is unconditionally valid, which is to say, a categorical imperative.
One might object, however, that these passages do not so much show the principle to be a categorically binding norm, but rather to be a transcendental principle in the more usual sense. In other words, one might worry that if the principle is indeed transcendental, as these passages are frequently taken to suggest, then this undermines the basic premise of the methodological reading altogether, namely that the principle is normative, and thus directed to guiding the activity of subjects (rather than to determining anything a priori in the nature of objects – or in our representation of them). So why take these passages to show that the principle is a categorically binding norm, and not, say, a transcendental condition on the representation of objects (in some suitably regulative sense)?
While it is not my purpose in this article to defend the methodological construal of the principle as normative, it will nonetheless be helpful to speak to these worries. In my view, the above objection fails in virtue of running together two distinctions that are in fact independent of one another, however poor a job Kant does of articulating this. There is, on the one hand, the question of whether reason’s principle is objective or subjective. There is also the separate question of the principle’s necessity: whether it binds transcendentally or methodologically. The passages I discuss above may coherently be taken to suggest that reason’s principle is a categorically binding norm – rather than, say, a transcendental condition on the possibility of experience (in some suitably regulative sense) – if Kant’s use of these distinctions allows for the following possibility: that reason’s principle is ‘subjective’ in the sense of being a norm directed to guiding the cognitive activity of subjects, and that it nonetheless binds on transcendental rather than empirical grounds, in the sense that it holds a priori for all such subjects. Now although categorical normative necessity of this kind may not typically be thought of as a form of transcendental necessity, one can appreciate why Kant may nonetheless wish to characterize it in this way if, in virtue of normatively binding all cognitive agents a priori, reason’s principle conditions the possibility of fully rational cognitive activity. One can also appreciate why characterizing the principle as transcendental in this sense would still be compatible with his claim that the principle is also merely ‘subjective’ insofar as it is directed to normatively guiding cognitive practice (without making any direct claim about objects).Footnote 22 Although the scope of this article prevents me from defending this possibility in greater detail, it nonetheless allows us to see how the passages I cite above need not be taken to indict the methodological reading tout court, with its insistence on the normativity of reason’s principle, but rather to indict the specific claim that the principle binds merely hypothetically, which is to say, on a posteriori grounds.
These textual considerations, however, are only part of the exegetical picture. They suggest that insofar as we grant that reason’s principle is normative, then Kant’s discussion in the Appendix supports the claim that it binds on categorical rather than hypothetical grounds. What I wish to show now, however, is that this construal of reason’s principle – as categorically binding – is one to which Kant’s larger, Critical reconceptualization of reason commits him in any case. It is to this larger interpretative question, concerning the meaning, status and indeed goodness of reason as a whole that I now turn.
6. The Principle of Systematic Unity and Kant’s Critical Reconception of Reason
The most serious problem with reading reason’s principle as hypothetically binding is that this implies a view of theoretical reason that conflicts with Kant’s larger account of reason as a whole. More specifically, this construal of the principle underestimates Kant’s critique of instrumental reason, while either ignoring or discounting his claim that all of reason serves a single practical telos. Via this later claim Kant attempts to vindicate reason, in all of its employments, as a self-legislating power. My argument, then, is not only that the hypothetical construal of reason’s regulative principle is conceptually incoherent and exegetically weak, but also that this construal is inconsistent with Kant’s Critical vindication of reason as such.
Kant is consistently clear that both our theoretical and moral uses of reason contribute in distinct ways to an overarching practical project that aims at a single supreme practical end.Footnote 23 Kant characterizes this end in a variety of ways, but at its most general as systematic unity itself (CPrR, 5: 119–21, 134–41, 146–7, 191; A680/B708, A738/ B766). This is widely noted in the literature. Kleingeld, for instance, writes that, ‘both theoretical and practical reason strive for systematization’, while Guyer observes that ‘reason has unity, in the sense of systematicity, as its special object, and practical reason, therefore, the unity or systematicity of purposes’ (Kleingeld Reference Kleingeld1998: 314; Guyer Reference Guyer2000: 87). In the same vein, Gardner claims that ‘theoretical and practical reason share an interest not just in their own systematic form but in systematic form in general’ (Gardner Reference Gardner2010: 268). To be sure, Kant also suggests that this supreme practical telos has a specifically moral character. Famously, in the third Critique he argues that theoretical and (narrow) practical reason are united in the highest good and in their mutual need to postulate the possibility of its realization in nature. One may wonder whether and how the highest good counts as a specification of the systematic unity reason aims for at the most general level, an issue that Guyer amongst others has discussed in detail (see 2000: 60–95). For present purposes, however, we need not decide this question here. What is salient is just that Kant thinks reason’s broad practical telos, however this is best construed, confers ultimate worth on all of reason’s activities.Footnote 24 He puts the point in various ways, for instance in writing that ‘the entire armament of reason … is in fact directed only at … what is to be done if the will is free’ and in saying that ‘in the end all the effort of our faculties is directed to what is practical and must be united in it as their goal’ (A800/B828; cf. CJ, 5: 206). These claims, coupled with Kant’s further assertion that only the ultimate end of pure practical reason has unconditional worth, leave us with the conclusion that all of reason’s activities have ultimate worth in virtue of serving this end. Though the scope of this article forbids me from engaging with these claims further here, it is nonetheless important to note that Kant’s Critical attempt to vindicate reason as such very much depends on this claim concerning its ultimate, practical fulfilment.
The hypothetical construal of reason’s regulative principle, however, is inconsistent with Kant’s thesis concerning the ultimate worth of reason as such, for it implies that in its theoretical employment reason has mere conditional value. Recall that hypothetical imperatives tell us only that a type of action is necessary as a means to an end, whilst remaining silent on the goodness and necessity of the end in question. If the command to seek systematic cognitive unity is a hypothetical imperative, therefore, this implies that there is nothing necessarily and unconditionally good about the unity and coherence of our cognition, and thus nothing necessarily and unconditionally good about the exercise of reason in its theoretical use. Rather, whatever goodness or necessity systematic unity might have would depend on whatever further goal the pursuit of this unity might instrumentally serve. And whatever further goal that might be – science, or systematic knowledge of nature, say – its goodness would not be something given as a priori necessary through reason itself. But this clearly contradicts Kant’s claim that all of reason has ultimate value in virtue of serving an unconditionally worthy practical end.
It may be objected, however, that in fact the text is more equivocal on this issue than I have suggested, and that in many places Kant at least seems to imply that theoretical reason has mere conditional value. For instance, he writes that ‘all interest is ultimately practical and even that of speculative reason is only conditional and is complete in practical use alone’ (CPrR, 5: 120–2). Similar passages surface in ‘What is Orientation in Thinking?’, where Kant suggests that theoretical reason’s ‘need’ to seek unity (inter alia, by postulating the existence of God) is merely conditional, depending as it does upon our desire to ‘judge about the first causes of everything contingent’ (WOT, 8: 139). In light of these sorts of passages one might be tempted to conclude that if the ‘interests’ and ‘needs’ of theoretical reason are merely conditional in this way, then surely the imperatives associated with these needs and interests are conditional as well. If this is correct, however, this may seem to support the hypothetical reading of reason’s principle. For, on this reading, theoretical reason has mere conditional worth in a double sense: first, because systematic unity is only worthy on the condition that agents set themselves the end for which seeking such unity is a necessary means, and second because that further end (however it is construed) can itself only be contingently valuable.Footnote 25 Thus, an objector might argue, the hypothetical construal of reason’s principle is in fact entirely consistent with Kant’s claims concerning theoretical reason’s value.
While I do not deny that the above objection gains some textual traction, I believe it fundamentally misreads Kant’s comments concerning the conditional status of theoretical reason and does so in virtue of missing the distinction Kant draws, however unclearly, between the broad and narrow practical. Whenever Kant suggests that theoretical reason has conditional worth he is almost always distinguishing between uses of reason in reference to their proximate aims (cf. CPrR, 5: 120–2; WOT, 8: 139). He is thus using the terms ‘practical’ and ‘theoretical’ in the narrow sense mentioned earlier, whereby they refer to two distinct, non-overlapping classes of rational activity, aimed at different goals. However, it must be remembered that Kant also considers theoretical rational activity to be an instance of practical rational activity in the broad sense. Indeed, it is only in light of this insistence that Kant’s further claim concerning the single practical telos that guides all uses of reason becomes intelligible. With these considerations in mind, Kant’s suggestions concerning the conditional worth of theoretical reason must be read differently. When he seems to imply that theoretical reason has conditional worth in reference to its proximate aim (‘cognizing the object up to the highest principles’, say) this should not be taken to mean that theoretical reason does not also have ultimate worth insofar as it a species of practical reasoning, broadly construed.Footnote 26 Indeed, I would suggest that the worth of theoretical reasoning (i.e. the worth of rational activity aimed at ‘cognizing the object up to the highest principles’) is conditional precisely insofar it rests on the condition that the activity in question is itself also broadly practical in character. That is to say, cognizing the object up to the highest principles is worthy on the condition that practical rational activity as such is worthy. On this reading, then, Kant does not mean to claim that theoretical reason has only conditional value. For this claim would contradict his affirmation that all of reason is ultimately worthy in virtue of serving an unconditionally worthy practical end. Rather, what Kant means to say is that the value of reason in its theoretical employment is conditioned by the ultimate value of rational activity as such, which Kant considers to be broadly practical. The upshot for present purposes is that the condition upon which the worth of our theoretical activity rests is internal to reason itself. Theoretical reason is not ultimately heteronomous, for its worth does not depend on anything alien to it.
If the above interpretation is correct, then we also have larger, architectonic reasons to reject the hypothetical construal of reason’s regulative principle. Quite simply, theoretical reason cannot serve a supreme practical telos, as Kant repeatedly insists it does, if in one sphere of its employment reason is not fundamentally a self-determining power at all, but is instead subservient to arbitrary, contingent ends. My claim, therefore, is not only that the hypothetical construal of reason’s chief theoretical principle is conceptually incoherent and exegetically weak, but also that this construal discounts Kant’s overarching attempt to vindicate reason as a free, self-legislating power. As Richard Velkley puts it, ‘[a]ll of Kant’s critical philosophy is a critique of instrumental reason, either from a moral or a theoretical standpoint’ (1989: 23). This can hardly be the case, however, if theoretical reasoning is itself an instance of instrumental reasoning, as the methodological reading implies.
7. Conclusion
Several commentators have likened reason’s principle of systematic unity to a categorical imperative – or indeed the Categorical Imperative.Footnote 27 Henry Allison, for example, states outright that the principle may be seen as ‘an intellectual categorical imperative’, while Michelle Grier picks up on a similar theme in writing that even though ‘the systematic unity of nature is “subjectively imposed” and subjectively necessary, it is a necessity that is objective … for us, as rational, discursive knowers [just like the moral law]’ (Allison Reference Allison2004: 52; Grier Reference Grier2001: 286).Footnote 28 By arguing that Kant’s way of vindicating theoretical reason in the Doctrine of Method is structured similarly to his vindication of practical reason through the Categorical Imperative, O’Neill implies a similar conclusion (O’Neill Reference O’Neill1992: especially 300–5). If the arguments I have provided here are persuasive, these observations should come as no surprise. For if the principle of systematic unity is indeed normative, then it cannot but be a categorical imperative for the conceptual, exegetical and architectonic reasons I have presented.
Needless to say, construing reason’s regulative principle as a categorical imperative disrupts the standard way in which the interpretative options tend to be conceived when it comes to conceptualizing reason’s authority in the theoretical domain, and raises more questions than it answers. How exactly ought we to conceive of the supreme practical telos of reason in all of its employments? Is this best understood as the highest good, as Kant suggests in the third Critique? Moreover, how does this broad practical telos function in guiding theoretical activity, and how does it relate to the proximate aim of such activity, which may be characterized as systematic knowledge of objects or ‘cognizing the object up to the highest principles’? Needless to say constraints of space forbid me from staking out answers to these questions here. I will close, however, by pointing in the direction in which I believe promising answers lie.
If it is assumed that the systematic unity of reason’s own activity is the supreme practical telos of reason in all its employments, then it would follow that in the theoretical sphere reason does not enjoin us to seek systematic knowledge of nature per se but rather to seek systematic unity in our way of thinking about nature. This would ultimately be a pure practical goal to do with the way we normatively organize our own (in this case, cognitive) activity. But does this mean that knowledge of nature has nothing to do with the exercise of theoretical reason? No. According to this proposal, systematic unity in our way of thinking about nature would necessarily result in systematic knowledge of nature under ideal success conditions, just as the systematic moral unity of agents’ maxims and ends would necessarily result in maximal happiness (apportioned to virtue) under ideal success conditions. That is to say, knowledge would ‘fall out’ of our supreme rational telos as it governs the theoretical sphere, just as Kant thinks happiness ‘falls out’ of this same telos as it governs the moral sphere. My suggestion, then, is that whether in thinking or in acting, we are enjoined to follow reason’s overarching categorical requirement by seeking the systematic unity of our own rational activity, and that this internal rational unity may then be seen to result in real-world success (in the form of happiness and knowledge respectively) under ideal conditions, conditions which themselves are objects of rational faith.Footnote 29 What I believe this rough sketch indicates is that there is at least one way of envisioning how an overarching categorical imperative may direct all rational activity in accordance with a single supreme practical telos, the differentiation of which telos in the moral and theoretical spheres jibes with Kant’s accounts of the determinate ends of moral and theoretical reasoning respectively. I trust that the consistency of this suggestion with the broad outlines of Kant’s moral and theoretical philosophies can be seen from afar even if the details remain to be explored on another occasion.
