The self, the ‘I think’, my existence as an intelligence, the thinking subject, apperception, the person—all of these and the relationships between them form the subject matter of Arthur Melnick's brilliant new book on Kant. Indeed, every one of these notions is brought into play by the end of the very first paragraph of the preface. This is a difficult book, primarily suited to the specialist, yet its difficulty only rarely lies in unclarity or confusion. The theory of the self that it outlines and explores is as subtle, intricate and forcefully argued for as it is original, stimulating and highly controversial.

The book is split into five parts and twelve chapters. Part I is a preliminary overview in which Melnick sets out what he takes to be Kant's basic view of the self (Chapter 1) and how it relates to transcendental idealism and the accompanying issues of spatial and temporal existence (Chapter 2). After this, two interweaved themes structure the rest of the book. The first is that of an incremental thickening of the notion of the subject that is under consideration, from the thinking subject and its attendant transcendental self-consciousness (Part II), to the cognizing subject and its apperception (Part III), to the person (Part IV) and finally to the embodied subject (Part V). Melnick moves from pure to empirical concerns, then from issues concerning single thought-episodes in isolation to those concerning a connected series of such episodes, and finally to the question of how all this relates to the organism and its place within the world in which it resides. In many ways this gradual thickening takes us some distance towards the self of practical philosophy, and this book will most certainly reward attention by Kant scholars of any ilk. Yet Melnick is restricted to a study of the self as it functions in the Critique of Pure Reason. Hence the second organizing theme of the book, running parallel to the first, is that of a detailed textual analysis of the sections of this work that are key to our understanding of Kant's theory of the self as it concerns philosophy of mind and metaphysics. Namely: the four A-edition paralogisms of substance, simplicity, personality and outer existence (Chapters 3, 4, 10 and 12 respectively); the A-edition doctrine of threefold synthesis as it connects to empirical apperception and schemata (Chapter 7); the B-edition doctrine of figurative synthesis as it connects to pure apperception and schemata (Chapter 8); and the so-called paradox of inner sense and its relation to the ontological status of the self as dealt with in Sections 24 to 25 of the B-edition transcendental deduction (Chapter 9). Melnick's aim in his exegeses is not only to colour in the picture of Kant's view of the self that was sketched in Part I. His method is one of indirect support for this picture, to show that it is the only one that makes good sense of the text and the arguments therein. Other interpretations simply do not stand up to scrutiny (Chapter 6 especially, but also throughout).

Melnick's central claims are these: Kant's reflection on the cogito yields conclusions of a distinctly ontological flavour, albeit not those reached by the rational psychologists themselves. The self for Kant is neither a thing in itself nor an appearance. In fact, it is not an entity at all. Nor, however, is it merely an abstraction. Rather it enjoys real existence, but as an abiding intellectual marshalling action for the unification of inner intuition. Thus the thinking self exists as an intellectual action exists. After Part I, the rest of the book is concerned to deepen, expand and apply these claims. Unfortunately, the first question that strikes one at this point is also a question that is never fully answered: What exactly is the kind of real existence that is enjoyed by an intellectual action?

We are told that it is not the same kind of real existence as that enjoyed either by things in themselves or appearances, and this negative aspect of the position is dealt with consummately. Kant's thinking self is not a thing in itself because it is temporal (contra, for example, Sartre's reading). But it is temporal only in the sense that it unfolds with or accompanies time, which for Melnick is the sheer progressivity manifested in the shifting of inner and outer attention. Yet for this very reason, the thinking self is likewise not an appearance. For unfolding with or accompanying time is not the same as arising in time, and only that which arises in time is intuitable, and only that which is intuitable is an appearance. So far, so good. But then this negative aspect of Melnick's position is its less original aspect. His explanation of what Kant denies (and why he denies it) at places like B157 and the footnote to B422 is both lucid and plausible. But surely some positive characterization of the supposed alternative is called for, some explanation of just what it means to affirm that the self as an intellectual action is ‘flat-out real’ (p. 62). After all, a clarification of this ‘third status’ is the professed aim of the book (p. vii).

Strange, then, that one finds virtually no clues until Part V and the very last chapter, where Melnick gives an excellent account of the relational, irreducibly transactional nature that phenomenal reality has for Kant. Here Melnick deepens considerably his conception of existence as it functions in the context of transcendental idealism, which in turn contributes implicitly but significantly to explaining the precise nature of the existential status of the intellectual marshalling action that is the thinking self. According to Melnick's strong metaphysical interpretation of transcendental idealism, space, as nothing but the form of intuition, exists solely in the outer shifting of attention. Thus what exists in space—material reality—is just what possibly or actually arises in the course of outer attending. In Melnick's terminology, then, material reality is a sheer or thin (i.e. not necessarily activated) potency or capacity for affecting us. In this way Melnick argues persuasively (against Langton) that the relational conception of reality is one that is entailed by transcendental idealism. Despite there being good reasons to leave this content until the end of the book—because it is closely connected to the final paralogism and because with a dispositional account of material reality it compliments and builds on the dispositional account of personhood given in Part IV—I cannot help feeling that some of it should have been at least mooted in Part I.

But even here things are not made altogether clear, for much of what Melnick explains about Kant's conception of existence seems only to pertain to physical action—after all, his discussion focuses on space and material reality. I take it that the thinking self cannot enjoy the same kind of real existence as that had by a physical action. The self is not an appearance because it cannot be intuited, but presumably we can simply intuit physical actions. Physical actions, like physical objects and events, are spatiotemporally located, can endure and interact unobserved, and might be deposable into parts that are themselves actions, objects, or events. So the problematic is specifically intellectual action. Perhaps, then, it is not so surprising that there is some obscurity here, for it looks very much like the question of the ontological status of Kant's self when it is seen as an intellectual marshalling action is none other than the question of the ontological status of the mental.

And, inevitably, obscurity regarding this central issue has repercussions. Melnick sets up his reading of Kant's theory of the self as falling between two extremes that are both commonly seen in the literature. On the one hand there are those who think Kant is forced, whether knowingly or otherwise, into noumenalizing the self. On the contrary, argues Melnick, Kant does not violate his doctrine of transcendental idealism and its attendant epistemological humility, for he in no way sanctions or unwittingly requires the idea of the self as a noumenal underpinning to space and time. The second chapter is for several reasons less exhilarating than the first. The discussion of the transcendental ideality of space and time is all too brief, although this is understandable given that Melnick has dealt with it at considerable length in earlier works. Similarly for the omission of any mention whatsoever of the problems of internal coherence that traditionally dog strongly idealist interpretations such as his. But in general, and insofar as I understand it, I agree that his reading of Kant's theory of the self renders it consistent with his reading of transcendental idealism, which fact I likewise agree is crucial. My dissatisfaction arises with his treatment of the interpretations at the other extreme, of those who think that in one way or another Kant empties the thinking self of any substantial content whatsoever, of those who argue it is an abstraction, merely logical, formal, intentional, semantic, representational, or functional. Given that the kind of real existence had by an intellectual marshalling action goes unexplained, it is not at all clear that Melnick's view of Kant's self as just such an action is not compatible with one or other of the austere interpretations he purportedly rejects. In particular, take functionalist accounts. Functionalists about a kind of action need not deny that it is real, and would it not be plausible to give the much needed constructive side of the explanation of intellectual action in terms of the role it plays in mediating between sensory stimulation and behaviour or judgemental disposition? If not, Melnick does not say why not. Of course doing so would not automatically solve our ontological question, but it would at least shift it to more familiar, indeed safer, ground.

Despite this problem with the positive aspect of Melnick's account of Kant's basic views on the self, Part I is exciting and sets the tone for the remainder of the book. Melnick's is a descriptive phenomenological account that evidently owes much to the phenomenological tradition and in particular to Heidegger. Yet it is in fact a particularly engaging feature of the book that citations from and discussions of such thinkers appear alongside and are given equal weight to citations from and discussions of more analytical figures like Sellars, Strawson, McDowell, Cassam and Aquila, to name just a few. Here is testament to the trenchant but somewhat unfashionable idea that the fundamental philosophical issues that herald from the Continent are quite simply the same as those of the Anglo-American world, and also that Kant, as part of our common heritage, is part of the reason for this. Moreover, I suspect it is the influence of the phenomenological tradition that has caused Melnick to so liberally sprinkle his text with such thoroughly concrete examples. This is so rare in books on Kant, perhaps because it is so rare in Kant, but the effect confirms that it really should not be. Of course the first Critique is avowedly abstract, and necessarily so, but chess games, baking cakes, marking time, perusing dogs, semantic Gettier cases, sugar dissolving in water, protons, electrons and billiard balls, all help to ground the discussion and ultimately prove extremely useful in aiding reader-comprehension. Although that is not to say that there are not moments when one suspects that too much in the argument is relying on certain details of an inevitably distorting example.

It is largely by way of phenomenological description that Melnick fills out his account of the intellectual marshalling action that constitutes the thinking self throughout a given thought-episode. Such description is invariably supposed to be, and most often is, both intuitive and informative. For example, he talks of a certain coalescence of an inchoate mass of peripheral thoughts, in varying degrees of readiness and with variously similar content, around a single focused thought as it is settled on, which coalescence signifies the holding or concentrating or focalizing action that I am in my thinking. Melnick's intellectual marshalling action, then, constitutes neither the self in isolation from what it thinks nor the thought in isolation from what thinks it. The distinction between thinker and thought is rather to be sought within the intellectual marshalling action as a whole. One cannot, as it were, separate them out and make them face each other. This is an important and attractive aspect of Melnick's view, not least because it paves the way for a novel account of transcendental self-consciousness and its contrast to inner intuition (in Chapters 5 and 9).

It is true that for Melnick the self is always, at the very least, the thinking self, but this is not to say that one cannot reflect on what one is thinking or indeed on the manner in which one is thinking. Transcendental self-consciousness, on Melnick's phenomenological account, is only an occasionally present facet of acts of thought such as ours, one that nevertheless exists entirely within intellectual marshalling action. Melnick's picture is of an added reportative element that is dynamically coupled to the ever-present focalizing element that is the thinker within the thinking self as a whole. But it is important to see that although this coupling is only ever contingently actual—we are not, as a straightforward matter of fact, always self-conscious—it is necessarily possible. Thus self-consciousness is reflective only in the sense that it involves a representation of (a report on) the very spontaneity of the thinking it accompanies. Transcendental self-consciousness is not reflective in the sense that it involves any kind of inner noticing, for only in the former sense is reflectivity always possible.

But what of that with which transcendental self-consciousness is contrasted? Melnick also gives a phenomenologist's account of inner intuition, and it is here that I think his examples cause confusion. Inner intuition—or as Melnick calls it, presumably to emphasize its dynamic character, inner attending—involves no additional manifold to that already involved in outer attending. Rather it consists solely in retracting one's attention inward while outer attending. Inner attending, precisely unlike transcendental self-consciousness, is a kind of inner noticing, a noticing of the character of one's own states. Again this view has its attractions, particularly in the context of Melnick's overall position. It fits perfectly with the description of a thought-episode as a more or less gradual coalescing of several more or less distinct thoughts around a single focal point. When one is engaged purely in outer attending, both one's focal and peripheral thoughts concern what is outer-perceived. When one is at the same time inner attending, one's focal point is the act of outer attending itself, whereas that which is outer-perceived now features at the periphery only.

However, I am not at all sure what it is to retract one's attention in such a way, or rather, I am not sure that doing so can really constitute Kantian inner intuition. By way of explication Melnick gives examples of attending to how one is looking at things, to the straining of our eyes, to how one is listening, to the tilting of the head upward (pp. 112–13). But surely such examples are all irreducibly bodily and thereby spatial, yet the form of inner attending is supposed to be time. What's more, it is not obvious that such cases give us a flavour of what is unique about inner attending. Presumably I can likewise attend to my manner of outer attending without inner attending at all. For example, am I not outer attending to the straining of my eyes if I look in the mirror and watch my pupils dilate? Melnick fully admits to relaxing the level of abstraction with his examples, but it is just not clear what would be left of the account if he did not do so. On the one hand, Melnick's account of inner attending as dependent on outer attending seems attractive for the very reason that it seems a promising place to find an explanation of Kant's notoriously problematic assertions to the effect that spatial and temporal experience are somehow interdependent (as opposed to the former relying on the latter but not so conversely). On the other hand, relaxing the level of abstraction as he does seems to blur the distinction too much. Melnick's apparent reliance on such examples makes one suspect that his account surreptitiously conflates Kant's conception of the line between inner and outer with the contemporary philosopher of mind's conception of that line, according to which it lies, say, at the skin. The latter is a line drawn within space and time, the former between space and time.

I have focused in this review on some points in Melnick's book where his account seems incomplete or unclear. This has meant that I have neglected a large amount of incredibly valuable material that is both controversial and clear. In particular, Melnick deftly explains the place of Kant's views in notable modern debates. To take two examples, with regard to the role of reality in perception (in Chapter 12) he suggests how, in the context of transcendental idealism, the immediacy of intuition amounts to a form of naïve realism (although I would be as happy to align it to externalism about mental states); with regard to persons he discusses (in Chapter 10) Parfit's psychological continuity view, and both fission and fusion and how they relate to the thorny matter of intellectual intuition.

There is also much of value that is more internal to Kant scholarship. With regard to the paralogisms that help structure the book, Melnick's readings are at the same time highly sensitive to the text yet innovative, ambitious yet persuasive. Each of the paralogisms follows the same pattern. They are syllogistic inferences that are rendered formally invalid on account of an equivocation that need not be tracked down to a single term—after all, ‘sophisma figurae dictionis’ (B411) does sound rather general. In each case, disambiguation reveals two distinct conclusions, both of which are of ontic significance, but only one of which is accepted by Kant as a conclusion that it is legitimate to draw from the cogito alone. Nevertheless the other, illegitimate conclusion is accepted, with a tightly circumscribed humility, as a noumenal possibility (particularly fascinating in this vein is Melnick's brief allusions to Spinoza's God and Leibniz's monads). And along the way, Melnick provides uncounted insights into issues such as Kant's conception of modality—a normative account of necessity and a non-counterfactual account of possibility; schemata—a kind of regulatory mental rehearsal of the conceptual rules with which they are coactive in thoughts; the difference between concepts and judgements—a deceptively simple issue that exercised Kant and his contemporary J. S. Beck, essentially a mere matter of positive assessment; and Kant's relation to Berkeley—both metaphysical idealists and relationalists about reality, but only the latter a phenomenalist to boot.

In short, there is far more to this admirable book than can even be hinted at in a short review like this, and it deserves to play a central role in the ongoing debate on Kant's theory of the self.