If you believe that there are moral truths and are working within the meta-ethical tradition shaped by the early twentieth-century demand to place ethics on a proper theoretical basis, then the first thing you have to do is to show how such truths are accessible to us. Stephen Engstrom invites us to take the longer view; he seeks to broaden the terms of reference of this debate by connecting it to a much older ‘cognitivist tradition in practical philosophy’ that goes back to Plato and to Aristotle and views morality as a type of practical knowledge, not just in the sense that its claims are truth-evaluable, but also in the sense that it can lead to practical ‘wisdom’ (vii). The aim of the book is to ‘recover this traditional understanding of morality's relation to reason’ (xi). This process of ‘recovery’ consists in a systematic defence of Kant's ethics but also, I think just as importantly, an unapologetically Kantian treatment of the puzzle of ‘practical knowledge’ bequeathed by Anscombe to those who grapple with Intention. As a reading of Kant, The Form of Practical Knowledge belongs with interpretations that take as their starting point the idea that ‘practical reason is a capacity for knowledge of the good’ (xi). The task then is to show that, to the extent that it is reason, practical reason is subject to a single set of rules and, to the extent that it is practical, that it is action-guiding. Engstrom's clarity of purpose in prosecuting this task allows him to cover enormous ground, offering in the process new insights and original solutions to old problems in Kant interpretation. As a contribution to meta-ethics and action theory, the book presents a bold central thesis, that morality is a type of practical knowledge and as such not only amenable to standards of rationality but also inherently favourable to the universalizability condition stipulated by the categorical imperative.

The argument is presented in three stages. In the first part, comprising chapters 2 and 3, Engstrom explains why we should be thinking of morality as a type of knowledge in the first place. These chapters set out the basic theoretical commitments that shape his interpretation of Kant's moral theory. The second part, chapters 4 and 5, contains the details of the claim that the moral law is the form of practical knowledge. Finally, chapters 6 and 7 aim to show that the theoretical and interpretative claims defended in the earlier chapters help resolve some long-standing problems of interpretation regarding the different formulae of the categorical imperative and usefully guide the determination of practical principles. By focusing on specific duties, Engstrom directly addresses criticisms about the vacuity and practical worthlessness of the categorical imperative.

The key thesis defended in the book concerns, as the title announces, the ‘form of practical knowledge’. We can get a provisional sense of what this thesis consists in by breaking the overall argument into the following premises: that willing is a type of practical knowledge; that there is a single form to practical knowledge; and that this form is the categorical imperative in its universalizability formulation. Because the singleness claim is about a specific application of the notion of universality, much of the probative burden of the third premise is carried by the second one. Then follow two ampliative premises that consist in a defence of the equivalence of the different formulations of the categorical imperative and various application issues (Engstrom's treatment of these thorny topics is exemplary). Jointly these lead to the conclusion that moral willing takes the form of the categorical imperative. Although this conclusion is clearly necessary for the success of the project as a whole, the argument for the ampliative premises is relatively self-contained. For this reason, and given the limitations of space, I will leave it out of this review, concentrating instead on the first two premises that spell out the core theoretical commitments of the book.

The specific problem which the first two premises address is a familiar one that has exercised Kant's supporters and critics alike: it concerns the connection Kant envisages between law-likeness and rational agency. Kant, Engstrom argues, seems to conflate two distinct claims, that as rational agent one is to will in accordance with a law and that one is under obligation to will the moral law, which stipulates that such and such maxim be universalisable (5, 6). From a practical and normative perspective the problem appears as a missing link between formal constraints and substantive moral demands (3, 6). The solution Engstrom proposes is that we take seriously the idea that morality is knowledge of the good. We know by means of judgements; we know practically by means of judgements that are orientated towards, or guided by the idea of, what is ‘good on the whole to do’ (50). The tricky thing is to show that what is being cognized as good, ‘the object of practical knowledge’ (14), is also recognized as something to be done, so that this exercise of reason is both contentful, insofar as it applies to specific judgements, and ‘efficacious’ (7).

The load-bearing concept is that of practical knowledge. Engstrom introduces this concept gradually beginning with an analysis of the ‘will’, which draws on features of Kant's conception of the faculty of desire (Begehrungsvermögen), of the power of choice (Willkür) and of rational willing (Wille). On Engstrom's reconstruction, the will is a practical thought that gains specification when it becomes an ‘intention’ (28). Whilst both intention and practical thought are efficacious representations, in the case of intention efficacy is part of the thought. The obvious reference here is not Kant but Anscombe, in particular, her observation that our knowledge of our intentions is not inferential. This is both eminently plausible and hugely puzzling, as subsequent debate has shown, for explaining how one knows in this case goes to the heart of the problem of figuring out what exactly ‘practical knowledge’ is. Engstrom partially absorbs the puzzle in the definition of intention as ‘self-conscious awareness of its own efficacy’ (30). He then proposes that the representations that are the content of the practical thought, and also of the intention, are conceptual (33, see too 39 and 55) and, as such, subject to rational requirements. So roughly the argument is that all thought is conceptual by virtue of being thought, and some thought is practically efficacious by virtue of being intentional.

The next step is a basic account of practical rationality: given that practical thought has conceptual content, then it is subject to rational requirements. These are set partly by the agent's theoretical knowledge. For example, the practical thought should not contain the representation of an action that it is not possible to perform (41). Unobtrusively, Engstrom places the fundamentally calculative idea that it is rational to do that which is possible to do in the domain of theoretical knowledge (albeit for practical use). We may then enquire after any other rational criteria that are proper to ‘practical’ knowledge. This leads to a more ambitious conception of practical rationality. To get to this, we need to conceive of the object of practical knowledge formally and as given a priori. There is no separate argument supporting this interim step (see 12), possibly because the order of exposition is ‘from below’, that is, from basic requirements of the exercise of our rationality in practical contexts (see 8).Footnote 1 Accordingly, the main focus of Engstrom's analysis is judgement, that is, something that is a basic component of the exercise of our rationality.

In Kant's epistemology, judgement plays a key role in cognition as the faculty of subsuming under rules (see A132/B171 for example), so its relevance to practical matters is evident: given some practical rule, we judge particular cases. In a dense but fascinating discussion, Engstrom goes in the opposite direction, using the very notion of judgement to get us to the relevant rule. He argues that judgement differs from thought in that ‘its combination of representations is nonarbitrary’ and ‘self-sustaining’ (1 0 1). This is a gloss on what Kant claims about judgements in the first Critique, namely that they are the means by which we organize our thoughts to make knowledge of objects possible – which itself relates to older conceptions of judgement as an actus of composition and of division (the actus consists just of the affirmation or negation, not of assent). However, as we shall see, Engstrom extracts from this basic idea a rule of judgement that contains the notion of agreement not just of consistency. On the theoretical model, to use Engstrom's example, when one judges something, for instance the ‘determinate magnitude of the sun’ to be such-and-such, then one's other judgements must agree with this (102). Presumably the idea is that if I judge the sun to be X, and also judge that Pluto is smaller than the sun, then I cannot also judge that the determinate magnitude of Pluto is X+1. Practical judgement retains this feature, but this time the object is not a contingent matter (as was the case with the sun's magnitude), so ‘nonarbitrary’ and ‘self-sustaining’ need to do more work. Engstrom uses them to get to ‘universality’. The move to universality is made through a claim about judgements, namely that they ‘sustain themselves individually through their agreement with themselves and collectively through agreement with one another’ (104). Retaining now from the more basic account of practical rationality the notion of possibility, we obtain the following additional, properly practical, criterion: that it be possible for every subject, who is capable of practical knowledge, to share ‘the practical judgement that every such subject is to act as determined in the particular judgement when in the conditions on which it is based’ (126). More concisely put: there is a ‘subjective’ universality across agents, which describes effectively intersubjective validity, and an ‘objective’ universality across cases (115). This double universality can be seen as a unifying notion for the practical object in question, i.e. the good, specifying what practical knowledge is knowledge of.

Double universality, and hence the ambitious conception of practical rationality, is presented as following from Engstrom's conception of judgement. However, it is underpinned by a specific view of agency, which Engstrom develops in parallel with this view of practical rationality and which has the task, as I see it, to secure a connection between mere capacity for practical knowledge and self-determination in accordance with such knowledge. Engstrom presents his theory of agency as an intrinsic part of his theory of judgement, to ensure both consistency in the ‘from below’ approach and also systematic unity. For analytic purposes, however, it is necessary, I think, to separate these two strands. The theory of agency is a variant of the self-constitution view. Practical thought constitutes the agent qua agent, it gives the agent's ‘practical self-conception’ (33). This self-conception consists initially in the thought that the object of practical thought is something that ‘can be produced as an effect by the subject’ (39). We saw earlier that practical thought whose efficacy is transparent to itself is intention. Now two more elements are introduced (as species of practical judgement): choice and wish. Wish is a potentially efficacious judgement about what is ‘simply good’ (68), for example I may have a wish to win a dressage competition, this may be unattainable but provided the end is not known to be impracticable the wish can generate all sorts of relevant practical deliberations and actions. Choice is an efficacious judgment about what is good ‘on the whole to do’ (67). Although to some extent both choice and wish coexist, giving content to our practical identity as agents, Engstrom suggests that wish can become choice with the addition of the notion of practicability, when the judgement becomes an all things considered judgement (i.e. about what is good on the whole to do). The final ingredient is the basic wish for self-sufficiency (89) which Engstrom deduces from the basic wish for happiness that, as Kant claims, animates all human agents. The self-constitution account gives us the subjective side of the theory: the concept of the practically judging subject, the determination of the agent who is the subject of practical knowledge. The upshot is the following conception of practical knowledge: knowledge of the good through the judgement of double universality is also a way in which the judging subject knows itself practically as self-determining (121, 122).

There are many captivating details in the argument, for instance the way in which, by introducing the notion of ‘wish’, Engstrom avoids some problems of Korsgaard's self-constitution view that seems to be committed to the attainability of the agent's end. There are also some details that I found puzzling. For example, I was not able to ascertain (a) what exactly distinguishes practical judgement from practical knowledge, in the sense of rational cognition, given the formal characteristics they share (i.e. if a practical judgement fails to meet the formal conditions of judgement, is it still a judgement?); (b) what exactly is ‘thought’, whether it is a component of judgement or something that comprises judgements – as in the thought ‘I hereby judge the magnitude of the sun to be such and such’; (c) how the strong claim about intention's awareness of its own efficacy is supported, for example whether something like Velleman's notion of prediction is applicable here (I suspect not but this further shows the need for a more detailed treatment of the idea of acting sub specie boni). But these are matters of detail, questions of clarification arising partly because of the ambition of Engstrom's synthetic vision, and partly, I have to say, his positively Aquinian delight in drawing terminological distinctions. I will conclude then by offering some critical reflections about the project as a whole.

Given the context of the discussion about practical knowledge, one aim must be to show the superiority of the double universality approach to other available alternatives, especially neo-Aristotelean and particularist. I imagine that particularists are likely to remain unconvinced by the assumption that judgements about the good have to have law-like form; and there are other ways of ensuring consistency in action without going all the way up to double universality. I think neo-Aristoteleans will have more difficulty piercing Engstrom's Kantian armoury, though I do think both the former's naturalization of virtue and the latter's rationalization of the good are vulnerable to questions of scope. Engstrom claims at the start that the account he develops does not aim to show that ‘any conceivable rational being’ must be bound by the categorical imperative but only that ‘any rational being who has the capacity to know the good’ must be so bound (8). This scope restriction serves only to state that we cannot get from a minimal conception of rationality to the fuller practical conception. This is perfectly true. But once this kind of sceptic is out of the room, the condition is meant to apply unrestrictedly to all human beings, an idea reinforced by the use Engstrom makes of the fundamental wish for happiness, which yields the self-sufficiency component of the theory and provides an immanent end for every person. This leaves another sceptic unrelieved of her doubts, this is the sceptic who doubts the reach of reason, given the possibility of being wrong in ways we simply cannot see. Here perhaps the promise of wisdom made at the start of the book acquires a consolatory function, provided we do not understand ‘wisdom’ as perfection of our exercise of our judgement (in the Mutterwitz sense Kant occasionally employs), but rather as knowledge of and by the highest causes.