A poem written by Toni Rønnow-Rasmussen's oldest daughter and remnants of a bookshelf made for him by his father have, he plausibly tells us, great value for him. Personal value, value for someone, he argues, is a category the theory of value has neglected in favour of such old (and not so old) favourites as value tout court, instrumental, extrinsic, intrinsic and final value. Philosophers from Schleiermacher, Simmel and Scheler to Peter Winch have occasionally defended the deontic counterpart of personal values: duties or obligations for someone. Perhaps the greatest fan of personal values, which he called individual (as opposed to general) values, was Max Scheler, who employs the category in his accounts of conscience, love, self-love, wisdom, vocations and many other things. The distinction between value tout court and value for has a counterpart in ontology, the distinction between general and individual essences. It is safe to say that the former is as controversial in value theory as is the latter in ontology. Understandably enough, Scheler's sketchy attempt to understand personal value in terms of individual axiological essences has rarely been taken seriously. Is there an alternative way of understanding personal value?

Rønnow-Rasmussen's very rich and fascinating exploration of personal value puts forward a buck-passing account of this type of value in which the category of attitudes towards – for the sake of – is central. An object x's value for a person a consists in the existence of normative reasons for dis/favouring x for a's sake. An object has final, impersonal value if there is reason to favour it for its own sake; it has final value for a person if there is reason to favour it for its own sake, for the sake of the person.

These constitution claims are intended to be neutral with respect to subjectivism and objectivism about value. The deontic claim in the analysans may be understood in subjectivist or objectivist terms. The proposal is an important extension of the buck-passing approach. This approach easily handles such value relations as betterness (to be better than is just for there to be reasons to prefer one thing or state of affairs to another) and value-height (for one value to be higher or more important than another is for there to be reason to prefer one value to the other). If Rønnow-Rasmussen is right, the approach can also deal with personal value.

Part of the interest of the proposal is that even a philosopher who rejects the constitution claim for personal value should accept that there is an intimate link between personal value and for-the-sake-of attitudes. For the buck-passer, for something to be valuable tout court is just for there to be reason to favour it. But even someone who rejects this claim should accept that something's being valuable is a reason to favour it (perhaps even what makes favouring it correct). Similarly, even the enemy of buck-passing should accept that the personal value of an object is a reason for favouring it for its own sake. Thus philosophers who want to understand personal value, whether or not they accept a buck-passing view thereof, should take seriously for-the-sake-of attitudes.

Rønnow-Rasmussen's pioneering account of these attitudes traverses a great deal of territory and illuminates many familiar claims. The distinction between personal and impersonal value, understood in terms of fitting attitudes, is, he suggests, not to be understood in terms of the distinction between agent-relative and agent-neutral reasons, since all reasons are agent-relative (chapter 9). Many objections to the very idea of personal value are carefully rejected (chapters 4 and 7). The relation between personal value and theories of welfare, hedonism and preferentialism are explored (chapters 8, 10). Along the way many important arguments and distinctions in general value theory are defended, for example the distinction between conceptions of instrumental value according to which what has instrumental value has value and conceptions according to which this is not the case (chapter 4), and the distinction between the roles of reasons in admiration and (personal, post-troubadour) love (chapter 5). Chapter 3 is devoted to the wrong-kinds-of-reason problem for buck-passing approaches.

Rønnow-Rasmussen distinguishes sharply between the reasons for (dis)favouring something and the reasons for (dis)favouring something for someone's sake, and refers to the reasons component and the attitude component of the latter in his analysis of personal value. But there is a serious objection to the account given of the attitude component, the view that favouring y for z's sake is one attitude with two objects. Instances of ‘x favoured y for z's sake’ are dominated by a hybrid connective or prenective, ‘for the sake of’, which takes a sentence and a name to make a sentence, and not by a psychological verb. If Toni wants to go to the theatre for his wife's sake, then he wants to go to the theatre and, if asked why, that is, for a reason, may reply: ‘for my wife's sake’ or ‘in part, for my wife's sake’. His wife plays something like the role of a reason here. It is true that a wife does not look much like what is often called a reason. (Compare ‘Sam makes it true that Sam exists’; Sam does not look much like what is usually called a ground.) He wants what he wants because of his wife, although ‘for the sake of’ does not mean ‘because of’. This strongly suggests that favouring for the sake of is not actually one attitude. Rønnow-Rasmussen introduces the distinction needed here in his discussion of ‘favouring something for its own sake, for someone's sake’. He notes that this might be taken to be one attitude or a complex or combination of attitudes. The same choice arises already in the case of favouring something for someone's sake. And here, I suggest, we clearly already have a combination of attitudes or a combination of an attitude with another act or state. Favouring something for someone's sake belongs to the same general family to which belong conjoined assertions (as opposed to the assertion of a conjunction), judging on the basis of a supposition (as opposed to judging a conditional content), and acting, judging or emoting on the basis of some fact.

If for-the-sake-of attitudes are really combinations of attitudes (acts, states), does this have any consequences for the claim that personal value is constituted by for-the-sake-of attitudes and reasons for such attitudes? One consequence, it might be thought, is that the constitution claim entails that value for is not after all a genuine relation, as are worseness and axiological height. If what looks like a relation is constituted by what is not really a relation, then the appearance of relationality is just a mere appearance. Of course, claims such as that x emotes y for the sake of z can be parsed in such a way that they are dominated by a relational predicate: Emotes for the sake of (x, y, z). But the claim dominated by the relational predicate only holds in virtue of the original claim, which is not dominated by a relational predicate but by a prenective. Emoting for the sake of can be treated as a relation, but at bottom it is no such thing. There is, it is true, at least one type of relation which is often said to be constituted by non-relations: internal relations. But it is not obvious how this strategy could be applied here. And are all personal value relations internal relations?

Two questions a theory of personal value should address, and which Rønnow-Rasmussen only touches on, concern thick values and goods. In a footnote he indicates that he thinks there are thick personal values and mentions ‘admirable for’ and ‘loveable for’. But ‘admirable’ and ‘loveable’ are really only place-holders: to be admirable is to be, for example, elegant. Is personal value ever thick? ‘Ugly for you’, ‘evil for them’, ‘unjust for us’ and ‘elegant/dumpy for her’ do not strike me as good candidates. But perhaps there are nameless thick personal values, just as, if Aristotle is to be believed, there are unnamed virtues. The second question concerns the relation between goods and being good for. If, as is often claimed, to be a good is to be a good for someone, what is the relation between being a good for someone and being good for someone?

Rønnow-Rasmussen's stimulating and often persuasive book brings into clear focus part of the geography of value which has long been overlooked. It is also an important and original extension of the buck-passing approach. It continues the great tradition of Scandinavian value theory.