his paper is essentially a dialogue between the two authors concerning which of two interpretations of preference utilitarianism (PU) is the better. The two versions of PU are the satisfaction and the object interpretations. The satisfaction interpretation (defended by Osterberg) states that what has intrinsic value is "the circumstance that our (intrinsic) desires and preferences be satisfied". The object interpretation (defended by Rabinowicz) states that what has intrinsic value are "those states that are the objects of our (intrinsic) preferences and desires" (p. 2). The article provides a useful summary of many of the main arguments for and against these two positions, and we shall not attempt, any more than do the authors, to deliver a final verdict. (Indeed, we find neither view plausible.) Our main worry is about the authors' agreed account of what motivates these two different versions of PU.

"To the satisfaction and the object interpretations of the preference-based conception of value correspond, we believe, two different ways of viewing utilitarianism: the spectator and the participant models" (p. 3) In the former the spectator detaches herself from her personal engagement in the situation; in the latter the participant identifies, in turn, with the point of view of each agent. The former requires that one step outside one's own skin; the latter that one imaginatively enters into the skin of others. The latter requires, "not a detached subjectivity, but a universalized objectivity" (p. 4). Clearly, such talk is metaphorical, and without more explanation of what each stance involves it is unclear what the consequences would be for the agent's view of values of adopting one or the other. How do the authors arrive at the conclusion that the spectator and the participant models respectively lead to the satisfaction and object interpretations?

We take the argument for the participant stance first, as the more persuasive of the two. "According to the participant model, a moral subject identifies with other persons and thereby takes over their preferences. Insofar as I make this take-over, the others' intrinsic preferences become my own and the objects for those preferences thereby present themselves to me as intrinsically attractive. The participant model, therefore, quite naturally leads to the object interpretation, which assigns intrinsic value to what is being intrinsically preferred" (pp. 5-6). What, from the inside, we find valuable is not the satisfaction of my desires, but the objects of those desires. This derivation is plausible, though perhaps not irresistible. It is not clear, for example, that Hume (who is cited as the classic source of this approach) drew this conclusion from the participant model. Sympathy might be thought to require only that one comes to share the feelings of, or have feelings similar to, those with whom one sympathetically identifies - not that one actually comes to see as valuable or attractive what the object of one's sympathy finds valuable or attractive. (The quotation from Hume's Treatise, Book II, Part I, Section XI which they give on p. 4 seems to support only the weaker form of identification.)

But it is the move from the detached stance to the satisfaction model which seems to us most problematic. The move is presented very briefly: "According to the spectator model ... the moral subject's attitude is one of benevolence: he wishes him well. But when I wish you well, this does not mean that I want what you want" (p. 6). (This last phrase must, presumably, be interpreted as meaning that I don't necessarily myself want whatever it is that you want, not that I don't want your wants to be met - for to want your wants to be satisfied is characteristic of the satisfaction interpretation.) We are given no guidance on what detachment involves. Detachment comes in different modes and degrees. We might here turn to Nagel's conception of the objective self for illumination, but in a footnote on p. 6 they appear to distance themselves from Nagel's account. Their only gloss is that "an ordinary person can approximate this attitude by detaching himself from his personal engagement in the situation" (p. 3). But while that tells us what the spectator is not to take into account, it does not seem much help in telling us what he is to take into account.

There is, of course, nothing in the claim that detachment (however understood) is the correct stance from which to make moral judgements that supports any particular view about what kinds of thing are valuable. Nor, pace the quoted remarks on p. 3, does a detached stance from within a utilitarian perspective favour the satisfaction rather than the object model. For what, as the authors rightly point out, the benevolent utilitarian spectator wants is that the lives of all go well. But the detached stance does not of itself tell us what it is for a life to go well; in particular it does not tell us that well-being consists in the satisfaction of desire, rather than in people's enjoying certain goods. They write: "because of the detachment that is characteristic to his transcendental perspective, he no longer has any particular conception of the good" (p. 14). But this is question-begging. To lose a personal conception of the good is not thereby to lose a particular conception. It is not even clear that, from within a preference utilitarian perspective, detachment would favour the satisfaction model. If I imaginatively disengage from my desires and concerns, rather than imaginatively identify with those of others, why should that disengagement prevent me from taking the objects of preference to be valuable, if that is a legitimate understanding of what is valuable from within the PU perspective? Conversely, if detachment "makes me lose the reason to want things that we human beings, with our definite positionings in the world, tend to want" (p. 6) why should the detached valuer find a resting place in the satisfaction of preferences? Why should he find any reason to care that humans get what they want, though he cares nothing for what they do want? Not only is it unclear why detachment should stop at this point, and not lead on to the vanishing of all value, it is also unclear why a concern for the satisfaction of desires would be the last thing to vanish as detachment led to evacuation of meaning.

We think that there is a more plausible source in the utilitarian tradition for the view that what is valuable is the satisfaction of preferences rather than the objects of those preferences. The utilitarian tradition holds, as the authors acknowledge, a broadly welfarist conception of value. There then arises the question: What is it for sentient beings, and particularly humans, to fare well? One answer is that to fare well is to have one's (prudential and rational) preferences satisfied. On this view, what is of value is the satisfaction of those preferences, whatever they might be. That it is the satisfaction of preferences that is of value, and not the objects of those preferences is explained, not by the adoption of a detached rather than an engaged stance, but by the fact that utilitarianism offers a welfarist account of the good. But if that is right, then the object interpretation is not a species of utilitarianism at all. It is a form of consequentialism, with an axiology in which what is good is identified by what suitably reflective persons in the actual world would prefer. It is not welfarist because it fixes what is valuable by (idealised) preferences in the actual world. Thus Rabinowicz writes in Section 3: "If we have a (considered) intrinsic preference for, say, the survival of rain forests, even for the hypothetical case in which we would not care for their survival, then the object interpretation implies that this state of affairs would have intrinsic value even though its realisation would then not satisfy any preferences" (p. 10). But in such a hypothetical case the survival of the rain forests would still be valuable even though, we are supposing, their survival does nothing to satisfy preferences. (The example might not be ideal because, firstly, there might be desires other than an intrinsic desire for the survival of rain forests that might be satisfied by their survival and, secondly, the preferences of the animals who live there would be affected by their survival. But the example could be amended to avoid these difficulties. The limiting case would be where we had preferences for a world in which there were no sentient beings)

Our insistence that the object interpretation of PU is not a form of utilitarianism at all is not just a case of the definitional sulks. It is quite wrong to see the two interpretations' as variant versions of a broadly welfarist conception of value. As Tannsjo is quoted as saying (p. 25) "the object interpretation is not really a version of preference utilitarianism but rather an objective list' theory of intrinsic value conjoined with a projectivistic preference-based meta-ethical view." Tannsjo goes on to claim that "the object-theoretician's axiology, unlike his meta-ethics, is not really preferentialist". This is not, perhaps, the happiest way of putting the point and allows the authors to reply to his criticism by doubting if a sharp line can be drawn between axiology and meta-ethics. The crucial point is the different role that the notion of preference plays in the two interpretations. In the satisfaction version what matters is that, in any world, the preferences of sentient beings in that world are satisfied. For the object version, preferences in the actual world serve to fix or determine what is valuable but neither they, nor their satisfaction, are (any part of) what is valuable. Indeed, a world in which there are no preferences to satisfy can be valuable.

A final comment. In defending the object interpretation, Rabinowicz goes for a rigidification move familiar from discussions of colour. Just as someone might claim that what is green is determined by human colour responses in the actual world, so the object-theorist can identify what is valuable with what is the object of (idealised) preferences in the actual world. On p. 18, in response to worries about the fact that other persons in other worlds might have different perspectives from ours, he replies that "the answer ... must be that the object interpretation presupposes a subjectivist (or projectivist') theory of value." It was unclear to us why he thought that the only option. Those who defend a dispositional account of secondary qualities, and who avail themselves of the rigidification manoeuvre, need not and standardly do not see themselves as holding a projectivist or anti-realist view of colour.