n this interesting paper, Rabinowicz and Österberg (R&O) ask what a "preference-utilitarian" moral theory should assign intrinsic value to, (i) the circumstance that our (perhaps idealized) intrinsic preference be satisfied, or (ii) the states which are themselves the objects of those preferences. Österberg prefers the former, Rabinowicz the latter. Each presents arguments for his favorite and against the alternative. Prior to presenting those arguments, they argue that each version of the theory is anchored in a different model of utilitarian evaluation, one an ideal spectator theory, and the other an ideal participant theory. The ideal spectator is motivated by generalized but detached benevolence to desire the fulfillment of everybody's wishes. The ideal participant identifies with the perspectives of others and takes over their preferences.
R&O think that the two utilitarian views are often confused (by Rawls for example), but it isn't easy to see how the two approaches as developed in the paper could easily be conflated, for the latter seems not really to be a welfarist and therefore utilitarian view at all. There are two related reasons for this, implicit in the two models of evaluation R&O propose:
(1) The models correspond to different underlying theoretical motivations. Satisfaction theorists seem to be moved by the thought that what is valuable is people's welfare, plus the idea that people's welfare consists in either their actual or fully informed preferences being realized. Object theorists, on one obvious interpretation, are motivated by the thought that people are the best judges of what is valuable, so that the best way to find out what is valuable is to defer to people's actual judgements or at least to actual people's judgements in suitably favorable conditions. While the theory thus takes people's preferences seriously, it does not take them seriously in the usual utilitarian way.
(2) In the course of defending the participant model, the object view is given an "actualist" interpretation, so that what is valuable is the objects we actually (would ideally) value, even in counterfactual circumstances where no one has those values. Hence, the two views differ in what they would recommend for situations where people do not prefer what they actually do prefer in the real world. The satisfaction interpretation recommends satisfying the desires the people in that situation (perhaps would ideally) have, whereas the object view recommends realizing the states which would satisfy (perhaps idealized) actual desires. Depending on what we actually desire, this latter view can actually recommend decidedly anti-utilitarian choices. Under those circumstances, it would not be a version of preference utilitarianism at all.
Perhaps the conclusion can be resisted. I can think of two rationales for viewing the object interpretation as a species of utilitarianism, but both fit best with a non-actualist version of the theory. And such non-actualist object interpretations are not ruled out by at least some of the considerations that seemed to move the authors to reject them. For example, the distinction between intrinsic qua final and intrinsic qua internal that the authors note in footnotes would indeed help them meet worries about the supervenience of the moral on the descriptive while defending a non-actualist object theory. But on to the rationales.
The first rationale begins with the same epistemic ideas that I claim motivate the ideal participant model. It is not really all that attractive to take people's unfiltered preferences as providing irrefutable evidence of what is valuable. The very moral practices that lead us to postulate that states of affairs or objects might be genuinely valuable, include our sometimes believing ourselves to have judged incorrectly. Presumably a participant would take over this aspect of our everyday judgement practices. Better then to think of the participant as tracking our idealized preferences, where the conditions of the idealization ensure that we are in fact tracking what is valuable. Thus we get familiar conditions requiring full relevant information, proper reasoning, appropriate emotional responsiveness, etc.. But once we build such conditions into the idealization, we have no grounds for thinking that anyone in those conditions wouldn't "get it right". If there is still disagreement between actual and possible individuals the actualist constraint would be unwarranted; we should be relativists, not actualists. On the other hand, if we get convergence of views under such conditions, the actualist restriction is unnecessary. If we get convergence on human well-being as the only intrinsic value, we would after all have a utilitarian object theory.
A different motivating rationale starts from the same welfarist concern as the satisfaction view, but gives it an object-theoretical twist. One might doubt that what is good for a person is the mere satisfaction of her desires. One might believe instead that the desired objects or states of affairs themselves constitute her good. While this is not the most popular idea, it isn't as crazy as it might seem. The fact that a person desires something (in suitable conditions) might either make it the case that, or be evidence that the desired thing is good for her. (Mill himself, in Utilitarianism, allowed virtue to constitute the happiness of those who desire it.) If so, a benevolent observer concerned for a person's welfare would wish for the states of affairs or objects which constitute her good. The observer would count as a participant insofar as she shared the perspective of others as to what is desirable, but her motivation would also be one of benevolence.
Given such a benevolent perspective, it is most natural once again to drop the actualist bias. If the world were different, such that it was populated by people disposed to desire differently when placed in conditions ideal for judging their own good, a benevolent person would desire the same objects they do. And this desire would not be merely instrumental, because the objects of those desires would themselves constitute the good of the people which a benevolent spectator desires. A non-actualist object theory would be the result.
If either of these rationales is plausible, object theories would after all be candidates for utilitarian conceptions of value.
*I'd like to thank Thaddeus Metz for helpful comments on an earlier draft.