However, it is important to notice that even though adopting a hedonistic life project may tend to interfere with realizing that very project, there is no such natural exclusion between acting for the sake of another or a cause as such and recognizing how important this is to one's happiness... while the pursuit of happiness may not be the reason he entered or sustains the relationship, he may also recognize that if it had not seemed likely to make him happy he would not have entered it, and that if it proved over time to be inconsistent with his happiness he would consider ending it.
So the sophisticated hedonist (SH) may take the goal of hedonism to regulate his other ends, but nevertheless regards those first-order contingent desires as non-instrumental for as long as he retains them. Railton continues (p.157):
It might be objected that one cannot really regard a person or project as an end as such if one's commitment is in this way contingent or overridable. But were this so, we would be able to have very few commitments to ends as such. For example, one could not be committed to both one's spouse and one's child as ends as such, since at most one of these commitments could be overriding in cases of conflict. It is easy to confuse the commitment to an end as such (or for its own sake) with that of an overriding commitment, but strength is not structure.
I don't think that is an "easy" confusion to make at all. It would be downright silly for someone to think that a desire must be instrumental merely because it was overridable. As I see it, the worry here is not that hedonistic concerns may outweigh SH's other desires; it is that they may extinguish them utterly. (Non-hedonistic ends seem to be treated as in some sense providing merely prima facie rather than pro tanto reasons.) We do not find this in ordinary cases of conflict: a parent will still care about their spouse, even as they favour their child. But SH, on my favoured reading, would cease to recognize an end if it proved clearly detrimental to his long-term happiness. So the structural relations of these desires is unusual, and this should be recognized.
We do not here have two first-order desires (on a structural par) weighing against each other. Nor - it is argued - do we have derived desires that are merely instrumental to one's hegemonic "true aim" of hedonism. Rather, in the case of SH we have first-order desires that are largely non-hedonistic, yet - despite being non-instrumental - they are contingent on the 'regulating aim' (let us call it) of hedonism. Hedonism is treated as a higher-order desire. It does not guide one's actions directly, but it guides the acquisition and maintenance of one's first-order desires. That's how I would want to explicate the idea, at least.
Railton's most vivid explication comes on p.159:
An individual could realize that his instrumental attitude towards his friends prevents him from achieving the fulles happiness friendship affords. He could then attempt to focus more on his friends as such, doing this somewhat deliberately, perhaps, until it comes more naturally. He might then find his friendships improved and himself happier. If he found instead that his relationships were deteriorating or his happiness declining, he would reconsider the idea. None of this need be hidden from himself: the external goal of happiness reinforces the internal goals of his relationships. The sophisticated hedonist's motivational structure should therefore meet a counterfactual condition: he need not always act for the sake of happiness, since he may do various things for their own sake or for the sake of others, but he would not act as he does if it were not compatible with his leading an objectively hedonistic [i.e. maximally happy] life. Of course, a sophisticated hedonist cannot guarantee that he will meet this counterfactual condition, but only attempt to meet it as fully as possible.
Discussing this in Michael Smith's seminar today, it was initially suggested that this 'counterfactual condition' - by implying that SH would never act on his non-hedonistic desires when doing so would be inoptimal - required a kind of overdetermination: although actually motivated by concern for others, SH's stronger hedonistic desire is waiting there in the background, ready to override the others just in case they fail to fall into line. But this just looks much like the simple hedonist. So I think we do better to interpret the sophisticated hedonistic desire as a purely higher-order one, which does not have any direct motivational force at all. (Note that the counterfactual condition may still be true due to finkish dispositions. Though it probably calls for a slightly looser interpretation in any case, i.e. SH may act inoptimally at times, so long as this doesn't too greatly undermine the happiness of his life as a whole.)
This is vital for seeing the difference between Railton's characters of John and Juan. Both feel great affection for their respective wives, and recognize the impersonal demands of utilitarianism as having ultimate moral weight in some sense. But John justifies his good treatment of his wife in directly utilitarian terms: "I've always thought that people should help each other when they're in a specially good position to do so. I know Anne better than anyone else does, so I know better what she wants and needs." (p.152) He thus seems troublingly 'alienated' from his personal concerns and relationships. Juan, on the other hand, responds simply: "I love Linda... So it means a lot to me to do things for her." (p.163) He adds the utilitarian justification only when further asked how in principle his marriage fits into the greater scheme of things.
Due to the assumption of structural parity, and thus motivational overdetermination, many in our class concluded that John and Juan were much the same, differing only in which of their two aligned desires were causally operative. But I don't think that's the right way to look at it. Juan isn't directly motivated by impersonal utilitarian considerations at all (we may say) -- not even waiting inoperative in the background. He has a quite different motivational structure, full of deeply personal and non-alienated concerns; it is just that these concerns are regulated by (or contingent on) a higher-order requirement that they align with the impersonal goals of morality.