One of the recent frames that's helped me reorganise my value-philosophy is that of rule-based versus utilitarian versus virtue ethics; I've been talking about virtue for awhile, as something complementary to my value metatheory, but the focus on the content of the actual value system as being components of virtue is helpful.
Getting close (in a meetup group) with 1-2 other philosophers (of the academic variety though; I'm the gentleman-scholar variety) who strongly identify with virtue ethics but have a very particular actualisation of that trend has been challenging, mostly in a good way (although I don't like needing to beat back their strong insistence on pushing their organisational and axiomatic choices; philosophy is too divergent to warrant the normal academic privilege to define these things). One of the ideas that stries me as strangest is the idea that the duty one has to others in society one also has to oneself. There is a certain practical elegance to the idea, in that it allows for a natural defense of not bending over backwards for the benefit of others; if one is obliged to respect oneself as much as the other person, one is therefore obliged not to do that (this is surprisingly relevant in some discussions I've been in where certain groups would like to ask other groups to take on social burdens to correct some real or imagined injustice). There's at least a question as to whether one's philosophy should treat oneself as just another human or be perspectived, and theoretically a nonperspectived treatment would generalise more smoothly to political philosophy. I don't take that tack though; I see philosophy as being necessarily perspectived, treatment of the self as being different in nature as treatment of others, and the idea of a duty to oneself to be almost entirely foreign. There is a gap between my political philosophy and my perspectived how-to-live philosophy. Here's a sketch as to why, and as to how this works out for the above traits.
The structuring comes from that we naturally are perspectived in everyday life; value-philosophy, as a default, should illuminate everyday perspectives and remain close to them except when distance from them is justified by valuable characteristics. While there is a parsimony in treating all humans the same way, it is a parsimony that is alienating, and for what I claim does not actually have useful/aesthetic effects.
As perspectived beings, we (can) have intimate knowledge of our values; some of them require some philosophical sophistication (and the process of achieving reflective equilibrium) to achieve, but we have at least primal responses to value-laden events and, when we are psychologically healthy, various desires that shape our lives. These differ somewhat between people; the differences are usually stronger in the preferences and the commits, at least within a given culture (by virtue of the fact that sufficient moral and ethical disagreements within society are difficult if large enough; a culture will usually inculturate strongly and slightly-fuzzily the moral and ethical conclusions that are common to it). Even as differences in value-conclusions are common, it's more common for a given value-framework, in application, to need to be specific to the desires of those involved at the time (as part of being parameterised by individual variance and situation) in order to be applied; notions of permission/desire are mechanisms for the application of the value-system of by actor to situations involving other actors and their desires. Freely-negotiated contracts are an example of formalisation of the expectations of everyone involved in ways that are theoretically equitable to all involved.
Rather than making a direct attack on the idea of duty-to-self at this point, I hope to present a compelling alternative, based on my idea of human nature. In contrast to both Locke (optimistic view of human nature) and Hobbes (pessimistic), I suggest a view that humanity is many layers of beast with some thinner layers of intellect riding on top of and controlling the beast; the beast cannot be entirely neglected because many kinds of happiness (and sanity) rely on the primal layers being at least moderately satisfied, but neither can the beast be a primary or sole kind of target for satisfaction, as doing so would neglect the potential, sanity, and happiness of the higher layers; tension between the layers of humanity is inherent to the human condition, with different people and different cultures striking different comprimises in their habits, institutions, public mores, and laws. So summarised, I first contend that any notion of self would by necessity need to be intensely personal (and so removing some possibilities of concept of self-duty) and then that any coherent concept of self-duty is necessarily intrusive to the negotiation that normally happens between the layers of self; instead, a notion of prerogatives that may be exercised, buttressing a legitimate interest in reasonable dignity. The reason that the intrusiveness is fatal to the concept is that such negotiations within the self usually happen in ways that are unquantifiable (if you wish to call this a form of Loki's Wager fallacy, I would not disagree, although in this case I feel that it's not actually a fallacy). The reason this is not a problem when dealing with other people is that the normal content of duty to others, that of opportunities and perogatives and the like collapses when applied to the self; perogative (or duty) is normally understood in terms of providing options that can be waived; this kind of structure adequately captures the great variety in desires-of-the-moment and desires-of-the-person that might make a choice to do A or !A part of the reasonable completion of one's life-story-framing or layers-negotiation for that person or situation. I do not address here other concepts of duty that can have significantly different content; were one presented I would need to provide a different analysis.
As for the political advantage of a duty-to-self view, I claim that is in practice illusory given that we rarely live in a society that matches our personal value-frameworks-for-society (if we have bothered to think them out); any such advantage must be purely theoretical and in practice would fail to extend to that layer anyhow because duty cannot capture anything consistent from the parameterisation process of the shifting desires/negotiations within the layers of a person, except in the form of prerogatives.
I view the idea of not-bending-backwards as a matter of duty-to-self a problematic rationality, in that it hides what should be a reasonable choice between entities, each with some self-interest and some public-interest, behind an alienation (in the same sense that alienating one's views to a creed, or a god, or similar is bothersome). Ordinarily the idea of legitimate demands we can put on another has the character of a worked out version of that tension, having a similar natural generalisability as self-duty without the complication.
(this is kind of rough; I find myself needing to leave where I'm typing this, and find this complete enough to post but probably in need of rewording and clarification in some parts; problems will be fixed in a future restatement)