The conference I attended was on social psychology and aging, and it included a brief mention of a recent finding regarding advance directives. The finding, like so many in social psychology, is not very surprising once one thinks about it and so gains its potential profundity from one's probably not having thought about it: evidence indicates that people giving directives in advance of an illness say they want less-heroic-measures than people actually in the midst of the illness.
I was reading a trashy mystery lately where the thirty-ish spunky detective said that once she needed a diaper in order to go to the supermarket, that was it, she wanted someone to kill her. One can imagine the detective being so adamant as to be willing to sign an advance directive to this effect. At the same time, the vast majority of older people who find themselves needing a diaper to go to the supermarket still would just as soon go on living. So the detective could find herself in that situation where she has signed a sheet of paper saying that she wants to die if she's ever in that situation.
Fortunately, under current laws, you cannot kill someone if they say they want to live, even if you have written permission to the contrary. Move to real cases, of course, advance directives are intended to kick in precisely when one is incapacitated and cannot express what they want. The problem is that advance directives, in terms of expressing actual wants, may actually only represent the lower bound in terms of the amount of heroic-energies that persons who are incapacitated would like expended on their behalf if their ability to express a non-advance-directive was not precluded by their being incapacitated. Which is, one might imagine, the opposite of the type of error we might want to make--that is, where we might think the best direction effort is to provide more-heroic-effort than the person wanted, advance directives might actually lead the medical system to under-expend.
A virtue of advance directives is that it spares the immediate relatives of the incapacitiated person all the troubles involved in speculating about what the incapacitated person would have wanted. I have been in that situation (the immediate-relative situation, not the incapacitated-person situation), and so I have some sense of the unpleasantness and uncertainty that surrounds decisions about life support when an advance directive is in place. In the absence of the present-self being able to express their wishes to us, it's seems obvious that we might pine for some guidance from a past-self. It's strange to think then there might be reason to wonder if this method for having the past-self speak in situations where the present-self cannot might actually provide a voice that is more naive than what it has been credited as being.