Jeff McMahan, in The Ethics of Killing, introduces the notion of a 'time-relative interest' in one's future life possibilities, the strength of which depend upon the degree of psychological relatedness that would hold between one's present and future self. This explains why early abortion is harmless (contra Marquis): the early embryo has no psychology, and hence no unity whatsoever with the future person. It thus has no time-relative interest in having that possible future be realized, so a death that "deprives" it of that future is not contrary to its interests, or in other words, does no harm. A later fetus or newborn infant, by contrast, would have some (slight) psychological connectedness with its adult future self, so is harmed by early death (at least slightly).
It's a neat account. One puzzle that emerges (which I haven't seen discussed elsewhere) concerns how we should aggregate time-relative interests. In explaining the wrong of pre-natal injury (which causes harm to the future adult), McMahan writes that "we must evaluate the act in terms of its effect on all those time-relative interests it affects, present or future." (283) That is, while the pre-natal being has little or no time-relative interest in avoiding the pre-natal injury, the future adult's time-relative interests would be gravely affected (we may suppose), which explains why the pre-natal injury is a morally weighty affair. (In case of abortion, by contrast, there are no future time-relative interests to be negatively affected; the abortion prevents those interests from arising in the first place.)
So far, so good. But how exactly are we to take all the time-relative interests into account? Utilitarian "equal consideration of interests", adding them all up, clearly won't work. That would cause massive over-counting. Compare the deaths of a pair aged 20 and 40. Suppose that each would have lived for exactly twenty more years of (equally) happy life if not for their current demise. But the 40-year old has twice as many time-relative interests that are thwarted, and many of them will be stronger (assuming stronger connections between his 21-yr old self and current self than are found between the twenty year old and her past infant self). So aggregating intrapersonal time-relative interests would lead us to the conclusion that the 40 year old's death is a much greater harm. But that isn't plausible.
Suppose instead that the (apparent) 40 year old instead popped fully-formed into existence as an apparent thirty-nine year old, one year ago. This should make no difference to how bad his death now is at forty (given that he would be deprived of just as good a future, to which he would be just as strongly psychologically connected, etc.). But he has many fewer past time-relative interests in avoiding this death. So if we are to add them all up, they will now add up to less.
Surely, a person's interest in some event should only be counted once, not once for every moment that they were alive. This suggests to me that the appropriate way to "take into account all the time-relative interests" in some event is not to count them all, but rather, to find the one strongest time-relative interest from across one's lifetime of moments, and count just that one. One's weaker time-relative interests in an event are to be subsumed within (not added to) one's greatest time-relative interest in it.
Typically, your strongest time-relative interest in an event will be concurrent with that event. So this explains why the harmfulness of an event typically doesn't depend upon the length of your life history. But in the rare cases when one's strongest time-relative interest is found at a different time (as in the pre-natal injury case), it can appropriately override the weaker interest one had at the time of the event in question.
Does that seem right?