Monday, June 24, 2019

Actualism, Evaluation and Prerogatives

In his new paper, 'An Argument for Objective Possibilism', Pete Graham argues that Actualism has trouble with cases of merely permissible (i.e. not required) beneficial sacrifice.

Suppose that the agent can now save two out of three lives, but knows that if they do, they will subsequently choose to save the third life at the cost of losing their leg.  Graham suggests that Actualism must treat this as equivalent to the choice of saving three lives at cost of losing a leg, or letting all three die, and that such sacrifice is not required.  But it's absurd to permit the agent to let all three die when they could have saved two lives at no cost.  So, Graham argues, Actualism is false.

It's an interesting objection.  I think the Actualist has a couple of different options for how to respond.

The Concessive Response would be to restrict Actualist normative assessment to the evaluation of impartial costs (or, more precisely, the weights of their normative reasons for action).  This covers the standard cases in which it'd be inadvisable for an agent to attempt to follow the ideal sequence of acts due to likely future deviations proving disastrous.  But while costs imposed by the agent's future choices get counted normally for purposes of assessing the weights of the reasons they now face (determining which of their current options is best, etc.), those future-chosen-costs count differently for purposes of generating prerogatives (permissions to do less than the best).  For whereas certain self-sacrificing (or morally fraught)* costs can excuse agents from doing what would otherwise be required, they can -- on this concessive view -- only do so when the costs are unavoidable for the agent in possibilist terms.  The fact that the agent will (even inevitably) freely choose in future to take on such costs if they now save two lives does not make it permissible for the agent to now refrain from saving the two lives.

* For the morally fraught case, suppose that after saving the lives you will subsequently turn a trolley, killing one to save five.  This is usually thought to be permissible but not required.

The Bullet-Biting Response maintains that Actualist normative assessment has universal scope, and instead seeks to defend the view that the agent's current choice to save two in this situation really is equivalent to choosing to save three at cost of their own leg, and hence really should not be viewed as morally required.  My old post on Deliberative Openness and the Actualism/Possibilism Dispute makes some moves in this direction by stressing how pathological it would be for the agent to truly be currently unable to secure the less-costly future choice.  If you imagine someone who truly deeply wishes that they could save the two lives and then stop, without sacrificing their leg for a third life, and is deeply distressed by their current inability to secure this outcome (no matter how hard they try, they know their future self will ignore their current resolutions), it no longer seems so problematic to permit them to do nothing.

I lean towards the bullet-biting response, but would be curious to hear what others think.  What is the best way forward for the Actualist here?


  1. It seems like the clearest way to describe the situation is that saving two people will cause you to lose your leg while saving a third. And if that's what it causes, I think it's pretty clear that it's supererogatory. I know the original wording says that it will cause you to *choose* the leg-sacrificing mission, but on any good theory of causation, the latter makes the former true. Nobody says that Hitler did't cause a war, because he only caused people to choose to drive tanks into Poland while shooting. That's what causing a war is like.

    Another way to think about it is that the charitable missions are causally coupled with a rationality destroyer, so the cost of embarking on a path is the destruction of the agent's rational judgment - which is why you will choose so poorly once you're two steps down the path. Once you know this, it's not much of a bullet to say that it's better to not start down the path. I'm saying that you chunk the whole sequence as a "do I or don't I let myself step down this slippery slope?" decision.

  2. [Pete Graham sends in the following replies:]

    Thanks again, Richard, for engaging with my argument against Actualism. I think you lay out the only two ways the Actualist can respond to my argument. I find the Concessive Reply unlovely and a bit ad hoc. If I were advising the Actualist, I'd suggest going for the Bullet-Biting Reply. Now, I don't find the Bullet-Biting Reply plausible, but I don't think I have much to say against it except the argument I offered itself. I don't think focusing on the distressed agent--the one who is distressed about not being able to ensure now that she doesn't act supererogatorily in the future--can really help the Actualist here. Because take the undistressed agent who in fact does dive in, save the two, and then save the third while losing a leg. The Actualist has to say of her that it would have been ok for her to not save the two. Just that, I'd contend, is implausible.

    Unknown: Thanks for engaging with my argument. I think the whole dispute between Actualists and Possibilists is a dispute about how the various causings are morally relevant to one's present obligations. The Actualist says that they're all the same, the Possibilist, by contrast, says that causings that go through the agent's own future agency are morally different than causings that don't. I point to a consequence of the Actualist's view that it seems to me is implausible. I can't really do a lot more against the Bullet-Biting Response than just urge a reconsideration of one's intuitions about the permissibility of letting the two people drown in the case as I describe it. Consider another version of the case, one in which the agent has contracted with the two "to save them if ever they need her to save them as long as she can do so without suffering a loss as great as a leg". Wouldn't the two have a legitimate grievance against the agent were she to let them drown? Wouldn't she owe them (or their families) compensation in such a case?

    1. Hi Pete, just to clarify, I don't take the distress itself to be what matters here. Rather, the thought is that the distress helps to bring to our attention that the agent truly lacks the current ability to influence their future behaviour. This is, I think, deeply unusual. So to simply present a case without dwelling on this point risks misleading us, if we implicitly imagine a more ordinary case where the agent simply fails to form an effective intention about their future behaviour despite being perfectly capable of doing so.

      "The Actualist has to say of [the undistressed helper] that it would have been ok for her to not save the two."

      Well, again, not in the most natural way to imagine the case. An ordinary undistressed helper would have been capable of resolutely choosing at t1 to save just two, so in the case where she's unwilling to sacrifice to save all three, this alternative is what's required of her. To create a contrast with Possibilism, we need to imagine a pathological case where she's truly incapable of such resolution. And then we arguably should say that "it would have been ok for her to not save the two". After all, she could have reasonably prioritized her leg over others' lives, and felt distress at her pathological inability to save the two lives without subsequently losing her leg. That would have been completely understandable and permissible. And that is the sense in which it "would have been ok for her to not save the two."


Visitors: check my comments policy first.
Non-Blogger users: If the comment form isn't working for you, email me your comment and I can post it on your behalf. (If your comment is too long, first try breaking it into two parts.)

Note: only a member of this blog may post a comment.