Saturday, January 11, 2014

Rationality and the Rooted Amnesiac

Suppose Amy has a deaf son Dale, whom she loves, and because she is so attached to Dale (and his actual personality) she reasonably prefers the actual outcome over an alternative world where she had a non-deaf son Hamish who grew up to be overall better-off than, but also very different from, how Dale actually is.  Call such partiality, rooted in one's attachment to an actual person, a rooted preference.

Some background:  I previously appealed to such rooted preferences as an objection to Caspar Hare's morphing argument, suggesting that a decent person might reasonably prefer a world where a counterpart of Dale exists over a world where a slightly better-off counterpart of that counterpart, but who no longer qualifies as a counterpart of Dale, exists instead.  Hare now responds by running his argument in terms of "active favoring", which is basically a special kind of counterfactual preference where we ask what the agent would prefer if she were agnostic about which world is actual.
There are a couple of different ways we could imagine this going.

(1) Suppose Amy retains all her (apparent) memories, etc., but just loses confidence in their veridicality.  Perhaps an evil demon tells her that there's a 50% chance that she's been massively deluded, and has actually led a very different life raising Hamish instead of Dale.  If so, she will soon have her "true" memories restored, etc.  Should Amy hope that her memories of life with Dale are a delusion, and that she will wake up to a happier Hamish instead?  I think the partialist can reasonably claim: No.  Given her apparent memories, providing grounds for a reasonable attachment to a child like Dale, Amy may (and perhaps even should) prefer Dale's existence over Hamish's, even though she is unsure which is her actual son.  So, on this interpretation of "active favoring", it does not help Hare's argument: it can be reasonably "rooted" in particular attachments just as old-fashioned preferences can.

(2) Suppose instead that Amy is a total amnesiac.  Not only is she unsure which world is actual, but we've also purged any memory traces (etc.) that the actual world had imprinted upon her.  Perhaps what Hare means by "active favoring" is what such a total amnesiac would prefer.  If all Amy knows is that she either has a deaf child called 'Dale', or a slightly better-off non-deaf child named 'Hamish', then it seems prima facie plausible that minimal benevolence requires that she prefer to have the better-off child.  After all, what basis could she have for preferring the worse-off child?

Well, let's develop the story a little further.  Recall that, prior to her amnesia, Amy loved her deaf son Dale very much.  She came to value his particular personality and way of life, including his deafness.  And let's suppose, more specifically, that she came to value Dale's personality and way of life in itself, and not just under the superficial guise of "being Dale's".  She comes to deeply value the deaf life, because her son is deaf; but what she ends up valuing here just is the deaf life and not the deaf life insofar as my son is deaf. The causal condition need not, and we now suppose does not, become embedded in the intentional content of what is valued.  Amy thus has a rooted preference for a deaf child, and one that seemingly could survive her amnesia.  If she no longer remembers Dale, then she won't fully understand why she came to have this preference for a deaf child.  But suppose she has it nonetheless, as a causal consequence of her forgotten attachment to Dale.  Is she unreasonable, or in violation of minimal benevolence or moral decency?  It seems harsh to say so.  But that is what Hare is committed to, if he is to rule out the permissibility of actively favoring the impartially worse outcome.

I'm less confident that rooted preferences are reasonable in the case of total amnesia (compared to the unreliable memory case).  Maybe the rooted amnesiac is unreasonable.  One reason to think so is that it would seem clearly unreasonable for someone to just out of the blue prefer to have the worse-off child.  But Amy the Amnesiac might be an internal duplicate of such an unreasonable person: there is nothing internal to her current mental state to establish that her preference was in fact causally "rooted" in concern for an actual person.  So if we are a certain sort of "internalist" about rationality, thinking that it supervenes upon the internal qualities of a person's mind at a moment, then it seems we're committed to thinking Amy unreasonable too.  (On the other hand, you might just take this case to constitute a counterexample to such an internalist view!)

I'm sympathetic to this form of "internalism", so perhaps the thing to say is that (i) Amy the rooted amnesiac is irrational to prefer a deaf child -- her amnesia stripped her of the grounds for her reasonably rooted preference, so its persistence in the absence of those grounds is no longer reasonable; but (ii) she has an excuse for this rational error, insofar as it is a natural continuance of what was a reasonable preference, with a respectable causal history.  So Amy thus doesn't warrant the kind of negative assessments that we might make of a person who arbitrarily, "out of the blue", prefers a deaf child.

Does that seem right?


  1. This is really interesting and subtle stuff! I am inclined to think that that doesn't seem right - that Amy isn't being irrational. It is through no failure of rationality on her part (or so I am inclined to say) that she got to where she did, and that seems important.

    This seems to me like it might be an interesting counterexample to the sort of internalism you talk about above, but I expect there may be away of retaining it while also avoiding charging Amy with irrationality.

    Another, more general thought - not sure if it'll be helpful, but here goes: what about the idea that rationality and irrationality are, at least sometimes, properties of *inferences* or more generally, formations of belief, which seem to belong to the category of processes rather than states? (One response, I guess, would be to just insist on distinguishing two notions here, and confining this internalism to the one dealing with states. But it might be argued against that that a unified theory is possible, or ought at least to be tried for. Also, it may perhaps be argued that irrational states are sometimes - or more ambitiously and probably less plausibly, always - irrational in virtue of their having been arrived at by way of irrational processes.)

    1. Thanks!

      "I expect there may be away of retaining it while also avoiding charging Amy with irrationality."

      If you have any thoughts as to how this might go, I'd be very interested to hear the details.

      In general, it seems possible to end up in an irrational state through no failure of rationality on one's own part. (Indeed, as Parfit taught us, sometimes it is positively rationally required to take, e.g., an irrationality pill). In this case, the cause of the irrationality is, in effect, the amnesia, as it deprives Amy of (what are arguably) the justificatory grounds for her rooted preferences. And it doesn't seem totally outrageous to claim that amnesia could, through no fault of the agent's own, render them -- in certain respects -- less than fully rational. Or does that still sound wrong to you?

    2. Taking the question at the end first: you're right, that sounds quite reasonable. I'm a bit more inclined than I was to accept that you can end up in an irrational state through no failure of rationality on your own part. But I'm not sure; couldn't we say that if, through amnesia or some other interference, you get into a state where your preferences (or credences for that matter) are "irrational-esque", your state doesn't count as irrational unless you fail - after having a reasonable chance - to update toward a non-"irrational-esque" state.

      So for example, if your reasons for believing P are suddenly wiped from your memory through no fault of your own, and even if it's usually irrational to believe P without reason, you can still escape the charge of irrationality by noticing this and abandoning the belief (or refurnishing a good reason for it). In this kind of case at least, I think it still seems more natural to me to say that the agent hasn't been irrational, than to say that they were briefly irrational through no fault of their own.

      Part of what bothered me about your case, I think, is that it deals with preferences rather than beliefs, and I guess I have hitherto tended toward a very minimal conception of what it takes for preferences to not be irrational; internal consistency being enough, for instance. (And this would be a broad and blunt way of retaining rationality internalism of the relevant sort while avoiding charging Amy with irrationality. But if a different viewpoint is taken, as it seems to have been in your post, I guess I have no acceptable suggestions for how to do that.)

  2. I would say that, if Amy's preference is causally rooted in some way to the actual existence of David, it is rational. If it isn't then it is probably a case of "right answer for the wrong reasons."

    Also, I would argue that even amnesiac Amy has a rooted preference for a deaf child, even if her amnesia is total and she thinks she prefers Hamish. Why? Well, most amnesiacs want to know who they used to be. They want to know what their previous life was like. They want to know what life goals and projects they used to have, and resume them if possible.

    -Amy has a preference to resume the life projects and goals she had before amnesia.
    -Before her amnesia one of Amy's life goals was the wellbeing of her deaf son.
    -Therefore Amy prefers her deaf son over a hypothetical healthy one.

    A good analogy would be a man who wants to drink a glass of water, mistakenly believing it is safe, when it is really poison. The man doesn't really want to drink the poison, he just mistakenly thinks he does. Similarly, if you completely erased Amy's memory so she didn't remember Dale at all, she wouldn't prefer Hamish to Dale. She would just mistakenly think she did.

  3. It seems totally unjustified to assume that there is a meaningful question of whether amy *should* hope that X.

    Is this a moral judgement? A claim that it would/wouldn't be (act) rational to so hope?

    If this is supposed to be a moral judgement than it has no obvious relevance to the concerns you raise. For instance, a utilitarian will simply say she should so hope just if that increases total utility. Other consequentialists will feel similarly and I suspect that only the most involuted (and most likely to be inconsistent) deontic theories will judge either way to be morally wrong.

    As far as rationality goes I don't think there is anything satisfying our intuitive notion of rationality. Our intuitive notion demands two things:
    1) If one rule for action (act-rationality) or belief (belief-rationality) *always* produces better results (more true...more desirable) than another the rule that produces better results is the one we should choose.

    Note that whatever rule you pick can't itself be justified from your evidence since things like basic logical inference rules can't be deduced from evidence and your prior probabilities are by definition what you believe before evidence.

    2) Rationality shouldn't prescribe doing anything in blatant conflict with what is intuitively prescribed by the evidence.

    Now the problem comes up with whether a mathematician should act (believe) as if a famous unsolved problem is very difficult when that is the view of all your colleagues but actually isn't true. Rule 1 says that the rational behavior is to act in accordance with with the truth which is also what you would believe given enough time to reflect (you can list all the proofs until you stumble across it). Rule 2 says that in this case we should clearly act in accordance with what our colleagues and our own heuristics indicate since we have no grounds to do any differently.


    I think this demonstrates that our intuitions about rationality are simply inconsistent. It comes down to the frequent fallacy that there is a principled notion of what counts as a rule and what counts as an exception.

    In any case the burden is at least on whoever wants to deploy the notion of rationality that there is a coherent notion that they are gesturing at. You don't get to simply assume that because we use the word and it feels like it's meaningful in every day situations that it's a meaningful notion even when applied to unusual circumstances.


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.