I haven't read the book, merely listened to a podcast about it and followed these summaries, but I had a similar reaction. I just don't understand the big deal about first person experiential evaluation, relative to 3rd person reports etc.
Here's a though! How would you decide what would it be like to become a bat?
Also, the curiosity-value doesn't help you when you're deciding between two unknown futures - should I have a biological kid or adopt? It also doesn't help with two bad options - should I get a transfer to Siberia or move my family into my parents' basement?
At which point I think the species just died out: people who don't want to be transformed shouldn't have kids, because it's transformative.
Thank god some responsible people will still accidentally get pregnant.
On rereading, something I thought and forgot to say is that I thought the nature of 'authenticity' was underexplored. It's clearly an important concept, as one of the main reasons that first-person experience is the only acceptable source of information for decisionmaking, but I don't have a solid sense of it.
Evaluating based on your previous transformative experiences without taking account of third party reporting seems to lead to the Russian Roulette problem: "It's been harmless and exciting so far; I think I'll pull the trigger again!"
Bluebeard's wife probably started by poking around in all the non-locked areas, then peering through the keyhole, then sliding a mirror under the door.
Dumbledore was copying Peter Pan? I didn't know that.
12 & 13 are both wrong! Delicious!
I don't think Paul is arguing that you need to completely ignore third-person perspectives. Just that for major life-decisions there needs to be a first-person component.
I rather liked this paper's approach to (basically) the same issue. Differences seem to be:
1- Ullmann-Margalit doesn't really address *epistemically* transformative decisions. I'm kind of dubious about the weight that Paul places on that category (the black-and-white-room-raised person now seeing colors example, etc.), so this isn't a problem for me.
2- Perhaps because of this, whereas Paul thinks the only meta-rational way of choosing is to use one's curiosity, one's preference for having-been-the-sort-of-person-that-transforms, as the guiding rule, U-M thinks it's perhaps rational to act in accordance with 2nd-order preferences about being a difference person. A materialistic executive might want to be a spiritual Buddhist monk--because despite fitting badly with their current first-order preferences about possessions and immediate fulfilment, it would fit better with their deeper judgments about the good life, and dissatisfaction about the one they currently lead.
One can gather a lot of third-party testimony about what it's like to be a parent, make judgements about the parental role in society, etc., and decide that a parent is or isn't something you would like to be, which partly takes into account how parent-you will experience life going forward, but is guided most directly by whether present-you views the expected parent-you life as one worth having, even if, to a certain extent, it's no longer you that will be having it. What makes such a choice an authentic one is whether you're using your current beliefs and values to decide whether parent-you would be a better you than present-you.
Paul might object that this approach fails when it comes to genuinely transformative experiences, where future-you is so far from present-you that it doesn't make sense to talk of future-you being a better or worse "you" at all. To which I think the response is:
a- I think it's an empirical, psychological question of just how far future-you can be from present-you and still feel, to the present-you who's trying to decide, like "still you". (And the philosophical rationality of the choice depends on that contingent, psychological feeling of connection, rather than on some deep truth of identity.) Moreover, I don't think there are all that many real-life cases where the perceived sense of identity breaks; I don't think the ones from the book I've heard mentioned would count (at least for me). I also think most of the cases where it does break down are often felt as unchosen conversions (Paul on the road to Damascus) rather than explicit decisions (this is another important distinction that U-M addresses up).
b- Even in cases where the sense that you are deciding *for yourself* breaks down, that doesn't mean you've left the space of reasons. Take suicide missions, or other cases of intentionally sacrificing yourself. Assume materialism/atheism. Obviously there's no more first-personal experience happening after death, and so you can't decide it's better to sacrifice yourself because of how things will go for you afterwards. But you can still evaluate how things will go for others, and that can be balanced against the value that present-you places on continued selfhood and your aversion to suffering, with the result generating an authentic judgment about whether the sacrifice is worthwhile. I believe I've read and/or seen a bunch of sci-fi stories that are very much like this--a character chooses to, say, merge their consciousness with an AI in order to save the world or something, and in those cases I think the natural reading is that 1- it's a sacrifice, insofar as it's a loss of selfhood for the chooser, but 2- it's rational, insofar as saving the world is in tune with their values, and 3- with the sense of continued identity broken, the experiences of the future merged-AI self might still count in the calculation of reasonableness, but only insofar as any other third-parties' experiences would count, and hence probably not a significant part of the decision.
---
As an aside, the U-M paper also makes some interesting points about other "reasonable if not rational" ways to confront these decisions: 1- trying to break it into smaller step-wise choices, so that at each point you're able to preserve a consistent set of 2nd-order preferences over identity; 2- arranging things subconsciously to "drift" from A to B without ever fully committing to the transformative choice, and instead experiencing it as a set of things that happens to you. I think it's worth thinking about the extent to which those are in fact reasonable strategies vs. unreasonable self-deception.
Take suicide missions
If you need the plural, you aren't doing it right.
Maybe I should have read the other threads before commenting on this one. Oh well.
U-M thinks it's perhaps rational to act in accordance with 2nd-order preferences about being a difference person. A materialistic executive might want to be a spiritual Buddhist monk--because despite fitting badly with their current first-order preferences about possessions and immediate fulfilment, it would fit better with their deeper judgments about the good life, and dissatisfaction about the one they currently lead.
This makes much more sense to me as an explanation of how people do choose to make transformative decisions.
19: You aren't a little bit curious what the other threads say?
arranging things subconsciously to "drift" from A to B without ever fully committing to the transformative choice
That's been how I handled my career. It hasn't been without its moments, but sometimes I wonder what I would have accomplished if I'd been more driven.
6: You could be more curious about the experience of adoption or Siberia-endurance-learning.
This whole idea of authenticity seems fundamental to Paul's emphasis on first-person input without direct future simulation, there's more on it in the afterword. However a sense of personal ethics is put aside to focus on decision theory. Which makes sense from a point of view of managing the scope of the book, but makes the examples seem a bit absurd. A first personal drive to do a particular transformative thing is related to future expectations, and usually driven by a passion. There's both a future evaluation there and a willingness to undergo a transformation.
[using 2nd order preferences] makes much more sense to me as an explanation of how people do choose to make transformative decisions.
Well, I think U-M actually thinks people in general try to avoid feeling like they're making big transformative decisions, and instead usually go for the drift or step-wise pathways. And she's not convinced that 2nd-order preferences always generate a consistent ordering. E.g.: what do you say about someone who didn't want to have kids because they feared becoming boring, ended up having kids anyway, became exactly the boring they'd feared, but was now okay with it? I think that's actually easy: it was an irrational decision at the time, because counter to 2nd-order preferences, but is not in hindsight a regrettable one. But U-M thinks cases like this, especially when there's uncertainty thrown in, are hard cases for rationality; the paper aims to make things problematic rather than offer solutions. So in that sense it's a lot like Paul's book, except it was published 10 years ago (and was a working paper 20 years before that), and only ever got 16 citations, whereas Paul's book already has a symposium devoted to it, though it seems to have not met with the approval of The Mineshaft.
Hmm. Poking around a bit, everything I've said seems to have been said in previous comments on other threads. Oh well.
25.last
Not at all, I'm finding your take refreshing and it's making me wish I'd been able to get the book in time to participate (as well as reading the U-M paper which I'm about to do).
I imagine that Sifu's right about the reason Paul resists higher-order preferences as a solution (mathematically intractable, at least in a general way). And if so, I suppose that illustrates where the book diverges from the sort of thing I'd like it to be (and perhaps what others want, too; e.g. Halford's "how will this help me run my business?" complaint). I can see how needing a fully general and mathematically tractable value function is important if we want to, say, develop super AI. I don't think it's necessary to the development of practical reasoning in actual human beings, even ones who want to live as well as possible.
Can we get Emerson to come back here to complain about analytic philosophy?
Paul's book already has a symposium devoted to it
12 notwithstanding, this is the part the makes me keep thinking that I'm somehow just fundamentally missing the point. I read the book and think it is basically an extremely long-winded mediocre undergraduate essay (polished into a well-above-undergraduate-level work product, but at its heart still just not saying anything very interesting), but the author seems to be a credible academic philosopher, so I keep expecting that it's me, and not the author, who is missing the point. In a way it's been encouraging that more or less everyone else here has been reacting to the book in the same way that I have, but in a way it's been maddening--I was really instead hoping that these comment threads would explain to me the brilliance I was missing in the book, and then I would better understand why it's getting serious attention.
He's already done that, so it wouldn't be transformative.
It's still a bit odd to me why she considers the how-transformative-do-you-wanna-be decision procedure (which I like thinking of as YOLO Decision Theory) as legitimate when she rejects the higher-order-preference one. As LB points out, YOLO Decision Theory seems to have some odd consequences of its own, and it's also just really unappealing as a governing rule, letting the tail wag the dog in a serious way. I wonder if it's simply that higher-order-preferences-as-regulative-ideal has been around long enough to have been critiqued to death, whereas YOLO hasn't.
If you accept her standards for legitimacy -- firsthand knowledge of an experience is the only rational basis for choosing it -- YOLO is pretty much the only option left standing. (Although to apply YOLO, you do have to irrationally choose a first transformation, or at least undergo one. Someone without a history of transformation has no legitimate options.)
Would anyone object if I just put a big dump of text in here?
No, huh. Here goes.
I simply do not understand how her conclusion is supposed to either comport with what's been said up to now about rational deliberation or be remotely satisfying.
Everything we've heard about "the normative standard" up to now, as far as I can tell, requires that we assign values and probabilities to outcomes. That seems to have vanished in favor of locutions like the following: "you choose it in order to discover who you'll become" (119, my bolding, her italics), "you are choosing based on the fact that you don't want to create and satisfy the new preferences you'd have" (119; my emphasis); "if you choose rationally, you choose on the basis of whether you want to discover new experiences or preferences or whether you want to forgo such a discovery" (120; my emphasis); "whether you want to discover how your life will unfold" (120). Why can't I just decide based on whether I want to do such-and-such?
A passel of first things: if we're just deciding based on desire like this, it seems that we couldn't be applying the normative standard as it's been consistently described (to the extent that it's been consistently described!); to do that we'd need to be deciding based on the value of discovering who you'll become, or the value of discovering whether and how your preferences will change. And we don't seem to be any better situated to determine what the value of those things might be than we were to determine what the values of any of the other things in question in the book. Paul is surely correct to note that such a discovery might be valuable even though the experience is bad (though I couldn't help thinking of Lichtenberg's remark that "the American who first discovered Columbus made a bad discovery"), but one might just think that that means that the experience, though bad, was valuable, refusing to divorce the value of the experience from the value of the discovery, by divorcing the value of the experience from its pleasantness or unpleasantness. But I think there's a problem about determining the value of the discovery independent of this. She refers at one point to having "the right kind of previous experience with making radically new discoveries" (115), which strikes me as oxymoronical.
Moreover, if it's hunky-dory, rationality-wise, to decide to have a kid because you want to discover whether and how your preferences will evolve, I'm a little unclear, and perhaps this shows that I've really missed the forest here, why you can't decide to have a kid because you want to have a kid. It's true that your desires and preferences regarding having a kid might change as you're having and raising the kid, but your desires and preferences regarding discovering how your desires and preferences change after having a kid might also change after having a kid. (Paul says that "if you choose revelation, you choose it for its own sake", but I'm not sure how that helps, or how it distinguishes it from the choice to have a child, which can also be made for its own sake.) Basically: if the first-order revelation strategy had the problem that "the revelatory experience of becoming a vampire also changes your core personal preferences, making you into a different kind of conscious being" (115), why doesn't the same apply to the higher-order strategy? It might be different if the discovery of whether & how your preferences change were instantaneous, I suppose, but it isn't; it's a process and your future self is certainly entitled to have preferences regarding it, too. (What if you were offered the chance to take a pill that would make you the sort of person who placed a large, or no, premium on transformative, revelatory experiences?) And if you can just say "I want X" in one case, why not another? I am very suspicious of solutions that rely on a higher-order instance of the same kind of thing, because the higher-order whatever is still just a whatever. Harry Frankfurt thought he could get around the supposed lack of authority desires had by appealing to desires to have desires, but guess what, desire to have a desire is still just a desire; it isn't a new kind of thing.
Previously Paul had talked about the value of revelation rather than your desire to undergo revelation and I was concerned because I thought that such a value was too nebulous and ill-defined, not to mention suspiciously factitious-seeming, to serve its purpose as a component of rational deliberation. It still seems factitious in its new role as the indirection of an indirection, and it seems to me that anyone deciding in this fashion would be indistinguishable from someone deciding according to the rule "do what you want, then tell a story about wanting to experience the consequences of doing what you want".
In this particular case the desire to know how your preferences change in response to having a child just seems like an unusually self-regarding way to desire to have a child. Appeal to "revelation" strikes me as bluster. Especially given that both choices can be made out as revelatory, not only when deciding between being a violinist or being a doctor, but even when deciding to have or not have a child. I can decide to experience the revelation of having a child, or the revelation of growing old without having had a child. (The way Paul frames the choice as revelation-with-child vs. reject-revelation-and-keep-on-with-what's-known-without-child strikes me as neither value-neutral nor particularly accurate as regards the second option. Doctors (still!) occasionally refuse to perform tubal ligations on women below a certain age because preferences and desires change also without children; I don't know how I'll feel about having no children in the future and one hears of men hitting forty or so and suddenly realizing that they want kids, etc.) Since Paul seems to think that we can just reinterpret the decision that faces us so as to make it tractable to one particular decision-theoretical apparatus, why can't I reinterpret it this way, where I still do choose revelation? (Doesn't that sound bold?) Can I reinterpret it any way?
The appeal to authenticity strikes me as both unexpected and underdeveloped, and weird given that much of the rhetoric in this chapter seems to embrace bad faith as an ideal. The idea that one chooses so as to discover who you'll become is bad faith in a nutshell. The idea that one might "consciously adopt the preference to evolve your preferences in whatever way they will evolve when you have a child" strikes me as borderline incoherent; if they will evolve that way anyway, then what does it mean to say that you will evolve them? Unless, of course, this is a roundabout way of saying that one is choosing to have a child. (As long as I'm talking about something Sartrean, I'll add that I'd have appreciated an account of the way in which a choice between "choosing authentically" and "choosing rationally" is Sartrean, since I haven't actually read Sartre in a long time. Even a footnote.)
(Also, when we read that "what you care about, when making a decision … is not the third personal descriptive knowledge … but your experience-based first personal knowledge …" (109), in addition to (again) registering skepticism that we care about that, we might recall that Paul has earlier explicitly said that she is not interested in how people actually deliberate but in working with a standard for deliberation. That means it is always open to her to say "you do care about this, but you shouldn't", which in turn means that when she notes putative features of our actual practice and just rolls with them, we ought to ask why.)
Who cares about (Paul's version of) rationality, anyway? Paul thinks "we need to preserve some version of decision theory" (112). Why? She repeatedly claims that we care about being rational and deciding rationally, but that doesn't mean we care about what she thinks being rational comes to. We care about "being rational", but that's a contested term; Blackburn in that same piacularity article had a nice formula for it: "the words 'rational' or 'irrational' are so often placeholders, doing little more than pointing us, peremptorily but obscurely, in the direction of some supposed dimension of virtue or fault that out conduct may have exhibited without actually telling us what that dimension might be" (226). Comes now Paul to tell us what the dimension is; it's open to us to respond: "oh, well, I don't care about that". And I think we mostly don't care about making decisions according to such a calculative view of rationality. I think Kolnai (in "Deliberation is of Ends") more accurately captured what we actually care about when he more vaguely claimed that "it pertains to our constitution to deliberate with a view to arriving, not just 'to a decision' but to 'a wise' or 'the right' decision. I do not mean 'a wise decision or none' but to a decision which shall be a wise decision. … 'To deliberate' is to confront, compare, forecast and argue in view of a happy choice." I had previously tended to think that Josh was being silly, or confused, in thinking Paul owed us an exploration of the option "how about just give up on rational deliberation", but I do think that, in fact, some explanation of why just this particular vision of rational thought is the one we need would have been nice. Especially since, if we are in a dilemma whose branches are "choose authentically" and "choose rationally", it would seem that there's another method of choosing ready to hand! (It seemed to me right up until I read that we have a dilemma that the problem with "choose authentically" is that it can't be done rationally, but if we really have two options, I think I missed why the first one is ruled out.)
I just opened a couple of windows on this bus, it was transformative and I experienced no hesitation! It's hot and stinky in here and several people have on their morning breath. Yuck. This book sounds .... not so compelling.
28: Well, I think it's possible that we really are all missing the point, given that she is a brilliant and respected philosopher (and friend of the blog once removed, etc., which makes me feel bad about being so critical). And it's not really surprising that a book by an academic philospher reflects the preoccupations and methods of academic philosophy, or that non-specialists don't share those (although Nosflow also didn't seem a big fan).
If we're wondering about why, e.g., U-M's work got so much less attention, my guess there connects somewhat (amusingly enough) to Kieran's work on hierarchy and citations within academic philosophy. Paul is an acknowledged star from a higher-status subfield (metaphysics), with mastery of higher-status, formal tools, taking the preoccupations and problematics of the higher status subfields (a requirement that candidate decision-theoretic procedures be fully general, the blind neuroscientist thought experiment as saying something important about knowledge) and applying them to questions of practical reasoning (lower-status subfield) that everyone has an opinion about and cares about (children! vampires!). So now you have someone very respectable writing about something very relevant, which is an invitation for other heavyweight MME-subfield types to weigh in. I think that might certainly explain why the book (& article) have provoked high-level scholarly attention in a way that U-M's various articles on the subject never quite did; my (very impressionistic) sense is that U-M was more in the political/social theory & applied ethics citation networks, rather than being a core part of metaphysics/mind/epistemology "core" philosophy. The two of U-M's papers on this general topic that did get a bunch of citations, interestingly enough, are both with (more famous, male) coauthors: Sidney Morgenbesser & Cass Sunstein, respectively.
I will say, though, that if "YOLO Decision Theory" actually became a term of art, that would be pretty awesome.
And it's not really surprising that a book by an academic philospher reflects the preoccupations and methods of academic philosophy, or that non-specialists don't share those (although Nosflow also didn't seem a big fan).
Well, actually, tbh this book does not feel to me as if it reflects the methods of academic philosophy to which I am accustomed and seems as if it can't decide whether it's written for an academic or a lay audience. It is also true that the overlap between things Paul specializes in and things I know about is probably very, very narrow, though one of the things I have found very strange about this book is the way it hardly addresses any of what I would consider the obviously relevant extant literature (with the caveat that I know and care little about the literature relating to decision theory). For instance, the U-M paper you mention.
Neb's comment is half as long as we are allowed for the main text portion of most journal articles we submit.
Science has cut my attention span, is what I'm saying.
Right, I keep wondering if an editor hacked the book up to help endear it to a lay audience. Specifically the bits about 'deciding authentically' and spending so much time on examples that didn't seem to go anywhere.
I am very suspicious of solutions that rely on a higher-order instance of the same kind of thing, because the higher-order whatever is still just a whatever. Harry Frankfurt thought he could get around the supposed lack of authority desires had by appealing to desires to have desires, but guess what, desire to have a desire is still just a desire; it isn't a new kind of thing.
My only-semi-thought-through view here is that higher-order preferences/values function like so: you keep climbing the ladder of higher-orderedness until you reach a point at which you feel comfortable declaring these are the ones that matter, that overrule those others down there, and are fundamental enough to need no more justification. How are you supposed to make that judgment rationally? Eh. Just a kind of all-things-considered sense of coherence, I suppose, and/or some sort of existentialist commitment. Why expect much more than that from rationality, anyway?
42: I don't think that helps in the present case. Your future self might differ from you anywhere in the hierarchy.
OK, I just read the (short!) U-M paper linked in 17 and it seems like such a vastly more clear, sensible, interesting, useful, and realistic treatment of these issues that now I'm twice as sour as I already was on Paul's book.
45: Sorry!
44: I think present-self's values determine the hierarchy, because present-self is deciding, and it's the decision whose rationality we're judging here. Future-self's values only matter insofar as present-self's higher order values say you should become very-different-future-self. Just as it's present-self's values that enable one to, quite rationally, sacrifice future-self entirely for some greater good.
Especially since, if we are in a dilemma whose branches are "choose authentically" and "choose rationally", it would seem that there's another method of choosing ready to hand! (It seemed to me right up until I read that we have a dilemma that the problem with "choose authentically" is that it can't be done rationally, but if we really have two options, I think I missed why the first one is ruled out.)
This was one of my quibbles that I forgot to put in the post -- Paul calls it a dilemma in the passage I quote, but I don't think she's describing the problems she's set up accurately. She's not talking about a choice between authenticity and rationality, she's claiming that any satisfactory choice must be both authentic and rational, but that there's a problem when you don't have the first-hand knowledge necessary to support authenticity. But I don't think she's set up any option where you could possibly make choices that were authentic but weren't rational.
(And then she solves her problem by saying that you do have the firsthand information necessary to choose or reject transformations on the YOLO basis, so that's both authentic and rational.)
In that case, though, you don't really need to invoke a hierarchy to solve any problems. Just say that the present self's desires count. Are they hierarchically organized? Great, let the consequences of that play out as they will. But the theorist doesn't need to care.
45: Yeah, I should really have emphasized that the paper's only 16 pages. Read it! Oh, except I fucked up the link. Here.
47: yeah I wrote in the margin that I didn't actually see how we had a dilemma based on what had come before, but oh well.
I would like to know more about how she thinks one can rationally choose death.
Given Paul's focus of epistemically transformative experience, any desire to have something which must be obtained through an experience that changes preferences is disqualified.
I said earlier this seemed tied to causality, maybe we draw that out. Paul treats transformative experience as a kind of short hop time machine, with mental side effects. Take the extreme case: the time machine takes you into the future and is guaranteed to make doughnuts your favourite food. You hate doughnuts now, and so choosing to eat them later is irrational for present you.
However choosing to switch preference - focusing on the transformation - is ok, because your preference is not at odds with your future self.
It suddenly seems like a very computational argument about pointer arithmetic and off by one errors.
You hate doughnuts now, and so choosing to eat them later is irrational for present you.
But, of course, the choice you're making isn't "eat doughnuts later or no", it's "eat doughnuts later and enjoy it or no".
Anyone on the fence on the doughnut question should read this: http://dish.allrecipes.com/meet-the-worlds-most-amazing-doughnut/
Something related, that I'm having trouble dealing with in Paul's terms, is thinking about the temporal extension of transformation.
At least some transformations are instantaneous -- Mary, either ordinary or vision scientist, becomes a person who has seen red the moment she steps from her locked room and sees red. Becoming a vampire is the same, I guess (depends on the details, but you are or you aren't). For that kind of instantaneous transformation, I can read the YOLO decision theory as about what the expected value is of the experience of making the leap -- not seeing red, but opening the door knowing it's out there; not being a vampire, but feeling the teeth touch your neck and knowing that in the next moment you will no longer be human. And that's the last moment, temporally, that has any expected value, because all future moments are experiences of someone with different preferences that your current self can't incorporate into your decisionmaking process, so your whole decision is based on that transformational moment. If you value 'fangs on the neck' enough you change, if not, you stay the same.
But real transformations are usually more gradual. There's not a 'fangs on your neck' moment when becoming a parent changes you into a different person. You look up a few months or years into the process and you've changed, but spotting the last moment when you were same kind of person you were before kids isn't practical. So under the YOLO decision theory, what exactly is the transformational experience that you're assigning an expected value to, for a process of gradual change?
Probably there was an exact moment of change but you're just too sleep deprived to notice.
One day, a few years ago now, I woke up and felt strange all day. I couldn't figure out what the feeling was. Then I realized it was simply being well rested. I'd forgotten what it felt like.
57: Wow! Does that mean you transformed back?
How would I know if I had, epidemiologically speaking?
Of course the other thing that makes this very program-y - object-oriented even - is the importance of encapsulation. Our future subjective preferences are private fields on an object of another type.