Are agent-action-dependent beliefs underdetermined by external reality?
6the gears to ascension
3Said Achmiz
2Said Achmiz
6the gears to ascension
2Said Achmiz
5papetoast
4Said Achmiz
4Vladimir_Nesov
2Said Achmiz
4Vladimir_Nesov
2Said Achmiz
4Garrett Baker
4Said Achmiz
0TAG
2Said Achmiz
2TAG
New Comment
https://www.lesswrong.com/posts/8dbimB7EJXuYxmteW/fixdt
it doesn't seem particularly "post-"rational to think about the potentially self-referential structure of action-dependent beliefs
it doesn’t seem particularly “post-”rational to think about the potentially self-referential structure of action-dependent beliefs
That’s as may be. I’m responding to a class of arguments that I’ve encountered exclusively in that sort of context.
Anyhow, thanks for the link; I’ll take a look at it and think about whether it’s relevant to what I’m saying here.
Ok, I’ve now read the linked post.
As far as I can tell, the account of decision-dependent beliefs described in that post is entirely compatible with what I say here.
(The account of “belief-dependent beliefs”, if you will, is a different matter; but I make no claims about that, in this post. Also, I think that the notion of “world reacts to agent’s beliefs”, as described there and elsewhere, is confused in an important way, but that’s a discussion for another time.)
On the whole, I must admit that I’m slightly confused about what you were getting at, with that link.
hmm, I did read you to be making claims about that, which is why I linked it. in particular:
As I’ve shown, there is little material difference between a belief that’s “about the future” and one that’s “about a part of the present concerning which we have insufficient information”
There seems to me to be a legitimate difference in processes which implement decision theories that choose actions by choosing belief-dependent beliefs, in that which belief is true happens computationally after the decision of which belief to hold. in some logics this is equivalent, but some of your emphasized statements in this post don't seem as obviously justified to me as you assert them to be. But if that doesn't justify the connection to you, then I don't think I'm interpreting my intuitions into precise claims correctly, and so we won't be able to do further evaluation of correctness of the intuitions I get from a difference I see between your post and that one, at least in this thread or until I can be more specific.
Meta note - it's possible I was irrationally reacting to your use of emphasis and asserting-before-justifying, in eg, the sentence
Of course any philosopher worth his salt will find much to quarrel with, in that highly questionable account of decision-making
"of course" and "highly questionable" don't hold in their syntax a statement of first person perspective, which my brain automatically translates as a peer-pressure-backed request to update before processing the rest of the claims. I generally prefer to avoid such claims and try to parameterize first-person-ness whereever I can. This is not material to your point, but seems to have affected some features in how I responded in ways that were avoidable by either of us, and is an example of the sort of thing I think is an unnecessary overhead in iterative-disagreement-based communication.
But if that doesn’t justify the connection to you, then I don’t think I’m interpreting my intuitions into precise claims correctly, and so we won’t be able to do further evaluation of correctness of the intuitions I get from a difference I see between your post and that one, at least in this thread or until I can be more specific.
Alright. If you have further thoughts on the matter in the future, I’ll certainly be interested in reading them.
“of course” and “highly questionable” don’t hold in their syntax a statement of first person perspective
Correct, it is not a first person perspective, but rather a reference to an ongoing debate in philosophy (of course, you can guess which side of it I come down on…). This post originally having been a comment in a thread that mostly wasn’t about this specific point, I didn’t add any links to point a reader to references for this; I agree that this is a lacuna, and I’ll see about adding a clarifying link or two.
However, please note that the rest of the post does not depend on this point! Take a closer look at the paragraph beginning with “Let us consider again the belief”; as you see, the logic that I outline in the last part of the post is agnostic about whether beliefs are prior to decisions, or vice versa.
(You’re at least the second reader to find this unclear, which suggests that it could use a bit of an edit…)
(EDIT: Edited.)
a … request to update before processing the rest of the claims
asserting-before-justifying
Please see this old comment for my stance on this.
The title contains a typo (undeterdetermined -> underdetermined)
[This comment is no longer endorsed by its author]
Embedded agency and beliefs also lead to some kind of embedded truth values, with all the strangeness of depending on possible values of deterministic things. Consider some complicated computation F that would compute a particular value F() such as 2 or 5, but we don't yet know which one. And consider a different computation G=F+10. Does the value G() of G depend on F()? Well, F() is some particular number, what does it mean to depend on it? And also we don't know the number. But there is still some sense in which G() depends on F(). Even though G() is also just some particular number. It doesn't make much sense to claim that 15 depends on 5, outside the context of the computations we are talking about.
Now, we can also consider truth value of a claim that G()>14. Does the truth value of that claim depend on F()? It seems that it does in some sense, in the same way that G() depends on F(), that is the claim G()>14 is true iff the claim F()>4 is true, and the truth value of F()>4 depends on the value of F(). Even though F() evaluates to some particular number such as 5, bringing about a different sense in which G() (which is just 15) or G()>14 (which is just true) don't depend on F().
Uh… this is interesting, but I’m not sure what it has to do with the post? Could you please spell out the connection a bit? What am I missing?
The complicated computation F could be some person making a decision F(), and the complicated computation G could be defining an outcome of that decision we are considering, so that G()>14 is a claim about that outcome with some truth value. If everything is deterministic, it might still make sense to say that G() depends on F(), and even that the truth of G()>14 depends on F(). And also that it's F that determines F(), and therefore that it's F that determines the truth value of G()>14.
(I think there is some equivocation about beliefs vs. decisions in the post, but it doesn't seem essential to the core puzzle it's bringing up. A decision is distinct from a belief about that decision, and if you are making decisions because of beliefs about those decisions, you run into Löbian traps, so it's not a good way of thinking about the role of beliefs about decisions.)
Hmm… well, I certainly agree that a decision is distinct from a belief about that decision (indeed, I explicitly argue against the opposing view), and that making decisions because of beliefs about those decisions is nonsensical.
I’m not sure about the rest of it. If you are making the assumption that everything is deterministic, it seems like that just gets you to @Shmi-style “decisions are not about changing the world, they are about learning what world you live in”, and (as I say in the post) that fully dissolves the “puzzle”.
For example, “beliefs are prior to decisions” is necessary in order for there to be any circularity, and yet it is, at best, a supremely dubious axiom.
This doesn’t seem that strange an assumption to hold? Pretty much all decisions I make happen because I believe something about the world. It’s hard to imagine what I would do if I believed nothing about the world, which seems the only way to make my decisions independent of my beliefs. I expect it would look like a seizure or coma.
It’s hard to imagine what I would do if I believed nothing about the world, which seems the only way to make my decisions independent of my beliefs. I expect it would look like a seizure or coma.
No, that’s not the only alternative, and certainly not anything like what I have in mind here. You seem to have understood that part of the post as saying something very odd—like perhaps that I was objecting to the notion that one must have any kind of beliefs at all, about anything, before one can make any kind of decisions at all? But of course nothing remotely like that is what I meant.
(As an aside before I continue: please note that the thesis of the post does not rest on this point. As I say in the post, we can grant that beliefs are prior to decisions, and yet the logic I describe still goes through.)
No, what I object to is “beliefs are prior to decisions” as an axiom, i.e. “beliefs are always prior to decisions” (in other words, the idea that beliefs cannot be a consequence of decisions). Put that way, the problem should be obvious: surely my belief that I will go to the beach this evening is an effect of my decision to go to the beach this evening, and not a cause of it! The reason why I believe that I will go to the beach this evening is that I’ve decided to do so.
This is nothing more than the perfectly ordinary, intuitive account of decision-making. I am not claiming anything weird here. On the contrary, I am saying that the converse—the idea that one first comes to believe that one will do something, and only then (and as a consequence) decides to do that thing (or perhaps that coming to believe that one will take some future action, and deciding to take that action, are actually one and the same)—is what’s unintuitive and unnatural. But that strange view is precisely the view that one must hold, to think that there’s some sort of circularity involved in the view that the truth of a belief like “I will go to the beach this evening” can be evaluated via the Tarski criterion. On the normal and intuitive view, there is no circularity at all. That’s all that I’m saying in that part.
What is the difference between this, and the quote above? Is merely the fact that “I will go to the beach this evening” is about the future,
No, it's also the fact that the future is taken to be unfixed. Future facts are knowable for a Laplace's Demon in a determined universe.
(Note that events that are undetermined because they depend on an agentive decision that hasn't been made yet are only a subset of undetermined events. You would also have difficulty fixing a belief about a random nuclear decay in the future)
And even free-will concerns evaporate if we adopt the perspective that decisions are not about changing the world, they are about learning what world you live i
Which is to say free will concerns "evaporate" if you assume determinism. This is not rationality. Rationality means believing the world is deterministic iff the world is deterministic ,believing it on the basis of a evidence...not assuming it to make problems go away.
Rationalists want as much as possible to be knowable, but believing things to be true because you want them to be is the essence of irrationality . Rationalists have to recognise that the nature of the world can constrain what is knowable.
(Indeterminism isn't the only problem area. There is also the indirectness of perception, which allows for simulation and other sceptical hypotheses -- you can't tell what is at the far end of a chain of a perceptual chain from the near end. And there is also the problem that a haphazardly evolved brain doesn't have any apriori guarantee to be able to.understand anything. And, quite possibly, an is-ought gap that prevents values being fixed by facts. ).
1And even free-will concerns evaporate if we adopt the perspective that decisions are not about changing the world, they are about learning what world you live i
Which is to say free will concerns “evaporate” if you assume determinism. This is not rationality. Rationality means believing the world is deterministic iff the world is deterministic ,believing it on the basis of a evidence...not assuming it to make problems go away.
The point of the quoted line is not that we assume determinism in order to make free-will concerns go away—as you say, that would be obviously irrational.
Rather, the point is that there are other reasons to assume determinism. (Those reasons are described in the linked post, which is why I linked it.) If, for those other reasons, we do indeed adopt such a deterministic perspective—which, again, there are (or so the linked post claims!) good reasons to do—then we will find that free-will concerns have also evaporated.
(If those other reasons do not convince you, and you think that we shouldn’t adopt determinism, that is fine. You will note that this is not in any way a load-bearing assumption of the post’s argument.)
The linked post gives one reason , which is debatable, since it's been debated. You don't seem to have complete confidence in it yourself.
Curated and popular this week