but if an unconscious superintelligence a billion light years away was asked to guess whether any entities had the property of there being something it would be like to be them (whatever that even means to the unconscious intelligence) there's a 0% chance it would say yes,
I'm not sure if you mean this literally, but there's no way this is true. A superintelligence that had any interest in possible aliens would think a lot about what sorts of evolved minds are out there. It would see how and why this was a property an evolved mind might conceptualize and fixate on, and that such a mind would be likely to judge itself as having this property (and even that this would feel mysterious and important). This just isn't the sort of thing a recursively self-improved superintelligence would miss if it was actually trying!
To clarify I'm talking about why consciousness is a possible thing to exist, in this universe or any. I'm not talking about a reason why if things can be conscious, they would evolve to be so, or regardless of whether consciousness can exist, they would evolve to act in a way similar to conscious beings.
Granted that given that this universe does have consciousness in it, hence there must be some explanation for consciousness, the superintelligence probably would predict it.
But I'm saying that none of the explanations for why consciousness is a possible thing to exist, feel like the sort of thing that would be convincing to an entity that has no idea it exists in the first place. Nor can I even imagine what sort of argument would be convincing (other than showing them that conscious beings do in fact exist).
Using the framework in this post, I suspect you’re in Camp #1 about consciousness i.e. you believe consciousness can be reduced to a functional/physical description of a system in a deflationary way.
Those in Camp #2 believe there’s some additional metaphysical property that constitutes “what it’s like” to be an entity. In this case, I don’t think there’s any obvious reason an unconscious superintelligence would postulate this property in alien minds. The superintelligence itself wouldn’t possess the property (as it’s unconscious by stipulation) so even if it discovered the functional/physical markers which give rise to “conscious utterances” it wouldn’t discover the phenomenal property that constitutes “what it’s like.” This is especially true if it never actually observes or interacts with conscious creatures.
I think the OP is pretty clearly in camp #2 which might be why you’re so surprised at the claim if you’re in camp #1.
I think of myself as in camp 2 — I believe there is a fundamental sense of experience which is metaphysically independent of the physical description, I just don't think it's very mysterious.
Regardless of which camp is right or what the right metaphysical property is, I claim that a superintelligence would be able to deduce that such aliens would have the camp 2 intuitions, and that they would postulate certain metaphysical properties which it could accurately describe in broad terms (it might believe it's all nonsense, but if it is true, then it would be able to see the local validity of it).
Being a superintelligence thinking about something is almost as good as actually observing and interacting with something when it comes to the broad shape of things.
I think of myself as in camp 2
Thanks for the clarification. I'm surprised by this, and I think it presents a problem for your view in the 0P/1P logic post you linked.
If I understand your view correctly, you want to say:
As far as I can tell, this is the Phenomenal Concept Strategy i.e. from purely 3rd person facts about a cognitive architecture, we're supposed to see that the system will form these primitive "experience" concepts, treat them as metaphysically independent and form hard-problem intuitions. My worry is that once you pair this with Camp #2 style phenomenal realism it's vulnerable to Chalmers' dilemma. Roughly, you have two horns to choose from
Horn 1: If phenomenal concepts are defined too strongly then a zombie won't have those concepts, in which case, these concepts are part of the hard problem and require an explanation.
Horn 2: If phenomenal concepts are defined too weakly then zombies can have them. In this case, they don't explain the Hard Problem of consciousness (since zombies don't have phenomenal experience).
To put this in the language of your 0P/1P post, if "possessing phenomenal concepts" just boils down to being the kind of system that needs 1P logic to reason effectively about its internal states then zombies would possess them and zombies would be conscious which is a contradiction.
You could get around this by saying zombies are not conceivable (because any entity capable of deploying 1P logic is conscious) but this would kind've defeat the purpose of our thought-experiment as the superintelligence wouldn't be unconscious if it can accurately reason about 1P logic.
That sounds about right. I simply disagree with Chalmers' dilemma (at least as you describe it).
In my view, this metaphysical fact is necessary but not sufficient for explaining the Hard Problem. It applies to "zombies" in a fairly trivial way. A phenomenal experience is a type of experience (in my 1P sense), and must be understood in this frame — but not all such experiences are phenomenal. I don't claim to know what exactly makes an experience phenomenal, but I'm pretty sure it will be something with non-trivial structure, and that this structure will sync-up in a predictable way with the 0P explanation of consciousness.
If you’re permitting a difference between 1P functional concepts and 1P phenomenal concepts then I’m happy to grant that an unconscious superintelligence would possess all the functional 1P resources and notice a kind of “functional analogue” of the hard problem intuitions due to the conceptual isolation of 0P/1P.
I’d push back if you’re making the stronger claim that the unconscious superintelligence would be able to fully grasp the actual hard problem of consciousness in anything like the sense that we do when we appeal to our 1P phenomenal concepts. By stipulation, it doesn’t possess 1P phenomenal concepts so it could never really “grok” the hard problem in the same way that we do. If it doesn’t possess the concepts I don’t see why it would be motivated to think evolved alien minds have genuinely additional metaphysical properties rather than just a certain kind of sophisticated self-model that lets them talk the way they do.
I don't claim to know what exactly makes an experience phenomenal, but I'm pretty sure it will be something with non-trivial structure, and that this structure will sync-up in a predictable way with the 0P explanation of consciousness.
I’m not 100% sure if I’m interpreting this correctly. If the claim is that an ideal 0P observer would, in principle, be able to tell which concepts were 1P phenomenal for a given entity purely from 0P information and absent any of its own 1P phenomenal data points then I disagree and this is a crux for me.
Alright.
I am making the stronger claim. I claim it could in-principle simulate us deeply enough to pull out the 1P phenomenal concepts, and could self-modify so as to legitimately experience them if it so chooses. It would be motivated to think this through carefully because it's a huge part of our values (at least as we understand them), as long as it was interested enough to try to understand us (including as a special case of generic aliens) as agents at all.
I don't believe there's anything metaphysically "magical" going on such that it couldn't or wouldn't see this. Probably why I feel camp 1-ish.
As for the last point, my point of view is that any agent has a "bridge prior" which allows them to connect their 0P models with their 1P model. So I claim that in a sort of trivial way... it will have some prior here, and whatever the bridges spit out will inform what it deduces about the 1P experiences at play. I additionally claim that simple bridge priors will be adequate for finding 1P phenomenalism, and that you would have to have a pretty unnatural one in order to avoid seeing this.
I claim it could in-principle simulate us deeply enough to pull out the 1P phenomenal concepts
I don’t think this is right. A simulation (even an extremely detailed one) is ultimately only telling you about relational/dispositional facts i.e. given a physical state P it will evolve to state P’. It doesn’t say anything about the associated phenomenal state Q.
I additionally claim that simple bridge priors will be adequate for finding 1P phenomenalism, and that you would have to have a pretty unnatural one in order to avoid seeing this.
Ok I think this is the heart of the matter. I read the OP’s original point (“an unconscious super-intelligence would not guess that alien minds are conscious”) as essentially saying the most natural bridge prior for an unconscious system to posit is a null one i.e. that no real bridge exists.
Why think an unconscious system would be motivated to posit a bridge prior in which phenomenal properties actually exist? A prior that connects functional states to real phenomenal states is more complex than a prior in which phenomenal states are not real properties. The only reason to introduce the more complex prior is if the system had access to data points which require it as an explanation I.e. if it had access to phenomenal states Q.
If the bridge exists like I think it does, then it does say something about Q. But yeah, the bridge is necessary for this to work.
Under my framework, the bridge is necessary for connecting indexical (1P) statements with classical logical (0P) ones. Any system that is actually instantiated somewhere and has sensors will metaphysically have both of these (as in, once you fully specify everything that 'actually instantiated somewhere' means, you will have to have posited both of these for it), and a null bridge does not explain its own sensory data.
So I see the unconscious system as already having all of this inherently, and that the phenomenal structure is perhaps foreign, but understandable within the 1P side the same way a cell is understandable to it on the 0P side. Even if it does not itself experience the 1P things here as true, it will be able to use the bridges to understand the perspective of another agent that does have the phenomenal structure.
This is a cool position. Thanks for taking the time to explain it in so much detail.
I think I can see where we’re diverging. You want to place the metaphysical bridge between the 0P -> 1P perspective and because there’s something metaphysically substantial happening in this bridge it’s a camp-2 position. But within the 1P side you’re treating 1P phenomenal states as a subset of the full 1P space and what links the phenomenal states to the other 1P states is some kind of structural relation. This idea is very camp 1-ish to me, the extra work done is being done by structural/relational links between phenomenal and non-phenomenal states in 1P.
By contrast, I want to place the metaphysical bridge between the physical and phenomenal states P -> Q. This means I’m rejecting the claim that Q is related to P by purely structural relations or dispositional properties. This is also why I said the most parsimonious bridge was a null one unless you had access to Q. I agree the null bridge is incoherent if what you’re talking about is the 0P -> 1P link that an instantiated agent needs to access is sensors. But that’s not the bridge the unconscious superintelligence needs. It needs a bridge to Q and since it doesn’t possess Q it could coherently postulate a null bridge between P and Q. From its perspective the null bridge would also be most parsimonious if it truly didn’t possess Q.
I’ve kind’ve sketched my reasons for thinking structural and dispositional properties don’t yield Q in the rest of the thread but I’ll throw in one more: from my perspective the structural/dispositional properties are inherently 0P there’s no 1P categorical/intrinsic properties which describe “what it’s like”
So when you say:
the phenomenal structure is perhaps foreign, but understandable within the 1P side the same way a cell is understandable to it on the 0P side
I reject the analogy. On my view, your link between the non-phenomenal 1P and phenomenal 1P is still structural/relational and these properties are always 0P in nature.
This might be a natural point to end the conversation as I think we’re at a point where our intuitions lie on opposite sides of a pretty large crux. But I’m happy to continue if you think there’s another angle I’m missing.
Thanks for pushing me to describe it better! This has been a lovely discussion.
I agree there is something very camp 1-ish about the idea (and just me as a person, frankly).
So your Q is not even a type of 1P thing, is that right? I'm not sure what sort of thing your Q is supposed to be, which I suppose is what my side of the crux looks like. (I kind of suspect that if you are right about Q, then I do not have access to it myself.)
I also (regardless of my other points and arguments) think you are wrong that structural/relational properties are always 0P! I think 0P can't actually even have a proposition like "I always see blue right after I see red", which still needs to use indexicals in order to refer. There's a similar seeming "Environment X has a red-blue light sequence" on the 0P side which is not actually the same (e.g. what if I'm not actually in that environment?).
To me, "what it's like" grounds to something like: an experience that there is something I observe which has its own 1P experiences (and a prediction of what those might be based on my observations). Phenomenal consciousness is then maybe something like: the observation that there is an observable entity 'self' such that 'what-it's-like_self(to see red)' implies 'to see red'. And this sort of fixed-point thing is inherently really weird and slippery just from a pure math point of view, e.g. Löb's theorem (imagine 'what-it's-like' as the box), which has the infamous Gödel's 2nd Incompleteness theorem as a special case. And all of this is inherent to the 1P side; only on the 0P side can you just reduce things to neurons or atoms or whatever (though I claim a simple bridge would still reveal the 1P structure just from the 0P side). This formulation is speculative and off-the-cuff, and only intended to gesture at the sort of structure I think is possible here.
And happy to leave the discussion here if you're done, but I am curious to know what you think of this idea.
Let me explain my view in a little more detail - it’s worth noting that I hold it pretty tentatively (around ~p(60%)) but I think you’ll find it appealing and hopefully see where it parts way with your view.
If you hold a blue image in your mind there’ll be something it’s like for you to experience blue. Call that Q. Now if you hold a red experience in your mind there’ll be a corresponding red experience that’s different to the blue one. Call it Q’. Hopefully you know what I’m talking about here! Some people with very strong camp 1 intuitions aren’t even willing to grant this, but I feel like if we’ve gotten this far in the thread we have some common ground here.
On my view, there is a fact of the matter about what blue and red look like for you and this is underdetermined by the physical/dispositional properties. The physical/dispositional properties could be held constant and these blue/red experiences could vary in principle. Granted, there’s a lot of structural constraints e.g. light cones in your retina, reflectance of surfaces, wiring of your brain etc.. But I claim that even if physics were fully fixed some aspect of your experience could vary in principle.
More precisely, a complete description of physics would tell you everything about the dispositional/relational properties of physical particles. Specifically, given a state P it will tell you how it evolves to P’. An example is an electron with charge q and mass m moving in an electric and gravitational field. The physics fully specifies the dispositional properties of the particles e.g. the electron will move in such-and-such a way. But this doesn’t tell you about any of their essential properties. If you switched the mass with something that played the same role but was intrinsically different (call is schmass) would that change anything? On standard physics, it wouldn’t matter what was playing the mass-role itself only that the structural form of the equations are intact.
On my view however, it does matter. The particles have an additional categorical/essential property that fixes something about the world. Importantly, these properties are physical in some sense (they’re all part of the same “stuff” that physicists talk about) but they’re not captured by the normal relational/dispositional properties of physics. This view is called Russellian Monism.
So with this formalism in place it actually connects up quite nicely with the other components of your view. The 0P/1P framework gives a nice overview of the difference between describing the disposition of the state (0P) and tokening the categorical essence of the state (1P). The hard problem intuitions just fall straight out of the difference between 0P/1P. Also on this view there are no zombies, since duplicating the physical particles necessarily duplicates their categorical properties — so there’s no gap between what I’ve been calling functional 1P and phenomenal 1P. As soon as experiences enter 1P they’re phenomenal.
Where this differs from your view is I think you need a categorical property to fix the phenomenal character of certain states. Whereas on your view it seems like you’re using the bridging law + structure to fix phenomenal character. In my previous comments I’m mostly pressing you about how much work structure is doing in your framework. If you’re happy for bridging laws to provide the jump then our views actually become really close.
Your point about Löb’s theorem is interesting and it seems like it could be a nice formalisation of the 0P/1P idea. I’d just emphasise that it’s still a structural argument for why 1P/phenomenal talk is really tricky - it doesn’t give you a metaphysical explanation for why 1P has a “what it’s likeness” in the first place. For this you need the bridging laws or the categorical properties.
Yes, that sounds right (minus the word metaphysical in camp 2).
To be precise: If you were to explain why, based on the laws of physics, I say the words "I Am Conscious" and otherwise act the way I do, I would still not feel like the mystery of consciousness has been explained, because there still doesn't seem to be any reason why there is something experiencing saying those words.
If you believe that an exhaustive explanation based on the laws of physics would still leave the mystery of consciousness unexplained then I'd urge you not to drop the word metaphysical from the description of camp #2.
Metaphysical doesn't necessarily need to mean "spooky" or "non-naturalist". There are plenty of coherent naturalist camp #2 formulations of phenomenal consciousness e.g. Russellian monism and property dualism. These both require additional metaphysical commitments to try to explain consciousness but they fit coherently within a naturalist viewpoint.
I'm not claiming that we need any extra laws of physics to explain consciousness. I'm saying that even if you showed me the equations that proved I would behave like a conscious being, I still wouldn't feel like the problem was solved satisfactorily, until you explained why that would also make me feel like a conscious being.
In my view: Qualia are a type of mathematical objects which depends only from themselves. This explains first two questions.
1 Only qualia exist which depends only from themselves.
2 Only consciousness content is real and the universe outside is just a way to describe how one qualia is connected with another (This view was suggested by Ernst Mach - I will make a post about soon).
I share your confusion about the first two, but I believe that the question about the subjective passage of time can be dissolved, unless I am misunderstanding your question.
As you mention, the laws of physics are reversible, but as you mention, entropy gives an arrow of time. That arrow of time causes it to be the case that your brain encodes memories of the past instead of the future. You are not a subjective observer outside of time. You are always experiencing the current moment from the perspective of your brain at that moment. Thus, you always have the experience of being in a brain that remembers the past and remembers the immediately prior moment as the immediate past. So your experience will always be that your current experience has moved forward in time from the past. As a thought experiment, imagine that the 4D block universe existed and that there is some process that evaluates slices at which subjective experience happens. Imagine that instead of that process moving forward in time, it is moving backward in time from the "end" of time to the Big Bang. What would your subjective experience be in that case? It would still be that you are traveling forward in time. Because, as you experienced each moment, you would do so with the memory of the past moment, not of the moment that was experienced sequentially before (which is the temporally future moment). There is therefore no question here except what gives rise to qualia, and perhaps whether the block universe view is correct, or the universe is actually an evolving system for which only some "now" exists for each point in space.
(In reality, I don't think a block universe makes sense. While I don't understand what gives rise to qualia, all evidence says that it is tied to the execution of the "thinking" algorithm of my brain. A block universe would have no "execution" and so I think would have no qualia unless qualia exist eternally at all places in the block universe where there is a conscious being.)
I don't understand what you mean.
The human brain takes time to process sensory signals so that the qualia experienced are slightly delayed from when the sensory input that gave rise to those qualia entered the brain. In that sense, experience happens over time. But at any moment, there is only the qualia that is being experienced. How could it be otherwise? If you say that you then recall that the qualia you had just before that was different. Well, that is a different instant in time in which you are experiencing recall of a memory from your brain.
This brings up an interesting consideration: Imagine two observers existing in the same universe, one being the time reverse of the other. From their internal perspectives, each views its own observations as defining an arrow of time, yet their arrows point in opposite directions. It would seem then, that the arrow of time resides in conscious experience itself, and is not a property of the universe. And yet it would seem to define a "canonical" direction to time, even if the external time parameter were reversed.
So would it even make sense to have an observer that can experience time from the opposite direction? Why can't I experience my death and then age backwards until my birth? It would seem like the key here is memory itself: that is, memory requires observation, which is inherently a time asymmetric process. But what is experience itself but a brain state, ie. an impression, or memory of the world at large? But then are we saying that qualia themselves have an arrow of time impressed upon them? That a universe that is running backwards nonetheless cannot have conscious beings in the sense that they can experience the same "flow" of time?
Imagine that this time reversed observer were only run partially to completion. That is, their internal states were only run until say the middle of their life instead of to their birth. In what sense do they experience their life? Would it not be the case that their life from birth to the middle of their life was undetermined? But what would it be like to have an internal experience that corresponded to this scenario? It would seem like the way out of this confusion would be the fact that experience only relies on the current state and not on any future or past state. That is, all the information necessary to render a conscious experience is contained in the present moment. This would be similar to a Boltzmann brain scenario where the impressed past is not the "actual" past.
And yet there is still something confusing about this picture, namely which of these moments should be the one being experienced as "now"? Do they all exist in some platonic realm? But then what prevents us from thinking that we are living in a perpetual groundhog day? No, it seems like something is missing from this picture. It definitely seems like time is flowing, that I have not yet experienced this exact moment before, yet it would seem like everything in our physical theories cannot rule out the fact that this could be deja vu, because physical theories are memory-less theories. That is, they can be fully determined from one time slice only and not on any future or past state.
But this cannot fully be the case, because the dynamical equations of motion involve first and second derivatives, which would involve at least two more snapshots of time, at least to get the ball rolling to a calculation of the evolution of the state. But to a first approximation, all the information necessary to determine the state of the universe is contained in only one time slice, and this is the key to the question of the problem of time reversal symmetry. For it would be trivial to break this symmetry if the universe depended not just on the present state, but on future or past states as well, giving them further "existence".
But then what would the laws look like if they depended not just on the present moment, but on other moments as well? Well for one, there would be no way to just "run the universe backwards" since this would assume you can just use the current state of the world to run the universe backwards, plus two time slices, to get the ball rolling.
But what does this entail? Well, if the physical laws depended on more than one slice of the past, then the time reversed implication would imply that the reversing the universe would require knowing future states that have not yet occurred.
And therefore if we are to believe that we are "extended objects" in time and not just one slice and are forced to trust our senses that we are not living in groundhog day, it would seem as if the future has yet to be "determined".
Like everyone there are millions of things that I don't know. There are even millions of things I'm certain I'll never know.
But only a few of them feel fundamentally confusing.
A confusing question is one where I can't imagine how an answer could possibly actually answer the question. At a meta level I believe an answer must exist, but looking at the problem itself, it seems impossible.
Now there's two general approaches that have effectively answered such confusing problems in the past.
The first is research:
How life could possibly exist felt equally mysterious 150 years ago, and was ascribed to gods and inexplicable life forces in equal measure. Then we worked it out, and whilst there's still a lot we don't know, it no longer has that mysterious quality.
The same applies to my individual understanding of computers. As a child they seemed like magic, then once I learnt about the details of computer architecture, it made enough sense that I was willing to leave the blanks left unfilled.
The second is dissolution:
When you force yourself to understand what you're actually asking in your question, and reply "and so?" to every reductio ad absurdum, you often find you don't actually have a question anymore. You also often come out significantly more enlightened than you came in.
The classic example is the problem of free will, but you can add lots of others, like morality or decision theory. Jessicata does a great job of dissolving some particularly knotty problems, like anthropics, here.
But even after that some problems seem to resist such approaches, at least for now. Here's a list of problems I still find confusing. YMMV.
There were definitely some more I thought of earlier, but it took me a while to get round to writing this post and now I seem to have forgotten them.
Either way it seems that all my remaining problems seem highly related to consciousness. So at least I've turned 3 big problems into one really big problem 💪.
God we don't have the vocabulary for these questions!