This emulation is not actually related to any phenomenal consciousness in reality, right?
Correct.
1: Yes, I believe the EM fields in the digital computer might have an experience, but the shape of that experience is not strongly correlated with the shape of the digital software. Or perhaps, more crudely: The digital software cannot be a p-zombie because the digital software does not really exist, except insofar as it has a physical manifestation (which would have a phenomenal experience related to the shape of that physical manifestation).
2: I’m not sure I follow but I recommend reading this part of the Lerchner paper and really wrapping your head around how alphabetisation is metaphysical or at least observer-dependent rather than physical.
3: I think that the entire body may have qualia! This doesn’t imply substrate-independence, though. Perhaps if the knee happened to implement an EM field structure with the same shapes/symmetry group as what we experience as colour then your knee could indeed experience red. However, I want the causal structure which is actually linked to talking, so we can correlate external observations with internal ones. In my ideal empirical programme this would involve picking a brain region and perturbing it and correlating this with direct reports. Susan Pockett proposed something like this.
On functionalism: Note that I am arguing against computational functionalism and not functionalism itself – that’s a different matter. Check here, I have been exploring with jessicata whether or not this qualia structuralist account is reconcilable with analytic functionalism. Discussions are ongoing. However, I still think you have to take a fine-grained physics level interpretation of the relevant causal structures, not a coarse-grained computational one, so you still wind up drawing the physicalist rather than computationalist-flavoured interpretations of subjective experience.
Why does the WBE talk about phenomenological consciousness, then, if it doesn’t experience it? That’s the real question I need an answer to… it means that the reason I talk about having phenomenological consciousness is not actually causally downstream of my having phenomenological consciousness, either (since a WBE of me would also talk about having it, despite not having it). That feels like very straightforward p-zombieism, right? If I am my qualia, then… my qualia is not actually able to cause me to talk about having qualia, and yet the p-zombie whose EM fields create me, the body I mistakenly think of as being ‘me’, talks about having qualia anyway? What do you think is going on there?
It seems to me you either have to say that 1) the WBE’s behavior will diverge from the bodily-instantiated counterpart, such that it does not talk about phenomenological consciousness, despite the fact that maxwell’s equations are being faithfully computed and the predictive model of the EM/neuron interactions is 100% accurate
or else 2) p-zombies
1) seems, to me at first glance, like an absurd thing for an intelligent person to believe (in a way that makes me think I must be missing something, not that you aren’t intelligent), but you reject 2). I think I predict that you’ll say 1), but I don’t know what form that answer will come in, and I could be wrong. Maybe there is some third option I’m not thinking of.
But I really do feel like… hm. Like, if the answer is 1), then we could examine the emulation and the real-world brain to figure out the exact moment they diverge, and then just… fix the emulation, so that it accurately modeled reality instead of failing to. And we could just keep iterating on that until we succeeded. And then we’d have (according to you? maybe?) a whole-brain emulation of a human which talked about experiencing phenomenological consciousness for the same reasons that I talk about it, but which did not actually experience it. and then we’re just back at p-zombies.
Yes, he is indeed that Herschel.
edit: I like the WBE formulation of the generalized anti-zombie principle because it’s an actual experiment that could theoretically be performed in real life, it forces the issue down to the level of empiricism (even if hypothetical, armchair empiricism)
it’s the actual point of friction in my mind, when I try to figure out how to adopt your perspective, to see what can be seen from it
I can agree that, as mentioned downthread, there are also contradictions in the computationalist theory of consciousness… but, idk. when I hear non-computationalists try to answer the GAZP, mostly what I hear doesn’t actually convince me that their understanding is better or more complete, instead it convinces me that we started from different priors.
Pointing out contradictions in the computationalist framing is definitely a valid move, I don’t mean to say that it isn’t. I don’t understand these matters and don’t claim to. But… idk. When I try to understand a system, the actual function that my brain performs is to imagine writing a program that models the inner workings of that system. I don’t know if I really have a way to try to understand consciousness that isn’t doing that. This might be a failing on my part, but it’s why the WBE thought experiment causes so much friction for me.
The WBE talks about phenomenological consciousness because it is simulating the kind of system which would do this. If we have a magic system which can perfectly emulate the EM field, then sure, these dynamics won’t diverge.
The WBE does experience phenomenological consciousness, but this is not correlated with the words it’s saying. Given a translation function we are confident in, what the WBE is experiencing is the output of that function as applied to its hardware. Which in the case of the physicalist function I proposed, would be what it’s like to be that hardware, not what it’s like to be the simulation – unless the hardware was designed in such a way that these correlate well with one another. Consider analog hardware in which the spatiotemporal topology of the causal relationships is the same in the WBE as it is in the original brain, for example.
You can be as confident as you like that your qualia is correlated with your actions because, well, you have direct access to this (though you couldn’t prove prove it to me).
With regards to both systems, the only way an independent observer could be truly sure what’s going on is by inspecting the physical structure of the (human, WBE) to the level of detail that they could apply their preferred translation function.
With regards to simulation divergence, consider the case where it is actually difficult to simulate the EM field (if physics was continuous, you’d require a hypercomputer, but if it isn’t, digital simulation is expensive anyway). Consider also that evolution could have recruited the EM field for the purpose of running efficient computations of some nature – this may be why evolution has ensnared consciousness at all.
With regards to your edit:
The actual function that my brain performs is to imagine writing a program that models the inner workings of that system
My recommendation is to try playing with both digital and analog models. (Actually, I could tell a really long story about how my electrical engineer friend jailbroke me out of a programmer’s mindset and into a continuous-domain mathematician’s mindset… that would take some time, but it’s a divergence in worldview that I notice when talking to people sometimes. There’s a Type of Guy who wants to translate everything into a short PyTorch program. I used to be that guy.)
>The WBE does experience phenomenological consciousness, but this is not correlated with the words it’s saying.
Does this not make it a philosophical zombie? If the qualia and the talking-about-qualia are not necessarily correlated with each other, how did such a coincidence come about originally, in the actual physical system?
If the relationship between me saying “I experience qualia”, and my qualia, is functionally isomorphic to the relationship between my emulation saying “I experience qualia” and my emulation’s qualia… does this not imply that my claim of experiencing qualia is similarly not correlated with the words I’m saying, for the same reason?
Separately: I like the possibility that the EM field just cannot be emulated in the necessary detail to get convergent behavior, that’s one which hadn’t occurred to me. But… this doesn’t exactly feel like it pushes against computationalism? It feels more like saying that, yes, qualia is a computation, but the only computer with sufficient computational power to compute the program happens to be computers made out of infinitely-continuous analog EM fields.
I also note that if that’s true, we really ought to figure out how to harness EM fields for our own mundane computation. This feels like evidence it isn’t true, but I admit that’s mostly about pessimism that such a cheap source of compute could exist.
If the relationship between me saying “I experience qualia”, and my qualia, is functionally isomorphic to the relationship between my emulation saying “I experience qualia” and my emulation’s qualia
Let’s say in the original brain there is a contiguous chunk of EM field which is what it’s like to be you. In the process of creating a digital emulation of the field, at the hardware level, the field and its causal pathways would get broken up and reshuffled in a way which does not preserve spatiotemporal continuity. Even modulo some diffeomorphism, the shapes in the WBE hardware would be different to the shapes in the original brain. They could have their own experience, but it would be a very different experience to the original brain.
But if you only look at the functional relationships between input and output, you would ignore all this intermediary detail.
Also, if you find the premise of hypercomputation with fields of qualia entertaining, check this out.
but this would not actually cause my WBE to not talk about having phenomenological consciousness, right?
The relationship between the simulated neurons and the simulated em fields would be identical, therefore my whole brain emulation would talk about having phenomenological consciousness for the same reason that I do, except it wouldn’t have it?
edit: coming back to this, I sort of anticipate that you will point to the fact that I’m just looking at the edges of the system, because that’s true in a sense
but if I actually had the physics simulator in front of me, and I could compare the wbe to the real world brain, then I would be able to find the point of divergence, and then I wouldn’t be talking about edges anymore, we’d be examining the nitty-gritty details
I want you to make a prediction about what would happen in that circumstance. would I be able to find the point of divergence and fix it? what happens if I did? if not, why wouldn’t I be able to?
Yes, it would not stop the WBE from talking about phenomenal consciousness.
The WBE would still have phenomenal consciousness (the entire point of my article is that everything does), but it wouldn’t have the phenomenal consciousness described by the simulation. Rather, it would have the phenomenal consciousness you’d expect if you analysed the WBE hardware running the simulation.
You wouldn’t find this divergence in inputs and outputs, rather you do need to look at the internal states.
Perhaps if we ran the WBE simulator next to a WBE simulator simulator (so we can nondestructively analyse both the simulated WBE and the simulated WBE hardware), if you ran our translation function on the simulated WBE and the simulated WBE hardware you’d get different internal states. But if you ran the translation function on the simulated WBE and the simulated WBE inside the simulated WBE hardware, you’d get the same internal states).
Out in baseline reality, if we applied our translation function (not possible nondestructively) to a brain and WBE hardware, we’d see similar divergence in internal states. Recall that our translation function can only return equivalent internal states for the human as well as the digital simulation of the human if you can figure out how to write one which can traverse the digital simulation abstraction stack. My argument is that even if you did, it would be so complex that I would find it implausible as a proposal about how the universe works (well, assuming the simulation hardware has an architecture similar to current human computers).
Regarding edges of the system, picking the edges of a system is an opinionated choice and not an objective fact of the universe (unless we are talking about something really low level and universal, like Andrés’ whole EM field topology thing). Those edges are not objective facts of the universe; you might pick edges around the (brain, WBE) which do not reveal a point of divergence in the inputs and outputs, but I could just as easily pick different edges. Better not to try to coarse-grain the system like this and to consider the system as a whole, including its internal states.
It may be illustrative to consider that I anticipate that analytic functionalism may well be equivalent in its implications to my preferred qualia structuralist interpretation of physicalism, but only if we examine reality at a fine-grained/high enough resolution. See this conversation.
I could try to deconstruct the psychological proclivities which lead humans to want to coarse-grain reality, but I believe that this is well trodden philosophical terrain...
How do I tell that I’m not in the same position as that WBE, who is talking about phenomenological qualia it doesn’t actually experience (since it experiences something wildly different)?
it sure feels to me like the reason I talk about having qualia is downstream, causally, from the qualia itself… but if I am correctly understanding your model, that’s not true. the interaction between my neurons, and my EM field, is isomorphic to the interaction between the WBE’s simulated neurons, and the simulated EM field, yes?
so if the WBE’s reason-for-talking-about-having-qualia is not actually that it has qualia, then neither can mine be?
I ask for clarification, because I used the word divergence to describe the difference between the two systems, base reality vs emulation, but you seem to be saying that both systems would diverge in exactly the same way, which makes me think you’re using the word differently… I can’t tell if you actually think the wbe will behave the same
i would probably get around the ‘edges vs interior’ problem by just postulating, for the sake of the thought experiment, that we emulate the entire universe. that way the edges *have* to be well-defined and agreed upon
How do I tell that I’m not in the same position as that WBE
You cannot really do this by trying to prove this to yourself. You can only do this by looking at your subjective experience and accepting this from first principles, given that it’s the only thing you’re experiencing.
so if the WBE’s reason-for-talking-about-having-qualia is not actually that it has qualia, then neither can mine be?
I think that if you are trying to infer the internal state of entities very different to yourself then reasoning processes which lean on talking-about-having-qualia are largely bunkum. Perhaps this might draw accusations of anthropic or biological chauvinism, but I don’t think it’s unreasonable to look at my own biology and the biology of others and accept that similar things might generate qualia in similar ways. Anyway, at no point have I made talking-about-having-qualia load-bearing to my argument, even though it is a popular subject of thought experiments. Instead I have proposed alternative reasoning processes.
I can’t tell if you actually think the wbe will behave the same
Haven’t we already specified that the simulation is perfect for the sake of argument? I’m sorry but we’ve been going long enough that I can’t remember now. My point is that even a perfect simulation might not have the same qualia, because in order to make predictions about qualia in unfamiliar systems, we have to look at the hardware before we can figure out what’s going on in the simulation, and going from the hardware to the simulation is an opinionated process. I could propose a translation function which does actually derive a qualia state from the hardware which is identical to the original brain’s, but any such function is going to be unfathomably baroque.
Perhaps it might be helpful to – rather than reasoning about this from a human’s perspective, imagine you are a divine creator who despite their omniscience has forgotten how the universe works. They can’t remember what function they used to relate the structure of matter to the structures in the minds of beings – and for this reason, they also only have access to the raw shapes of matter, because in order to access human knowledge and understand the ways that humans categorise raw shapes of matter into reified concepts, they first must derive human internal experience from raw shapes of matter. When trying to rederive the translation function they came up with back when they raised the firmament, what’s your prior that our amnesiac God would construct something so complex that it could also pull some highly specific qualia out of complex digital hardware?
Also, going meta for a moment (having a hard time putting words to this):
I sometimes feel there is a mismatch that happens when I talk to functionalists, which is that I get the impression that they are considering the “whole functional system”/inputs and outputs all the way out at the very edge of the system (like, red light entering the eyes, the word “red” being spoken, etc), and this seems arbitrary and not grounded in internal observations.
And if I am being honest, working with tokens at the level of “red” feels coarse/crude, I prefer to work at a level of detail that’s actually informed by my qualia, like I’ll be thinking about minute red fluctuations in the visual field and what neural structures might underly them. I’d not be working with red in isolation because I don’t think you can even have colour qualia without them being embedded in a field (Richard P. Stanley talks about this sort of type error in this paper).
So, I am trying to find some physical structure whose momentary shape is similar to whatever is happening in the visual field, and this might have an input/output boundary around it equivalent to the boundary between phenomenal and access consciousness – and I am inclined to be skeptical that this extends all the way out into neighbouring cortices, let alone the eyes. But maybe it does, who knows (Emmett Shear once claimed that the brain sends more messages to the retina than the other way around, but this didn’t pass my friend’s fact check).
It’s an outside in, behaviouralist perspective rather than inside out one, and it leads to very different intuitions, but I think they can be worked through. Lmk what you think.
No sweat, thanks for the long comment.
Correct.
1: Yes, I believe the EM fields in the digital computer might have an experience, but the shape of that experience is not strongly correlated with the shape of the digital software. Or perhaps, more crudely: The digital software cannot be a p-zombie because the digital software does not really exist, except insofar as it has a physical manifestation (which would have a phenomenal experience related to the shape of that physical manifestation).
2: I’m not sure I follow but I recommend reading this part of the Lerchner paper and really wrapping your head around how alphabetisation is metaphysical or at least observer-dependent rather than physical.
3: I think that the entire body may have qualia! This doesn’t imply substrate-independence, though. Perhaps if the knee happened to implement an EM field structure with the same shapes/symmetry group as what we experience as colour then your knee could indeed experience red. However, I want the causal structure which is actually linked to talking, so we can correlate external observations with internal ones. In my ideal empirical programme this would involve picking a brain region and perturbing it and correlating this with direct reports. Susan Pockett proposed something like this.
On functionalism: Note that I am arguing against computational functionalism and not functionalism itself – that’s a different matter. Check here, I have been exploring with jessicata whether or not this qualia structuralist account is reconcilable with analytic functionalism. Discussions are ongoing. However, I still think you have to take a fine-grained physics level interpretation of the relevant causal structures, not a coarse-grained computational one, so you still wind up drawing the physicalist rather than computationalist-flavoured interpretations of subjective experience.
Btw, that’s not this Herschel is it?
Why does the WBE talk about phenomenological consciousness, then, if it doesn’t experience it? That’s the real question I need an answer to… it means that the reason I talk about having phenomenological consciousness is not actually causally downstream of my having phenomenological consciousness, either (since a WBE of me would also talk about having it, despite not having it). That feels like very straightforward p-zombieism, right? If I am my qualia, then… my qualia is not actually able to cause me to talk about having qualia, and yet the p-zombie whose EM fields create me, the body I mistakenly think of as being ‘me’, talks about having qualia anyway? What do you think is going on there?
It seems to me you either have to say that 1) the WBE’s behavior will diverge from the bodily-instantiated counterpart, such that it does not talk about phenomenological consciousness, despite the fact that maxwell’s equations are being faithfully computed and the predictive model of the EM/neuron interactions is 100% accurate
or else 2) p-zombies
1) seems, to me at first glance, like an absurd thing for an intelligent person to believe (in a way that makes me think I must be missing something, not that you aren’t intelligent), but you reject 2). I think I predict that you’ll say 1), but I don’t know what form that answer will come in, and I could be wrong. Maybe there is some third option I’m not thinking of.
But I really do feel like… hm. Like, if the answer is 1), then we could examine the emulation and the real-world brain to figure out the exact moment they diverge, and then just… fix the emulation, so that it accurately modeled reality instead of failing to. And we could just keep iterating on that until we succeeded. And then we’d have (according to you? maybe?) a whole-brain emulation of a human which talked about experiencing phenomenological consciousness for the same reasons that I talk about it, but which did not actually experience it. and then we’re just back at p-zombies.
Yes, he is indeed that Herschel.
edit: I like the WBE formulation of the generalized anti-zombie principle because it’s an actual experiment that could theoretically be performed in real life, it forces the issue down to the level of empiricism (even if hypothetical, armchair empiricism)
it’s the actual point of friction in my mind, when I try to figure out how to adopt your perspective, to see what can be seen from it
I can agree that, as mentioned downthread, there are also contradictions in the computationalist theory of consciousness… but, idk. when I hear non-computationalists try to answer the GAZP, mostly what I hear doesn’t actually convince me that their understanding is better or more complete, instead it convinces me that we started from different priors.
Pointing out contradictions in the computationalist framing is definitely a valid move, I don’t mean to say that it isn’t. I don’t understand these matters and don’t claim to. But… idk. When I try to understand a system, the actual function that my brain performs is to imagine writing a program that models the inner workings of that system. I don’t know if I really have a way to try to understand consciousness that isn’t doing that. This might be a failing on my part, but it’s why the WBE thought experiment causes so much friction for me.
Here is your third way:
The WBE talks about phenomenological consciousness because it is simulating the kind of system which would do this. If we have a magic system which can perfectly emulate the EM field, then sure, these dynamics won’t diverge.
The WBE does experience phenomenological consciousness, but this is not correlated with the words it’s saying. Given a translation function we are confident in, what the WBE is experiencing is the output of that function as applied to its hardware. Which in the case of the physicalist function I proposed, would be what it’s like to be that hardware, not what it’s like to be the simulation – unless the hardware was designed in such a way that these correlate well with one another. Consider analog hardware in which the spatiotemporal topology of the causal relationships is the same in the WBE as it is in the original brain, for example.
You can be as confident as you like that your qualia is correlated with your actions because, well, you have direct access to this (though you couldn’t prove prove it to me).
With regards to both systems, the only way an independent observer could be truly sure what’s going on is by inspecting the physical structure of the (human, WBE) to the level of detail that they could apply their preferred translation function.
With regards to simulation divergence, consider the case where it is actually difficult to simulate the EM field (if physics was continuous, you’d require a hypercomputer, but if it isn’t, digital simulation is expensive anyway). Consider also that evolution could have recruited the EM field for the purpose of running efficient computations of some nature – this may be why evolution has ensnared consciousness at all.
With regards to your edit:
My recommendation is to try playing with both digital and analog models. (Actually, I could tell a really long story about how my electrical engineer friend jailbroke me out of a programmer’s mindset and into a continuous-domain mathematician’s mindset… that would take some time, but it’s a divergence in worldview that I notice when talking to people sometimes. There’s a Type of Guy who wants to translate everything into a short PyTorch program. I used to be that guy.)
>The WBE does experience phenomenological consciousness, but this is not correlated with the words it’s saying.
Does this not make it a philosophical zombie? If the qualia and the talking-about-qualia are not necessarily correlated with each other, how did such a coincidence come about originally, in the actual physical system?
If the relationship between me saying “I experience qualia”, and my qualia, is functionally isomorphic to the relationship between my emulation saying “I experience qualia” and my emulation’s qualia… does this not imply that my claim of experiencing qualia is similarly not correlated with the words I’m saying, for the same reason?
Separately: I like the possibility that the EM field just cannot be emulated in the necessary detail to get convergent behavior, that’s one which hadn’t occurred to me. But… this doesn’t exactly feel like it pushes against computationalism? It feels more like saying that, yes, qualia is a computation, but the only computer with sufficient computational power to compute the program happens to be computers made out of infinitely-continuous analog EM fields.
I also note that if that’s true, we really ought to figure out how to harness EM fields for our own mundane computation. This feels like evidence it isn’t true, but I admit that’s mostly about pessimism that such a cheap source of compute could exist.
Let’s say in the original brain there is a contiguous chunk of EM field which is what it’s like to be you. In the process of creating a digital emulation of the field, at the hardware level, the field and its causal pathways would get broken up and reshuffled in a way which does not preserve spatiotemporal continuity. Even modulo some diffeomorphism, the shapes in the WBE hardware would be different to the shapes in the original brain. They could have their own experience, but it would be a very different experience to the original brain.
But if you only look at the functional relationships between input and output, you would ignore all this intermediary detail.
Also, if you find the premise of hypercomputation with fields of qualia entertaining, check this out.
but this would not actually cause my WBE to not talk about having phenomenological consciousness, right?
The relationship between the simulated neurons and the simulated em fields would be identical, therefore my whole brain emulation would talk about having phenomenological consciousness for the same reason that I do, except it wouldn’t have it?
edit: coming back to this, I sort of anticipate that you will point to the fact that I’m just looking at the edges of the system, because that’s true in a sense
but if I actually had the physics simulator in front of me, and I could compare the wbe to the real world brain, then I would be able to find the point of divergence, and then I wouldn’t be talking about edges anymore, we’d be examining the nitty-gritty details
I want you to make a prediction about what would happen in that circumstance. would I be able to find the point of divergence and fix it? what happens if I did? if not, why wouldn’t I be able to?
Yes, it would not stop the WBE from talking about phenomenal consciousness.
The WBE would still have phenomenal consciousness (the entire point of my article is that everything does), but it wouldn’t have the phenomenal consciousness described by the simulation. Rather, it would have the phenomenal consciousness you’d expect if you analysed the WBE hardware running the simulation.
You wouldn’t find this divergence in inputs and outputs, rather you do need to look at the internal states.
Perhaps if we ran the WBE simulator next to a WBE simulator simulator (so we can nondestructively analyse both the simulated WBE and the simulated WBE hardware), if you ran our translation function on the simulated WBE and the simulated WBE hardware you’d get different internal states. But if you ran the translation function on the simulated WBE and the simulated WBE inside the simulated WBE hardware, you’d get the same internal states).
Out in baseline reality, if we applied our translation function (not possible nondestructively) to a brain and WBE hardware, we’d see similar divergence in internal states. Recall that our translation function can only return equivalent internal states for the human as well as the digital simulation of the human if you can figure out how to write one which can traverse the digital simulation abstraction stack. My argument is that even if you did, it would be so complex that I would find it implausible as a proposal about how the universe works (well, assuming the simulation hardware has an architecture similar to current human computers).
Regarding edges of the system, picking the edges of a system is an opinionated choice and not an objective fact of the universe (unless we are talking about something really low level and universal, like Andrés’ whole EM field topology thing). Those edges are not objective facts of the universe; you might pick edges around the (brain, WBE) which do not reveal a point of divergence in the inputs and outputs, but I could just as easily pick different edges. Better not to try to coarse-grain the system like this and to consider the system as a whole, including its internal states.
It may be illustrative to consider that I anticipate that analytic functionalism may well be equivalent in its implications to my preferred qualia structuralist interpretation of physicalism, but only if we examine reality at a fine-grained/high enough resolution. See this conversation.
I could try to deconstruct the psychological proclivities which lead humans to want to coarse-grain reality, but I believe that this is well trodden philosophical terrain...
How do I tell that I’m not in the same position as that WBE, who is talking about phenomenological qualia it doesn’t actually experience (since it experiences something wildly different)?
it sure feels to me like the reason I talk about having qualia is downstream, causally, from the qualia itself… but if I am correctly understanding your model, that’s not true. the interaction between my neurons, and my EM field, is isomorphic to the interaction between the WBE’s simulated neurons, and the simulated EM field, yes?
so if the WBE’s reason-for-talking-about-having-qualia is not actually that it has qualia, then neither can mine be?
I ask for clarification, because I used the word divergence to describe the difference between the two systems, base reality vs emulation, but you seem to be saying that both systems would diverge in exactly the same way, which makes me think you’re using the word differently… I can’t tell if you actually think the wbe will behave the same
i would probably get around the ‘edges vs interior’ problem by just postulating, for the sake of the thought experiment, that we emulate the entire universe. that way the edges *have* to be well-defined and agreed upon
You cannot really do this by trying to prove this to yourself. You can only do this by looking at your subjective experience and accepting this from first principles, given that it’s the only thing you’re experiencing.
I think that if you are trying to infer the internal state of entities very different to yourself then reasoning processes which lean on talking-about-having-qualia are largely bunkum. Perhaps this might draw accusations of anthropic or biological chauvinism, but I don’t think it’s unreasonable to look at my own biology and the biology of others and accept that similar things might generate qualia in similar ways. Anyway, at no point have I made talking-about-having-qualia load-bearing to my argument, even though it is a popular subject of thought experiments. Instead I have proposed alternative reasoning processes.
Haven’t we already specified that the simulation is perfect for the sake of argument? I’m sorry but we’ve been going long enough that I can’t remember now. My point is that even a perfect simulation might not have the same qualia, because in order to make predictions about qualia in unfamiliar systems, we have to look at the hardware before we can figure out what’s going on in the simulation, and going from the hardware to the simulation is an opinionated process. I could propose a translation function which does actually derive a qualia state from the hardware which is identical to the original brain’s, but any such function is going to be unfathomably baroque.
Perhaps it might be helpful to – rather than reasoning about this from a human’s perspective, imagine you are a divine creator who despite their omniscience has forgotten how the universe works. They can’t remember what function they used to relate the structure of matter to the structures in the minds of beings – and for this reason, they also only have access to the raw shapes of matter, because in order to access human knowledge and understand the ways that humans categorise raw shapes of matter into reified concepts, they first must derive human internal experience from raw shapes of matter. When trying to rederive the translation function they came up with back when they raised the firmament, what’s your prior that our amnesiac God would construct something so complex that it could also pull some highly specific qualia out of complex digital hardware?
Also, going meta for a moment (having a hard time putting words to this):
I sometimes feel there is a mismatch that happens when I talk to functionalists, which is that I get the impression that they are considering the “whole functional system”/inputs and outputs all the way out at the very edge of the system (like, red light entering the eyes, the word “red” being spoken, etc), and this seems arbitrary and not grounded in internal observations.
And if I am being honest, working with tokens at the level of “red” feels coarse/crude, I prefer to work at a level of detail that’s actually informed by my qualia, like I’ll be thinking about minute red fluctuations in the visual field and what neural structures might underly them. I’d not be working with red in isolation because I don’t think you can even have colour qualia without them being embedded in a field (Richard P. Stanley talks about this sort of type error in this paper).
So, I am trying to find some physical structure whose momentary shape is similar to whatever is happening in the visual field, and this might have an input/output boundary around it equivalent to the boundary between phenomenal and access consciousness – and I am inclined to be skeptical that this extends all the way out into neighbouring cortices, let alone the eyes. But maybe it does, who knows (Emmett Shear once claimed that the brain sends more messages to the retina than the other way around, but this didn’t pass my friend’s fact check).
It’s an outside in, behaviouralist perspective rather than inside out one, and it leads to very different intuitions, but I think they can be worked through. Lmk what you think.