For some people, that's very difficult. They experience speaking as "I let my mouth say something." Even asking them afterwards "what did you just say" can get an "I dunno." This is what happens when you spin off parts of your mind into "semi-autonomous entities" and give them feedback to modify them.
I think it's more like "If I don't know what I'm going to say next, I can laugh at my own jokes." This provides valuable feedback on "bringing the funny."
Every jazz musician feels that regularly, I'm "speaking faster than I think" any time I take a solo. Very similar to when I'm making an extemporaneous argument in front of a jury, you're just... doing... the performance, and the active narrator in your head is more like a monitor that notices when a problem's about to come up, "this argument you're speaking right now could lead to a contrary inference you need to address before it takes root", "you're getting too chromatic, find some safe chord tones here".
What was the way in which LLMs helped you realize this, just from witnessing their ability to do that performance without having a consciousness to direct it?
If you refactor your world-model enough, you can see your "self"/soul as a sort of container of "lifelines", where a lifeline is a connection to the outside world (e.g. a social relationship, an ability that can be applied to places in the world, or an object you own). In this model, one of the main things you can do is balance competing pulls in your soul, which I call the autistic state of mind. But as far as I can tell, many people defer this balancing act to others, which gives up most of their agency; I call this the schizophrenic state of mind.
I should say my model disagrees with Scott's post here in that I regard most people as having pretty severe cases (often more schizophrenic, but in a way that is buffered against clinical pathologies), whereas he regards most people as basically healthy rationalish individuals. And I'm generally unimpressed by biomarkers.
If there’s just a continuum, and you define a different point as the zero point, then the zero of the other scale can look like a really high or really low number.
Psychometrically it'd look like two continua, but psychometric measurements are logarithmically related to real magnitudes, so in practice you're going to have one that is much bigger than the other, and "which is biggest?" gives you a discrete typology.
An important piece of prior work on this "internal model of thoughts & self" perspective is Michael Graziano's theory of consciousness (which Steven Byrnes cites as largely consistent with his view):
This seems a plausible model of multiple personality disorder.
Buddhist texts often point out the evidence that you don’t have a coherent self. Multiple personalities is another model you might appeal to to explain this incoherence that the Buddhists are going on about, especially if your psychiatrist is telling you that MPD is a real thing.
A relatively common situation: a depressed person sort-of wants to get on and tidy up their room, but also sort-of doesn’t. How can it be that we can both want and not want something at the same time?
At this point, vajrayana Buddhism might talk about demons to explain the phenomenon, with the understanding that demons dont really exist.
> A relatively common situation: a depressed person sort-of wants to get on and tidy up their room, but also sort-of doesn’t. How can it be that we can both want and not want something at the same time?
It’s because we have two different decision trees to reconcile within ourselves; one springs from our viscera and the other springs from our conscious modeling.(aka; rationalizing They’re not always on the same page are they? A simple model would be a parent telling a small child to pick up his toys and the child not wanting to. Those two forces combine in us as we become adults and if they aren’t on the same page, well things can go pear shaped .
Memory is very dependent on context (thing X reminds you of related thing Y) so it is not totally out of the question that you might only remember thing Y while inhabiting persona X.
Though, we might suspect something closer to: the memory can’t be integrated into a particular persona’s sense of self, and so is repressed.
Also for when religious people talk about hearing the voice of the devil telling them to sin etc. they have a model where the devil can be a voice in their head and interpret thoughts within that
Is there something special that distinguishes the sorts of beliefs and models that are bistable - that we can suddenly snap out of - from the more common sorts of beliefs and models that seem to gradually change over time?
Things like phobias or habits, for example, that take gradual exposure and consistent effort to change.
The vast majority of optical illusions only exist because they are two dimensional representations of something that we might encounter in three-dimensional space. The primary abstraction of two dimensional perspective and motion is the the first illusion, which we have become rather comfortable with since the birth of perspective drawing. The AB checkerboard illusion works because each square is surrounded by squares that are lighter or darker than it which forces the eye to recognize the pattern and dismiss the truth, which is that those two squares are the same gray.
How do the rest of you perceive the "upside-down plates" picture? For me, it's very resistant to fully flipping.
If I start at the bottom left, I can easily get the three leftmost plates to flip inside out; I can do something similar from the right-hand side, moving along the top-right corner. I can kind of flip that whole set of 8 along the left-top-right edges, though the circular one at the top is tricky.
The multi-compartment ones take more effort, and not only do they not trigger a cascade, sometimes one compartment will pop out without affecting the other compartments in the same dish.
It seems like my brain is strongly inclined to see them as right-side-up, flipping them requires focus and effort, and at that point I'm not really seeing the image as a whole. I haven't tried viewing it from a long distance, but rolling my chair back a bit didn't make a lot of difference.
To the limited extent that I can flip the divided ones, what seems to work is consciously thinking "okay, if this was the other way around, what would these regions of colour actually represent?" Basically I have to actively reason my way to seeing them differently. Maybe you've already done that, but if not, I'd be curious whether it makes a difference.
Im not sure its even possible for a real object to correspond to the flipped image here. The bright left rims, and the visible line between the wall and its shaded ground, with the wall brighter, I dont think those are possible inverted. Paying attention to those prevents me from flipping anything.
Ill note that I have never seen a flipping illusion using a real life picture, that wasnt actually resolvable with information from within. The only one I can think of that most people dont get right at a glance is The Dress, and I did get it, so skill issue.
The shaded bottoms-now-tops of the small top right bowls couldnt be parallel to those on the other dishes either, so that might be what kicks you out of flipping fully.
This maps well to Michael Levin's model of memory (section 4 of https://www.mdpi.com/1099-4300/26/6/481 and https://thoughtforms.life/some-thoughts-on-memory-goals-and-universal-hacking/, also talked about in some podcasts and presentations he has done). In this model, memory isn't some high fidelity thing, but rather a highly compressed/encoded bit of information that, when run through a specific decoder, will result in a particular memory. What is key/critical here is that the decoder plays a very significant role in determining how that compressed information is decoded.
If you encode a memory using decoder/encoder (coder) A, and then decode it using coder A, you'll get out basically the same information that went in. However, if you decode it using coder B you'll get out different information than was put in.
Under normal circumstances, your coder changes very slowly over time as you experience life, and during that time you are constantly remembering things which causes them to be decoded and then re-encoded (so to speak) using the latest version of your coder. However, one can imagine a person having two distinct coders in their head at the same time, and in such a situation encoding something with one coder and then decoding it using a different coder could result in complete failure to decode (no memory), or a different memory than what actually occurred.
This maps well both to the hypnosis situation you have described here, as well as things like split personalities. An individual may remember things differently than they actually occurred because the coder used to encode the event is not the coder used to decode the event, or they may "forget" what happened. It can also explain things like being able to remember correctly if you are put back under hypnosis or if your personality switches back to the one that originally experienced it.
I think it also may have some explanatory power for retrograde amnesia. This could be caused by damage to the coder that needs to be "rebuilt" or "repaired", which is why your memories will sometimes return later. Perhaps older memories are encoded in a more "generic" way, while newer memories are encoded in a way that are are more tightly coupled with the current version of your brain's coder.
That’s interesting. That also explains why, when thinking about things I did in the early 2000s, I sometimes remember myself having looked up directions on my smartphone, because that sort of behavior is part of “find out where I am and where I’m going” in the decoder I’ve used for the past fifteen years, even though it wasn’t part at the time I formed the memory.
This feels like a really great companion piece to a lot of what Sam Harris writes in his book "Waking Up".
One of the claims in there, made by him and other meditators, is that meditating on your own thoughts enough or in a particular way eventually causes the illusion of the Self to disappear ("You search for it and fail to find it").
In that book, the perspective shift is just given as an empirical claim to tested by the reader, via meditation practice. In this post, it feels like we get a satisfying and plausible mechanistic theory for it.
I think that's a very convincing interpretation of self-consciouness/ ego. It's like concurrent interpretations of quantum mechanics : for some like relational quantum mechanics (RQM) vs Everett's many worlds, it's unlikely that you could experimentally prove one over the other but your mind can switch from one to the other to explain what you see. It could be a quasi solution to the so-said "hard problem" of consciousness. If consciousness is itself an interpretative model, a paradigm, it's no surprise you can't really prove it nor infirm it, or reach any kind of definitive answer on that subject. Now the question arises : is the Buddha a philosophical zombie ?
By the by, but Deutsch quite strongly believes that the truth of Everett's MW is experimentally provable. I don't have a strong enough physics background to evaluate the claim but I think it also rests on his disagreements with Bayesian approaches to probability (which itself is reinforced by his MW views). If someone knows more please chime in.
Edit: If I recall correctly his possible proof relies on self-reporting quantum computers- but the point is that he believes it is absolutely provable *in theory*, which I assume many would dispute. Whether that can disprove RQM specifically, I'm unsure.
I'm not sure either but I wonder if the idea isn't to experimentally disprove wave packet collapsing. If that so it would not disprove RQM that is also an interpretation without collapsing.
The first thing to note is that MWI is more than one theory. What splittng is...how complete and irrevocable it is ... varies between particular theories. So does the rate of splitting, so does the mechanism of splitting.
The second thing to note is that many worlders are pointing at something implied the physical formalism and saying "that's a world"....but whether it qualifies as a world is a separate question from whether it's in the formalism , and a separate *kind* of question, from whether it is really there in the formalism. One would expect a world, or universe, to be large, stable, non-interacting, and so on . It's possible to have a theory that has collapse , without having worlds. A successful MWI needs to jump three hurdles: empirical correctness.
mathematical correctness and conceptual correctness -- actually having worlds
---- Coherence versus Decoherence----
There is an approach to MWI based on coherent superpositions, and a version based on decoherence. These are (for all practical purposes) incompatible opposites, but are treated as interchangeable in Yudkowsky's writings. . Decoherent branches are large, stable, non interacting and irreversible...everything that would be intuitively expected of a "world". But there is no empirical evidence for them (in the plural) , nor are they obviously supported by the core mathematics of quantum mechanics, the Schrödinger equation.Coherent superpositions are small scale , down to single particles,possibly observer dependent, reversible, and continue to interact (strictly speaking , interfere) after "splitting". The last point is particularly problematical. because if large scale coherent superposition exist , that would create naked eye macrocsopic scale:, e.g. ghostly traces of a world where the Nazis won.
We have evidence of small scale coherent superposition, since a number of observed quantum.effects depend on it, and we have evidence of decoherence, since complex superposition are difficult to maintain. What we don't have evidence of is decoherence into multiple branches. From the theoretical perspective, decoherence is a complex , entropy like process which occurs when a complex system interacts with its environment. [Decoherence isn't simple)(https://www.lesswrong.com/posts/Atu4teGvob5vKvEAF/decoherence-is-simple). But without decoherence, MW doesn't match observation. So there is no theory of MW that is both simple and empirically adequate, contra Yudkowsky and Deutsch
I had never heard of Lehar or this site and just checked it out. Thank you for the link! (And I wonder what my eight-year-old will think about it when I share it with him...)
This made me think of how quantum spin in entangled particles can collapse and reverse when observed and then the spin reverses even if the objects are far distances.
I don't know if this has any experimental support I could point to, and it's been a long time since I read it, but Douglas Hofstadter wrote (in either The Mind's I or I Am A Strangle Loop, I don't remember which) that the 'self' concept might be like the "this" pointer in a data structure in C++. Literally just a pointer that points to its own address, but otherwise empty.
In which case, sure, we tie a lot of things to our identity, but they aren't part of the core 'self' concept. And the instant bistable switch of satori is something like noticing this fact?
Great stuff, my qualm is just that even saying perception is indirect is kind of wrong. Suppose there's video evidence in a courtroom, if someone points to the screen and says "that's the killer right there," are they wrong? (And a video isn't much different from any internal model.)
Or to put it another way, the idea that we're "perceiving a model" isn't really correct. We're HAVING a model and at all times directly perceiving a WORLD (and that world as it really, really is, to boot, see bottom of this post).
The trick is, I think, that words like "seeing" and "perceiving" are third person terms - so they're something a third person observer says when (e.g.) they see a tiger perceiving its prey, they're part of that objective, third person STORY.
What the tiger experiences directly (its internal model of the prey) isn't itself yet a perception, it's only granted the grace of being trully callable a "perception" IF AND WHEN a third person can observe both the tiger licking its lips in anticipation, and that there is actually a deer over there.
So neither of the "flips" are a perception until there's been a checking process, and that checking process doesn't change the subjective experience, it just corrects an error in the truth claim about what that subjective experience portended.
We are always and at all times directly perceiving the world, because we are always and at all times perceiving an objective "slice" of the world that's a secret treasure that the world gives of itself to us, and to nobody else, at just that point in time and space.
The explanation of hypnotism feels completely unsatisfying to me. If you ever get an opportunity to talk with a friend who has been deeply hypnotized, you learn there is something else going on. It is like a bizarre co-opted dream state. The subject isn’t just going along with the hypnotist because they believe they’ve lost control of their body; that’s not how it works at all. The subject completely, irrationally believes the scenarios introduced by the hypnotist, and cannot be convinced otherwise.
Relevant to the multiple-personality angle, tulpamancy (qua mental hobby, not supposedly-supernatural woo) is specifically the practice of training yourself to "flip" to an intuition that some of your thoughts come from a second homunculus on purpose, consciously, aware that what you're doing is flipping some switches back and forth. It does indeed feel very much like flipping from duck to rabbit etc.
Interesting! I greatly enjoyed Jaynes’s book but always thought he oversold his idea by claiming that people in pre-Classical times were not conscious; it seemed to me more likely that they conceived a new understanding of consciousness, maybe analogous to Adam Smith’s new understanding of commerce.
I became convinced that something like the homunculus account of consciousness must be the case back when I used to rock climb (before I broke my neck two years ago in, erm, a rock climbing accident).
On many occasions, I would observe my body begin to execute physically demanding, precise technical moves in fear inducing situations before I had consciously "told" my body to start moving. Of course my conscious mind would rapidly catch up and experience between conscious awareness and movement would quickly become synchronic. But it happened so often I became quite familiar with the thought, "oh we're going are we?" as I began to execute the required physical moves.
The way I came to think of "myself" is as being like Maggie Simpson in the introduction to The Simpsons. She steers the steering wheel and beeps the horn, and is under the illusion that she is driving the car. But really, Marge is driving the car. Maggie as homunculus.
I'm sure this must be very common, I can perceive it whenever I do a sustainable fast-walk pace for some period of time (say 4mph for 15+ minutes) then get off the treadmill and my legs just seem to move on their own. It sort of feels like you're just "steering" the direction of a Flintstones car from the homunculus seat. Being that it's a compound movement, and not merely a steady contraction/expansion of some muscle fiber, it's makes less sense to be an involuntary motion than the "lean against this wall while you try to raise your arm against it for 30 seconds then step away" thing everyone knows from grade school.
Trying to mess around with one's perception of consciousness on purpose seems just as dangerous as rock climbing in its own way. Seems like an easy way to break your brain. (Sorry to hear about your accident, and hope you were able to make a recovery.)
Well, I'm a quadriplegic now, so not much of a recovery. But given that by any reasonable metric I should have died, some would say I should be grateful.
The essay feels a bit incomplete to me. Particularly the ending, I actually wasn't even sure it was the ending when I was reading the blog on Gmail.
As for the topic itself, I dunno, while I don't find anything particularly objectionable it doesn't really seem like this model gets us anything. What are the benefits of this over a model like Plato's tripartite soul? I think this is what's missing from this essay is more commentary on the other theories of consciousness and how Byrnes' theory differs.
This model-flipping optical-illusion account fits my experience with Mahamudra (a Vajrayana Buddhist meditation practice). We often meditate with live instructions, which often have the form "take the view that..." (except it's about mental phenomena and the architecture of awareness, rather than twirling dancers).
Thomas Metzinger has a version of this account which he calls the "transparent self-model" (explained in his book "Being No One" or any number of YouTube lectures). He distinguishes between internal models that we see as models, vs. "transparent" models where we relate to the model contents *as if* it were the real world. On this view (as philosophers like to say), a first-person perspective requires we have a *transparent* world-model which includes the self as a specially-indexed object.
This transparency idea highlights how weird some of these "model-flipping" practices can get, especially with the self-model. It's not abstract propositional beliefs that are changing (which we're used to as rationalists), rather our experience of the world.
I think it's well-established that stage hypnosis is basically fake -- people play along because it would be awkward not to, nothing more. So using stage hypnosis (a fake thing) as evidence for this theory is kind of anti-convincing to me.
I was thinking about your review of Jaynes the whole time, I was very glad that it was eventually mentioned. That’s the most interesting part to me: just like in the optical illusions, the initial states are just the most common and most likely to make sense (upside down stairs are rarer than regular ones), but when it comes to our mind, there’s not as many guardrails. If all or most of us end up with roughly the same homunculus model, it’s probably because of the way we hear our parents talk about their own self-perception, but it’s not hard at all to imagine how different cultures could end up with very different models.
I completely understand your point about skipping the free speech part, but it still makes me a bit sad. People really add a whole lot of crazy epicycles to their world models to make space for it. I think it would be very healthy for a lot of people if we finally learned to deal with that idea.
Since I am autistically obsessed with language, I will point out that this
>the predictive processing interpretation is that we perceive our models of the world.
is misleading, insofar as it's Scott's interpretation or whether people working in predictive processing all agree with it. Perceive is an ordinary language world defined by its usages in ordinary contexts. We do perceive the world, because that is the correct way to use the word as it is embedded in everyday life. We do not perceive a model of the world, because "models" aren't imposed between us and the world in our ordinary linguistic behavior.
You might say that perception consists in our brains interpreting various other cognitive models and so we can be wrong about what the world is like - but it's strictly speaking false to say that we perceive those models. Nobody ever speaks of perceiving models irl when they see an illusion or hallucination or regular apple or whatever.
Now, since this is just an autistic objection about language use, it could be immaterial. But it could also mislead people by appropriating a technical term and giving people the misleading impression that there is the world, then their model, and what they perceive is the model, when that's not right. Perception is a complex behavior and can be given deep explanations, but it is not accurate to use the word that what you perceive is sense data or qualia or an inner world or whatever, because nobody irl speaks this way and the technical usage potentially or implicitly adopted here is not congruent with other ordinary usages.
> If the homunculus-self is a mostly-accurate but not-directly-perceived-and-real model of mental processes, then a person whose mental processes often flip between two or more dramatically different states (for example, borderlines, who are notable for very strong emotional states and “splitting”) may gather evidence for and eventually flip to a model of themselves as multiple different homunculi. This is especially true if they’re primed with the suggestion that this is a likely way for the inside of their mind to be (for example, by a psychotherapist who believes in multiple personalities).
An uninformed take like this makes me wonder if Scott is just clueless about DID specifically or is everything else of his I am reading and nodding along is subject to the Gell-Mann Amnesia effect.
The most unfortunate part is him uncritically repeating the discredited iatrogenic model of DID in the parentheses.
This topic deserves a much more in-depth discussion, suffice it to say for now that not one person with DID (as opposed to the ticktok multiplicity crowd) want to have it. They desperately want to "flip to a model of themselves" where they have a single unfractured self. Most try very hard to do so, repeatedly and in various ways, invariably failing. Most people with clinical DID symptoms have multiple nuanced alters and not the media-beloved one good and one evil, or the Borderline "strong emotional states".
What must it be like to predict the next thing that's going to be said by you versus actively saying it?
For some people, that's very difficult. They experience speaking as "I let my mouth say something." Even asking them afterwards "what did you just say" can get an "I dunno." This is what happens when you spin off parts of your mind into "semi-autonomous entities" and give them feedback to modify them.
What's that line, something like "sometimes I start a sentence without knowing where I'm going and I just hope I end up where I wanted."
I think it's more like "If I don't know what I'm going to say next, I can laugh at my own jokes." This provides valuable feedback on "bringing the funny."
It just seems obvious to me that I speak faster than I think, and LLMs helped me make sense of this.
Every jazz musician feels that regularly, I'm "speaking faster than I think" any time I take a solo. Very similar to when I'm making an extemporaneous argument in front of a jury, you're just... doing... the performance, and the active narrator in your head is more like a monitor that notices when a problem's about to come up, "this argument you're speaking right now could lead to a contrary inference you need to address before it takes root", "you're getting too chromatic, find some safe chord tones here".
What was the way in which LLMs helped you realize this, just from witnessing their ability to do that performance without having a consciousness to direct it?
I just finished giving a lecture and it was exactly like that.
If you refactor your world-model enough, you can see your "self"/soul as a sort of container of "lifelines", where a lifeline is a connection to the outside world (e.g. a social relationship, an ability that can be applied to places in the world, or an object you own). In this model, one of the main things you can do is balance competing pulls in your soul, which I call the autistic state of mind. But as far as I can tell, many people defer this balancing act to others, which gives up most of their agency; I call this the schizophrenic state of mind.
Interesting choice of terms.
A spectrum of mind from Autistic to Schizophrenic is definitely provocative.
If you're curious, Scott has written about this (theoretical) spectrum before https://slatestarcodex.com/2018/12/11/diametrical-model-of-autism-and-schizophrenia/
I should say my model disagrees with Scott's post here in that I regard most people as having pretty severe cases (often more schizophrenic, but in a way that is buffered against clinical pathologies), whereas he regards most people as basically healthy rationalish individuals. And I'm generally unimpressed by biomarkers.
If there’s just a continuum, and you define a different point as the zero point, then the zero of the other scale can look like a really high or really low number.
Psychometrically it'd look like two continua, but psychometric measurements are logarithmically related to real magnitudes, so in practice you're going to have one that is much bigger than the other, and "which is biggest?" gives you a discrete typology.
Interesting piece I hadn't seen before. I absolutely love the word 'schizotypy' BTW. Had never seen that before either.
Now how am I going to work that into a conversation with my wife today?
"Hey honey you know that new guy who bought the place across the street, you know, the one whose trip we still haven't figured out..." :)
An important piece of prior work on this "internal model of thoughts & self" perspective is Michael Graziano's theory of consciousness (which Steven Byrnes cites as largely consistent with his view):
https://www.goodreads.com/book/show/43726566-rethinking-consciousness
This seems a plausible model of multiple personality disorder.
Buddhist texts often point out the evidence that you don’t have a coherent self. Multiple personalities is another model you might appeal to to explain this incoherence that the Buddhists are going on about, especially if your psychiatrist is telling you that MPD is a real thing.
A relatively common situation: a depressed person sort-of wants to get on and tidy up their room, but also sort-of doesn’t. How can it be that we can both want and not want something at the same time?
At this point, vajrayana Buddhism might talk about demons to explain the phenomenon, with the understanding that demons dont really exist.
> A relatively common situation: a depressed person sort-of wants to get on and tidy up their room, but also sort-of doesn’t. How can it be that we can both want and not want something at the same time?
It’s because we have two different decision trees to reconcile within ourselves; one springs from our viscera and the other springs from our conscious modeling.(aka; rationalizing They’re not always on the same page are they? A simple model would be a parent telling a small child to pick up his toys and the child not wanting to. Those two forces combine in us as we become adults and if they aren’t on the same page, well things can go pear shaped .
Whether or not the selves share memories is pretty crucial to MPD.
Memory is very dependent on context (thing X reminds you of related thing Y) so it is not totally out of the question that you might only remember thing Y while inhabiting persona X.
Though, we might suspect something closer to: the memory can’t be integrated into a particular persona’s sense of self, and so is repressed.
Also for when religious people talk about hearing the voice of the devil telling them to sin etc. they have a model where the devil can be a voice in their head and interpret thoughts within that
Is there something special that distinguishes the sorts of beliefs and models that are bistable - that we can suddenly snap out of - from the more common sorts of beliefs and models that seem to gradually change over time?
Things like phobias or habits, for example, that take gradual exposure and consistent effort to change.
The vast majority of optical illusions only exist because they are two dimensional representations of something that we might encounter in three-dimensional space. The primary abstraction of two dimensional perspective and motion is the the first illusion, which we have become rather comfortable with since the birth of perspective drawing. The AB checkerboard illusion works because each square is surrounded by squares that are lighter or darker than it which forces the eye to recognize the pattern and dismiss the truth, which is that those two squares are the same gray.
How do the rest of you perceive the "upside-down plates" picture? For me, it's very resistant to fully flipping.
If I start at the bottom left, I can easily get the three leftmost plates to flip inside out; I can do something similar from the right-hand side, moving along the top-right corner. I can kind of flip that whole set of 8 along the left-top-right edges, though the circular one at the top is tricky.
The multi-compartment ones take more effort, and not only do they not trigger a cascade, sometimes one compartment will pop out without affecting the other compartments in the same dish.
It seems like my brain is strongly inclined to see them as right-side-up, flipping them requires focus and effort, and at that point I'm not really seeing the image as a whole. I haven't tried viewing it from a long distance, but rolling my chair back a bit didn't make a lot of difference.
I can’t flip them. The ones with the divided trays just refused to flip for me.
To the limited extent that I can flip the divided ones, what seems to work is consciously thinking "okay, if this was the other way around, what would these regions of colour actually represent?" Basically I have to actively reason my way to seeing them differently. Maybe you've already done that, but if not, I'd be curious whether it makes a difference.
If I work really hard, I can hold that illusion for a few moments, but it never settles for me
I deliberately did not want to try that one. I felt that if I got them to flip, I might not be able to flip them back, which I found very disturbing.
What worked for me was looking at Scott's square and then rolling my phone left so it's as if the light is coming from me instead of from the left.
Im not sure its even possible for a real object to correspond to the flipped image here. The bright left rims, and the visible line between the wall and its shaded ground, with the wall brighter, I dont think those are possible inverted. Paying attention to those prevents me from flipping anything.
Ill note that I have never seen a flipping illusion using a real life picture, that wasnt actually resolvable with information from within. The only one I can think of that most people dont get right at a glance is The Dress, and I did get it, so skill issue.
The shaded bottoms-now-tops of the small top right bowls couldnt be parallel to those on the other dishes either, so that might be what kicks you out of flipping fully.
This maps well to Michael Levin's model of memory (section 4 of https://www.mdpi.com/1099-4300/26/6/481 and https://thoughtforms.life/some-thoughts-on-memory-goals-and-universal-hacking/, also talked about in some podcasts and presentations he has done). In this model, memory isn't some high fidelity thing, but rather a highly compressed/encoded bit of information that, when run through a specific decoder, will result in a particular memory. What is key/critical here is that the decoder plays a very significant role in determining how that compressed information is decoded.
If you encode a memory using decoder/encoder (coder) A, and then decode it using coder A, you'll get out basically the same information that went in. However, if you decode it using coder B you'll get out different information than was put in.
Under normal circumstances, your coder changes very slowly over time as you experience life, and during that time you are constantly remembering things which causes them to be decoded and then re-encoded (so to speak) using the latest version of your coder. However, one can imagine a person having two distinct coders in their head at the same time, and in such a situation encoding something with one coder and then decoding it using a different coder could result in complete failure to decode (no memory), or a different memory than what actually occurred.
This maps well both to the hypnosis situation you have described here, as well as things like split personalities. An individual may remember things differently than they actually occurred because the coder used to encode the event is not the coder used to decode the event, or they may "forget" what happened. It can also explain things like being able to remember correctly if you are put back under hypnosis or if your personality switches back to the one that originally experienced it.
I think it also may have some explanatory power for retrograde amnesia. This could be caused by damage to the coder that needs to be "rebuilt" or "repaired", which is why your memories will sometimes return later. Perhaps older memories are encoded in a more "generic" way, while newer memories are encoded in a way that are are more tightly coupled with the current version of your brain's coder.
That’s interesting. That also explains why, when thinking about things I did in the early 2000s, I sometimes remember myself having looked up directions on my smartphone, because that sort of behavior is part of “find out where I am and where I’m going” in the decoder I’ve used for the past fifteen years, even though it wasn’t part at the time I formed the memory.
This feels like a really great companion piece to a lot of what Sam Harris writes in his book "Waking Up".
One of the claims in there, made by him and other meditators, is that meditating on your own thoughts enough or in a particular way eventually causes the illusion of the Self to disappear ("You search for it and fail to find it").
In that book, the perspective shift is just given as an empirical claim to tested by the reader, via meditation practice. In this post, it feels like we get a satisfying and plausible mechanistic theory for it.
In Buddhist meditation, you are supposed not to find it. In vedantic mediation , you are supposed to realise it is the entity doing the looking
I think that's a very convincing interpretation of self-consciouness/ ego. It's like concurrent interpretations of quantum mechanics : for some like relational quantum mechanics (RQM) vs Everett's many worlds, it's unlikely that you could experimentally prove one over the other but your mind can switch from one to the other to explain what you see. It could be a quasi solution to the so-said "hard problem" of consciousness. If consciousness is itself an interpretative model, a paradigm, it's no surprise you can't really prove it nor infirm it, or reach any kind of definitive answer on that subject. Now the question arises : is the Buddha a philosophical zombie ?
By the by, but Deutsch quite strongly believes that the truth of Everett's MW is experimentally provable. I don't have a strong enough physics background to evaluate the claim but I think it also rests on his disagreements with Bayesian approaches to probability (which itself is reinforced by his MW views). If someone knows more please chime in.
Edit: If I recall correctly his possible proof relies on self-reporting quantum computers- but the point is that he believes it is absolutely provable *in theory*, which I assume many would dispute. Whether that can disprove RQM specifically, I'm unsure.
I'm not sure either but I wonder if the idea isn't to experimentally disprove wave packet collapsing. If that so it would not disprove RQM that is also an interpretation without collapsing.
Quantum computation works by coherent superposition, but that doesn't get you all the way to many worlds as usually understood
The first thing to note is that MWI is more than one theory. What splittng is...how complete and irrevocable it is ... varies between particular theories. So does the rate of splitting, so does the mechanism of splitting.
The second thing to note is that many worlders are pointing at something implied the physical formalism and saying "that's a world"....but whether it qualifies as a world is a separate question from whether it's in the formalism , and a separate *kind* of question, from whether it is really there in the formalism. One would expect a world, or universe, to be large, stable, non-interacting, and so on . It's possible to have a theory that has collapse , without having worlds. A successful MWI needs to jump three hurdles: empirical correctness.
mathematical correctness and conceptual correctness -- actually having worlds
---- Coherence versus Decoherence----
There is an approach to MWI based on coherent superpositions, and a version based on decoherence. These are (for all practical purposes) incompatible opposites, but are treated as interchangeable in Yudkowsky's writings. . Decoherent branches are large, stable, non interacting and irreversible...everything that would be intuitively expected of a "world". But there is no empirical evidence for them (in the plural) , nor are they obviously supported by the core mathematics of quantum mechanics, the Schrödinger equation.Coherent superpositions are small scale , down to single particles,possibly observer dependent, reversible, and continue to interact (strictly speaking , interfere) after "splitting". The last point is particularly problematical. because if large scale coherent superposition exist , that would create naked eye macrocsopic scale:, e.g. ghostly traces of a world where the Nazis won.
We have evidence of small scale coherent superposition, since a number of observed quantum.effects depend on it, and we have evidence of decoherence, since complex superposition are difficult to maintain. What we don't have evidence of is decoherence into multiple branches. From the theoretical perspective, decoherence is a complex , entropy like process which occurs when a complex system interacts with its environment. [Decoherence isn't simple)(https://www.lesswrong.com/posts/Atu4teGvob5vKvEAF/decoherence-is-simple). But without decoherence, MW doesn't match observation. So there is no theory of MW that is both simple and empirically adequate, contra Yudkowsky and Deutsch
Reminds me of Steven Lehar's work, especially his cartoon epistemology is a wonderful introduction to our perception and the homunculus: https://qualiacomputing.com/2022/12/28/cartoon-epistemology-by-steven-lehar-2003/
I had never heard of Lehar or this site and just checked it out. Thank you for the link! (And I wonder what my eight-year-old will think about it when I share it with him...)
I also enjoyed his blog on the perceptual origins of mathematics: https://perceptualoriginsofmath.wordpress.com/chapter-1/
and his Harmonic Gestalt YouTube video, and QualiaComputing has lots of interesting further reading material.
This made me think of how quantum spin in entangled particles can collapse and reverse when observed and then the spin reverses even if the objects are far distances.
I don't know if this has any experimental support I could point to, and it's been a long time since I read it, but Douglas Hofstadter wrote (in either The Mind's I or I Am A Strangle Loop, I don't remember which) that the 'self' concept might be like the "this" pointer in a data structure in C++. Literally just a pointer that points to its own address, but otherwise empty.
In which case, sure, we tie a lot of things to our identity, but they aren't part of the core 'self' concept. And the instant bistable switch of satori is something like noticing this fact?
Great stuff, my qualm is just that even saying perception is indirect is kind of wrong. Suppose there's video evidence in a courtroom, if someone points to the screen and says "that's the killer right there," are they wrong? (And a video isn't much different from any internal model.)
Or to put it another way, the idea that we're "perceiving a model" isn't really correct. We're HAVING a model and at all times directly perceiving a WORLD (and that world as it really, really is, to boot, see bottom of this post).
The trick is, I think, that words like "seeing" and "perceiving" are third person terms - so they're something a third person observer says when (e.g.) they see a tiger perceiving its prey, they're part of that objective, third person STORY.
What the tiger experiences directly (its internal model of the prey) isn't itself yet a perception, it's only granted the grace of being trully callable a "perception" IF AND WHEN a third person can observe both the tiger licking its lips in anticipation, and that there is actually a deer over there.
So neither of the "flips" are a perception until there's been a checking process, and that checking process doesn't change the subjective experience, it just corrects an error in the truth claim about what that subjective experience portended.
We are always and at all times directly perceiving the world, because we are always and at all times perceiving an objective "slice" of the world that's a secret treasure that the world gives of itself to us, and to nobody else, at just that point in time and space.
(cf. "Process Externalism" (Manzotti), "Affordances" (JJ Gibson), etc.)
No offense, but this point of view is called ‘naive realism’ by the majority of cognitive scientists.
The explanation of hypnotism feels completely unsatisfying to me. If you ever get an opportunity to talk with a friend who has been deeply hypnotized, you learn there is something else going on. It is like a bizarre co-opted dream state. The subject isn’t just going along with the hypnotist because they believe they’ve lost control of their body; that’s not how it works at all. The subject completely, irrationally believes the scenarios introduced by the hypnotist, and cannot be convinced otherwise.
Relevant to the multiple-personality angle, tulpamancy (qua mental hobby, not supposedly-supernatural woo) is specifically the practice of training yourself to "flip" to an intuition that some of your thoughts come from a second homunculus on purpose, consciously, aware that what you're doing is flipping some switches back and forth. It does indeed feel very much like flipping from duck to rabbit etc.
by feel, do you mean that the concept is reminiscent of it, or that you have tried tulpamancy to some extent?
Interesting! I greatly enjoyed Jaynes’s book but always thought he oversold his idea by claiming that people in pre-Classical times were not conscious; it seemed to me more likely that they conceived a new understanding of consciousness, maybe analogous to Adam Smith’s new understanding of commerce.
Jaynes did his best to convince but in the end I put it in the ‘likely malarkey’ file.
I became convinced that something like the homunculus account of consciousness must be the case back when I used to rock climb (before I broke my neck two years ago in, erm, a rock climbing accident).
On many occasions, I would observe my body begin to execute physically demanding, precise technical moves in fear inducing situations before I had consciously "told" my body to start moving. Of course my conscious mind would rapidly catch up and experience between conscious awareness and movement would quickly become synchronic. But it happened so often I became quite familiar with the thought, "oh we're going are we?" as I began to execute the required physical moves.
The way I came to think of "myself" is as being like Maggie Simpson in the introduction to The Simpsons. She steers the steering wheel and beeps the horn, and is under the illusion that she is driving the car. But really, Marge is driving the car. Maggie as homunculus.
I'm sure this must be very common, I can perceive it whenever I do a sustainable fast-walk pace for some period of time (say 4mph for 15+ minutes) then get off the treadmill and my legs just seem to move on their own. It sort of feels like you're just "steering" the direction of a Flintstones car from the homunculus seat. Being that it's a compound movement, and not merely a steady contraction/expansion of some muscle fiber, it's makes less sense to be an involuntary motion than the "lean against this wall while you try to raise your arm against it for 30 seconds then step away" thing everyone knows from grade school.
Trying to mess around with one's perception of consciousness on purpose seems just as dangerous as rock climbing in its own way. Seems like an easy way to break your brain. (Sorry to hear about your accident, and hope you were able to make a recovery.)
Well, I'm a quadriplegic now, so not much of a recovery. But given that by any reasonable metric I should have died, some would say I should be grateful.
The essay feels a bit incomplete to me. Particularly the ending, I actually wasn't even sure it was the ending when I was reading the blog on Gmail.
As for the topic itself, I dunno, while I don't find anything particularly objectionable it doesn't really seem like this model gets us anything. What are the benefits of this over a model like Plato's tripartite soul? I think this is what's missing from this essay is more commentary on the other theories of consciousness and how Byrnes' theory differs.
This model-flipping optical-illusion account fits my experience with Mahamudra (a Vajrayana Buddhist meditation practice). We often meditate with live instructions, which often have the form "take the view that..." (except it's about mental phenomena and the architecture of awareness, rather than twirling dancers).
Thomas Metzinger has a version of this account which he calls the "transparent self-model" (explained in his book "Being No One" or any number of YouTube lectures). He distinguishes between internal models that we see as models, vs. "transparent" models where we relate to the model contents *as if* it were the real world. On this view (as philosophers like to say), a first-person perspective requires we have a *transparent* world-model which includes the self as a specially-indexed object.
This transparency idea highlights how weird some of these "model-flipping" practices can get, especially with the self-model. It's not abstract propositional beliefs that are changing (which we're used to as rationalists), rather our experience of the world.
I think it's well-established that stage hypnosis is basically fake -- people play along because it would be awkward not to, nothing more. So using stage hypnosis (a fake thing) as evidence for this theory is kind of anti-convincing to me.
I was thinking about your review of Jaynes the whole time, I was very glad that it was eventually mentioned. That’s the most interesting part to me: just like in the optical illusions, the initial states are just the most common and most likely to make sense (upside down stairs are rarer than regular ones), but when it comes to our mind, there’s not as many guardrails. If all or most of us end up with roughly the same homunculus model, it’s probably because of the way we hear our parents talk about their own self-perception, but it’s not hard at all to imagine how different cultures could end up with very different models.
I completely understand your point about skipping the free speech part, but it still makes me a bit sad. People really add a whole lot of crazy epicycles to their world models to make space for it. I think it would be very healthy for a lot of people if we finally learned to deal with that idea.
Since I am autistically obsessed with language, I will point out that this
>the predictive processing interpretation is that we perceive our models of the world.
is misleading, insofar as it's Scott's interpretation or whether people working in predictive processing all agree with it. Perceive is an ordinary language world defined by its usages in ordinary contexts. We do perceive the world, because that is the correct way to use the word as it is embedded in everyday life. We do not perceive a model of the world, because "models" aren't imposed between us and the world in our ordinary linguistic behavior.
You might say that perception consists in our brains interpreting various other cognitive models and so we can be wrong about what the world is like - but it's strictly speaking false to say that we perceive those models. Nobody ever speaks of perceiving models irl when they see an illusion or hallucination or regular apple or whatever.
Now, since this is just an autistic objection about language use, it could be immaterial. But it could also mislead people by appropriating a technical term and giving people the misleading impression that there is the world, then their model, and what they perceive is the model, when that's not right. Perception is a complex behavior and can be given deep explanations, but it is not accurate to use the word that what you perceive is sense data or qualia or an inner world or whatever, because nobody irl speaks this way and the technical usage potentially or implicitly adopted here is not congruent with other ordinary usages.
> If the homunculus-self is a mostly-accurate but not-directly-perceived-and-real model of mental processes, then a person whose mental processes often flip between two or more dramatically different states (for example, borderlines, who are notable for very strong emotional states and “splitting”) may gather evidence for and eventually flip to a model of themselves as multiple different homunculi. This is especially true if they’re primed with the suggestion that this is a likely way for the inside of their mind to be (for example, by a psychotherapist who believes in multiple personalities).
An uninformed take like this makes me wonder if Scott is just clueless about DID specifically or is everything else of his I am reading and nodding along is subject to the Gell-Mann Amnesia effect.
The most unfortunate part is him uncritically repeating the discredited iatrogenic model of DID in the parentheses.
This topic deserves a much more in-depth discussion, suffice it to say for now that not one person with DID (as opposed to the ticktok multiplicity crowd) want to have it. They desperately want to "flip to a model of themselves" where they have a single unfractured self. Most try very hard to do so, repeatedly and in various ways, invariably failing. Most people with clinical DID symptoms have multiple nuanced alters and not the media-beloved one good and one evil, or the Borderline "strong emotional states".