10 ms·
I have a PhD in biology and I work in machine learning today.
After years of study my takeaway is that reductionist analogies are helpful for giving folks some
I have a PhD in biology and I work in machine learning today.
After years of study my takeaway is that reductionist analogies are helpful for giving folks something to grasp on, but at a certain level, you have to step back and appreciate the complexity of certain things. At some point, only machines will be able to fully grok what is actually going on. Does that mean reductionist thinking is bad though?
How does GPT4 think? A piece of data getting multiplied by a gazillion parameters you will never understand.
How does one cell become a human? A gazillion chemical reactions happen in parallel you will never understand.
What is consciousness? The firing of a gazillion synapses you will never fully understand.
The three challenging phenomena I described may seem intractable to reductionism, but have seen the greatest progress from reductionist approaches. Most of the greatest and most useful knowledge will be very reductionist in nature. How many transcription factors are needed to turn this skin cell into a stem cell? About 4. What is the best architecture for machine translation? Fits on the back if a postcard? What chemical transmitter underpins human happiness? Can be synthetically manufactured in your mother's basement...
Can a reductionist approach be good enough to control life independently? Or do we need to emulate the full process? I am not sure if the point of the article is to bash reductionism, but maybe push for full understanding. Kind of like Newton's law fits well within Genera Relativity. A first step.
>> to push for full understanding.
The point of the GP is that that full understanding can't happen in a human brain, but reductionism may be a lever good enough for that human brain to use.
>> do we need to emulate the full process?
Probably not, but let's assume 'yes'. It seems likely that it would possible to optimize the emulation to a point where it becomes practical. We do something similar today with weather simulations; it's not possible to emulate each molecule in the air, and yet meteorological models manage to account for an insane amount of processes and--short term--produce accurate predictions.
In a more subjective note, I believe the methods and ways of thinking in biology need to shift from "let's understand everything" to "let's produce a forecast and maybe fix some stuff". Pharma companies are already in that track, but it could be that the general public and people in academia are still in the full-complete-very-detailed-understanding wagon.
Aside from consciousness, you can understand those processes by studying the respective fields. It's true for non-experts that don't have the foundational knowledge.
I believe the argument is not that any part of those fields is incognoscible - "too complex for a human brain to grasp", but instead that the whole field is massive - "the amount of assertions and relationships required to understand the whole process is bigger than what the human brain can hold". That it is a problem of quantity, not quality; that even experts will need to abstract away and statistical average many details on these fields.
They're incognoscible in the sense that our models are very complicated metaphors, and that meta-modelling process runs out of steam at some level which we don't understand.
It's philosophical Dunning-Kruger. We don't know what we don't know. Our models are patchwork associations that have some experiential consistency, and appear "logical" based on our subjective experience of what logic is.
But that's all they are. There is no automatic implication they're complete, or even that it's possible for them to be complete.
We usually assume that limitation doesn't exist. But that seems very naive and optimistic.
I see, in that case I would grant it for biology but not so much for AI, at least not yet.
> What is consciousness? The firing of a gazillion synapses you will never fully understand.
This doesn't fit with the other examples. Consciousness is not a problem of scale, but a problem of kind. We can see from simple models how you might make things more complex until they become something like GPT 4.
So too with human cells, we can see analogous engineering feats that make it possible to imagine that there could be a way to have a system that can produce a human from small beginnings. Again, simple systems show us how the complex might be possible.
But there are no simple building blocks that we can point to and say "yes, I can see how more of these would produce consciousness".
Reductionism is useful because we need to find manageable pieces of the system so we can model it, and manipulate it. Without reductionism we wouldn't have modern medicine.
It is important to acknowledge tho that it can never produce the whole story. There's nothing wrong with it. It's just not sufficient for a complete understanding.
I don't have a PhD in biology (or in anything), but isn't temperature a single number "explaining" a gazillion interactions we will never "understand"?
In biology, being amenable to reductionist explanations is an evolvability trait. Repeating building blocks can be re-used, combined and repurposed with small changes. That's the fundamental reason why so much in biology does make some sense. On the other hand, we are of course biased towards understanding things that are easy to understand. Who knows how many things are overlooked because they're just too messy!
> How does GPT4 think?
It doesn't think. It only processes.
And as far as I grasp it, its process can essentially be understood, not least because they were designed and documented.
It is several big mistakes to use "you will not understand" complexity mysticism equally when referring to biological processes and human-designed computer processes. Not the least of which that it is patronising and reductive.
That's right. It is not human.
How do humans think? We do not know.
AI is merely a facsimile of human thought. We don't know how that works, but we can certainly make a cult of it.
"It doesn't think. It only processes."
Human thinking is frequently described as 'processing', a 'process'.
Guess, that is problem with using Metaphors and analogies.
When it comes to Brains and AI, a lot of the same metaphors and analogies have been used over the centuries, so now, which one is 'more accurate'. Can't really say, it depends on what each person meant by that metaphor.
> What is consciousness? The firing of a gazillion synapses you will never fully understand.
But then you realize that transcripton factors (genes that activate or inhibit other genes) form networks with feedback loops that are conceptually equivalent to neural networks, and some cells grow camera-like eyes complete with a retina and a lens [1].
It isn't about neurons per se, more about the way information is treated by these systems.
1. https://en.wikipedia.org/wiki/Ocelloid https://en.wikipedia.org/wiki/Ocelloid
What makes you think that a machine any less complex than the entire organism is capable of "grokking what is actually going on"?
I agree with you that humans may never be able to understand most of what's going on in biology. But I'm even less enthusiastic about gradient descent. The interesting results in machine learning for biology all seem to amount to better ways of indexing the data we already have (like crystallographically-discovered protein shapes) and interpolating between the existing matches nearest to a query term.
I don't claim to know what "understanding" really truly is, but I do know what it isn't.
> What is consciousness? The firing of a gazillion synapses you will never fully understand.
Hmm. How do you know this is the case if you don't understand it? Magical thinking?
Complexity is not useful because it can't be used to fix things
That’s basically a cop out I think it’s called ? Some kind of cheap magic trick fooling the audience by pretending that if we haven’t explained xyz it’s only because of complexity.
This kind of thinking is deeply entrenched in physicalism.
And please, GPT does not think.
As Alan Watts beautifully said « science is the art of prediction « To forget that is naïve at best (meaning, any amount of untangling the complexity will not explain where the universe comes from, nor how consciousness arises from inert matter)
> Most of the greatest and most useful knowledge will be very reductionist in nature.
More likely the greatest and most useful knowledge will simply not be understandable by our primate brains.
We'll either upgrade our brains to understand, build machines to understand it, or it'll be left not understood.
It's a miracle our primate brains can understand relativity or calculus.
[dead]
This actually creates a good argument for considering current LLMs and other generative AI as steps in the right direction. Specifically, while we have "no idea of what basic organisation (of synapses) it stems from", we do know a few things about this organization which creates consciousness - such as:
- It is simple enough to be reachable by a dumb, brute-force, random-walking, incremental process of evolution;
- It scales well, conferring survival benefits from the start, and at every point of the way, from the first neuron, up to the human brain;
With those in mind, it absolutely makes sense for us to find a relatively simple compute/organizational system and scale it all the way to consciousness (or a set of those that scale together) - because that's exactly how evolution must have done it, since it's structurally the only thing it can do.
> It is simple enough to be reachable by a dumb, brute-force, random-walking, incremental process of evolution;
I'm not an expert in this field, but from what I heard brute-force natural selection is the naive explanation we're given in school. There are many more factors at play other than random chance: e.g. there is also sexsual selection. Sexual selection selects some characteristic that isn't necessarily an advantage in the current environment, but it is somehow preferred by the opposed sex. According to some research, the reason why we lost the bone that other primates have in their penis is due to sexual selection.
One intuition that helps is to see evolution as a random search through solution space. Another thing to realize is that -in evolutionary algorithms- the distribution of random trials will be strongly biased around existing solutions.
It helps to understand that in some situations, a search algorithm with some level of randomness can arrive at a solution faster than a systematic/random approach, on average. In other situations, a search with added randomization might be slower, but its ability to escape local optimae (to some degree) means that it is much more likely to find the global optimum (or at least a better local optimum :-P ).
Compare also: Monte-Carlo methods.
https://en.wikipedia.org/wiki/Evolutionary_algorithm https://en.wikipedia.org/wiki/Evolutionary_algorithm
Interesting points. But I would add that sexual selection can select for some trait that is actually counterproductive, a famous example being peacock tails. Female peacocks use them to evaluate the health of potential partners, so it is an indirect measure of fitness to the environment. But at the same time, it is clear that healthy peacocks would still be better off without carrying around such big tails. If tails weren't so important for reproduction, they would have likely shrunk by now. I wonder how a comparison with the methods you mentioned would capture this.
Having done a bit of biology, I would not want to use the word "evolution" and "simple" in the same sentence. ;-) Certainly not things that have evolved over many generations.
(Even artificial) evolution sometimes does really weird things.
https://www.cse.unsw.edu.au/~cs4601/refs/papers/es97thompson.pdf https://www.cse.unsw.edu.au/~cs4601/refs/papers/es97thompson... (page 399) "Fig 7 [...] The cells
shaded gray cannot be clamped without degrading performance, even though there is no connected path by which they could influence the output"
I don't mean to diss evolution. It is simple in terms of a program, the simplest it could possibly be - but it's also massively parallel; every molecule of or around any living thing participates in it. All life, constantly, everywhere, all at once. And then it had literal billions of year to run on this rock, to get to the point of complex life in complex environment.
Like a brute-force enumeration running on a supercomputing platform, it's mighty, but it's also simple.
Also, the process is simple. The output not so much.
We both agree on the output being non-simple and, to be fair, that's what I was thinking about when I wrote it. ;)
Taking it a step further though; Of course evolution happened to the very mechanisms by which biological organisms evolve.
Because of course it would.
(eg.
* Eukaryote -> Meiosis : This is obviously a means to create/tailor Variation. (Evolution ~= Variation->Selection )
* Reproduction in Metazoa in general, with all kinds of complex methods to exchange DNA.
* Vertebrates. Just. Vertebrates. (specifically: Sexual reproduction thereof.)
* Simple but still strange: Viruses with multiple reading frames. Sure they successfully evolve, but it's never going to be straightforward!
)
So I was watching a clip about black holes and entropy today that seems to fit in this somewhat. That complexity behaves like entropy.
https://www.youtube.com/watch?v=yLOHdW7dLug https://www.youtube.com/watch?v=yLOHdW7dLug
It’s also very much not just about the neuronal synapses.
I agree with you that the brain is nothing else but its physical makeup (whether that's synapses or more). But that's not the point of contention -- the point of contention is whether those synapses are enough to explain (or explain away) consciousness. And to me it is plainly obvious that physical stuff (synapses or whatever) can never explain (or explain away) consciousness.
(side note: there is absolutely, undeniably, a correlation between brain states and mental states -- but that does not entail that mental states reduce to or are explained entirely by brain states)
> there is absolutely, undeniably, a correlation between brain states and mental states -- but that does not entail that mental states reduce to or are explained entirely by brain states
It's not plain correlation. It's long been experimentally demonstrated that you can circuit-bend[0] an animal or a human live, by poking in their brain[1]. If in those cases, altering brain state creates/affects behavior and mental states, then there isn't really a place for a non-physical-stuff consciousness. If it existed, it's demonstrably overridable by purely physical interventions, which implies you could just make a physical-stuff consciousness, at which point why postulate a non-material one?
--
[0] - https://en.wikipedia.org/wiki/Circuit_bending https://en.wikipedia.org/wiki/Circuit_bending
[1] - Then there's obviously alcohol, tobacco, and other drugs, that have a slower onset.
You can give alcohol to a driver and it will affect their driving.
However, if you're an alien and you don't know the exact workings, you might conclude that alcohol affects cars.
Hence my point about circuit bending. We're at the stage equivalent to that alien a) being able to electroshock the driver, and b) observe them getting out of car and functioning independently, therefore c) conclude that the driver is the thing controlling the car.
Poking at the brain or consuming certain substances causes changes in the phenomena experienced in consciousness.
But doesn’t seem in any way explanatory of consciousness itself, i.e. the owner of the brain’s awareness of those state changes.
I think this points at the somewhat nebulous definition of consciousness, which at its core seems categorically separate from the phenomena that can be experienced (and manipulated by poking at the brain) within it.
This is why some have theorized that everything is conscious, and the human experience of consciousness just happens to be particularly rich and complex due to the complexity of the human organism. This would suggest that a sufficiently advanced computer could also be conscious, with the specific experience mediated by the architecture of the computer.
(I’m not endorsing this position, but it’s an interesting thought experiment to help separate experiences in consciousness from consciousness itself).
I never denied the claim that changing the brain can change mental states. I think that the experiences we have are ones we have because of physical states (primarily, the brain). I just do not think that the experiences themselves are physical things, and on a purely naturalistic explanation of the universe experiences would not exist.
You are right that on naturalism there isn't really a place for non-physical-stuff consciousness. And yet, the existence of non-physical-stuff is the one thing I am most sure of, the one and only thing I have access to. So the challenge is to explain it (or explain it away). As you say, there's no place for it in a naturalist view of the world, and I agree.
By analogy/homology:
Technically, software isn't exactly physical either, of course. It works with information contained in physical states, but isn't -technically- itself physical.
Arguably this can be demonstrated by moving it to a different physical representation (magnetic, optical, diode charges, dynamic memory cells; little endian, big endian; AMD, Intel, ARM, Turing Machine In Game Of Life, Minecraft Turing Machine) without affecting its behavior or functionality.
I don't think there's anything particularly mysterious about this kind of phenomenon, per-se.
You could grok the behavior of the software by poking at the hardware its running on for long enough; but it might be more useful to split the problem into separate layers that are easier to reason about. (Hence the need for reduction and emergence)
> I just do not think that the experiences themselves are physical things, and on a purely naturalistic explanation of the universe experiences would not exist.
But they could be. Elsewhere in this thread, I mentioned computer science is effectively a branch of physics now, because we can connect computation with thermodynamics, and e.g. put lower bounds on amount of energy required to perform certain computation. Therefore, if you consider experiences to be "execution", some specific computations done by our brains, then they still are physical things. Might be tricky to point and prod at them in the structure of our brain, but physical concepts that are smeared over space or time are nothing new - think e.g. waves, whether EM waves or mechanical waves or virtual waves, think of the boiling soup they make out of their medium, and how e.g. Fourier transform can tease them all apart anyway.
> If in those cases, altering brain state creates/affects behavior and mental states, then there isn't really a place for a non-physical-stuff consciousness.
This does not follow. The altering of brain state affects behavior; from this we can deduce that there is a correlation between the two, and that there is a causal link between the two. We can't deduce from this that they're the same thing or that one is reducible to the other: this is illogical.
More generally, it is illogical to claim that brain states are the same as mental states (thoughts). There are several reasons for this. One is that reducing thoughts to brain states means a thought cannot be correct or incorrect. For example, one series of mental states leads to the thought "2+2=4"; another series leads to the thought "2+2=5". The correctness of the former and the wrongness of the latter refers only to the thought's content, not the physical brain state. If thoughts are nothing more than brain states, it's meaningless to say that one thought is correct -- that is to say, it's a thought that conforms to reality -- and that the other is incorrect. A particular state of neurons and chemicals cannot per se be incorrect or incorrect. If one thought is right (about reality) and another thought is wrong (not about reality), then there must be aspects of thought that are distinct from the physical state of the brain.
If it's meaningless to say that one thought is correct and another is incorrect, then of course nothing we think or say has any connection to reality. Hence the existence of this disagreement, along with the belief that one of us is right and the other wrong, presupposes that your position is wrong.
No amount of neurological study will undermine this argument. More generally, describing one aspect of reality in greater and greater detail doesn't gainsay those who claim there are other aspects. You can describe the inner workings of the brain in as much detail as you like -- it is nonetheless logically certain that human thought is not reducible to brain states, even though it obviously correlates with them.
> One is that reducing thoughts to brain states means a thought cannot be correct or incorrect. For example, one series of mental states leads to the thought "2+2=4"; another series leads to the thought "2+2=5". The correctness of the former and the wrongness of the latter refers only to the thought's content, not the physical brain state.
> If one thought is right (about reality) and another thought is wrong (not about reality), then there must be aspects of thought that are distinct from the physical state of the brain.
Of course there is. Computation. Observation. A correctness of a thought isn't a function of thought itself, it's a function of thought and reality. Correctness of a thought is not a physical label. It's not a state. It's something you evaluate, something you compute with respect to knowledge you observe.
The way I see it, you're confusing a thought with computed properties of it. There is no privileged notion of correctness and nature does not put XML tags with correctness score on neurons.
> (side note: there is absolutely, undeniably, a correlation between brain states and mental states -- but that does not entail that mental states reduce to or are explained entirely by brain states)
How does this make sense? There is some divine being that only sometimes looks at the actual synapses to decide if it should feel nauseous or happy?
No. Nauseousness or happiness are states that can only happen when mind and matter are intertwined in the form of a living brain.
Much like there is no divine being that decides "I'm going to make this dense star a black hole", a black hole and the attached phenomena are consequences of matter and forces interacting.
Emotions are thoughts (brains and the attached phenomena) are consequences of matter and mind interacting.
What OP is describing is the strong form of emergence, that emergent behavior cannot be fully understood or predicted by understanding each of the individual parts. Aka the whole is more than the sum of the parts. If you fully understood the laws of physics at an atomic level, would it be sufficient to recognize a cpu as turing complete if you see one?
It's a debatable position in philosophy against the reductionism OP is describing, and maybe a bit magical but I'd find it the more interesting of the two positions to hold.
So reduction and emergence are obviously each other's complement. (A bit like differentiation and integration)
> If you fully understood the laws of physics at an atomic level, would it be sufficient to recognize a cpu as turing complete if you see one?
This is testable if you take an alternate universe with very few laws. Somewhat canonically, do you understand a Turing machine built in Conway's Game of Life if you understand the 2 rules of that game. (Or 4 rules, depending on how you count).
The only way to understand it is to play the game.
If a group of scientists in 1923 were given a working iPhone, would they have a chance at figuring out how the ChatGPT icon is able to launch something that talks or where Grand Theft Auto is in the machine?
Is trying to figure out how the brain works by looking at synapses and blood flow and electric fields similar to trying to see how an iPhone works by looking at transistors and electric fields?
Others have had similar lines of thought!
* https://www.cell.com/cancer-cell/fulltext/S1535-6108%2802%2900133-2 https://www.cell.com/cancer-cell/fulltext/S1535-6108%2802%29... Can a biologist fix a radio?
* https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1005268 https://journals.plos.org/ploscompbiol/article?id=10.1371/jo... Could a Neuroscientist Understand a Microprocessor?
This is a crucial point to make. Even if every natural phenomenon can be reduced to very early-level processes (e.g., physics), earlier levels of analysis have little predictive value at later levels of analysis.
Does anyone have a term for this from philosophy of science literature?
If I understand what you’re asking, the term would be emergence. In philosophy this means that that behavior, property, or attribute arises from parts or components that do not show that behavior, property, or attribute themselves either in part or in whole.
I don't know the term for it, but it's the most common non-strawman version of the anti-reductionist argument.
If you want to ruin your day, you can try and parse through the SEP page on reductionism. It's a doozy. https://plato.stanford.edu/entries/scientific-reduction/ https://plato.stanford.edu/entries/scientific-reduction/
3y ago
> the brain is nothing else but its physical makeup
> physical stuff can never explain consciousness
These statements seem contradictory to me, can you elaborate? Also, you don't provide your reasoning for the second statement. Why do you believe that physical stuff can't cause consciousness? What is your definition of "physical"?
I wholeheartedly agree. For those interested check out the hard problem of consciousness
And I have a thought I would like to share here, an attempt at arguing for why it might be impossible for a physical pure materialistic model of the world to explain consciousness
In a pure physical materialistic world, we can represent everything as laws, interactions between materials, guided by a set of rules. That's it.
If we had a absolute knowledge, we can represent those rules, all of them, as a set of mathematical and logical equations.
I reduce the material world to a mathematical world of rules and laws, an abstract world.
Saying consciousness could arrise from a pure material world, is akin to saying the mathematical equations and laws, combined, can be conscious. As in, there is a system of rules that if tuned in a certain way can think. Since rules are abstract, and by definition rules don't think. No combination of them no matter how complicated can think
The initial rebuttle I have in my mind is that "well you are conflicting the system with the objects governed by it", I feel there is a better way to phrase the initial proposal to address this issue, but I just wanted to throw this idea out there
I think you're on the right track here. I usually try to reason down a similar line. I wouldn't say that the claim is that the equations and laws combined can be conscious. Instead, I would say that none of the equations or laws that we come up with involve mental attributes (such as consciousness or qualia) as either inputs or outputs.
You look at all the proposed laws of the universe, and they involve things like energy, mass, time, position, work, etc, as inputs or outputs. No equations have inputs of these sorts of natural quantities and outputs that entail consciousness or that first-person experience.
We can see from these equations how things like tree leaves blowing in the wind, or mountains, might arise. These things involve motions and mass and things interacting with each other. While we might not be able to paint the full picture, we can see that the inputs and outputs of those equations are the right kinds of things that might plausibly lead to mountains and leaves blowing in the wind, at the right scale.
We can't see from these equations anything that might plausibly lead to the first-person experience, the what-it's-like to be something.
Note that the naturalist cannot then just inject these mental/consciousness related parameters into their equations. That would amount to a form of dualism, and destroy their project. What I think the naturalist needs to do is somehow explain away consciousness, because I don't see any path they could take to explain it in purely natural terms. They need to deny that there's anything here that needs explaining at all.
That’s not true at all. The most interesting research into components of consciousness involve astrocytes, glia, and formerly assumed “helper cells” throughout the brain. It isn’t “all synapses.”
Source: neuroscience MS (dropped out after defending to become a software engineer)
This is unacceptable level of reasoning about such a topic, almost comically illiterate.
The moment you've used the term “brain”, you'd already detached that thing from the rest of the body (without which it can't function — just like a spherical cow can't really live in a vacuum), drawn its borders, set limits to its functions, and so on. The same with “synapses”. So by stating that those mental objects are “real”, you've simply found that X equals X, that those things are exactly the same as they are defined. Their evident presence here is no different from the evident presence of demons inside the possessed people in the past.
But unless you're a dualist, you have to admit that our biological brain is a massive bundle of cells from which consciousness somehow arises. It must be those mushy tangled cells, there's nothing else there. It probably is a problem of scale along with a philosophical problem around consciousness likely not quite being what we intuitively imagine it to be. (e.g. visual consciousness seems continuous but we know from saccades that it can't be)
Consciousness is unique, because it gives rise to the problems of qualia. None of the other examples face this issue. A.k.a. "Why is there a "likeness" of subjective experience?" and "why am I me and not somebody else?". Saying that consciousness is merely a side effect of neurons kicks this proverbial can down the road.
> unless you're a dualist, you have to admit that our biological brain is a massive bundle of cells from which consciousness somehow arises
This is indeed a consequence of a physicalist/naturalist reductionist view. On that view, there is only natural entities and causes, and everything else reduces to just natural entities or causes. Therefore, as you imply, consciousness must somehow be explained by our biological brain or other natural entities/causes (note that I didn't say "arise", since some naturalists deny there's actually any such thing as 'consciousness' to be explained).
But you can also approach this from the other direction: it is impossible for naturalism to explain (or explain away) consciousness, and consciousness is real, therefore naturalism is false. That's my position. There isn't even the seed of an explanation given naturalism, and I think the naturalist's best position is to simply deny there's anything to explain. The whole project, given naturalism, is hopeless.
Again, it's not a problem of scale but rather kind. Naturalism just doesn't have the right building blocks to explain (or explain away) consciousness.
> But you can also approach this from the other direction: it is impossible for naturalism to explain (or explain away) consciousness, and consciousness is real, therefore naturalism is false
So you believe in a soul or what is the practical consequence?
> it is impossible for naturalism to explain (or explain away) consciousness, and consciousness is real, therefore naturalism is false. That's my position. There isn't even the seed of an explanation given naturalism, and I think the naturalist's best position is to simply deny there's anything to explain. The whole project, given naturalism, is hopeless.
Materialism (the much more common name of "naturalism") can, in principle, very much describe consciousness, if we understood the functionality of the brain enough. We already have an inkling, due to Turing: the human mind is quite possibly a computation, a mathematical object. Consciousness could then be an identifiable aspect of this computation.
I do agree though that in general, the materialist position is simply that consciousness is just not that deep a mystery, and some of the "deeper mysteries" like qualia are nothing special that even needs any explanation.
> there's nothing else there
By way of analogy, if a civilisation had found an antenna, but has not discovered radio they might think that the internet and it's content is coming from within the antenna, because that's all there is.
They may also have thought that the content is being fed in by a giant purple bunny that lives on the moon.
I was going to counter that we can circuit-bend humans enough to know that the processing is local, unless you're going to claim that every little piece of a human is connecting remotely to the magic-land - but then I remembered that this is exactly how "Internet of Things" starts to look like: tons of devices sitting next to each other, but for no fucking reason, talking through servers on the other side of the planet.
So who knows, maybe god is an adtech corporation.
You can also say 'consciousness doesn't exist, we're all P-zombies'.
Or even the more limited version: consciousness exists, but doesn't have a lot of the magical powers usually ascribed to it (free will, etc).
We currently don't have a definition of consciousness on which people agree and which allows you to determine whether something is conscious or not. It is therefore not surprising that you don't believe that scaling simple systems produces "consciousness", but unless you think that consciousness is some kind of magic that does not follow from the laws of physics, there must be some kind of reductionist thinking can explain the phenomenon, given sufficient computational power to do the math.
> But there are no simple building blocks that we can point to and say "yes, I can see how more of these would produce consciousness".
If you are open to have your view on this changed, I urge you to read "Vehicles: Experiments in Synthetic Psychology" by Valentino Braitenberg (1986).
Before we had LLM it wasn't at all obvious that the simple ones we had would work as well as those do.
It's not clear what you define consciousness to be but for general intelligence it's easy to imagine how something would would with enough resources.
We can imagine how animals could evolve to be smarter and smarter over time.
A old school symbolic AI with near infinite power could create and sci-fi tyle old school AI.
We can imagine a super advanced LLM mixed in with logic reasoning.
My own view is probably atypical among dualists and idealists, and hard to explain. I'll just state what I think is true, without justifying why I think these things are true or compatible with each other. I think:
- Most or all of what makes us human, like love, pain, sense of belonging, seeing, hearing, find almost all of their explanation in the physical. I include even things like thinking, reasoning, beliefs, and so on, as finding almost all their explanation in the brain.
- The experiential/phenomenological side of these things (love, pain, sense of belonging, seeing, hearing, etc) are non-physical mental/consciousness things. Everything else besides the phenomenology finds its explanation in our brains (or some wider physical states than just the brain).
- I am not surprised that LLM's could be as powerful as they are today (though I am surprised that I was alive to see that), and I can see them getting better.
- The foundation of reality is ultimately mental/consciousness and not physical, and the physical (to use a philosophical term) supervenes on the mental.
Regarding definition of consciousness: I'm not so bothered with how we define words. I think a lot of philosophy is pointless arguing that amounts to no more than people disagreeing about how we should define a word. I see no point in that. Instead, let's just stipulate what we mean by a word, and get on with the meat of the discussion. In this particular instance, when I talk about consciousness, I'm talking about the thing that the 'hard problem of consciousness' is talking about. I'm not talking about, e.g., consciousness in the sense of wakefulness (where being asleep or under a general anaesthetic would count as unconscious).
Before planes people imagined machines that would flap their wings like birds.
It's very possible you're wrong.
Yes. Now that we have machines that people are making claims like this for, we really should stop using the word "thinking" so lazily.
Human or animal thinking is a process, but it is an innate process of instinct, hypothesis, testing, prediction, overrule, and introspection. (Whether there is metacognition or not).
GPT "thinking" is only "thinking" in the colloquial sense of "wait a sec, the computer is thinking about it".
GPTs don't think. And they especially don't think in some imaginary complex way that we can't understand.
necessity and sufficiency is insufficient (and not necessary) in biology. I've attended talks where the system was complex enough that something (a molecular factor) was sufficient, but not necessary! Pure logic isn't useful in biology because it's a feedback-controlled system with an enormous number of internal states.
Does it not illustrate that the person making the assertion either hasn't considered that their explanation is not necessarily exhaustive (in fact, as opposed to according to current theories within a popular, not entirely genuine/consistent in behavior at this point in time ideology), or would like you to believe that it is?
If there is a way to prove it is necessarily exhaustive, I would like to hear it.
Not quite. What you prove by circuit bending is that you can alter behavior through physical means, no one's denying that. But the conclusion that because of that, then the physical means are also creating consciousness, that's not proven.
It's like assuming that the driver creates a spark in order to initiate combustion within the car, because the aliens haven't observed an instance of the car starting without the driver present.
> Therefore, if you consider experiences to be "execution", some specific computations done by our brains[...].
Which it seems to me to boild down to the assumption that intelligence and/or consciousness are computable. Which is a possibility, but we should keep in mind that we don't have any proof of this yet, only plausible-sounding arguments.
> if you consider experiences to be "execution"...
Why would anyone think that is true? The problem I find with responses like yours is that I think you haven't understood or fully appreciated what it is I'm pointing to that needs explaining. There's something real here that needs explaining or accounting for in a full theory, and an explanation like this doesn't even touch on it. In the full story of "what is experience", there is absolutely without doubt a computational side. But there is an aspect of experience, the 'felt'-ness of it, or the phenomenology, the qualia, that is not explained even in part by reference to computations or any other physical process.
There's just nothing like that, no precursors of it, to be found in the physical world. It's not the computational side of experiences, but rather the what-it's-like-ness of it.
Here's an example of it: what, given naturalism, explains the redness of seeing red? This is not a question about the wavelengths of light, or light hitting the retina, or processing in the brain that leads to the brain states that correspond to having that experience. This is a question about what it's like to see red -- the experiential side of the experience, as opposed to the physical causes of it. Why, given naturalism, would we think that anything experiences (first person perspective) anything at all?
> A correctness of a thought isn't a function of thought itself, it's a function of thought and reality.
I agree with this, although I think it's more precise to say a thought is correct insofar as it conforms to reality.
> Correctness of a thought is not a physical label. It's not a state. It's something you evaluate, something you compute with respect to knowledge you observe.
This appears to be illogical. The correctness of my thought (that is to say, whether it conforms to reality) is one thing; my evaluation of its correctness is entirely different. Further, correctness must be a 'state', or property, of a thought: otherwise you undermine your own argument (see below).
> nature does not put XML tags with correctness score on neurons.
I agree with this. But this supports my argument. An arrangement of neurons (or anything physical) cannot be correct or incorrect. But a thought can be correct or incorrect (you must think that, because you must think you're right and I'm wrong about this disagreement). Therefore a thought cannot be reduced to an arrangement of neurons (or anything physical).
I have a vague familiarity with that view, and that is definitely not the view I hold.
That view states, as you describe, that somehow the whole is more than just the sum of the parts. I reject that. Instead, my claim is that if the only parts you have are physical parts, you will not get consciousness. You need other parts too -- specifically, mental/consciousness parts.
(although, this sounds closer to a dualist view, while my view is more idealist)
It's logically impossible for brain states (configuration of atoms) to be the same as mental states (thoughts). They are correlated, but they are not the same. Please see my reply to Temporal above.
> How does this make sense? There is some divine being that only sometimes looks at the actual synapses to decide if it should feel nauseous or happy?
One possibility: the experiences that God gives me at any moment are what they are because of the state of my brain. This isn't a sometimes thing, but an always thing. I see or experience what I do because of the state of my brain, but not only because of the state of my brain.
The problem is that the physical states alone don't entail anything about the experiential -- nothing about the phenomenology. The physical explanation leaves out things that are real and need explaining.
> If a group of scientists in 1923 were given a working iPhone, would they have a chance at figuring out how the ChatGPT icon is able to launch something that talks or where Grand Theft Auto is in the machine?
Would they get another iPhone if they break the first one?
The task is no different than any other scientific endeavor, from biology to physics: they'll keep poking and prodding at it in different ways, making observations, accumulating and discarding hypotheses, until some kind of coherent and useful model becomes apparent. Hell, given sufficient supply of iPhones, this would be much easier than figuring out the brain - there's lot of value in short feedback loops and being able to do meaningful experiments on the cheap.
> Is trying to figure out how the brain works by looking at synapses and blood flow and electric fields similar to trying to see how an iPhone works by looking at transistors and electric fields?
They would be looking at that (after speed-running the relevant part of the "tech tree" to understand fields and transistors), and importantly, how it correlates with observed behavior such as ChatGPT saying something, or crashing the weird fire-breathing demon plane???! simulacra in Grand Theft Auto V. Same is the case with intelligence, except we're not free to circuit-bend people the same way the 1923 scientists would be with their iPhones, due to ethics and stuffs.
Overanalysing this perhaps, but they might need some philosophical tools.
Would it still be possible without Boolean Logic (1847)? Of course they'd have to realize that it applied. [1]
Earliest possible date they'd be successful might be post-Leibniz in the 1690's. [1]
Of course, there's an argument to be made that it'd not be possible before Alan Turing's work in 1935/1936-ish. On the other hand, in 1923 you'd already have most of the ideas that lead to Turing's insights.
[1] https://en.wikipedia.org/wiki/Boolean_algebra#History https://en.wikipedia.org/wiki/Boolean_algebra#History
Can a computer be conscious?
If I take a camera, a microphone, a speaker, and some other sensors and feed them into a CPU/GPU and self train it to have an understanding of its capabilities (internal/self) and the world around it (the external), this this consciousness or not? If I light a fire near this 'smart' computer via it's sensors it can detect the heat and move farther away. If I give this computer a complex task that it has to calculate multiple steps to achieve before it acts, is this not mental work?
In LLM based systems we can't really figure out how this occurs because the computational complexity of the operations is too high, much like brute forcing encryption, getting to the answer of how it's working isn't impossible, you'd just have to burn the visible universe to figure it out.
> since some naturalists deny there's actually any such thing as 'consciousness' to be explained
> it is impossible for naturalism to explain (or explain away) consciousness, and consciousness is real, therefore naturalism is false
I’ve always found the denial of the existence of experiential consciousness to be kind of funny. If you removed all my senses and put my brain in a vat to keep it alive, what would be left? As far as my brain is concerned, consciousness is the only thing for which I can be definitively certain of its existence.
It’s a solipsistic viewpoint, because I can only prove that consciousness is real for me. And since I can’t prove that it is real to anyone else, it’s not “science” in the sense of the societal process that requires reproducibility among peers, but perhaps it is still science at the personal level since I can reproduce the observation of its existence to myself at any moment that I choose.
> I’ve always found the denial of the existence of experiential consciousness to be kind of funny. If you removed all my senses and put my brain in a vat to keep it alive, what would be left? As far as my brain is concerned, consciousness is the _only_ thing for which I can be definitively certain of its existence.
Yes exactly, it is the thing that is most familiar to us, most accessible, and the only thing of which we can be certain. Descartes in his Meditations went down this path, starting with that foundation.
But I don't think you need to land on solipsism, because we don't have to believe only those things which can be proved. To me, it seems a reasonable hypothesis to suppose there are other consciousnesses. For an experimental line of thinking that leads to that conclusion, I would recommend the aforementioned Meditations.
Many are tempted to think that the physical world is the certain thing that we have, missing the fact that we experience the physical world, and so it is experience that we have access to, not the physical world itself. That I am conscious and have experiences of such and such is the thing I am most certain of. Whether there are physical things that correspond to those experiences is an extrapolation.
Honestly, that would be great, because we could eventually reverse that connection, and take over heaven.
That's not a good example because we could easily measure the modulation of the signal reception by simply covering and moving the antenna.
Clearly, we know reality through our senses and our ideas. That is uncontroversial. The problem is when one says (as I think you are) that our sensual perceptions and ideas are all we know, and that we can reasonably doubt that they ever refer to reality.
We know reality by means of our senses and ideas. We don't know our senses and ideas as such.
To repeat myself: if your claim is true, then the person who says "reacting hydrogen and oxygen will produce water and heat" can be no more, and no less, correct than the person who says "reacting hydrogen and oxygen will produce jello in the shape of George Washington's nose". Because if, as I think you claim, we know nothing but the "mental formulation of our mind", then our thoughts don't point to reality; and if this is true, we can't predict anything about reality. And therefore each of these two claims about chemistry is as valid as the other. Do you agree with this?
> If however, there were hypothetically some other observer that did not agree with you on these basic beliefs, such that they said that the detector turned on at time 2 and not 1, then what is the reality? What method do you have to confirm what is reality when two observers disagree on what was observed?
This is a red herring. We may never know who is right; that is different from saying that there is no reality to know.
>Clearly, we know reality through our senses and our ideas. That is uncontroversial. The problem is when one says (as I think you are) that our sensual perceptions and ideas are all we know, and that we can reasonably doubt that they ever refer to reality.
We know reality by means of our senses and ideas. We don't know our senses and ideas as such.
I am failing to see the distinction here. If your only conception of "reality" is through your sensual perceptions, how can you know anything beyond that? How does reality then differ from mental conception of your sensual receptors? To me this is equivalent to saying, "there is a spiritual world that doesn't interact with anything and we cant prove it but it exists". The only thing I can think of is that reality is shared among several observers, who all agree on it.
This is why I brought up the hypothetical scenario of two scientists who disagreed on whether the detector turned on at t=1 or t=2.
If you were the only being on this planet so that no one could disagree with you, and you for some reason were high on LSD for the entirely of your life and saw flying rocks, would that not be your reality? Every scientific test you do would fit your findings, as perceived and evaluated by you. If there was a community of such people who never saw a rock standing still, what would be the real truth?
And if you say "but there is still a reality, none of those people would just know it", then I can ask, how do you know we don't fit that same bill, where some other observer could look at us and say "those guys are all seeing an illusion, its actually this way"
>This is a red herring. We may never know who is right; that is different from saying that there is no reality to know.
But what determines who is right? That is my question. What answer can you give me besides "a conscious observer?", or "a collection of conscious observers who all agree".
And if you can't determine who is right, what meaning is there to it being reality? Why is one conscious experience any more correct than the other?
Let's work out precisely where we disagree:
1. Do you agree that it is rational to think that burning hydrogen in oxygen will produce water and heat (prediction A), and irrational to think that it will produce a juicy stake that can play tennis (prediction B)?
2. If yes to 1, do you agree that A is correct (or at least closer to being correct), and B is incorrect (or at least further from being correct)?
3. If yes to 2, do you agree that there must be something against which A and B are both measured that makes one right and the other wrong?
4. If yes to 3, do you agree that the thing against which A and B are measured must be independent of them both?
5. If yes to 4, do you agree that there is a reality independent of our observations, and which our observations presuppose and are caused by?
I think you're saying that we can reasonably think that our sensual perceptions and ideas are all we know, and that we can reasonably doubt that they ever refer to reality, but I want to confirm this.
> And if you say "but there is still a reality, none of those people would just know it", then I can ask, how do you know we don't fit that same bill, where some other observer could look at us and say "those guys are all seeing an illusion, its actually this way"
See point 1, above. I suggest that nobody really systematically doubts their observations. But if you honestly disagree with point 1 then let me know.
> But what determines who is right? That is my question.
Obviously that is sometimes impossible to answer, depending on the situation. But sometimes (more often) it's not. Again, do you disagree with point 1 above?
>who disagreed on whether the detector turned on at t=1 or t=2
This is an already answered question by relativity and causality.
Since light has a constant speed and also a propagation speed there are operations that observers will not agree on the order of... but they will always agree on the order of causality. The cup will always get pushed off the table first before it's broken on the ground.
No that's not what is being said. The laws of physics are a mental model. There is something they map on to, but we can only form an incomplete map. They're a reduction in order to understand, and thus as much as they're formed by our observations, they're equally informed by our biases and limitations.
You're making a fundamental mistake in comprehending this. You're confusing the map for the territory. When I say the laws of physics I mean the map, and you're thinking about what they point to, the territory.
What is the most fundamental fact? What is the first step, from which everything then can be derived?
That first step is the fact that there's something observing. That observer can construct a theory.
That theory can posit "actually, there's matter, and there seems to be matter which does not possess this same capacity of observation, therefore this inert matter must have formed first, and then [insert hypothesis of your choice] that's how the observer arose".
That's valid.
However, it's not a fact, and it's not the only theory.
The theory I'm entertaining instead, is this: the observing phenomenon is as much a fundamental part of the universe as gravity. The reason some things don't seem to possess it is that we can't directly perceive other's observing phenomena, and it's only when this phenomena binds with matter in such a way (time, appearance, scale, complexity, frequency) that it can communicate with similar binding phenomena, then there is recognition: "you have a face, you're smiling, you seem to be alive, so I'm going to assign the observer trait to you".
Now that I recognize other observers we can communicate and form a theory, and see what we can agree on. Then and only then I can say "well let's pretend that there are no observers and see what we can agree on". And thus the laws of physics are born, as a mental model.
> You're making a fundamental mistake in comprehending this. You're confusing the map for the territory. When I say the laws of physics I mean the map, and you're thinking about what they point to, the territory.
I agree with the map-territory distinction -- but the laws of physics (or, better, the "way things behave") is primarily in the territory, and only derivatively in the map. Our map may be more or less accurate, but the map presupposes the territory. Otherwise, nothing we can say about reality is more or less true than anything else, and absurdity results (like the hydrogen-oxygen example I gave before). If the way things behave is solely a property of the map, not of the territory, then it must be the case that we can't predict what will happen when we react hydrogen with oxygen. But we can predict this, and therefore the way things behave is a property of the territory. The map that says hydrogen+oxygen=water+heat is a better map -- that is, closer to the territory -- than one that says hydrogen+oxygen=chocolate bunny. But again, better vs worse map presupposes the territory, and presupposes that the stuff on the map points to real stuff on the territory, however imperfectly.
Are you in agreement with this as far as it goes? Do you agree with my premise that we can predict what will happen when we react H and O?
> What is the most fundamental fact? What is the first step, from which everything then can be derived? That first step is the fact that there's something observing. That observer can construct a theory.
Surely the fundamental fact is existence? Without this, there can be no observation, no observer, and no observed object. These things, like any individual thing, presuppose existence.
Further, your statement presupposes other things: for example, the law of non-contradiction (which is a facet of existence). Your statement says that there is a 'thing', that it observes, and that it constructs a theory. It must therefore take for granted that there is not nothing; that the 'thing' doesn't not observe; and that it doesn't not construct a theory. Otherwise the statement is meaningless.
This is just the beginning of what I could say, but your argument, like any argument, presupposes existence, the laws of logic, non-contradiction, the distinction of one object from another (the law of identity), etc. So we can't take an observer as a fundamental fact. More generally, we can't take any particular thing as the fundamental fact. Universal knowledge precedes particular knowledge. Existence in general precedes existence of the particular.
I'm not sure if this directly affects our disagreement, but it might help understand where I'm coming from :-)
> Are you in agreement with this as far as it goes? Do you agree with my premise that we can predict what will happen when we react H and O?
I do agree with the second premise, but I'd say that's just saying we have a really good map for that particular area of the terrain. If you're going to your aunt's and you have a really good map, you're not going to show up in hong kong (I'm mapping the example onto your hydrogen-oxygen case).
But fundamentally, "the laws of physics" are still information as digested by the human brain. At one point our map of physics didn't include relativity. Now it does. The terrain did, but the map didn't.
With this in mind, it's crucial to remember that our current map has nearly no information about consciousness. It's like you have really detailed information about your aunt's neighbor in Chicago, and you know that Hong Kong exists and is Eastward. So you just assume that it's a straight road. Someone might suggest there's a few mountains in the way, but fundamentally you have no idea how to actually connect both points of the map.
Materliasm is essentially that to me, looking at the very detailed map of current physics, wiping their hands and saying yup that should be enough, it's a straight line from here.
>Surely the fundamental fact is existence?
I try to refrain from making an observer-observation distinction. To quote Krishnamurti "the observer is the observed".
Moving the discussion in this direction feels like another map, the map of logic. Within the framework of logic, I do agree with those terms. However, I don't think reality obeys logic in so much as logic is another way for us to make sense of the phenomena around us.
I hope I don't annoy you by another analogy: Imagine we're watching a wolf and his cub, and I ask which comes first. You say the father of course, because he was born earlier. This is true. However, I was asking which of them was ahead, that is in the present moment which of them will arrive first. You were answering from knowledge, not from direct observation.
Maps, models, and logic, are within the realm of thoughts. But experientially, without following and validating the narrative of thoughts, can you directly observe that there is existence without observation? Your thoughts will immediately tell you that this is already a contradiction: can you observe that there is no observer? Can you discern in your own experience "this is observation", "this is existence". Could they be one and the same?
This is where I'm coming from. Thank you for engaging in these topics which are tricky to discuss :)
I think the initial problem with your argument remains. If the 'laws of physics' (honestly I'm not keen on that term except for casual use -- can we say 'way things behave'?) only exist in the human brain, and not in reality, then we have the problem that our map does not point to any territory. All maps are imperfect, but the 'map' that includes relativity is more accurate than the one that includes Newtonian mechanics, which in turn is more accurate that the one that includes Aristotelian physics, which in turn is more accurate that the one that says a giant dragon in the sky controls everything, etc. But again, the map presupposes a territory. The fact that one map is better than the other presupposes something independent of any map that each map must be measured against to determine its accuracy. Given that each map is attempting to describe the 'way things behave', it implies that a territory, independent of any map, exists; and further, that the 'way things behave' exists on the territory. Agree?
And therefore, the original statement that I took issue with, which was "[t]he laws of physics are 'universally' agreed truths about certain mental phenomena experienced by the conscious observers that conscious observers have so far interacted with" must be false.
> Materliasm is essentially that to me, looking at the very detailed map of current physics, wiping their hands and saying yup that should be enough, it's a straight line from here.
Oh I totally agree with you about that. I'm not a materialist at all, as my comment history will reveal. Materialism cannot account for the obvious fact of consciousness, or the fact that the human mind is capable of reasoning. In general, any philosophy that tries to reduce all reality to a single principle (like matter, number, energy, power, dialectic, the self, etc) is suspect, IMO. Starting with 'existence' as the first principle avoids this problem.
> I try to refrain from making an observer-observation distinction. To quote Krishnamurti "the observer is the observed"... I don't think reality obeys logic in so much as logic is another way for us to make sense of the phenomena around us.
This must imply that someone who says 'the observer is not the observed' (assuming he means these terms in the same sense as Krishnamurti) is wrong, but this only makes sense if one accepts logic as an unbreakable and truly-existing principle.
So I would disagree that logic is a map; instead, logic must be presupposed by any statement or any belief; it must be presupposed by any map. (I think logic is 'baked into' existence itself -- it's inseperable from existence -- in some sense it is existence.) Again, if we deny the law of non-contradiction (one of the first principles of logic), anything we say, believe or think can be true or false in the same sense at the same time, so any attempt to make any sense of reality will inevitable not even get to first base. Hence, nothing we say would have any meaning whatsoever. So hence, I disagree that logic is within the realm of thoughts (except insofar as it's something we grasp or see - meaning it's only derivatively in the realm of thought).
> Can you discern in your own experience "this is observation", "this is existence". Could they be one and the same?
I'd respond by saying that all these questions presuppose something, which is existence, for the reasons I gave in this and my previous post. The fact that you can ask these questions, and that there is any meaning in the questions, presupposes that (for example) the law of non-contradiction exists independently of you (and me, etc).
> Thank you for engaging in these topics which are tricky to discuss :)
Sure!