Hacker News Reader

Comment by quantadev | Hacker News Reader

Computer chips don't have any brain waves. Even some theoretical computer chip able to run a "Perfect in Every Way" simulation of a brain including even simulating the physics of the actual brain waves...still has no brain waves. Nothing but 1s and 0s turning on and off, and not even turning on and off in anything resembling a wave pattern at all.

This is why people who believe the "Consciousness is a Computation" view never want to discuss brain waves, despite the overwhelming evidence that brain waves is precisely the thing out of which consciousness is composed.

View on HN

Why would a perfect simulation of a brain not have brain waves? They're not magic. I'm perfectly happy to talk about brain waves. I suspect anyone else in the "Consciousness is a Computation" camp is happy to as well, I have no idea where you got that idea from.

I'd be interested in seeing any of this overwhelming evidence, but it's also not relevant, unless you're arguing that brain waves are supernatural in origin and can't be simulated. I think the crux of your argument can be boiled down to: "What's inferior about a perfect brain simulation, regardless of medium?"

I can write a computer program that analyzes, or even simulates, massive magnetic fields, but that program doesn't create any actual magnetic fields, right? Sometimes people say don't "Confuse the Map with the Territory" right? You get what that means surely.

You're correct about the magnetic fields[0]. But you can write a program to control things that do make magnetic fields.

But I don't think this is a relevant analogy. It seems that you're arguing a specific form of the embodiment hypothesis; that consciousness requires a body. Surely this is a sufficient condition (and is why I think roboticists have a good chance at cracking AGI), but is it a necessary condition? I'm not sure anyone has any convincing evidence that cognition requires "a body" (is a standard computer "a body"? What about a laptop that has a camera? What body parts are necessary?) and if it does, it is unclear what constitutes the minimal necessary requirements. It is quite possible that this is all "software" (e.g. a person in a vegetative state may be completely oblivious to their surroundings but still may have conscious experiences. Most people do this every night!).

I think you are erroring too far in the other direction. While I highly disagree with the person you're responding to, I do no think you are making strong arguments yourself. With so little known about what actually constitutes consciousness, I think it is difficult to make bold claims about what it is (though we've ruled out things that it isn't; e.g. a rock. Which is our typical method of doing reasoning in science; we generally disprove things, not prove. Like you said, the map is not the territory, and in this say way our models (yes, plural) of physics is not physics itself).

I think the is evidence to believe that consciousness may be a "software" but that is not proof. Similarly I think we have lots of reasons to believe that embodiment is highly beneficial to creating consciousness, as not just by nature of examples, but our understanding of consciousness is in part about responding to environments and stimuli. But that's clearly not sufficient either and pinning down where this is, is quite complex (even our definition of life is ill-defined). I'm highly convinced that embodiment is the best path to pursue to achieve AGI, but I'm not convinced it is required either.

[0] If we get pedantic, we should note that any simulation has to run on hardware and that creates magnetic fields but these are not in fact necessarily the fields being simulated. I don't think we need to pursue this avenue for this conversation.

I don't think consciousness/qualia requires a body, but definitely a brain (excluding plantlife to keep this short). I think it's a dynamic phenomenon involving waves. As I've said, all neuroscience data is consistent with it being waves.

I think neurons do 4 distinct types of things. They 1) gather data (sensory inputs) and route it to 2) locations where it's oriented into 3D spatial shapes that function essentially like "Fractal Radio Antenna" [for lack of better description] where the 3) waves resonate to create what we call qualia, and then 4) the waves induce a current/action going to motor neurons. It works similar to a television set working in reverse, with the screen being an eye, and everything going backwards in time, not forwards. Pixels are sensors, not lights, in the TV analogy. Thousands of pixels are input (rods/cones) and it all routs to (not from) an antenna.

You can explain all of that with Maxwellian Electrodynamics. There's no "woo woo" there. It's physics. If you look at it like this, you can see that most of the neural network wiring is I/O and signal routing, except for the antenna-like structures, the best of which being the Corpus Callosum and Hippocampus, which evolved first, which is why they're deepest in the brain. Their physical shape is critical, because they are both indeed functioning as a kind of resonator/antenna, and are getting memories (what most of consciousness is built on is memories) thru wave effects.

CORRECTION: You don't need a biological brain for qualia, but you need something that generates the same actual wave mechanics in actual physical reality. 0s and 1s in a computer ain't that.

  > but definitely a brain (excluding plantlife to keep this short)

"brains" are quite diverse and come in many variations. Squids have donuts. Sea Squirts devour their own brains. Leeches have multiple brains, Octopi (cephalopod) don't have a centralized brain and it's distributed, mostly in their tentacles. They also don't show cognitive decline when losing limbs. Maybe Wittgenstein can't understand a lion, but certainly a cephalopod is completely alien.

  > I think neurons do 4 distinct types of things.

I'm a bit curious, how much have you read about brains? What about signal theory?

I ask because I suspect you're trying to solve the problem yourself and not building on the wide breadth of literature and research we have.

Brains do not operate in waves the same way a wave of water or electromagnetic wave does. Though anything can be approximated by a wave too, but signals need not be waves. Yes neurons conduct electric signals, but neuron to neuron communication (synaptic transmission) is chemical. These are discrete particles. But as I mentioned before, if we're talking about the electrical transmission inside the neuron, these are spikey[0,1]. When you look at neurons firing, there is a digital like signal. If anything, these look like Gaussian mixtures.

  > There's no "woo woo" there.

I agree that there's no magic, and so I ask that you don't introduce it. You're imbuing waves with magic while not recognizing what waves actually are. Analogue and digital are both waves. They both can integrate, resonate, interfere, and all that jazz. The same wave mechanics work. Your call to Maxwell is meaningless because ALL OF ELECTROMAGNETISM (minus quantum) is explained through those 4 equations (and have been extended to incorporate the latter). And just think about what you are suggesting. That digital signals are not described by physics? That software can't be described by physics? It's all physics. Sure, the abstract representations of these things are not physics, but that's true for any abstract representation.

So don't try to dispel people who are calling to magic by making a call to a different kind of magic. I don't think you're dumb, but rather it is easy to oversimplify and end up with something that requires magic. It's FAR easier to do this than to get the right answers. There's a reason people study very narrow topics for decades. Because it is all that complicated and trying to reinvent their work (while a good exercise) is just more likely to end up making many of the same mistakes that were then later resolved. We have the power of literature and the internet makes it highly available (though can be hard to differentiate from hogwash and it can be hard to find truth). Use the tools to your advantage instead of trying to do everything from scratch. Do the exercise, but also check and find the mistakes, improve, and do this until you catch up. Hell, you may even find mistakes the experts made along the way! The good news is that its SO much easier to catch up than it is to extend knowledge. You can learn what took thousands of years to get to in a week! It's hard, but in comparison catching up is trivial. So don't squander humanity's greatest innovation just because you worked hard at building your model. But refine it and make sure it doesn't invoke magic.

[0] https://en.wikipedia.org/wiki/Biological_neuron_model

[1] https://en.wikipedia.org/wiki/Soliton_model_in_neuroscience

Every sentence of that was pure strawmanning, so this is where we stop talking; but you were a good sounding board for a few back and forths up until here.

Asking you to step up your game is not strawmanning. But if you're unwilling to then I guess the other comments should have made it clear this would happen

There was no strawmanning

Strawmanning is when someone intentionally invents an incorrect interpretation of someone else's words, and then disproves or shows the faults with that wrong interpretation. That's precisely what we saw above.

Like this? https://news.ycombinator.com/item?id=41440133

Asking clarifying questions is the exact opposite of Strawmanning. Rather than assuming someone is wrong, it's better to ask for a clarifications first. That's what's happening in #41440133.

  > This is why people who believe the "Consciousness is a Computation" view never want to discuss brain waves, despite the overwhelming evidence that brain waves is precisely the thing out of which consciousness is composed.

Fwiw, I do actually believe that the brain is a computer. What I don't like about "consciousness is a computation" is it's utterly meaningless. Everything is a computation because computation is a fairly abstracted concept (same with simulation. Yes, we live in "a simulation" but that doesn't mean entities on the "outside" programming it. A ball rolling down a hill is also a simulation...)

But I really disagree with your argument. The difference doesn't come down to analogue vs digital and that seems like quite a bold assumption, especially considering 1) digital can get quite approximate to continuous signals and it can represent continuous signals 2) the brain doesn't function entirely continuously. Neuron signals are quite spikey. The neuromorphic people drew their inspiration from these and other aspects of the brain. But the brain is not operating on just continuous signals. There's a lot going on in there.

But if your true complaint is about how few ML people and far fewer ML enthusiasts will passionately discuss these subjects but also refuse to read literature from neuroscience, mathematical logic & reasoning, and other domains, then yeah, I also have that complaint. When you're a hammer I guess....

Do you think a computer simulation of an EMF wave is an actual EMF wave? If everything, according to you, is just a computation aren't you also claiming that actual EMF waves are "just computations"? Also does the substrate for the computation matter? What if I'm moving beads on billions of abacuses (as a thought experiment), are those abacuses therefore identical to an EMF wave if they're calculating wave mechanics? How does this abacus know I'm doing EMF wave calculations instead of, say, quantum probability wave calculations?

  > Do you think a computer simulation of an EMF wave is an actual EMF wave?

Of course not. I'm pretty sure we've also established this.

  > aren't you also claiming that actual EMF waves are "just computations"?

No. But actual real world EMF waves perform computation.

You are confusing what simulation means, which is why I explicitly put it in quotes. There's a dumb "simulation hypothesis" (the likes of which people like Elon Musk discuss) and there's a "yeah the terms are broad enough so it can be viewed as a simulation - simulation hypothesis". The latter isn't really contested because it also isn't very meaningful in the first place. But it does matter here for what you're talking about and why we brought up that a rock has "memory" since we can store information in the atomic structure. You're missing the abstraction of the stuff and how we're just talking about entropy.

Look at it this way, how does nand flash store memory? You trap electrons in states, right? This is to say you isolate charge to one side of a transistor gate from another. You can also do this with magnetic storage. There's plenty of people who build macro computers with things like water, and in those you're separating buckets of water. Physicists love to build computers out of things, and there are even thermal computers, using heat flow for computation.

That's the thing, computation is just about differentials in entropic states. So literally everything is a computer and everything is computing. Thanks emergence (another extremely misinterpreted word that people believe holds much more meaning than it does). But this does not mean there's some programmer, god, or any of that. It just means that information exists... You're making the same mistake these Musk like people talking about "living in a simulation" make, conflating the computer you're sitting in front of now that you control and have power over with just information flowing (obviously you can affect things in the physical world, and that's why experiments are physical simulations -- since they are controlled and limited settings -- but good luck programming "the universe" -- whatever the fuck that means. You're going to need more energy that it's got...)

So view physics as the instruction set of the universal computer or whatever. We're dealing with huge levels of abstraction so you can't go about understanding these things with reference to how we normally operate (this is why physics is so fucking hard and why advanced math -- well beyond calculus -- gets so crazy). Just like you can't work with high dimensional data by trying to relate to things in 2 or 3d, because the rules are very different at those abstraction levels (for example, the concept of a distance measure becomes useless. This is the curse of dimensionality that people very often misinterpret due to looking at it in a limited scope).

I think this field of discussion needs a new word for a new type of fallacy I'll name "Appeal to Base Reality" (ABR).

When someone makes a conjecture about physical reality like "Consciousness is made of EMF Waves, not Computations" (like I did) and then someone else tries to refute that with an appeal to the Simulation Theory (like you did) then that's called "Appeal to Base Reality".

Even if we're living in a simulation we still should consider EMF waves a very different phenomena than computations. Computations are not a phenomena of physics. Computations can be done on an abacus.

Reply on news.ycombinator.com