Ask HN: Is the Brain a Computer or an Antenna? | Hacker News Reader

Ask HN: Is the Brain a Computer or an Antenna?

14 pointssamuel232 comments

When building artificial intelligence systems, is it enough to consider consistencies at the atomic level as building blocks (computer), or do we need to dig deeper into lower level structures that might enjoy non-local interaction (antennas)?

(In my view the reality is made of onion like slices, each layer is made of feedback loops interacting together while preserving certain properties - invariants that are used for the next layer to build up higher order consistencies. Feedback loops develop and change over time due to disability to reproduce perfectly as they need to rely on a semi-stable environment - other feedback loops.)

View on HN

A model is only as good as its explanations of previous events and accuracy of future events.

What does your model predict and explain better than any other model? Is your model a simpler explanation?

> onion like slices,

So, nested hierarchies? They're pretty useful. But why the onion metaphor? Is there a centre? Do they have thickness? Is there an ultimate end or ultimate beginning? Are there other onions? Is there no cross talk between non-neighboring layers? How deep does your model go?

That is a good point.. these are just my conceptual thoughts (at this stage without any real experiments to back up my claims). The thing with nested hierarchies is exactly at the core of my question. If some lower layers are not hierarchies in sense of local interaction, we might not be able to simulate them on layers that "understand" only hierarchical structures. (for instance the main strategy for humans to deal with complexity is by decomposition into nested hierarchies)

It's exactly the model I'm exploring in http://lambdaway.free.fr/lambdawalks/?view=esprit_matiere or in http://lambdaway.free.fr/lambdawalks/?view=memoire : according to this model data are not stored in our brain - which is nothing but but a transmitter-receiver station - but in a kind of cloud (somewhere in a parallel universe) where the memory of the world since the beginning is stored.

We have private access to it thanks to a DNA type key and sometimes, when this key fails, we can have access to portions of memory belonging to others. This model makes it possible to describe quite reasonably paranormal phenomena, telepathy, talking with the dead, near-death trips, dreams,... For example, two distant twins communicate ... because they share the same memory area to which they have the key. If I see my long dead mother in a dream, it is not because she would be alive somewhere, but because, in the secondary state of sleep and because our DNAs are close, I had a furtive access to a portion of her memory and, when I woke up, I painfully reconstructed a scenario in which she would have said a few words to me. Why not?

And if you wonder about the nature of this parallel space where the memory of hunanity is located, think about this ocean of electro-magnetic waves in which we are immersed and to which we only have access with the help of transceivers like our smartphones, radios, televisions, computers, ... which do not store anything.

What do you think of that?

honestly a bit far fetched, my model tries to unify and build on scientific research without trying to make big claims about things we have not formalized yet. I think information is encoded into consistencies (we can and or cannot observe).

I agree with you, it's a simple reflexion about paranormal things. Microsoft plays also with that (https://www.forbes.com/sites/barrycollins/2021/01/04/microso...). I just guess that our brain does exactly the same, an empty engine interacting in real time with data in a cloud, somewhere in a twinned universe of information. When Maxwell wrote his 4 equations between electricity and magnetism he didn't only unified a lot of strange phenomena but he was leaded to discover electromagnetic waves running in an unknown continuum called "éther". Nobody knows what is actually this continuum but who cares. It works!

There are "brain areas"(local brain chunks specialized in something) and "brain networks"(non-local connections like the default network). NNs seem be focusing in modeling specific "brain areas"(static capabilities) but brain is highly modular, 'neuroplastic' piece of machinery that is dynamic and evolving with brain networks being central to its 'learned capabilities'. so i'd have to say inter-area communications are more important aspect that is being neglected as the 'brain networks' are more complex to probe vs static brain areas. https://www.pennmedicine.org/news/news-releases/2020/februar...

I think it's mainly a competitive & cooperative self replicator. The brain exists because previous generations did well. Do we need to infuse "antenna magic" into our brains, aren't they marvelous enough?

I found this video about open endedness (contrasting evolution vs optimization) in the brain and AI very inspiring: https://www.youtube.com/watch?v=lhYGXYeMq_E

The "marvelous enough" argument always feels disappointing to me. Sure, Occam's razor, Russell's teapot etc etc. But it feels like a blunt tool, too easy to wield.

Yes, I'm only using it because I detected the desire for more "marvelousness" in the original post.

I actually think AI advances could give us much better insight, such as how perception works, how world models are created, how taking decisions is learned through experience. After clearing all these out to some degree I don't need to take my marvelous or magical hat when thinking about the brain.

Also in my model intelligence is a property of layers, defined by how well they can simulate other layers using an API available to them (think about brain simulating physical interaction - lower level or simulating how the society might develop in the future - higher level). The more intelligent, the better the understanding of the surrounding reality. (I use predictive power as a measure of effectiveness of a theory / software that is instantiated on the higher level API). But if we build up artificial layers starting at an arbitrarily selected level - creating a different (unnatural) kind of feedback loops (species), can they simulate as broad range of layers as humans can (be as intelligent / think as much out of the box)? At which level do we need to start, to get close to what natural species are capable of? How much more complicated does the (AI) software needs to be to have equivalent predictive power considering it runs on hardware that provides a very restrictive version of the set of instructions natural species utilize?

(hardware refers to to the lower API a layer uses, software are feedback loops of the layer)

Maybe the mind is partially made of dark matter or influenced by it? There are current research explorations of whether dark matter and normal matter interact in ways other than via gravity:

https://www.sciencedaily.com/releases/2018/07/180713093545.h...

So if the two kinds of matter can interact it may be possible that the mind is either partially made of dark matter or influenced by dark matter. Maybe our minds experience dark matter as a kind of "weather" that has some effect on our thoughts. Maybe it is responsible for times of widespread rationality or irrationality. As (or if) we learn more about dark matter it will be interesting to think about how it may interact with our reality, including our minds.

If you want to track consciousness track the data. Vision has the highest bit rate and is easy to track, your consciousness has to exist somewhere that can receive all that vision data, enough pixels to cover what you see. Not the raw data, but the modified data with the vision "glitches" we experience. That data exists in a very limited part of the brain, if the brain is an antenna it has to be sent from there, and sending it has to be structured enough that it hits exactly one consciousness and not the billions/trillions of others living nearby.

Whether you think that is possible is up to you. If that part of the brain had such a high bitrate antenna to send stuff we would probably have noticed by now, so I don't think it has that function.

Not that I buy into this, but I can imagine being able to communicate using low bit-rate representations of a few archetypes and patterns/flavors of relations. Each receiver would reconstruct a different reimagining but should share some key elements when there's a 'connection' otherwise not.

Compressive sensing

The brain is a brain. All natural brains have stuff attached. If you want to attach stuff to your AI, go wild.

But if you imagine that to work it needs an antenna to pick up woo waves, my advice is to stop spending money on AI and take up another field.

The thing that bugs me about people comparing technologies to how the brain works or saying their technology is built like the brain is that yes, the brain has processing units that are connected, but the timing and speed of those connections is critical in the brain working properly. In large part this is mediated by the myelin sheath. If something happens to the myelin, like in multiple sclerosis, the brain starts malfunction even though the same connections are still there.

I'm open to the possibility that the mind could be a sophisticated communication channel. I recall reading about mothers feeling the distress of their child when they weren't physically near them. A woman I know had a similar experience.

The idea of the mind as a sort of antenna wouldn't be controversial to me.

Except that mechanism would have to have some sort of measurable effect in the world. To date, has there been any evidence of this?

Do we have enough knowledge about the universe to claim there isn't anything like that? Do we have sufficient methods to measure everything that is going on?

We do not know everything about the universe, but we know some things. We also know there is not much room in the theories of what we do know to support any claims like this.

It is certainly possible, but it would require a rethink of our current understanding of nature. That has happened, but most of the time when presented with anecdotes like this... it does not happen.

Why would it "have" to be measurable with today's technology?

They didn't say anything about current technology, they simply asked if here was current evidence.

Well sure, and evidence is collected by various means, all of which are temporally bound to the present :)

remir says>"The idea of the mind as a sort of antenna wouldn't be controversial to me."<

It would be controversial unless proof was provided otherwise.

It would be extremely alarming b/c every spammer in existence would immediately resort to "brain spamming" as a way to short-circuit e-mail and our current relatively protective spam filters. We would have to begin wearing anti-spam helmets to save our sanity and our thoughts. Criminals would (effectively) shield themselves from prosecution by claiming that acts of crime were transmitted into their brain.

IIRC Philip K. Dick had something to say about all this.

I think both? It is a computer that emits and receives signals. It has IO interfaces that interconnect the entire body. It can even receive signals passively through induction and process them. That to me suggests it should be considered both antenna and computer and probably quite a bit more.

What frequencies does it transmit/receive on?

Delta (1–4 Hz), Theta (4–8 Hz), Alpha (8–12 Hz), Beta (13–30 Hz), Low Gamma (30–70 Hz), High Gamma (70–150 Hz) [1] And here [2] is a different method of brain interface using MEG/MRI.

[1] - https://en.wikipedia.org/wiki/Neural_oscillation

[2] - https://en.wikipedia.org/wiki/Brain–computer_interface#MEG_a...

The onion analogy sounds accurate to me, based on random things I've read about the brain.

Not sure what you mean by antennas. Connections to the outside world? Connections that jump across the brain? Quantum effects?

I was thinking about antennas as layers that consist of feedback loops that employ interaction outside of any reasonably small bounds (the more abstract layers onwards are therefore also affected, but the error might seem to be negligible). The question is, if the error we make by simulating brain on a higher level (physical) API can destroy the intrinsic properties of intelligent beings.

I think you need to re-word this in simpler more concrete language. My mapping of words to concepts is not resulting in a coherent idea of what you’re thinking.

> layers that consist of feedback loops that employ interactions outside any reasonably small bound

I’m interpreting this as: Maybe there are interactions between part s of the brain that occur using a method we are not currently measuring (for example, maybe quantum probabilities interact in another dimension), so what seems to be a random process is in fact influenced by something outside of the “physical” world.

I personally “believe”, that consciousness operates outside of the physical world (or in a dimension we don’t directly observe), and interacts wit the physical world through the quantum boundary, influencing probabilities and observing the states of atoms. By paying attention to certain atoms, it is able to create an API that allows it to interact with the physical world through our common modes of cognition, mediated through the nervous system. I’d say the nervous system itself is the physical manifestations of the API, and the brain contents is another part of the API, and the user of the API is consciousness, observing through the quantum doorway.

So to bring this back to what you were saying, I think the brain in a way is an antenna in that it mediates interaction between consciousness and the physical world.

What the physical world is, is a whole different kettle of fish.

If you want to discuss much further, I can setup a throwaway/anonymous email. I love this stuff!

I think the source of randomness is the imperfect reproduction of feedback loops (at each layer), so basically I think randomness is just interaction of great complexity we cannot comprehend - it is a direct opposite of consistency. In my view there is just one "physical" world - the reality and all interactions happen inside this one frame. Consciousness is one of the layers, it uses a consistent but highly expressive API, that makes it possible to reach the level of predictive power we enjoy. Society might be considered a higher level, building on APIs of consciousness humans. If we were to use "trivial" self-made APIs, could we create a software that would match our level of predictive power? (see also the other comment I wrote)

An antenna is not a non local interaction. You mean interaction via a medium between non local things ? Eg having eyes , ears, etc?

I also don’t see the either or case making any sense here.

Reply on news.ycombinator.com