The Nature of Consciousness/Self

+DarkMatter2 01:04:24
where instead of the pure chaos of the firewall theory or the preservation of information paradox we have what we have, which is a universe governed by its own weird laws
consciousness is different
consciousness appears to exist separately from matter
and it is a crazy mystery
our will shapes the path of energy and neuron function
@Sookie 01:05:42
hehe
+DarkMatter2 01:05:42
I need to poop out dinner
innit 01:05:44
Why does consciousness seem different?
I don't think consciousness is mysterious.
I think a high percentage of people who think consciousness is spooky are theists.
+DarkMatter2 01:06:44
what do you believe consciousness is?
and I do believe G-d exists, but not as a divine voyeur, I have no idea what the nature of a being like that is
innit 01:07:30
I think what we call consciousness is a sufficiently complex learning system, instantiated in the brain, that has self-representation.
+DarkMatter2 01:07:31
I can barely understand wtf an electron is
innit 01:08:22
What we call consciousness is an information pattern in the brain that has long-term memory with an extracted Self that emerges as an invariant perception across many experiments infants perform.
For example, push hand, block fall over. Push hand, block fall over. Push hand, block fall over.
+DarkMatter2 01:08:42
ok, so what is spooky to me is how that learning system works on a subatomic level
innit 01:09:02
Infants perform these experiments. The above experiment infants perform help them learn about centers of gravity of composite objects.
The way in which the blocks fall features some invariant principle that we call gravity, which infants learn about.
+DarkMatter2 01:09:31
what we're talking about as perceived reality being a 4d model built in your mind that you interact with
innit 01:09:52
Then, you perform experiments in other situations without hands or blocks and more experiments, eventually, you realize there is an invariant in all the experiments do the experimenting and the Self emerges.
So, once you have that learning system, the step to Self doesn't seem that difficult.
So, we can stop talking about 'consciousness' now and talk about that learning system.
+DarkMatter2 01:10:43
using that to make decisions that then, on a subatomic level, send energy flowing in different directions that in turn affect reality on higher levels
innit 01:10:54
Yes, the brain constructs a 3d-time model of reality around the Self.
The brain is complex @ your last post, that's for sure.
It's not mysterious, though.
If you also describe laptop computers in the way you just did for the brain/mind, you'll also be overwhelmed.
Notably, laptops use quantum effects to allow me to type this conversation with you.
@ "subatomic"
+DarkMatter2 01:12:15
yeah but you can trace the path of an electron through a laptop
innit 01:12:25
Laptops are much less complex than human brains.
+DarkMatter2 01:12:35
you're not really flipping a switch when you make a decision
innit 01:12:52
What we call 'decisions' consists of changes in many neurons, yes.
+DarkMatter2 01:12:53
It's more like the double slit experiment in a far more complicated form
right
innit 01:13:08
'Decisions' are changes in internal states resulting from calculations don't in networks of neurons.
done*
+DarkMatter2 01:13:26
so on a subatomic level, when we are talking about "making a decision"
innit 01:13:28
I don't see the relevance of the double-slit experiment.
I think you're taking on too much.
Your mind doesn't have sufficient working memory to understand the mind in real time.
Things cannot represent themselves with loss in fidelity.
+DarkMatter2 01:14:11
the relevance being, in that experiment, to insert a particle detector in order to collapse the wave function of the particle
innit 01:14:32
Collapse of wave functions is just one possible model of quantum mechanics.
And arguably an antiquated one.
But, again, that's not relevant.
(Copenhagen interpretation, specifically.)
+DarkMatter2 01:14:59
well, in realtime, we have the capacity to at least imagine a large amount of the possible outcomes for whatever is being thrown at us
those of us capable of it anyhowe
innit 01:15:22
You should experience no surprise at the perception you can't fully understand yourself.
No system can.
+DarkMatter2 01:16:08
yeah, if I didn't have a decently intelligent circle of friends to bounce things off of, I think I'd probably have a rougher go of things
I tried to explain that to the people at the data centers I worked in
they tended to have a policy of "someone thought of this so we're trying to cover it" instead of "lets actively find and eliminate threats"
innit 01:17:00
It's not precisely known the architecture of the learning system in the human mind, but it seems to be possibly using Monte Carlo rollouts in Bayesian inference.
+DarkMatter2 01:17:17
I have to admit, that's over my head
innit 01:17:26
Some of the statistics of neural learning match machine learning models using those approaches.
+DarkMatter2 01:17:38
That sounds much more like statistical analysis than what I'm familiar with
innit 01:17:45
Monte Carlo rollouts is where you do simulations of the future using your model of causality.
We can do a simplified version in a decision tree, e.g. in chess.
+DarkMatter2 01:18:12
does that give you a good idea of best/worst case outcomes?
innit 01:18:17
I go there, you go there, I go there, you go there -- estimate result.
Do that again. Do it many times. What percentage of rollouts ended with you better?
Use that to estimate the current state.
+DarkMatter2 01:18:51
that sounds like stocker :)
https://github.com/WillKoehrsen/Data-Analysis/tree/master/stocker
+pela 01:19:11
[ Data-Analysis/stocker at master · WillKoehrsen/Data-Analysis · GitHub ]
innit 01:19:12
Monte Carlo is just using, strictly speaking, sampling from a probability distribution to estimate a high-dimensional integral, generally.
It originated in the development of nuclear fission by von Neumann and pals.
+DarkMatter2 01:20:02
well, with the exception of true random number generation, isn't that what the basics of the math involved with quantum physics is all about?
innit 01:20:02
Bayesian inference is one of the two paradigms in statistics, the other being frequency-based statistics.
In Bayesian inference, there are competing models of the system you put prior probabilities on, then you collect evidence and update the posterior probability via Bayes' rule.
The SOTA learning systems haven't yet identified self for probably several reasons.
One, they're extremely rigid compared to the cerebral cortex' architecture.
Two, they generally have extremely low capacity compare to the cerebral cortex.
There's some meta-models like networks of networks in cross-domain learning that seem to begin to very slightly do learning similar to the human brain.
So, you shouldn't worry about consciousness -- just worry about how the brain became such a good learning system.
Humans can do things like zero-shot and one-shot learning that SOTA machine learning models cannot even remotely touch.
innit 01:24:30
The brain is such a good learning system, apparently, because evolution is a powerful algorithm for optimization.
+DarkMatter2 01:25:08
Bayes theorem looks a lot like the slope formula for derivatives
innit 01:25:13
When you're born into Homo sapiens in ~2000, it all seems magical, but if you saw humans wandering around 50k years ago, it would seem less magical.
+DarkMatter2 01:25:22
admittedly, you're way over my head with this
innit 01:25:35
And if you watched the evolution of brains over the last few million years, it would seem fairly straightforward.
I have a background in statistical learning theory.
+DarkMatter2 01:26:15
I'm worried about consciousness so far as trying to quantify what it is - it isn't like there is a subatomic consciousness boson hanging around
innit 01:26:34
And if you saw the emergence of specialized cells into neurons organizing into neural clusters into brains over the last few hundred million years, it would all seem quite reasonable.
+DarkMatter2 01:26:38
yes, I understand that it exists within the brain and that we can observe the affects of human choice
innit 01:27:02
I told you what consciousness is.
+DarkMatter2 01:27:18
you defined it as a learning system
innit 01:27:44
It's a combination of an invariant self over disparate experiments coupled with a learning system that has created a model of causality surrounding the agent.
+DarkMatter2 01:27:47
and I am strictly talking about it as a physical system
innit 01:27:59
All of that emerges from the learning system.
The learning system emerged from evolution.
+DarkMatter2 01:28:39
right - lets ignore the fact that it's even learning - or even time as a factor in this for a second
innit 01:28:41
When you 'think' or 'visualize', you're using your learning system to run simulations of reality or you're calling up encoded memories.
+DarkMatter2 01:29:04
lets just take a snapshot of a human head - frozen in time - and zoom in
when I ask "what is a thought" for instance
is this a pattern of energy, matter, or both that we're going to be looking for?
how small would I have to zoom in to be able to separate and quantify the singular mechanism that is causing energy to flow down one neuron vs another
innit 01:31:21
It's a complex system, and there are different ways and different levels of abstraction at which you can understand the system.
+DarkMatter2 01:31:40
right, I'm talking about Layer 1, if this was the OSI model
innit 01:31:45
Your brain cannot understand the entire brain's functioning in real time, so you'll have to use lossy abstractions.
+DarkMatter2 01:32:03
well, my brain can understand whatever I can give it enough data to understand
just saying "you can't understand this" is bullshit
innit 01:32:17
I didn't say that.
+DarkMatter2 01:32:18
this isn't the dark ages
innit 01:33:13
There are choices for Layer 1, I would say. I'd probably go with neurons and other brain cells with Layer 1, not atoms, or sub-atomic interactions.
This seems to closely mere Layer 1 in the OSI model.
+DarkMatter2 01:33:54
but that's ignoring all the basic physical layer underneath
innit 01:33:59
They, too, don't worry about quantum effects.
+DarkMatter2 01:34:07
neurons and brain cells are switching
innit 01:34:13
The physical layer consists of the electronic circuit transmission technologies of a network.[2] It is a fundamental layer underlying the higher level functions in a network, and can be implemented through a great number of different hardware technologies with widely varying characteristics.[3]
The physical layer defines the means of transmitting raw bits[4] rather than logical data packets over a physical data link connecting network nodes. The bitstream may be grouped into code words or symbols and converted to a physical signal that is transmitted over a transmission medium. The physical layer provides an electrical, mechanical, and procedural interface to the transmission medium. The shapes and properties of the electrical connectors, the frequencies to broadcast on, the line code to use and similar low-level parameters, are specified here.
The OSI model doesn't concern itself with physics.
+DarkMatter2 01:35:14
it does in the matters of what is being put on the lines, which lines are used and why
innit 01:35:18
It takes for a given the electronic properties of the computational components.
+DarkMatter2 01:35:23
frequencies, voltages, etc
right
innit 01:36:04
Then what you're aiming for is much more comprehensive than the OSI model on a much more complex system.
+DarkMatter2 01:36:28
but what I'm talking about is trying to figure how far down the model the physical mechanism of decision making is occurring - if it was a group of three neurons, physically what is occurring when a decision is made
← maxxiii has left (Quit: Connection closed for inactivity)
innit 01:37:48
There are probably some neurons that make what you'd call a 'decision'.
Different neurons learn different abstractions.
Complex learning systems decompose perceptions into multiple levels and perceive them all simultaneously.
The canonical example of this is the vision system.
Some neurons recognize composite object like specific faces, i.e. your mother.
Other neurons detect only very basic feature.
+DarkMatter2 01:40:06
great - so "decision A" is made - physically, what occurs inside of a neuron that causes energy to transfer from it and what decides the pathway?
lets say we could take a bunch of snapshots of that mind over the time of a decision being made
zoom WAY in to the relevant ones - or at least ONE of the relevant ones - and watch it throughout the process of a decision being made
+YourMom 01:41:23
I like beer
innit 01:41:46
The classic/core physical difference in a neuron that makes a decision is the action potential.
Are you familiar with that?
Signals are received on the soma and dendritic network that results in an increase in potential.
+DarkMatter2 01:42:46
No, is this a chemical potential or more of a schrodinger's cat scenario?
innit 01:42:57
It's an electro-chemical potential.
+DarkMatter2 01:43:00
ok
innit 01:43:50
Neurons have a resting voltage of like -70mV.
When the potential inside the soma (central unit) raises above a threshold, a positive-feedback cycle occurs, leading to a spike in voltage.
This is mediated through proteins in the plasma membrane allowing the diffusion of electrically charged ions in both directions.
It is believed that chains of action potentials is the code of the mind, analogous to the bits in a computer.
Just last week or so there was a decoding of the spike train in a human during an activity.
+DarkMatter2 01:46:00
is that a binary function?
our neurons are either on or off?
innit 01:46:21
They are usually modeled that way, yes.
+DarkMatter2 01:46:25
weak
innit 01:46:25
In the classic models.
+DarkMatter2 01:46:33
that's so disappointing
innit 01:46:44
But the code is complex in regards to the action potential pulses.
+DarkMatter2 01:46:49
I bet
You need it if you're just a giant binary computer
innit 01:47:34
Indeed.
The neurons are not truly binary, as the physical structure is complex with lots of other cells doing subtle modulations, but that may not be necessary to model.
+DarkMatter2 01:48:08
well, I appreciate you laying that out for me in laymens terms
if they're NOT binary, it would make a lot more sense to me
innit 01:48:49
You're welcome. :)