so if the human brain is sending around electrical signals it

it's constantly changing and adapting to stimulus too

>>8818654

to argue with the picture you posted: the brain CAN remember basic math facts, it is perhaps only the case that the synaptic pathways formed in learning said basic math facts is not strong enough.

you can learn something more strongly through repetition, for example

Sparse Distributed Representations

>>8818677

interesting, seems like a work in progress but they could be on to something.

>>8818654
That claim seems really overestimated

I've heard before that the brain's data capacity is around 1TB, which is more plausible to me

>>8818654
Java

Imagine a line that goes through the origin of some euclidean space (lets say 2D). on one side of the line there are 1's and on the other side of the line there are 0's. Posed the question "am I a 1 or a 0?" by a coordinate pair (x,y) we could determine which side of the line it is on and know the answer. this is called classification.

0000/1
000/11
00/111
0/1111

you could figure out the slope of this line by trying one random slope and counting how many 1s are on the wrong side and how many 0s are on the wrong side.

000|01
000|11
001|11
011|11

without loss of generality, if there are more 1s on the wrong side you would guess a new slope that moves in that direction. now do this a couple times and you'd be damn close to having the perfect slope and thus the ability to classify any future input. if you don't have many 1s and 0s its a bit harder to find the exact slope because there might be some gaps that let the line wiggle a bit without changed the number of 1s and 0s on each side.

but lets say there are more things at stake here than just 2D plane of ones or zeros. you could fathom a whole bunch of lines all moving to figure slopes on different axes for all the different pairs of 1s and 0s in some hyper-dimensional space. all you would do is guess a value that separates the 1s and 0s appropriately and fix it if its wrong. turns out this is enough to figure out ANY function of 1s and 0s. given some input coordinate (x,y) in ANY dimension, you could figure out which side of the line it is on and this classify it as a 1 or a 0.

you know anything can be encoded as 1s and 0s (input "3 + 3" output "6" would be something like 11, 11 --> 110) so that means you could figure out how the slopes to ANY mathematical (or even visual, cognitive, emotional) function. you just need a bunch of those slopes.

chain these little fuckers together in a connected format and call them neurons and you have the brain. this is also how deep learning works.

>>8818716

Then how come i dont have to pick up a cup 100 times until i get it right?

>>8818720
basically theres overlap in examples. if you witness a ball fall you're seeing gravity. if you witness marble fall you're also seeing gravity. there is overlap there and you actually learn a model about how things physically work. this means when you later see something else fall you can use all those other similar examples to predict what happens.

you'll know a cup is a solid with physical properties similar to the stupid shit you picked up as a kid (over 100 times cuz ur a spastic fuck)

Kurzweil did a fine job of dumbing down within "how to create a mind". If you have time for a "hhmm" explanation. Deep learning goes a bit further than stacking blox. We are the best machines for now.

File: AnalogComputerKitcalculo1959.jpg (78KB, 640x480px) Image search: [Google]

78KB, 640x480px

>>8818654
> I cannot imagine how something can possibly be programmed without math or language in built into the programming somehow.

There were analog computers that could solve complex math equations without any sort of programming language, code, or number crunching. You'd build an electrical circuit that behaved the same way as some other physical system. For example a graph of the voltage across a resistor as a capacitor discharges through it follows the same curve as a piece of radioactive material losing mass.

Our brains are probably closer to those than the digital computers you're familiar with. They aren't digital. Any sort of comparisons like "the brain is a trillion gigahertz and five petabytes" is popsci writers pulling numbers out of their ass.

>>8818744

wait is this just like a watch or something?
It will move a dial depending on something electrical thats setup to happen the same speed as the radioactive material loses mass?

I dont see how our brain can work like that as it has to adjust to new things all the time.

Or am i misunderstanding that machine?

>>8818804
They're potentiometers. Turning them changes the resistance of a resistor. They let you put different variables into the analog computer like starting mass or half-life of radioactive material.

I wasn't trying to say that our brain acts exactly like an analog computer; just that if even computers don't have to use programming languages and math then you shouldn't think our brains have to.

>>8818654
brain code is probably written in like matlab desu

>>8818654
Assigning a data capacity to the brain is nonsensical and the only reason I can imagine anyone would do this is either stupidity or curious facts outreach shit.

Anyway, you should look into machine learning. After having spend some time with that you'll have a nice intuition for the kind of stuff the brain does. Whatever is not instinct is simply statistics. Your brain gets stimuli and fits the state of its neurons to result in the best action to maximize some unknown loss function. There's really no complex math involved, everything is ad hoc.

>>8818674
aka working

>>8818701
The problem is that people often mix data and information.

1TB of pure information seems plausible but the encoding/compression format required for data to represent this information with exactly 1TB of data is very hard to find, even for the brain.

>>8818654
look up artificial neural networks
different from brain but an example of 'programming' without coding.

PHP

>>8818896

https://en.wikipedia.org/wiki/Machine_learning

As of 2016, machine learning is a buzzword, and according to the Gartner hype cycle of 2016, at its peak of inflated expectations.[12] Because finding patterns is hard, often not enough training data is available, and also because of the high expectations it often fails to deliver

/sci blown the fuck out, its in the first paragraph too.

>>8818654
>Brain/computer analogies

These really need to stop, no onec working in the field thinks the brain can be described like a computer. OP pic presents a good example of where the analogy fails.

Most likely more than one. Different parts of the human brain use different languages to communicate between one another. It's how we're set up so there might be a good reason for it. Maybe..

>>8818824
>muh index

I thought of this, and tried to imagine how some of the algorithms implemented in the brain might work.

>>8818716
so a support vector machine

>>8818654
mini brain system #425483234525 got a signal so it should do X

its just a massive cluster of autistic motors that do one thing
no rhyme or reason behind it

>>8818744
To add to this anon's post, consider purely mechanical computers: https://www.youtube.com/watch?v=s1i-dnAH9Y4

These machines take input and produce output, without the use of any programming language. In fact, they have an advantage over digital computers: although both mediums have a closed range of inputs, the distribution of possible inputs and outputs on the mechanical computer is continuous, whereas the set of possible values in a digital medium is discrete (your computer allocates only so many bits for floating point numbers). As far as numbers go, mechanical computers do not have the problem of narrowing.

While I'm no cognitive scientist, I'd imagine the brain has this advantage as well. While it does use electrical impulses, it is likely an analogue machine as >>8818744 explained, and this doesn't even touch upon neurotransmitters. Furthermore, when you consider the biological aspects (brain growth, decay, and neuroplasticity) the psych101 comparison of brains to computers crumbles to dust.

>>8818654
>>8818670

This. Whatever system our brain uses will be absolutely insanely complicated and acting over multiple levels simultaneously, nothing at all like a human-made programming language.

>>8818720
>Then how come i dont have to pick up a cup 100 times until i get it right?

You DO, you've just forgotten about it. Watch a baby try to interact with the world, it's always dropping and fumbling with things.