CPU/Human brain question (ot)

[iggi]

Does anybody know why a human brain can do multitasks at the same time while a cpu has to quickly switch from one task to another?

For instance, you can lift both arms at the same time. Does your brain give the lift signal to each arm alternatively really fast or is it simultaneously? How can it be simultaneously if you only have one brain? If the brain can do tasks at the same time, why can't computers?

[Slapdragon]

<BLOCKQUOTE>quote:</font><HR>Originally posted by iggi:

Does anybody know why a human brain can do multitasks at the same time while a cpu has to quickly switch from one task to another?

For instance, you can lift both arms at the same time. Does your brain give the lift signal to each arm alternatively really fast or is it simultaneously? How can it be simultaneously if you only have one brain? If the brain can do tasks at the same time, why can't computers?

<HR></BLOCKQUOTE>

From my high school biology:

The human brain is actually an acreation of 12 parts (six on each side of the head) that function as an integrated whole. IPeople who have their corpus callosum cut can do odd things like draw with one hand and work a calculator with the other. Parts of your brain can do CPU like things even when the rest of your brain is damaged or ill. A CPU is not a real good brain, it is just a sort of kind of brain, better to think of your whole mother board as a brain. Your mother board can transfer graphics, write to disk, run audio out, dump data to your firewire chain, and do traditional CPU tasks all at the same time.

[iggi]

I see the analogy. But if we stick with the example of just raising two hands at the same time, not talking and walking at the same time, there must be a section of the brain that is generating 2 signals simultaneously? No?

[bredberg]

Hi,

If you need a CPU which (apparently) handle multiple tasks use a workstation and not a PC.

Can a brain simultanously do several things?

Do you realy lift your arms at the same time.

A CPU can do different things within a couple of CPU-cycles, i.e. a couple of nano-seconds.

Would you see the difference if the left arm is raised even a micro second before the right arm.

.........

[iggi]

bredberg, I'm sure you would not notice the difference but is that what is happenning?

Sorry for the OT post. I was just curious.

[Slapdragon]

The action of raising the hand is controlled by your spine. The brain only has to give one command to get the action started. This is why you can drive a car, and why "practice makes perfect" because the functions do not occur in the brain but in other areas of the nervous system, only kicked off by cognitive thought in the cortex.

A better example is this. Try to write two different letters at the same time.

[Jeff Heidman]

It is an interesting subject, and one we probably cannot really do justice to here.

It is the case that a computer can do more than one thing at a time, it just needs more than one processor.

The issue with your arms raising may just be your brain sending a single signal that gets sent to two different places. That's just a guess though.

Certainly a computer could send out one signal to write to a disk, and switch then sends that signal to two different disks and they can both work at exactly the same time.

It is an interesting issue though. On a related topic, I thought it was interesting that Deep Blue did not beat Gary Kasparov by being "smarter", it beat him by being bigger. Deep Blue does not play chess like a human does at all. It just sits their and proceses every possible combination several million combinations per second. it was not smarter than Kasparov, just faster and with a better memory.

Jeff Heidman

[Guest Rollstoy]

I guess the bottleneck is the connection (bus) between CPU and memory, which can transmit only a limited number of Bits at any given time.

Note, however, that MMX for example is a kind of parallel multi-tasking as it applies the same operation to several bytes of data simultaneously (if I am not completely wrong).

Not to mention parallel computers in which the CPUs really work in parallel and simultaneous.

Nothing compares to the billions of cells in the human brain, though ...

Regards,

Thomm

[tss]

iggi wrote:

Does anybody know why a human brain can do multitasks at the same time while a cpu has to quickly switch from one task to another?

A brain is a big network of connected cells. Each nervous cell two kinds of connections, incoming and outgoing. There may be lots and lots of connections coming to and going from one cell.

If enough incoming connections are activated (that is, some other cells are sending a signal to the cell), the cell "fires" and sends a signal to all outgoing connections. There's no central controller that dictates when a cell may send or receive signals but they all operate independently.

So, in practice the "multitasking" only means that there's activity on several separate locations of the brain.

The examples of walking and lifting both arms are actually poor examples since the processing of sensory input takes much more effort and needs much more parallelism.

There are also fancy names for each part of the cells but since my relevant books are at home, I can't give them.

- Tommi

[iggi]

Slapdragon said "The brain only has to give one command to get the action started". You know, I kinda guessed that that might be happenning but then I thought that you can move both hands up and down and at different speeds. So just one command to the spine for every infinite possiblities of motion for both hands? Guess that's it.Thanks.

Jeff Heidman said "Deep Blue was not smarter than Kasparov, just faster and with a better memory." If only we can take a kind of catalyst pill that would do just that.

[iggi]

tss, hmm so our brain is not running a program, it is the program. The arrangement of cells is the code.

[Jarmo]

As for lifting 2 hands simultaneously, am I correct in remembering the

left side of the brain controls the right hand and vice versae?

And since we are talking about brains, I must recommend reading books by Sacks (Sachs? something like it). Especially "Mies joka luuli vaimoaan hatuksi" "Man who mistook his wife for a hat" (might or might not be the correct english name)

[This message has been edited by Jarmo (edited 01-10-2001).]

[Slapdragon]

<BLOCKQUOTE>quote:</font><HR>Originally posted by Jarmo:

As for lifting 2 hands simultaneously, am I correct in remembering the

left side of the brain controls the right hand and vice versae?

And since we are talking about brains, I must recommend reading books by Sacks (Sachs? something like it). Especially "Mies joka luuli vaimoaan hatuksi" "Man who mistook his wife for a hat" (might or might not be the correct english name)

[This message has been edited by Jarmo (edited 01-10-2001).]<HR></BLOCKQUOTE>

You are correct about hand control.

[iggi]

So then , the left and right areas of the brain fire off a message at the same time to the spine. At exactly the same time. For those two halfs to fire a message at the same time, they must also be stimulated at the same time,..by what? Thier stimualtion comes from where?

[jasoncawley@ameritech.net]

Computers can be built and programmed to do several tasks at a time. Supercomputers like the "connection machine" prototype were built to explore this idea, with one computer able to access hundreds of CPU chips in parallel. But the problem is that programming a massively parallel computer is rather difficult. Unless it is doing pretty simple things over and over (and comparing results, e.g.), it can quickly become more trouble than it is worth, compared to fast central processing and loads of memory to store results. But it is useful for some problems (estimating fluid flows with dynamic models, modeling a crystal lattice of atoms to investigate stresses and cracking).

These days one can do some similar things with networks of seperate computers, if the connections are fast enough. The programming problem is still there, so serious uses of the idea have not grown much. It has stayed a super-computer for special cases thing so far, and there is reason to believe it will remain so.

Brains on the other hand are massively parallel in a way that makes even computers built that way look centralized. Each neuron is much slower than chips, with less it can do. But everything is wired higgedly piggly to everything else, making the number of connections rise extremely fast as you look at more and more neurons (and there are billions of 'em).

No engineer could tell such a structure what to do, because he wouldn't know how. That is why it is a hard problem just to figure out how the brain even works in the first place. We understand the basic idea, which is the same one that computers exploit - that the number of arrangements of a limited number of things is enourmously larger than the number of items (plus some other more complicated things, to be sure). As an example of that fact, there are more distinct, legal games of chess than particles in the universe. Without using lots of parts, or making more than one move at a time.

For what it is worth...

[iggi]

"there are more distinct, legal games of chess than particles in the universe." By legal games I guess you mean a legal setup and a legal move. Very interesting.

[jasoncawley@ameritech.net]

I mean starting from the very begining and going all the way through to mate, both sides trying. An upper bound on particles in the universe is around 10^90. If an average game of chess runs 45 moves on each side and at each move a player has 10 legal moves, you get that many legal games of chess, since the possibilities go as (possibilities per move) to the power of the number of moves. And in fact, 40-60 moves is quite typical for chess games, and the number of options is higher than 10 for most of the game (it is 20 on the first move and can be as high as 30 on the second).

Incidentally, Go "blows up" even faster than chess. There is no complexity about how pieces move, but the larger board and higher number of moves more than makes up for it. Strategy games are for all intents and purposes arbitrarily complicated beasties, beyond the simplest (solvable) kids games. In fact, realism, following known working strategies, a limited number of dominant ideas that work in particular common situations, and other such rule-of-thumb means of lopping off huge parts of the "possibility tree", all simplify more than they complicate. The underlying complexity is still there, even when all those things aren't.

[curih]

Just thought I'd relate Jason's post a little more closely. Go becomes so complex that thus far no one has been able to write a simulation that can stand up to real high level players. But obviously the human brain is capable of it, at least to a better extent.

[Slapdragon]

The thing about the computer is that to play well, it need to compute those huge trees, humans do not since they can do instant reprogramming, and they have a skill that computers do not involving random choice. The "what the hell" random move.

[Guest aka PanzerLeader]

Ah! Very interesting topic, let's just hope it doesn't get padlocked by Madmatt

And while we're talking about coordination, is there a cure for people who don't use the correct hand? Example: you have a can of beer in your right hand and a piece of paper in your left. You walk pass the thrashbin, throwing away...the full can of beer!

It happens to me almost every day, quite embarassing in fact. Where does that problem come from? Nervous cells sending the wrong kind of stimulation? Or just mind "laziness"?

Just curious about it...