The Brain

  • #26
Nereid
Staff Emeritus
Science Advisor
Gold Member
3,367
2
Guybrush Threepwood wrote: Is it possible for us to make a functional model of the human brain?
This was the original question.

So, a few WAGs:

1) in terms of hardware, the human brain has ~100 billion neurons, and ~150 trillion synapses. At an OOM level - maybe 2 or 3 OOMs - this corresponds to ~100G of RAM, and ~150T of ROM ('hard drive') in a computer.
Well, I reckon we could easily construct a computer with that much hardware today.

2) for cycle time (or clock speed), the comparison isn't exact, but if the brain were a synchronous machine (it isn't), its clock would run at ~10Hz, which is ~9 orders of magnitude *slower* than today's computers. Of course, a more realistic comparison, involving relevant comparable domains for example, would reduce the gap, but not make it go away.
Again, human brains can be matched or bettered in this respect.

3) turning to resilience and robustness. This is certainly an area where the brain's capabilities are poorly understood. At a very high level ... brains suffer gradual decline in functional capability and performance, and can easily work continuously for 100 years. A significant percentage (1-5%?) do suffer systematic failures which can be compensated for to some extent. In contrast, today's computers are ~1 to 3 orders of magnitude less resilient. Methods and approaches to improve this are known (e.g. how the key Shuttle computers are designed).
However, it's likely to be a decade or two before we come close to matching the human brain's capability here.

4) architecture. At one level we know how the brain is wired; that's where the name 'neural network' came from. In other respects - as zoobyshoe pointed out - we are still likely very ignorant of all that the brain does, let alone how it's wired to enable and perform these activities. We can already build neural networks in silicon, or similate them with code; we can also handle the inherent parallelism, at least in principle.
For what we know today, an artificial brain could be built today, architectually. Say two decades before we discover ~80% of all the key brain architectures, and another decade to emulate these.

5) apps. Some of the brain's apps we know reasonably well (e.g. vision), some we're still likely decades away from even outlining (e.g. personality, social interactions), and some are in-between (e.g. language).
WAG: we might get to an '80%' brain in a couple of decades; this is perhaps similar to zooby's 'very damaged human brain'.

6) integration. IMHO, this is the area we know least about, and don't even know how important it is, in terms of the objective as Guybrush set.
Since we don't know what we don't know, this could take a mere five years to clarify, or over 100.

My conclusion: "Is it possible for us to make a functional model of the human brain?" Not today, but maybe in a limited sense by 2020-2030.
 
  • #27
21
0
I think that we will be able to model the brain as soon as we have perfected nanoTech. My reasoning behind this is that we would be able to inject nanites into the brain. From there that could each attach to a neuron and map its inputs and outputs. And as the brain is working, it would be able to determine how it responds to stimuli both artificially and from other neurons. Depending on the technique used, we could then both extract the nanites and have them reassemble themselves 3-dimentionally. Or, (taking a Borg like approach) they could assemble some sort of transmitter. If they made the transmitter, it might be possible to actually gather real time data from the brain.

Furthermore, I do not really care too much for true computer AI. I think it would be very nice to learn enough about the brain to build mind machine interfaces, or some sort of other augmentation.

I heard from someone that one on the applications of nanoTech was brain augmentation. The idea is to make nanites with very small transmitters and receivers. These would allow the brain to make connections between neurons at the speed of light, as well as many more connections. Currently the brain can only send a signal at about 200mi/hr and even then, the neuron must be reset. I do not know of any projections of how much better off the brain could be with these enhancements, but I believe that if everyone had them there would not be anyone without a 200 IQ (that is a bit optimistic, but oh well).

Going back to talking about windows, I think I heard that win XP has 3 billion lines of "C" code if you include all of the utilities that go with it. However, the important thing to note is what programming language those lines of code are. For example,... 3 lines of visual basic COULD translate to 9 lines of Visual C. 9 lines of Visual C COULD translate to something like 54+ lines of Assembly code (also called machine code).

In addition, some one stated that the operational speed of the brain is something like 10 Hz. Well im not totally disagreeing, but you have to multiply that by the brain cells in use, as well as, all of the other brain waves. That is, ive done research into binaural waves and using sound waves you can stimulate your brain to change its frequency. Therefore, more accurately, the brain is like a computer where the processor speed readily fluxuates. You also have to take into account the individual processing power of a neuron. I heard some statistic placing a single neuron in the optic nerve as having the processing power greater than a Cray super computer. I heard another statistic that the brain, as a total, has an equivalent processing power of 5exo-hertz. In addition, if you think about it, it kind of makes sense. Right now computers find it very hard to recognize objects 3-dimentionally. An example of this would be asking a computer to use images from security cameras to identify a telephone. Now if it is a phone it is seen before it should be able to spot it right away, however, what about different models of phones?

.. Any way, im just rambling now...
 
  • #28
Unless humans are artificial, I don't think you can ever model a "human" brain. The closest to come to it is to recreate a model of the whole universe to authentically recreate a human brain.
 
Last edited:
  • #29
8
0
INX990, What source did you use for your facts on the brain's processing speed? In researching my next SF novel, I came across a statement that the brain thinks at 1.25 times the speed of light. As the entire premiss of the book hinges on that fact, I need to be absolutely sure of it's truth. Do computers think faster than us? Obviously I have a lot more research to do. Any help would be greatly appreciated.
 
  • #30
1.25 the speed of light? doesn't that violate the laws of physics?
 
  • #31
21
0
yeah, that should. im not completely sure wot he meant by it thinks at that speed. If your refering to pure thought you cannot realisticaly put a measurment of speed on it. However, measuring the chemical signaling of the brain, it is much slower. um, im going to brush up on my neuroBiology ang get back to youall on the exact processes required to make this happen.
 
  • #32
8
0
speed of light

From my master (research) notebook: "Scientists have found ways to break the speed limit. In one experiment, published in the May 22 issue of Physical Review Letters, scientists at the Italian National Research Council of Florence shone light beams at a curved mirror. The mirror then shot the beams back at the instrument that measured the ray's speed. The beam coming from the center of the mirror was measured at 5% to 7% faster than light speed. The authors said this effect only works over short distances, such as the one meter used the the Italian researchers."
 
  • #33
I seriously doubt it.... could you provide a link?
 
  • #35
21
0
neuron mechanics

The diagram (attachment) is self-explanatory.

Nevertheless, essentially the nerve impulse is a sudden rush of sodium ions. The rush then triggers the next few proteins to let in more ions. This process continues in a relay fashion until the pulse reaches the presynaptic membrane. As soon as the pulse passes an indicated area of proteins, they will immediately stop allowing sodium into the pathway. Then active transport kicks in via the sodium potassium pump and resets the nerve. This entire process occurs at about 100m/sec. (much slower than the speed of light).
For those of you that are curious about claims that nerves are electric well, here is the explination. The axon has positive ions on the outside and negative ions on the inside. This creates an electric potential. So then, why doesn’t the pulse travel at the speed of electricity? Well, the protein gates that keep the ions out have different ways of being activated, chemically, and electrically. The main pulse is mostly controlled by electrically activated proteins. However, they have a high "Tolerance.” That is, while they can "feel" the pulse coming via the weakened electric current, they do not activate until the pulse is near to them. Ultimately, the signal sent across the synaptic cleft is chemical. Thus, it would seem to be inefficient to keep changing the signal medium. However, there are some organisms that do have nerve that rely solely on electricity.
This electric method is fast but it does not integrate information well. Electric transmission across synapsises is very fast and can proceed in either direction. However, they are less common in vertebrates and other organisms that have complex nervous systems. First, electrical continuity between neurons does not allow temporal summation of synaptic inputs (one way that signals are integrated). Second, an effective electrical synapsis requires a larger area of contact between the cells. This makes it impossible to have thousands of connections coming from one cell (witch is common in vertebrates). Third, electrical synapses cannot be inhibitory (not allowing for complex brain chemistry); and fourth, there is little plasity (modifiability) of connections (much harder to learn). Essentially, all there good for if speed, but not so good for complexity.

The statistic I got for brain speed came from a friend of mine, ill try to contact him about his source whenever he is back from the holidays.

The question "do computer think faster than us?"
Well, yes and no. They are capable of transferring information faster than we are, and are faster at processing in a linear fashion. However, they are much slower when it comes to nonlinear systems and processing also known as thought. Simply computers "think" what we program them to. So yes, computers process linear equations faster than we process, but could never understand their environment (at least with current tech).

As far as either building or emulating a brain, I think we should start small. We should try to create something called a ganglion. A Ganglion is simply a small cluster of nerve cells that carry out a specific task. These tasks could be something simple like detecting danger then relaying the signal, of cause the creature to disengage itself from prey. This simple basis of neuro-networking would not allow the subsequent computers to gain emotions or anything and would be reacting on instinct. Thus it wouldn’t decide one day that it no longer likes the scientists and lock them out side the space craft..... “HAL. Let me in”.....

Another interesting thing to put to the test would be what the DNA says about creating a brain. I mean, it eventually has to do it some way or the other.
 

Attachments

  • #36
8
0
Fascinating! I would call the "wiring" in the brain associative wiring, which makes sense. Also, thinking about nanite mapping, each individual brain would need to be mapped as we are like snowflakes. But an overall, general map could be acquired, but I wouldn't want it used on my surgery for fear I'd come out someone else. Hey, let's say in 30 years or so we advance molectronics enough to build a brain-interfacable computer. I've giving this a lot of thought, but what do YOU guys think? Would using this bioware restructure the brain? How might it change us? How would it alter society? 'xcuse the pun, but pretty heady stuff.
 
  • #38
8
0
Thanks for the link. Highly interesting that the writer sets a timetable of only 30 years. I think he's being optimistic, but that's just my opinion. 100 million gigaflops is a measure of quantity, not speed. I still need a confirmation on the brain's signaling speed. Even if we find a documented speed, it's my theory that the actual speed will exceed the findings due to the associative nature of neuronal interaction. Another rung in my ladder of research.
 
  • #39
21
0
the statement that it will take 30 years is a little optimistic, but not impossible. the only thing would be if anyone would do it. I mean, we have the technology to do somany things, yet that dosent mean we have. The technology to create a human equivalent brain will be there but there is a certain problem. The human brain rewires itself to learn stuff. Thus the robotic brain would proly end up running some sort of emulation program.

as for the 100million gigaflops being a measure of quantity vs. speed.. well it is a measure of speed. Its the unit of measure used to judge supercomputers, largly because they have so many processors. An example of this would be the comparison between Intel and AMD chips. An Intel P4 running @ 1.6 GHz will process information at about the same rate of a 1.4 GHz AMD chip. Thus, the measurment of flops is much more accurate.

as for the source of the brains signaling speed, i got the info from my AP Biology book [Life, the science of biology 5th edition. William K. Purves. Copyright 1998 by Sinauer Associates, Inc. pg. 910, 915-6] just incase you wanted to look it up yourself.

Furthermore, could you more thoroughly explane your position about the speed of the brain.
 
  • #40
2,193
2
The human brain, like anything "finite", can be modeled. If current technologies and resourses can achieve this is, well, another matter.
 
  • #41
8
0
As to mapping the brain... Image mapping of the brain was a key component in developing speech recognition software, according to Microsoft. So it's already been done. My interest is in what we do next. I overheard on CNN that with nanotechnology we could build a cell phone small enough for an ant to use. OMG Now, which egghead thought that might be a viable goal for such technology? I'd rather see communication devices nanoscaled and implanted for direct use in the brain. By programed firing of neurons the brain could experience a total virtual world without external devices. The liberty I'm taking in my current SF fiction work incorporates this and goes a bit beyond to explore social issues.

As to previous questions about the computational speed of the brain... My research suggests that the brain actually works at the quantum level, where speed can't be measured because everything is in an infinite-motion state. We can, of course, measure stimulus and response, and we can measure the all-over activity of the brain at any given time. However both of these sums will not give us the speed of thought, for thought preceeds both.
 
  • #42
21
0
Originally posted by evernow
As to previous questions about the computational speed of the brain... My research suggests that the brain actually works at the quantum level, where speed can't be measured because everything is in an infinite-motion state. We can, of course, measure stimulus and response, and we can measure the all-over activity of the brain at any given time. However both of these sums will not give us the speed of thought, for thought preceeds both.
Neural science suggest that the brain does not work at the quantum level. Neural connections communicate through firings. The higher the frequency of firing the more excited the neuron, the limit to the frequency of a neuron is about 25 HZ, so it's extremely slow by modern processors standards. The fact that there are billion of neurons in the brain allows for it to perform enormous amounts of processing. Neurons migrate together to form circuits that learn stimuli. The chemical conditions of each connection determine its conductivity and therefore its contribution to a task. So the means by which neurons react to their environment pretty much disqualifies a quantum process for consciousness.
 
  • #43
Averagesupernova
Science Advisor
Gold Member
3,673
655
I watched something on Nova on PBS that suggested that the human brain had more possible pemutations than there are atoms in the universe. Now is that impressive or what?

I think if we are a long way from recreating a human brain. We need to stop and realize why the human brain is the way it is and how the human being developes from a learning standpoint. I suspect that the human brain is the way it is partially due to the social nature of humans. We have a need to interact. It is hard wired in our genetics. The brain through evolution has wired itself to what it is today to continue to fulfill the basic need. Now if we could create simulations or machines to have that basic need to learn and interact, and accelerate their evolution, could we gain some answers? Maybe we would end up with something TOTALLY different than the brain, but with just as much if not more mental power than we have ourselves. Who knows...
 
  • #44
8
0
The Brain

To Squeeze:
It sounds like the old chicken and egg quandry.
Does thinking inniate pattern firing, or does pattern firing produce
thougt? If the former, from where does thought originate? If it's the latter, who begins the thought proccess?
It's to the latter that I posited a quantum state in which lies
the unconscious strucutes of the brain.
Evernow
 

Related Threads on The Brain

  • Last Post
Replies
4
Views
3K
Replies
7
Views
12K
Top