“Classes will dull your mind; destroy the potential for authentic creativity” –
Nash [A Beautiful Mind]
It is not possible for a man to stick tight for more than 20 minutes and concentrate on something. Initially you try to get the grasp of it but the time you lag behind your brain packs its bag and goes to sleep. Same goes for lectures at university. The other day while the lecture was under commencement, I was hovering somewhere else in world, a strange idea instantiated in my mind. Why can’t a human brain replace a microprocessor? With this question I started comparing the actual computer architecture with our brain. I made the following comparisons in my mind:
Brain
|
Microprocessor Computer
|
Memory
|
· Develops over time
· Volatile/Non Volatile
· Decays
· Nurtures with experience
· Dependent on number of neurons
|
· Static
· Non Volatile
· Non Intelligent
· Available Always
|
Processing
|
· Slow in mathematical calculations
· Fast in image/sound processing
· Builds up on requirement
· Improves by use
|
· Very Fast processing of Mathematical/logical problems
· Dumb
· Can’t relate previous events
· Parallel
|
Interfaces
|
· Biological based, very fast
· Operate on nervous system
|
· I/O devices with speeds of the order of 10s of megahertz
|
Backbone connection
|
· Nervous system
|
· FSBs, UARTs, USBs, IDEs
|
I came up with a formula that the brain only lacks storage space but can perform extremely well in terms of processing as the number of processing units in our grey matter is of the order of billions. Also it has been established [Einstein had more connected brain than ours] that the more connected your brain is from the inside the smarter it functions. However this was just my speculation during the class along with other scribbling and doodling information on a piece of paper.
The idea appealed to me so much that I searched for it online. My research online revolved around brain, computers, the function of brain, brain as a microprocessor and new developments in areas of computing.
The function of the brain isn’t that simple as I thought initially. But being a going-to-be-engineer it is easy to model things as subsystems. Therefore I found out that brain is one central controlling authority that deals with individual subsystems to make things work for us. The brain is linked with our body through a stream of fibers bundled together to form a network of devices dependent on brain. The entity which lays the foundation of our brain is called a neuron. On average a human brain possesses around a 100 billion of them. Our brain is generally divided into four parts depending on the function and the portions of the cranium [The stuff that makes your head hard]. These parts are:
· Cerebrum – Voluntarily movement
· Limbic System – Survival instinct and emotions
· Brain Stem – Vital supporting mechanisms e.g. heart rate blood pressure etc.
· Cerebellum – Motoric movements
These parts are further divided into many different portions depending upon their function but that is an entire subject and cannot be discussed over here. These parts are interconnected to each other via connections of the order of 1000’s of trillions. However these connections vary from requirement and gender basis.
Ok now we know how our brain looks like let’s have a look at the numbers now. The technique which makes brain function faster than a computer in some applications is adaptive parallelism. The neurons in our brain join forces to complete the task earlier and faster than a normal computer. A brain fires approximately 10^3 neurons per second, which is quite less than the clock cycles of a CPU – normally 2 GHz – , but the fact that the number of neurons performing the task multiplies this number by many folds. For example if a neuron fires an 10^3 neurons per second, multiply this by 100 billion, also the neighboring neurons also fire at the same rate so we multiply another 1000, the result comes out to be of the order of 100’s of millions of billions neurons. That’s what makes the brain so fast. However this number is only calculated by raw hypothesis so it may vary by magnitudes. The brain carries a lot of memory with it as contrast to my initial belief that we are short of memory; actually it does have a memory equal to 1000’s of normal hard drives of a normal computer. The memory is also dependent on number of neurons therefore ours is only a rough estimate. All in all a brain uses its resources much more efficiently than a computer does.
Now we come to computer and see how similar is the architecture. A typical computer system consists of the following:
· CPU – Central Processing Memory
· RAM – Random Access Memory
· HDD – Hard Disk
· FSB – Front side Bus
Now the CPU is the brain of computer, the RAM is a faster memory while the Hard Disk is permanent. Front Side Bus is the backbone of computer. In computers the Memory is static, i.e. it can’t be expanded without physical modifications. CPU operates at GHz speeds while there are at most four CPU’s operating in parallel In home PC’s. Notice the architecture is very similar to our brain. The brain acts like a microprocessor, It has got everything embedded into it e.g. memory, ALU and CU. Therefore it is established that we can replace the Processor of a PC with a human brain. The story doesn’t end here, like every other idea conceived, there is work going on it some other part of the world. So I searched for the current trends in the field, here is what I found;
MonoLayer
Researchers from Japan and Michigan Technological University have come up with the circuit that mimics the behavior of neurons. This circuit consists of parallel layes of switches known as monolayers. These switches can evolve like neurons and solve complex problems. It has been published in the paper “Massively parallel computing on an organic molecular layer”. The circuit also has the ability of self-healing like the neurons. An MRI of the human brain in comparison with this circuit is shown which clearly shows how similar the patterns are.
Memristors
The concept of memristor had been around In late 80’s but nobody actually produced it till 2008. A memristor is a circuit element in which resistance is strictly dependent in current and voltage feedback. The memristor also is capable of creating complex structures to speed up processing power of a processor. This CE is also capable of maintaining states without consumption of power. It’s also seen in memristors that they seem to have more than two states therefore this can entirely revolutionize the process of computing, therefore a future computer may not require programming but paradoxically it programs itself. Recently HP which is one of the prominent organizations working in this area has reported that their Titanium based memristor is capable of handling logic, therefore it’s a microprocessor in itself – Remember how neurons also are microprocessors – .
Memristors are essentially the next big thing in the semiconductor hardware market. Their ability to self-program makes them a hot topic of research.
image Source: Wikipedia
NOMFETs
Nanoparticle Organic Memory Field Effect Transistors are organic devices and are essentially composed of molecules. These are based on organic semiconductor known as pentacene and gold Nano particles. The NOMFET mimics a human synapse. Synapse is the gateway through which a neuron communicates to its surroundings by sending electrical signals. NOMFETSs operate on the principle of firing an amplified spike if the time distance between two incoming pulses becomes shorter and vice versa, this is called plasticity. NOMFETs use plasticity, which in turn will be basis for faster image processing applications.
Brain Computer Interface
The first thing that comes in mind when you think of brain as a device which is based on charges, is can we externally interpret the charges using some device? In fact yes, Brain computer Interface deals with this subject. Scientists initially tried it on animals like rats and monkeys, and later they devised devices which could be implanted in humans. In experimentation on a monkey a test was conducted by reading the brain signals of the animal. At the back they had a robotic Arm which moved with the movement of monkey’s arm. The Monkey was given a computer mouse; it had to drag it to a target on a computer screen. When the animal did it successfully it was rewarded with a sweet. The animal quickly got the hang of it and started doing it again and again. This gave the scientist enough time to monitor the brain signals of the animal responsible for the action of his arm. Sometime later the scientist disconnected the mouse secretly and kept the exercise going. This time the cursor moved because of the brain signals only. Amazingly the monkey got the hang of it as well, after sometime the monkey was just sitting and moving the cursor with its brain, WITHOUT any arm movement.
BCI is also used in people with blindness, electrodes are inserted in their brains and they are connected to a camera and a device which converts the image into neural signals, result is that the patient can get a very vague idea of the surroundings around him.
Conclusion
It is evident from the above discussion that science is progressing along the lines of analogue computing. We still are waiting for a breakthrough from this industry but the revelations that have already been made, makes the idea look promising. A computer with human brain would require very sensitive instruments to monitor the most delicate and intricate part of our body. Or in search for it we might be able to replicate the human brain altogether, who knows.
So keeping all the above discussion in a nutshell don’t you see intelligent computers come into existence in next twenty years? It’s alright if you don’t; probably you are taking too many classes.
References:
· Researchers develop new brain-like molecular processor – http://www.zdnet.com/blog/emergingtech/researchers-develop-new-brain-like-molecular-processor/2213
· Massively parallel computing on an organic molecular layer
· Anirban Bandyopadhyay1, Ranjit Pati2, Satyajit Sahu1, Ferdinand Peper3 & Daisuke Fujita
· The amazing memristor – beyond Moore’s law and beyond digital computing http://www.zdnet.com/blog/foremski/tha-amazing-memristor-beyond-moores-law-and-beyond-digital-computing/1309
· An organic transistor that mimics a brain synapse http://www.zdnet.com/blog/emergingtech/an-organic-transistor-that-mimics-a-brain-synapse/2095
· Energy Limits to the Computational Power of the Human Brain
http://www.merkle.com/brainLimits.html
· The Brain As An Information Processor http://www.racai.ro/~dragam/The_brain_as_an_information_processor.htm
· BBC Science and Nature http://www.bbc.co.uk/science/humanbody/body/interactives/organs/brainmap/index.shtml
· Brain Structures & their functions http://serendip.brynmawr.edu/bb/kinser/Structure1.html
· Wikipedia http://www.wikipedia.org