Human brain versus IT machine

Artificial neuron’ is the man-made object that comes

nearest to competing with the human brain. It is the basic computing unit in the

Artificial Neural Network –ANN. The Biological Neural Network–BNN created by

the almighty–while losing out in speed–beats the best Artificial

Intelligence (AI) system in a few facets. ANN is a set of processing units

assembled in a closely interconnected structure that exhibits similarity of some

features of BNN. Soon computer will be called as an Information Technology or IT

machine.

The Cyborg



Professor Kevin ‘Cyborg’ Warwick, University of Reading, UK, makes

`intelligent learning robot’–which learn by trial and error and ‘not by

fully programmed computing’. He works in Mobile Autonomous Devices–MAD-laboratory

and estimates its cost as $ 10 million. A human brain is, perhaps, cheaper.

Kevin believes in embedded systems–literally. He gets sensors embedded into

his body to let tissues grow around it. His body communicates with computer.

Lately he coaxed his wife to get embedded too–perhaps they communicated

through sensors and not by the elements as vision, touch, aural, olfactory and

taste.

The brain 



The number of neurons in the human cortex is estimated to be 10 raised to

the power 11, equivalent to the arithmetic digit ‘one’ followed by eleven

zeroes. The cell body of a typical neuron measures from 10 to 80 micron or micro

meter. The total number of interconnections between the neurons is estimated to

be ten thousand times the number of neurons. The ‘interconnect’ is called

‘Axon’. When axon extends to other nerve cells, it is ‘inter neurons’;

when axon reaches muscle fiber, it is ‘motor neuron’. A third type of neuron

receives information from muscles or from sensory organs. It is ‘receptor

neuron’.

This infrastructure facilitates the complex ‘pattern

recognition task’ that a brain is capable of. Complexity of the brain is

further compounded–computing in brain takes place at three locations:

• In the cell body called ‘soma’ of each neuron

• Outside the cell body in the ‘dendrites’ which is a
  
  tree-like nerve of fibre

• Outside soma in the ‘synapses’ which are junctions of
  
  axons

Pattern and data



Brain has an ability to deal with a variety of ‘data situations’. The

network can deal with information that is fuzzy, noisy, probabilistic and

inconsistent too. This is because human brain perceives everything as a ‘pattern’.

For IT machine everything is ‘data’. If there is no pattern, it is very

difficult for a human brain to remember and to reproduce data later. Brain tends

to see a pattern even in routine data consisting of arithmetic integers such as

telephone, bank account and car numbers. Storage and recall operations are

different in a human compared to a computer.

The brain is robust too. Because of the vast BNN, the

information is distributed in the connections throughout the network. Robustness

and fault tolerance are because of pattern nature in storage and recall. The

brain is far superior to IT machine in pattern recognition. Brain is fault

tolerant. Any slight decay of some nerve cells does not affect the performance

significantly. Ability to compromise helps to optimize. The sense of vision can

be taken as one example.

Value addition



‘Thalasima’ is an interface between the retinal nerve endings inside the

human eyeball and the vision related neurons in the brain. Light rays pass

through the aperture called pupil of the eye to reach the retina. In reality, it

is only a hazy representation–of data, text, image–that is formed. It gets

interpreted into a clear pattern. Thalasima does ‘value addition’ to the

data collected by the retina. Perhaps it is true that beauty lies in the eyes of

the beholder–value addition is indeed different, say, in each of the nine

judges in a beauty pageant. In the future, one may propose an optimal limit

through eye test for the judges before they sit on the panel!

Learning by experience and example is a derivative of logical

thinking. From childhood, human brain is adept at this. Thematic Apperception

Test and psychological test of logic, rely on this faculty. Sensory perception

by vision, sound and hearing, smell or olfactory sense, touch and taste

understand input data in the form of number, text, shape, picture and audio.

Some brains exhibit extra sensory perception too. A prodigy has perhaps

compressed logical steps–to access data and solve abnormally fast.

Kevin says, "Humans do not sense the world in other

ways."

Any interference in the sensory perception is ‘noise’.

Even in a noisy environment with deformed or partial data, the brain can derive

an overall and complete pattern to make sense. But AI needs algorithmic

instructions. IT machine can be made to sense not only the ‘five’ but, more.

For instance, the machines can be made to sense ultra violet rays, infra red

signal and ultrasonic too, adds Kevin. If wanted patterns are mixed with others–which

is noise–the brain is discerning, it recalls the required ones. Kevin also

says that brain can get added ‘intelligence’ with embedded implants!

Brain makes up



Brain exhibits flexibility. The network automatically adjusts to a new

environment without using preprogrammed instructions. It can function in a

non-programmed environment. The best computer always needs programs and

instructions. The brain ‘understands’ patterns whereas, computer ‘recognizes’

data. Brain can get the whole object in the available data even if, there is no

clear indication of sub-pattern in the data. As one example, a name written in

handwritten cursive script may not be clear; some alphabets may be hazy. Yet the

brain understands or makes out the name. Similarly, speech is understood even

when it is slurred.

Intuition is God’s gift to human. BNN uses electro-chemical

signals and the five sensors generally convert them into mechanical–loco motor–activity.

IT machine use electrical signals; deprived of it they cannot exist in

semi-alive


form.

IT machine need a memory to store information. Addressing

system is necessary to locate and retrieve any information. A new information

stored in the same location destroys the old one. BNN stores information in the

‘strengths of interconnect’; new information is ‘added’ by adjusting the

strengths. Old information remains. Information in the brain is adaptable and in

IT machines it is strictly replaceable. But IT machines can be built with

enormous amount of memory.

Even if a few connections are snapped or if a few neurons are

not functioning, in BNN the information is still preserved due to the

distributed nature of the encoded information. In contrast, computers are

inherently not fault tolerant; corrupted information in the memory cannot be

retrieved. Each neuron in BNN acts on the basis of the information locally

available. Output is transmitted to the connected neuron in the neighborhood.

Whereas control unit rules the roost in a computer, there is no systematic

control to process information in the brain.

Computer is faster



In computers, the cycle time to execute one step of a program in the CPU is

in the range of a few nanoseconds. The cycle time of a bio neural event prompted

by an external stimulus takes milli-seconds. IT machines are a million times

faster. A large number of instructions are involved in most of the computer

programs and they operate in a sequential mode. But the brain operates with

massively parallel operations, each of them having comparatively fewer steps.

Human information processing is superior for certain tasks, despite of being

several orders of magnitude slower as compared to IT machine.

Variety of communication by machine is evident. In the brain

there is no ‘full duplex’ system in communication–one medium cannot carry

transmission and reception signals simultaneously. Some junctions are ‘half

duplex’–either transmission or reception takes place, one at a time. The

rest are one way medium–either it facilitates only transmission or only

reception. Axon is the transmission medium and the dendrite receives signals

when interacting with neighboring neurons.

Human brain is habituated to see the ‘space’ only in

three dimensions–four if ‘time’ is added. Space is multi-dimensional.

"Thought in human has not graduated to feel multi-dimensionality of the

universe", says Kevin. "The brain is upgradable with extra memory and

even X-ray vision", he adds.

Biological virus strikes due to hacking, which is a disease.

The worst kind of hacking on the BNN happens when the brain causes at least two

distinct and contradictory behavioral patterns towards the same fellow being–schizophrenia.

It will be akin to a helpful robot turning destructive all of a sudden in the

same environment–by learnt behavior or otherwise!

Sqn Ldr BG Prakash