A team of Japanese scientists from Toyohashi University of Technology is constructing mathematical models each for three levels of vision which includes individual nerve cells called neurons, the complex circuit of neurons in the eye's retina and processes involved in final vision.

The retina contains a complex network of neurons, fibre cells that receive light inside the back of eye, and converts optical images formed on the eye lens into electrical signals that travel to the brain. Besides these cells, the retina also has a complex computing system for these signals.

Some of the Japanese team members are modeling single neurons, while others are modeling the complex neural networks involved in visual functions such as ability to see form, colour andmovement.

The virtual retina will have ion channels and nerve junctions called synapses similar to those in the natural eye, Dr Shiro Usui from Toyohashi University of Technology reported at the ongoing International Congress on Brain Research here.

Shiro said these virtual devices could be implanted in the eye. But scientists are yet to work out how to connect the silicon chips to the natural neural networks in the brain and there is still a long way to go before a device hits the market, Shiro said.

The work is part of Japan's Neuroinformatics Research Invision (NRV) project. Neuroinformatics -- literally a combination of information sciences and brain science -- is the latest frontier technology that aims at understanding the electrical and chemical signals in the brain to process and represent information.

Shiro said scientists in the next century will focus on understanding the complicated and elaborate functions of the brain such as vision, hearing and touch, recognition, memory andemotion.

The human brain is a kind of a computer, explained J Ascott Kelso, from Centre for Complex System and Brain Sciences at Florida Atlantic University.

The brain has billions of neurons and each neuron in turn has ten raised to the power of 23 -- more than a billion molecules.

Computing realistic brain models is very complex, Kelso said adding even with the remarkable increase in computing powers, mathematical models that incorporate all the known biophysical details of the brain "are out of question".