Chip in the head

Harnessing the power of the computer to augment the human mind has always been one of the hallmarks of science fiction. But now, the correction of memory using a chip in the brain is approaching the actual experimental stage 

Anna Griffith, Scientific American

Harnessing the power of the computer to augment the human mind has always been one of the hallmarks of science fiction. But in fact, scientists have already made several steps towards brain-machine integration. In the spring of 2007, a team from the University of Southern California intends to replace damaged rat brain tissue with a neural prosthesis.
Over the past few years, researchers have demonstrated their ability to translate another creature's thoughts into action. In 2000, Duke University neuroscientist Miguel Nicolais attached an electrode to the brain of a monkey that controlled a robotic arm with the power of his mind. Brain-machine interfaces, developed by neurologist Niels Beerbaumer of the University of Tübingen in Germany, are already helping some paralyzed patients move the computer cursor using brain waves and use it to select letters on the screen to write messages.
Theodore W. Berger and his colleagues at the University of Southern California (USC) have developed a brain-machine interface that for the first time communicates back to the brain. In January 2006, they used a silicon chip to mimic the actions of biological neurons in tissue sections of the hippocampus, the memory sorting and storage node, removed from the brain of a rat. The chip replaced part of the hippocampus that was surgically removed and restored its operation by processing neural signals and translating them into the appropriate output with a 90% accuracy level.
For several years now, the biomedical engineers have been on the verge of testing a chip in hippocampus sections, but they have encountered roadblocks that have slowed their progress. The electrode array technology did not work properly in tissue sections, a failure that forced the researchers to develop their own electrodes. There was also difficulty in cutting the hippocampus in a way that would keep the neural pathways intact.
The construction of a chip with an area of ​​one square millimeter costs tens of thousands of dollars and takes several months. Therefore, the test in the spring of 2007 is actually based on a model of such a chip, that is, on a larger component of the FPGA type - field programmable gate array that can be programmed and connected to a computer. The component will allow researchers to easily examine and modify their mathematical model, which describes the neural communication in a live rat, before burning it on a chip. Sam Deadweiler, a professor of physiology and pharmacology at Wake Forest University and co-author of the study, demonstrated how stimulating the hippocampus of live rats with a specific activity pattern improves their performance in memory tasks, such as remembering which pedal to press to get water. In a few months he will use a mathematical model programmed on an FPGA to predict the activity in the hippocampus. If the model turns out to be correct, the artificial implant could restore the memory of rats suffering from drug-induced amnesia.
In complex model animals, physicist Armand Tanguy from the University of Southern California suggests implanting a multi-chip unit that will facilitate the transfer of information. Light rays will transmit signals between neural units located on multiple chip layers. Unlike wires, light rays pass through each other without interruption, a feature that allows for many more connections. And the result: a network of light rays connecting silicon chips and simulating a dense neural network.
"Researchers are expected to face many challenges when their research moves from the laboratory to the body of the rat," says Grace Peng, program director in the Department of Scientific and Technological Discovery of the US National Institutes of Health (NIH). In fact, staff members don't know for sure what to expect when they start working with animals. Chemist Mark Thomson of the University of Southern California says that to prevent rejection by the immune system it may be necessary to anchor on the chip molecules that normally allow cells to stick to each other. In this way, the surface of the implant will resemble the tissue of the body. The neural plasticity, or the ability of the brain to recognize the connections found in it, may also pose a difficulty because the brain may prevent the creation of stable connections between the nerves and the chip. "In other areas of application, such as motor control or perception, the flexibility and adaptability of the brain usually help to achieve results through artificial interfaces," says Feng optimistically.
Another difficulty that may arise if similar implants are tested in humans is the possibility that the bypassing of damaged nerves in the hippocampus may also bypass other neural connections in the brain that filter our memories. In other words: the brain may no longer be able to distill the memories. If that happens, the implant will indeed be a truly unforgettable component.