Researchers at MIT have unveiled a computer chip that mimics the human brain. Its purpose will be for studying how the brain's neurons respond and adapt to new information, a phenomenon known as plasticity. This process is believed to underscore many human brain functions such as learning and memory.
The computer chip uses about 400 transistors to simulate the activity of a single brain synapse, which connects two neurons and is what allows information to flow between them. There are close to about 100 billion neurons in the human brain, each forming synapses with many other neurons, through ion channels that control the flow of charged atoms. The computer chip mimics the ion channels using an analog method instead of binary so that parameters can be tweaked to match specific ion channels of the brain.
Circuits have been built previously that fire off actions but did not take into account the differences in the intracellular processes involving ion channels. Dean Buonomano, a professor of neurobiology at UCLA, says that this new computer chip is a "significant" advancement and that "the level of biological realism is impressive."
The MIT researchers intend to build systems using this chip for specific neural functions such as visual processing. A system built with this chip could function much faster than digital computers, which can take hours or days to mimic a simple brain circuit even on a high-capacity computer. The analog chip can actually perform even faster than the biological system.
Additionally, researchers can use the chip to build systems to communicate with biological systems, such as to enable control of neural prosthetic devices like artificial retinas. Usage in artificial intelligence is another possibility later on.
[via MIT News]