Artificial intelligence: a chip designed to simulate brain activity
Last reviewed: 23.04.2024
All iLive content is medically reviewed or fact checked to ensure as much factual accuracy as possible.
We have strict sourcing guidelines and only link to reputable media sites, academic research institutions and, whenever possible, medically peer reviewed studies. Note that the numbers in parentheses ([1], [2], etc.) are clickable links to these studies.
If you feel that any of our content is inaccurate, out-of-date, or otherwise questionable, please select it and press Ctrl + Enter.
For many decades, scientists have dreamed of creating a computer system that could replicate the talent of the human brain to explore new challenges.
Scientists from the Massachusetts Institute of Technology have now taken an important step toward this goal by developing a computer chip that mimics the mechanism for adapting brain neurons in response to new information. This phenomenon, known as plasticity, is believed by scientists to underlie many brain functions, including learning and memory.
About 400 transistors and a silicon chip can simulate the activity of a single brain synapse - a connection between two neurons, which facilitates the transfer of information from one neuron to another. Researchers expect that this chip will help neuroscientists learn much more about the work of the brain, and can also be used in the development of neural prostheses, such as artificial retina, says project manager Chi-Sang-poon.
Simulation of synapses
In the brain, there are about 100 billion neurons, each of which forms synapses with a large number of other neurons. Synapse - the gap between two neurons (presynaptic and postsynaptic neurons). The presynaptic neuron secretes neurotransmitters such as glutamate and GABA, which bind to receptors on the postsynaptic membrane of the cell, activating the ion channels. Opening and closing of these channels leads to a change in the electrical potential of the cell. If the potential changes dramatically enough, the cell triggers an electrical impulse called the action potential.
All synaptic activity depends on the ion channels, which control the flow of charged ions, such as sodium, potassium and calcium. These channels are also key in two processes known as long-term potentiation (LTP) and long-term depression (LLC), which respectively strengthen and weaken synapses.
Scientists have developed their own computer chip, so that transistors can mimic the activity of various ion channels. While most chips work in binary mode - "on / off", the electrical currents on the new chip flow through the transistors in analog mode. The gradient of the electrical potential causes the flow to flow through the transistors in the same way that the ions pass through the ion channels in the cell.
"We can adjust the circuit parameters for concentration on a specific ion channel," says Poon. "Now we have a way to capture every ionic process that happens in the neuron."
The new chip represents "significant progress in efforts to study biological neurons and synaptic plasticity on a CMOS [complementary metal-oxide-semiconductor] chip," says Dean Buonomano, a neuroscience professor at the University of California from Los Angeles, adding that "the level of biological realism , is impressive.
Scientists plan to use their chip to create systems for modeling specific neural functions, such as a visual processing system. Such systems could be much faster than digital computers. Even on high-performance computer systems, hours or days are required to simulate simple brain circuits. With the analog chip system, simulation is faster than in biological systems.
Another potential application of these chips, adjustment of interaction with biological systems, such as the artificial retina and the brain. In the future, these chips can become standard blocks for artificial intelligence devices, says Poon.
