Neuroscience Technology

Synapses for AI Made in Lab – Simulates Human Brain

brain synapses

One of the biggest problem hampering artificial intelligence development is the lack of understanding of the human brain and trying to figure out how to mimic it. In an article published in ACS Nano, a group reported that they have managed to develop artificial synapses that can simulate one of the fundamental functions of the nervous system — the release of both stimulatory and inhibitory signals from the same “pre-synaptic” terminal.

Synapses are structures that enable neurons to pass chemical and electrical signals to each other. The human nervous system contains more than 100 trillion synapses. These synapses can inhibit and initiate biological messages in mammals. Many synapses are simply used to relay one type of signal, while others can either convey both types at the same time, or are able to switch between the two types. While developing artificial intelligence systems that mimic human learning, image recognition and cognition better, researchers typically imitate synapses in the lab by using electronic components. Most artificial synapses developed thus far are however only capable of delivering one type of signal. Jing Guo, Han Wang, and their colleagues looked for ways to create an artificial synapse that could be reconfigured to send noth inhibitory and stimulatory signals.

The team have developed a synaptic device that reconfigures itself depending on the voltage applied at the input terminals of the device. A junction manufactured from tin selenide and black phosphorus enables switching between the inhibitory and excitatory signals. The device is versatile and flexible, both characteristics that are highly appropriate for artificial neural networks. The artificial synapses may also in future simplify the functions and design of nervous system simulations.