Trans - European synapse
The natural neuron was connected to silicon via the Internet
Alice Bakhareva, N+1
Scientists have created a system that connected artificial and living neurons through a memristor. It is noteworthy that three elements of this hybrid were located in different parts of Europe and were connected via the Internet.
A drawing from the University of Southampton New study press release allows Brain and Artificial Neurons to link up over the web - VM.
Despite the distance, the network functioned and displayed the properties of living neural systems, for example, long-term potentiation. The article was published in the journal Nature Scientific Reports (Serb et al., Memristive synapses connect brain and silicon spiking neurons).
The functioning of the brain is based on neural networks. Synapses – connections of neurons - play a key role in the transmission, processing and storage of information between the cells of these networks. Modern technologies make it possible to create artificial neurons and synapses, as well as connect the brain and computer.
Memristors allow simulating such a property of a biological synapse as long-term potentiation: with prolonged exposure of one neuron to another, signal transmission between them becomes more efficient. These are electronic elements that change their resistance depending on what charge flows through them.
A group of scientists from the UK, Germany, Italy, and Switzerland, led by Alexantrou Serb from the University of Southampton, created a system that connected artificial neurons with biological ones using a memristor and could transmit signals in both directions. The first element of the network is a silicon neuron, which is an integrated circuit consisting of millions of transistors. This device generated electrical impulses that were transmitted to a memristor, and then through a microelectrode to a mouse hippocampal neuron isolated in culture. The voltage applied to the nerve cell resembled excitatory postsynaptic potentials (VPSP), from which neural impulses are formed in the brain. Such a hybrid synapse was called a synaptor.
The general scheme of the circuit. AN – artificial neurons, MR – memristors, ABsyn and BAsyn – synaptors. Here and below are the figures from the article by Serb et al.
In order to simulate the plasticity inherent in any synapse, a signal was sent to the memristor through two poles. The first served as an analogue of presynaptic stimulation, it received excitation from an artificial neuron. The second served as a postsynaptic input and returned a response from a biological neuron to the memristor.
Electrical circuit diagram. AN – artificial neurons, MR – memristors, ABsyn and BAsyn – synaptors, CME – microelectrode, BN – biological neuron.
The second part of the system was created to transmit a signal from a living cell to a silicon one. The neuron pulses were recorded using the method of local potential fixation (patch-clamp), then they were fed through the microelectrode to the second memristor and through it to the artificial neuron. The result was a hybrid circuit that transmitted a signal from one silicon nerve cell to another living neuron.
Even more exotic to the study is the fact that the elements of the system were located in different parts of the world: silicon neurons were in Zurich, memristors were in Southampton, and the culture of mouse neurons was in Padua, Italy. The system used the UDP protocol to transmit data over the Internet.
To demonstrate the properties of synaptors, the researchers decided to simulate the long-term potentiation of glutamatergic synapses of the hippocampus on them. The first artificial neuron performed the function of a rhythm driver: it produced electrical signals of a certain frequency. Memristors played the role of a postsynaptic membrane, which carries the function of plasticity in the brain. They were programmed to change the resistance in response to the discharge frequency of the biological neuron, which was recorded through the postsynaptic input. AMPA glutamate receptors of hippocampal cells work on the same principle. The second artificial neuron of the network worked in spontaneous discharge mode – it spontaneously gave out impulses without a set frequency, and the biological cell influenced its activity through a memristor.
As a result, in response to a period of high-frequency impulses set by an artificial neuron, a living cell increased its activity and maintained it even after a decrease in the frequency of irritation. This also led to an increase in the spontaneous activity of the third element of the chain. With a decrease in the frequency of discharge of the pacemaker, long-term depression developed, in which the activity of both the second and third neurons of the system decreased.
Long-term potentiation (LTP) and depression (LTD). From above – the discharge frequency of the first neuron, in the middle – the second (biological) neuron, from below – the resistance of the memristor.
This is the first network of this type, in the future it can be improved and applied to such medical tasks as therapy of cardiac arrhythmia, hypertension, spinal cord injuries and Parkinson's disease.
The properties of memristors are also used to create chips that can be used in machine learning. For example, in 2015, a neural network consisting of memristors was created.
Portal "Eternal youth" http://vechnayamolodost.ru