Mobile Menu

New simulation model suggests synthetic “elementary neurons” may be able to store memories

Written by Miyako Rogers, Science Writer

A new study has developed a simulation model which describes a device that exhibits similar properties to a human neuron. By leveraging advances in nanofluidic technology and using ion carriers instead of electrons, the first components towards a synthetic “elementary neuron” could be built.

The memristor effect

Researchers built simulation models of an “aqueous electrolyte confined into a quasi-2D gap between two graphite layers”. In other words, an ionic solution sandwiched between two layers of graphite. This layer is so thin that it confines the solution down to a single molecular layer, hence why this layer is described as quasi-2D.

Current understandings of ion dynamics in these systems are limited, so researchers developed a new analytical framework to simulate these conditions. According to their simulation models, this system exhibits conductance properties in line with the “memristor effect”.

Figure 1 ¦ Illustration modelling the hypothetical ion-based memristor, an aqueous electrolyte confined between two graphite layers

Memristor, a portmanteau of memory and resistor, describes a device in which electrical resistance can be programmed and “memorised”. Voltage-gated ion channels also change their resistance by opening or closing in response to certain stimuli (ligands, or changes in voltage across the membrane) and this response is “programmed” and “memorised”. Essentially, memristors are the electric equivalents of voltage-gated ion channels. Therefore, devices exhibiting memristor properties can be used as building blocks for the creation of artificial neurons, which can remember responses to certain stimuli.

Neuron-like properties of the electrolyte monolayer

The simulation models were used to investigate the properties of the 2D electrolyte monolayer. Researchers found that unlike in normal solutions, in a monolayer the electric potential is much stronger. They also found that the transport of ions is highly non-linear giving this system’s specific current-voltage properties. These characteristics indicate electrical conduction is mediated by the breakage of ion pairs, much like how a human neuron generates electrical changes through controlled changes in ion concentration.

Another hallmark of memristor behaviour is shown by the formation of pinched loops for frequencies larger than a “threshold frequency”. This results in an alternating electric field, and these voltage spikes are is similar to what we see in an action potential firing in a neuron. A fraction of ions also form “polyelectrolyte structures”. The history of conduction in the electrolyte monolayer causes changes in these structures, thus behaving as an internal “memory” system.

Future directions

This study lays out a theoretical framework for future research to build on. If future studies successfully build an ion-based memristor, this could lead to the development of artificial neurons, synapses and networks. Furthermore, despite advancements in computer processing, modern-day computers still pale in comparison to the processing power and energy efficiency of the human brain. Computer systems using artificial synapses could be much more powerful, as well as better model the human brain, which has big implications for computational biology, neural networks, AI, and more.