New advance in brain-machine interfaces: Artificial neurons command Venus flytrap to snap shut

Experts believe that within a decade brain implants and sensors will be commonplace. Neurotechnology – the field that develops tools to interact directly with the nervous system – makes incredible promises and causes understandable misgivings. Its advance is unstoppable and it is nurtured with billions of dollars from investors such as Tesla founder Elon Musk and Facebook. But neuroscience is so complex it must move in very small steps towards that future. Now there is a new focus that could prove to be of huge benefit to this path: an artificial simulation of a neuronal connection made of organic material, which will allow for a better connection with living cells.

This organic artificial neuron was able to take control of a carnivorous plant and, by sending the correct sequence of electrical impulses to its cells, cause it to close its insect-trapping mechanism. Using this circuit, impulse spikes can be modulated, which represents a “significant achievement,” one that adds another option to the toolbox of devices capable of simulating neural functions, according to the researchers who developed it. “It will potentially allow us to build the basic components of our brain: neurons and synapses,” says Simone Fabiano, one of the authors of the paper, which has been published in Nature Communications.

Carnivorous plants have been controlled in the past with electrical stimuli; the team used the process as a model because it was important to test if the artificial organic neurons were able to bio-integrate with living tissue. “Venus flytraps are easy to handle and were an easy choice for an initial demonstration. However, the possibility of modifying the electrophysiology of the living systems via artificial neurons could be expanded to other biological systems, and we are investigating that with animal models,” says Fabiano, of Linköping University in Sweden.

Having tested their device in this simple scenario, the researchers are now considering possible future uses such as brain implants and connecting humans to the internet of things. In the short term, it could be used “to detect, process and order a specific action,” such as moving the Venus flytrap. In the longer term, there is talk of artificial neuron connections to achieve machine learning in computers. “In the future, they could connect directly with biological neuronal networks for brain-machine interfaces,” Fabiano explains. That is the new Holy Grail of Silicon Valley: connecting human brains directly to computers and the internet.

Javier DeFelipe, a neurologist at the Spanish National Research Council (CSIC), says that “they are trying to imitate what the circuits of the brain and the nervous system do. But they have to take it step by step, because we don’t know exactly how they function. If you achieve something that can be reproduced, a small function as in this case, it is a significant advance: it is achieving biocompatibility that is closer to what is a biological cell.”

However, DeFelipe, who did not participate in the study, adds a note of caution: “Now they put an electrode in the brain and stimulate it with an electrical current and you hear a voice or move a muscle, but that does not imply that the entire circuit is understood to be able to reconstruct it entirely. It is one thing to intervene in the function of this neuron in a more organic way and another to recreate all of its complexity.”

The artificial neurons that have been developed in Sweden are based on components that can carry ions and electrons, the elements that transmit the basic impulses of neuronal communication. They can be molded at room temperature on plastic or paper and printed using cheap screen-printing materials, of the kind used for t-shirts. “This would be inconceivable with silicon-based electronics,” says Fabiano. These types of silicon circuits, like the classic computer chips, are what the researchers are attempting to circumvent in this case, as they do not integrate as well with biological organisms.

“The basic components of our artificial neurons allow for direct sensorial fusion between the neurons. This will enable us to develop systems that can feel, process and act, and as such introduce the decision-making process in devices. This could be used to monitor health, in brain-machine interfaces and in robotics,” says Fabiano. The next step in the process, he explains, will be to make his devices achieve the frequency and energy efficiency of real biological neurons.

According to DeFelipe, though, science is still a long way from developing an artificial neuron: “We are still attempting to understand how it works; this is another step in the development of these tools, which will be better adapted to the brain than those made of silicon, but one thing is the material with which the circuit is constructed; making it behave like a real cell is another,” he says.