Newly developed artificial neurons receive signals and then send special impulse patterns themselves: researchers have now successfully transferred the complex behaviour of neurons in the nervous system to semiconductor chips for the first time.
The researchers have also succeeded in keeping the energy requirements of these artificial neurons very low. The new technology could thus lead to the development of implants that restore the functions of disturbed neural circuits in the nervous system, such as in neurological diseases or injuries.
What happens in detail during control processes in our body is highly complex. The smallest units in the corresponding signal system are the neurons. They are linked to other nerve cells via extensions and form units that are in turn linked to each other and form the entire nervous system of the body, including the brain.
The neurons of the respective subgroups take on special functions: They receive electrical impulses from higher-level units, to which they in turn react with a special signal behaviour. In the case of the heart, special neurons receive signals from the nervous system, which they convert into control impulses for the optimally adapted performance of the pumping organ.
Repairing neural circuits
If such neuronal control systems are disturbed, physical problems arise. For example, there are heart diseases in which neurons do not react properly to signals from the nervous system and are therefore unable to regulate the pumping power optimally. The same is also true for other medical problems in which nerve functions are impaired by degenerative, pathological or injury-related processes. For this reason, scientists have been working for some time on methods that will enable the repair of defective neural circuits.
However, the development of artificial neurons that react to electrical signals from the nervous system like their natural counterparts has proven to be a tricky challenge. But now a team of researchers led by Alain Nogaret from the University of Bath are reporting a breakthrough in this field.
They have initially succeeded in determining exactly how certain neurons react to electrical stimuli from other nerves. They have then modelled the corresponding behaviour by means of equations. As they point out, this was tricky because the neurons' responses to impulses are complex: If a neuron receives a signal that is twice as strong, this does not necessarily mean that its response is twice as strong - a completely new impulse pattern can result, the researchers say.
Nevertheless, in the end they were able to precisely simulate the natural reaction behaviour of nerve cells. This is a breakthrough because we have now developed a method to reproduce the electrical properties of real neurons down to the smallest detail, says Nogaret.
Silicon chips programmed with 'neural software'
The researchers then developed silicon chips that can be programmed with the reaction behaviour of natural neurons. They have now first given them the characteristics of hippocampal neurons and respiratory neurons of rats. Tests have shown that the chips react to a variety of impulses in the same way as their natural counterparts. As the researchers emphasize, the artificial neurons have another important feature: They only require 140 nanowatts of energy. This makes these artificial neurons well suited for the development of bioelectronic implants, says Nogaret.
The researchers already have concrete applications in their sights. For example, it might be possible to develop intelligent pacemakers that can stimulate the heart not only to beat evenly. Using artificial neurons, these devices could adapt the heart's performance to the requirements in real time, as is naturally the case. In addition, applications for the treatment of neurodegenerative diseases such as Alzheimer's are also conceivable, the scientists say. The replication of neural responses through bioelectronics, which can be miniaturised and implanted, opens up enormous opportunities for the development of intelligent medical technology.