Biological/organic computing has long been the stuffs of science fiction. But as we have all come to see, what is science fiction one day, does become science fact the next.

Researchers have made a small breakthrough in the terms of biological computing. Specifically, the possibility of human and animal neurons have great potential in terms of computing speed. Neurons perform some very impressive tasks in terms of visual sensory input and pattern recognition, and all in real time (or pretty darn quick). If some form of neurons could supplement a conventional computer system, the power of that system could be increased tens, if not hundreds or thousands fold. Another benefit is the small size, three dimensional configuration, and parallel computing aspects that living neurons have.

A first of its kind, this recent breakthrough could be the first small step towards biological computing. The scientific journal Physical Review E has just published a paper (read it here) that highlights some of the new possibilities of using neurons for biological computing.

Researchers involved in this experiment needed to be absolutely certain that living neurons would function in a consistent and predictable manner. By using a small petri dish studded with an array of electrodes, they were able to track the firing patterns of neuron samples when invoked to fire a signal. And as many had found before, neurons that are grouped together will fire in a synchronized bursting event (SBE). When one nerve fires a signal, it would trigger those around it, sending that signal through the network of nerves. A very natural reaction for organic nerves, and a very predictable one for any given type of nervous system.

Unfortunately, nerve cells do not generate these SBE firings with any consistent pattern when left on their own. Researchers needed to train independent nerve cells to act in a more predictable fashion. The first step was to try and invoke some type of memory in the nerve cells by treating them with calcium. Calcium will cause nerves and cells to react and or fire. Their subsequent tests with a calcium solution did not yield any promising results, as the nerves would revert to their random state after the calcium reagent was removed.

The researchers theorized that inhibitory neurons were blocking the learning process from their neighbor nerve cells. So, they introduced a chemical that inhibited the functions of the inhibitory nerve cell, allowing the other cells to learn. By repeating this step (introduction of the inhibitory chemical) every twenty seconds, they were successfully able to teach the cells several different SBE firing patterns. And by dosing different collections of cells, they created different SBE groups without affecting any of the other cell clusters. Simply put, they could now layer different memories onto a single collections of cells. A biological Quad Core processor in a sense; multiple threads across the same system.

It may not sound like much, but it is the first step. We won't have living computers anytime soon as more elaborate functions must be taught to neurons and cells to even begin to perform the most basic of fencings. And odd as it may seem, the majority of this research simply relied upon the natural connections that neurons made when placed in a culture. Just like a dog, these neurons want to learn (on a very basic level), all we have to do is teach them.