Scientists have successfully connected electronic chips to neurons

Apr 2, 2006 19:41 GMT  ·  By

Marvin Minsky once presented the following argument: Suppose you replace one neuron in the brain with a silicon chip. What happens? Well, given that you have at least one hundred billion neurons, nothing special. Then, suppose you replace two neurons. What happens? Again, nothing special. But then suppose you replace them all. Would you lose anything? Minsky thinks not.

This is an argument for the idea that, in principle, computers can have the same mental abilities humans have. When Minsky presented this argument it wasn't actually possible to replace a neuron with an electronic chip. But now, European scientists have managed to do precisely that.

Connecting life and electronics

The main problem they had to face when trying to connect a neuron to an electronic chip was that the electric current of neurons is of different nature from that of electronic chips. The electric charges used by neurons are ions, while the silicon chips use a flow of electrons. The scientists have now overcome this problem by using a special protein which is naturally found in the brain and which can both glue the neuron to the chip and provide the needed two-way transmission.

The proteins "provided the link between ionic channels of the neurons and semiconductor material in a way that neural electrical signals could be passed to the silicon chip," said study team member Stefano Vassanelli from the University of Padua in Italy. This way, brain cells were able to communicate with the electronic chip: electrical signals from neurons were "sensed" using the chip's transistors, while the chip's capacitors were used to stimulate the neurons. Each chip contains more than 16,000 electronic transistors and hundreds of capacitors on an area of just 1 millimeter square.

What do neurons do?

To rephrase the question, what kind of behavior should the silicon chip emulate? Neurons are, of course, living creatures and have a staggering internal complexity, but their action is quite simple: they receive signals from other neurons and, if the sum of all these signals is higher than a certain threshold, they send another signal themselves. The complexity of human thoughts and feelings does not arise from the neurons themselves, but from the complexity of their networking. Or, at least, this is how some of our mental complexity arises.

So, the transistors and capacitors inside the chip have to be put together in such a way as to mimic this simple neuronal behavior, but in the same time to allow each chip to be connected with a large number of neurons or other chips.

Is this enough?

Enough for what? Enough for eventually simulating the actions of human brain. The answer is no.

Neurons (that is, natural neurons) do not bind together directly like the elements of an electric circuit, they bind via a chemical mechanism. Inside the synapse (the binding between two neurons) there are some chemical substances, called neurotransmitters, which facilitate the transmission. The reason why a neuron "feels" that a certain signal has been sent to it is that neurotransmitters bind to certain receptors it has. This binding of neurotransmitters to the neuron causes the electric current inside the neuron.

But when scientists connected the chips with the neurons, they have overridden this chemical mechanism. They have connected the chips directly into the electric circuits of neurons. So what? Well, a lot of things that happen in the brain and implicitly in the organism are due to the neurotransmitters and not to the neuronal network - certain processes in the organism produce neurotransmitters, others change the neuron's ability to bind to certain neurotransmitters and so on. To name only one example, depression is correlated with the neuron's ability to bind to serotonin, a neurotransmitter.

What's next?

So, although this paves the way for both chip transplants in the brain or, conversely, computer systems using live neurons, the researchers say that such advancements are still decades in the future. For example, one issue they have to solve more properly is the chip ability to send signals to the neurons without damaging them. I suspect they might be overoptimistic when they think only about a few decades.

However, on the shorter term, this technology is expected to provide an advanced method of screening drugs for the pharmaceutical industry. "Pharmaceutical companies could use the chip to test the effect of drugs on neurons, to quickly discover promising avenues of research," said Vassanelli.