Silicon Synapses


A new kind of computer chip mimics the way a neuron learns, a new study reports.

Behold! The mighty Synapse Chip!

The 400-transistor chip simulates the activity of a single synapse – a connection between two neurons. Because of the chip’s complexity, it’s able to mimic a synapse’s plasticity – its ability to subtly change structure and function in response to new stimuli.

For example, a synapse that repeatedly responds to an electric shock might, over time, become less sensitive to that shock. Thus, synaptic plasticity forms the basis of neural learning, well below the level of conscious processing.

The human brain contains approximately 100 billion neurons, and more than 100 trillion synapses. Ever since the anatomist  Santiago Ramón y Cajal discovered the function of neurons back in the early 1900s, scientists have dreamed of building a device that replicated the behavior of even a single synapse. For decades, they had to content themselves with mathematical models and digital simulations.

But now, as the journal Proceedings of the National Academy of Sciences reports, a team led by MIT’s Chi-Sang Poon has constructed a working silicon model of a synapse in the physical world.

The chip uses transistors to mimic the activity of ion channels – small “ports” in the cell membranes of neurons that allow various amounts of neurotransmitter chemicals and ions (positively or negatively charged atoms) to pass in and out of the cell. These channels form the basis for synaptic communication – and for some of the most hotly researched topics in neuroscience.

While ordinary transistors act as binary “on/off” gates, neural synapses conduct signals along fairly smooth gradients, allowing the signals to increase in strength until they finally trigger the neuron to “fire” a signal on to its neighbor(s). It was this gradient property that Poon’s team sought to mimic with their new chip:

While most chips operate in a binary, on/off mode, current flows through the transistors on the new brain chip in analog, not digital, fashion. A gradient of electrical potential drives current to flow through the transistors just as ions flow through ion channels in a cell.

Since the researchers can also tweak the chip’s properties to mimic different kinds of ion channels, this provides one of the most realistic models yet for studying how individual synapses actually work.

The researchers have already used the chip to study long-term depression (LTD), the process by which some ion channels can weaken the synaptic activity of others. They also hope they’ll soon be using chips like this one in conjunction with lab-grown biological neurons, to discover all kinds of exciting new things about how cells behave “in the wild.”

Who knows – by this time next year, we may be watching nature documentaries about Neuron Cyborgs – but my guess is that the SyFy channel will get there first.

Brief and tangentially relevant side-note: Connectome posts may be somewhat spotty over the next few weeks, as I’m currently launching a new project, the details of which need to be kept under wraps (or “on the DL,” as the kids say) for the time being. I’ll do my best to report on neuroscience breakthroughs as often as I can during this period, and things should be back to (relatively) normal soon. Thanks for stickin’ with me.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s