If We Only Had a Simulated Brain

Neuroscientists at the University of Waterloo’s Computation Neuroscience Research Group (CNRG) have built the world’s largest functional model of the human brain. The Spaun (Semantic Pointer Architecture Unified Network) simulated brain has been touted as the most complex computer-generated version of grey matter ever. It can learn, remember, and recognize items, and even write down numeric sequences using its robotic arm.

That controller arm, by the way, was built using Maplesoft’s MapleSim simulation software. Spaun uses a digital eye to “see,” and the robotic arm to write responses down on paper when the system is instructed to perform certain tasks. Using the simulation software, the design team constructed an arm model with nine muscles and three links (shoulder, elbow, wrist) in MATLAB.

Using specific brain regions as a map, the system’s creator (Chris Eliasmith) organized Spaun’s 2.5 million simulated neurons to mimic the basil gangilia, thalamus, and prefrontal cortex. While there are other electronic brains out there (like IBM’s Watson and Blue Brain projects), what makes Spaun unique is that its brain-like computing power is linked to specific behaviors.

“This is the first model that begins to get at how our brains can perform a wide variety of tasks in a flexible manner—how the brain coordinates the flow of information between different areas to exhibit complex behavior,” said Eliasmith, who serves as Director of the Centre for Theoretical Neuroscience at Waterloo.

Eliasmith also wrote a book about the project, How to Build a Brain. You can also read a nice description of how the system works over at Popular Science.

Below, I’ve pasted a video of Eliasmith giving a TED talk on the subject. You can see other videos about how Spaun works here.

Source: Maplesoft