Local researchers have developed a simulation of a functioning brain — technology that can be used to study ailments and illnesses like brain damage or Parkinson’s disease.
The technology, designed by a team at the University of Waterloo’s Centre for Theoretical Neuroscience, simulates 2.5 million neurons making it the world’s largest simulation of a functioning brain.
While other even larger models of the brain exist, none are able to replicate the brain function, said Chris Eliasmith, director of the centre and Canada Research Chair in theoretical neuroscience.
“It really bridges this brain-behaviour gap where we have this complex model of the brain . . . and it also (creates) the behaviours we think brains are producing,” he explained.
The model, called Spaun (Semantic Pointer Architecture Unified Network), illustrates how brain activity affects the behaviour of a human or animal.
By understanding that connection, researchers can test how changes or damage to the brain will affect behaviour.
“Models like Spaun can give us hints as to how changing part of the model, like damage, introducing chemicals or drugs or another stimulation, how those changes in the underlying biology affect what we see on the output side, the behaviour side,” Eliasmith explained.
These hints can provide insight to addiction, neurological diseases and the effect of dying neurons in process of aging. It can also provide the means to test the effects of treatment.
To simulate the connection between brain activity and behaviour, the network of neurons in Spaun is stimulated by visual images — letters, shapes or numbers — and then required to complete one of eight tasks.
As an example, one task requires Spaun to remember a series of images and write them down.
Like a human brain, Spaun doesn’t always remember each image it is shown in a series.
Despite being the most realistic simulator of a functioning brain, the model is far from complete. There are still areas of the brain — such as the hippocampus that is important in studying Alzheimer’s — it cannot replicate.
Eliasmith said he intends to continue building on the current model to include more brain areas, more sophisticated tasks and simulate learning.
He also plans to collaborate with researchers in neuroscience who are designing more superior computer technology that would speed up the simulation.
“When we run Spaun right now it takes two and a half hours to simulate one second of time,” Eliasmith explained. “Going to specialized hardware would be a natural thing to draw in.”
Eliasmith and his team’s recent findings through Spaun were published in the journal Science this week.
Share this Post