Back in 1992, a group of researchers at WashingtonUniversity in St. Louis, Missouri, formed the Computational Neuroscience Research Group (CNRG) with the ambitious goal of developing a unifying mathematical framework for understanding complex neurobiological systems. The group went on to simulate and evaluate a variety of neural circuits by combining computational and biological approaches.
Nine years later, Chris Eliasmith, a student member of that Missouri group, established at the University of Waterloo in Ontario a group that is based on the same methodologies and principles. In the 5 years since the group was founded, the Waterloo group has advanced the theoretical framework considerably by constructing simulations of perceptual, motor, and cognitive systems.
The Waterloo group is now preparing to move to a new building where they will set up a new Centre for Theoretical Neuroscience. The move will bring with it some new career and training opportunities, including faculty postings and studentships. “It’s exciting to launch a program unlike any other in Canada and get young researchers to participate,” says Eliasmith, director of the CNRG at Waterloo. But Eliasmith's biggest obstacle means opportunity for aspiring computational neuroscientists. “The challenge," he says, "is trying to get people to come here.”
Since starting up the Waterloo team, Eliasmith has co-authored, with his mentor Charles Anderson of Washington University, a textbook outlining their mathematical framework for describing single cells. He is now trying to extend that theory to account for more complex systems. Recently, his approach has been applied to rat navigation, lamprey locomotion, and the control of eye movement. “We are trying to build more biologically plausible models that account for everything we can actually see,” he says.
The work is promising if not yet definitive. “My sense is that Chris is highly respected in the field, but it’s not yet a mainstream approach," says David Van Essen, chair of the department of anatomy and neurobiology at Washington University. "They are still developing the infrastructure and convincing the field that this is a powerful way to go, but there are alternative approaches in computational neuroscience that are also making progress. ... It's not a monolithic enterprise by any means.” Van Essen adds, “The proof will be in the pudding in terms of which types of modeling efforts will make the fastest progress.”
The funding climate
Eliasmith’s group has received funding from several of Canada's traditional science-funding sources, such as the Canadian Foundation for Innovation and the Natural Science and Engineering Research Council of Canada (NSERC). But getting funding for computational neuroscientists in Canada is harder than it is in other countries, Eliasmith says: “There are a number of funding agencies in the European Union and the United States that have targeted theoretical neuroscience specifically, but there is no such thing in Canada yet. I think Canada might catch up, but for now my applications have to go through the main NSERC competitions.”
Without targeted funding for computational neuroscience, the Waterloo researchers and their Canadian colleagues are forced to compete for funding from sources such as NSERC’s interdisciplinary fund, which Eliasmith says is like a catch-all. "If you’re not doing one of their prenumbered areas of research, then you have to go somewhere else. And unfortunately, you don’t get as big a grant in interdisciplinary funding because selection committees are often less familiar with it and it’s harder for them to adjudicate,” he says.
“It’s a bit nerve-wracking, not because there are no special grants for it, but because there’s no special expertise on the review panels,” says Leonard Maler, a professor in the department of Cellular and Molecular Medicine at the University of Ottawa. "Without expert reviewers, it’s difficult to get a fair evaluation." Funding struggles are symptomatic of the need to move computational neuroscience into the mainstream, and the key to that is more people with better training. "I don’t think we need a special pot of money," Maler says. "We just need a larger pool of expert reviewers. I think we’ll see that in the next 5 years.”
Moving into the mainstream
Eliasmith says there are probably about 100 Canadian researchers studying computational neuroscience, but he emphasizes that this is only a rough estimate. “Because there are no departments or programs specific to computational or theoretical neuroscience, people are spread across other departments, with some in engineering, physics, or in the neurosciences, making it hard to know how many there are,” he says. One consequence of the wide dispersion of computational neuroscientists is that the Canadian computational neuroscience community lacks cohesion--and a high profile, which, Eliasmith says, can make recruiting difficult.
“I know a lot of undergraduates who would prefer to stay in Canada if they could but often can’t because there are no programs being offered,” Eliasmith says. “We’re hoping that Waterloo, given that it has a lot of technical resources and that it’s strong in computer science, biomedical engineering, and psychology, can attract Canadian students who don’t want to leave the country.”
Meanwhile, other institutions are getting with the computational neuroscience program. Universities in Quebec, Ontario, and Alberta are considering setting up courses in computational neuroscience that would serve engineering, math, physics, and biology students. Maler has started a course in computational neurosciences at the University of Ottawa, which has generated a lot of interest, with students and faculty from Queens University, the University of Montreal, the University of Laval, McGill, the University of Alberta in Edmonton, and the University of Winnipeg attending the course. “Up until the recent efforts of a few universities, there really has been no way to get training in computational neuroscience in Canada--but things are finally looking up,” says Maler.
“There is no doubt that this is a hot area and people are getting hired,” Eliasmith says. “But things will balance out when more centres are up and running and more graduate students enter the field.”
Staying in Canada --for now
Waterloo offered Bryan Trip, now a third-year Ph.D. student, an opportunity to apply his experience in a completely new context. He never imagined that his job writing patient-tracking software for a large metro hospital in Toronto would lead to the opportunity to model the changes in basal ganglia that result from Parkinson’s disease--but it all fits together, he says: “I didn’t know that there were that many people working in this field, but it turns out that it’s quite a healthy field with lots of jobs.” Still, he expects that he will probably end up doing some of his training internationally. “My best option I think will be to do a postdoc, maybe outside of Canada, and then come back and get an academic position.”
And there are academic positions, in Canada and abroad. Over two recent days, Eliasmith reports, he saw ads on the comp-neuro mailing list for 24 postdoc and assistant professor jobs from across North America. Four more from Canada are sitting on his desk. At Waterloo, two Canada Research Chair positions have remained vacant since the fall of 2005. Offers were made, but the preferred candidates were lured away by U.S. universities.
But for those who manage to find the right training, it’s a seller's market. “If you want to know how the brain works or really understand a complex system, how it can be adaptive, how it can learn--computational neuroscience is the way to go,” says Eliasmith. “There are so many open problems and ways to contribute; you don’t have to be Einstein. As long as you are extremely interested in these questions, there’s a lot of work left to be done.”
Andrew Fazekas is a correspondent at Next Wave and may be reached at firstname.lastname@example.org.
Comments, suggestions? Please send your feedback to our editor.