Theoretical Neuroscience

The emerging field of theoretical neuroscience is uniquely focused on developing new ways of understanding the unrivalled complexity and flexibility of the behaviour of biological systems. Researchers in theoretical neuroscience use quantitative tools to study neural systems, which include single neural cells as well as small and large networks of cells (including the whole brain). The main methods used in such studies are mathematical analyses and computational simulations, which employ dynamic systems theory, information theory, signal analysis, control theory, complexity theory and so on. Historically, theoretical neuroscience (also often referred to as computational neuroscience) has its roots in artificial neural networks (ANNs). However, the focus of theoretical neuroscience has become more biological, and hence more concerned with reverse-engineering actual biological systems. This focus contrasts with that of the ANN community, which applies ANNs in an attempt to solve various engineering or artificial intelligence problems, or uses ANNs as abstract models of cognition.

In short , theoretical neuroscientists are centrally interested in understanding the computational and mathematical principles behind brain function. Practically speaking, this research is important for both our understanding health-related issues and the development of future technologies. Already, a better understanding of neural systems has lead to important breakthroughs in treating brain-related illnesses. For instance, theoretical models allow us to understand and tailor medical interventions, such as the treatment of Parkinsonian tremor with deep-brain stimulation. Similarly, theoretical neuroscientists model the effects of various pharmacological interventions, improving our understanding of their precise effects (e.g., serotonin reuptake inhibitors used to treat depression and other anxiety disorders). Other health-related applications include brain-machine interface development, resulting in the construction of more effective prosthetics. New theoretical insights into neural coding have recently allowed monkeys to control robotic arms through direct neural implants.

In the context of information technologies, the insights gained from studying a successful, complex and behaviourly sophisticated system should prove ground-breaking for developing the next generation of intelligent machines. It has been clearly established that we do not have appropriate tools for engineering large, complex systems (for instance, only 30% of large software projects are successful). However, nature has solved such problems in ways we do not currently understand. The study of theoretical neuroscience provides new insights into nature’s solutions. In some areas (e.g., face recognition), neurally inspired algorithms are the state-of-the-art. However, there is much more we can learn regarding complex control, pattern recognition, learning, and dynamical systems from reverse-engineering the brain.

In recognition of the importance of theoretical neuroscience, significant international commitments of research funding have recently been announced, including programs from the European Union (Cognitive Systems Program, $72 million/5 years), the United States (Defense Advanced Research Projects Agency (DARPA) (Biologically Inspired Cognitive Architecture Program), National Science Foundation, National Institutes for Health (NSF/NIH) (Collaborative Research in Computational Neuroscience, $6 million/year), and Japan (RIKEN Institute, Creating the Brain Division $9 million/year). Unfortunately, Canada does not have similarly targeted funds, although Natural Sciences and Engineering Research Council of Canada (NSERC), Canadian Institutes of Health Research (CIHR), and Canadian Foundation of Innovation (CFI) fund research in the area through their general programs. As a result, there are few institutionalized organizations for theoretical neuroscience in Canada.