The vast amount of information about our brains that is generated everyday is giving neuroscientists a bit of a headache. The fragmentation of the field into specialisms as diverse as genetics, psychology, and behavioural studies has caused them to lose sight of the brain as a whole. To cope with the collection and analysis of huge amounts of neurological data a new discipline is emerging, combining the knowledge of neuroscientists and computer scientists. Neuroinformatics is the study of information processing in the brain. It is concerned with the development of new computational tools such as databases and modelling systems, but it also encompasses research into the similarities and differences between artificial and natural intelligence.
Part of the broader field of bioinformatics, Karl Friston, principal fellow of the Wellcome Institute of Neurology, believes that neuroinformatics "represents one of the most fundamental endeavours of science, namely to understand how the brain works. This has been facilitated enormously by recent advances in functional neuroimaging and computing power" that allow for more biologically plausible models of neuronal computation.
You may think that this important area of research is only open to those with years of experience in theoretical mathematics or obscure neurological backgrounds. However, almost anyone with computing, maths, or life science training can move into this novel area of study. "Neuroinformatics is a new field that attracts people from diverse backgrounds. Sometimes it is more important to have the right questions than the right training," says Friston.
Neuroinformatics research falls into four main areas: the development of neuroscience databases; the development of new tools for the retrieval and manipulation of data; the improvement of experimental and theoretical tools for the analysis of information processing in the brain; and the development of better robots and machines by studying the nervous system as well as building robots that mimic animals to further understand the nervous system.
Kyran Dale is studying for a PhD at the Centre for Computational Neuroscience and Robotics at Sussex University, where he is researching insect navigation using evolutionary robotics. The artificial neural networks which evolve can be used to guide robots in a simulated environment. Conversely, Tim Chapman is using robotics to help understand the nervous system. He is studying the cricket escape response for his PhD at Stirling University. "The emphasis is on trying to use robotics to model the biology, rather than trying to take ideas from biology and use them to develop better robots," says Chapman.
The researchers came to the field from contrasting backgrounds too. Having moved into neuroscience with virtually no science training (a BA in philosophy and politics and an MSc in knowledge-based systems), Dale has had to teach himself a lot of computer programming and feels that "at this level and in this area everyone is pretty much self-taught. Unfortunately there aren't any off-the-shelf solutions." Chapman, who graduated from Nottingham University with a BSc in psychology with artificial intelligence, thinks having a background in psychology has helped him considerably. "So many people who work in robotics are computer scientists or electronics engineers who have no background in the biological/psychological side of things, which for me is the more interesting problem because technology is just technology," he says.
The excitement of working in an emerging, multidisciplinary field was one of the main factors Dale considered when choosing to do his PhD. "There seemed to be a good opportunity to apply some of the new techniques currently being generated by artificial life/evolutionary robotics research to an area that seems to be crying out for new modelling approaches," he explains. Chapman also enjoys working in a domain that bridges the gap between disciplines. "One advantage is that you can learn a great deal by applying the ways that one discipline looks at their world to other disciplines," he suggests. "Computer scientists and roboticists have learned a lot from biology, for example, neural computation techniques and clever techniques for exploiting the environment in the way that animals do. Biology benefits by using robots to test out hypotheses that would be impossible to test in the animal--due to the damage caused by the surgery necessary to implant a recording electrode, for example. With a model you are able to record from every neurone in the simulated nervous system, whilst it is performing a behaviour," he says.
It's certainly not all esoteric, blue-skies pondering. The development of intelligent, adaptable robots built on the principles of neurobiology will have a major impact upon industry. And neuroinformatics is key to further development of brain imaging as a medical diagnostic device. Huge numbers of brain images have to be stored and handled on sophisticated databases. Software and biomedical companies are interested in the development of commercial systems for brain imaging to facilitate the manipulation of thousands of patients' brain scans.
One thing is for sure. This is only the beginning--so, think on. See a list of online resources to get you started. ...