Cross-Disciplinary Feasibility Account : Computational Statistics and Cognitive Neuroscience

The capabilities of functional magnetic resonance imaging (fMRI) and magnetoencephalography (MEG) are providing tantalising new frontiers of investigation for clinical and cognitive neuroscientists studying normal and diseased brain function. In Glasgow cognitive neuroscientists are undertaking internationally pre-eminent research into auditory cognition, face recognition, and sensory integration all of which need to exploit the potential of fMRI and MEG to the full. At the same time Glasgow computing scientists and statisticians working with life-scientists are leading the way in multi-disciplinary research developing the novel methodologies necessary to meet the challenges of the systems view required to elucidate complex biological mechanisms. Whilst cognitive neuroscience and systems biology seem quite orthogonal domains of research they share the need of evidence-based scientific inference. They also often share common data structures, involving large p (number of variables) and small n (number of cases) regimes. This therefore provides an exciting opportunity to explore the potential of inferential techniques developed in one domain to problems in the other. There is therefore an urgent need for the acceleration of strong engagement in cross-disciplinary research activities of the statistical inference, computational science, and cognitive neuroscience research groups to ensure that the UK maintains an internationally leading research profile. Novel experimental procedures can now be developed to study brain function given the unique resource of dedicated fMRI and MEG research facilities at Glasgow. However the computational statistical methods of data modelling and analysis necessary to make sense of and interpret the resulting data need to be researched and developed in parallel and in partnership with the neuroscientists. We are on the verge of making significant breakthroughs in our understanding of processes such as auditory cognition but the advanced inferential machinery must be developed and made available in a systematic and coherent manner. We wish to be established as a leading international facility, which has cognitive, computing, and statistical scientists working in concert driving forward systems cognitive neuroscience. Given our successful track record in multi-disciplinary research with life-scientists ( see group webpage www.dcs.gla.ac.uk/inference) the research group is seeking to make a major impact in this new multi-disciplinary research endeavour.

This collaborative venture has arisen at a very opportune moment, as the need for computational statistical methodological developments to support current cognitive neuroscientific enquiry is very pressing. There are already plans to develop these collaborations further; one at the institutional level is the joint appointment of the PI in the Computing Science and Statistics departments, the others are at the research level where formal discussions are taking place with the department of Psychology to institute cross-departmental research programmes. The three CDFA investigators are at the forefront of this effort and the award of the CDFA will prime existing aspirations to establish a cross-disciplinary internationally leading expertise at the crossover between Psychology, Statistics and Computing Science. The existence within the University of very strong research groups in all three of these areas provides an outstanding and unusual opportunity to create an internationally leading centre of expertise in this key topic. Each of the existing constituent research groups has a wide network of international contacts, which will provide the new research activity with a set of significant potential collaborators. We aim to identify those where collaboration would be most effective, with the strategic aim of placing our research group as a major player on the international stage. Careful thought will be given to any potential routes to commercial exploitation of this proposal, with the advice of the University's Research and Enterprise office. However, it is the scientific exploitation and application of our research which we believe will be most immediate and most valuable and this will therefore form the primary focus of our exploitation activity. We expect that potential for significant advances will be identified in each of the three contributing areas. In neuro-imaging we plan to demonstrate through publications what the new tools can achieve and to promote and support their use through contact with key neuro-imaging groups internationally. In the computational arena the principal contribution will be in the development and assessment of algorithms which can be adopted by researchers tackling similar problems in different application areas. In statistical methodology it is likely to be the contributions to spatiotemporal modeling which are of principal benefit to the community. This will include not only statistical researchers but also a wide variety of other researchers, for example in areas such as environmental science, who are actively engaged in the critical activity of quantitative modeling of scientific data. Areas of potential impact will be identified through the in-depth discussions of the project team, with each of the contributing partners assessing advances through the possible benefits in their own disciplines. We expect that the benefits in cognitive neuroscience will be most obvious and immediate but we will be equally active in identifying benefits elsewhere. In addition, we aim to include within the agenda of the final workshop a specific component of exploitation, where new areas of application can be identified. This proposal aims to explore, in an adventurous manner, potential new directions in the analysis of cognitive neuroscience. A clear indication of the potential benefit in doing so is indicated by the recent creation of the UK Node of the International Neuroinformatics Coordinating Facility (INCF; www.incf.org), which coordinates neuroinformatics activity on a global scale and which is funded by EPSRC and other research organisations. One of its current activities is to construct a roadmap for future research directions. New ideas and methods for the analysis of brain images therefore clearly have the potential to exert a very positive influence on the future research of a very large and highly active research community.