Research

Research_Icon1_NeurocompNeurocomputational mechanisms for visual face and object recognition. What is the featural code underlying human face and object recognition? How are objective image properties mapped onto neural representations? What computational principles and algorithms account best for this mapping? Our research addresses these questions in the context of a comprehensive empirical and computational framework. Concrete instances of this line of research concern and serve to establish: (i) the image-based code of neural representations; (ii) the statistical structure of human face space; (iii) the role of surface properties such as color in visual recognition, and (iv) the relationship between different visual categories with respect to their perceptual processing (e.g., faces and visual word forms).

 

Research_Icon2_recoveryVisual recovery after severe brain injury and visual disturbance. How does the visual system cope with perceptual deficits whether developmental or due to brain injury? What neural mechanisms underlie visual recovery (or lack thereof)? The investigation of such issues can illuminate core principles of neural organization and plasticity in the human brain and also facilitate potential strategies for rehabilitation. We examine these issues in patients who have undergone a hemispherectomy procedure (i.e., patients who had one hemisphere surgically removed or functionally disconnected) as well as in individuals with more specific visual deficits (e.g., ‘face blindness’/prosopagnosia). This line of research is based on a collaboration with Drs. Marlene Behrmann and David Plaut at Carnegie Mellon University.

 

Research_Icon3_methodologyDevelopment of new methodology for the analysis of neuroimaging data. How can we best elicit, extract and interpret the relevant signal present in neural data? While neuroimaging research yields a wealth of empirical data, the conclusions of this research and their support are only as strong and reliable as our methods allow them to be. Our work investigates novel approaches to the analysis of fMRI and EEG data using techniques originating in machine learning and applied statistics. As concrete instances, we explore and assess multivariate methods for functional brain mapping, connectivity and neural-based image reconstruction.