Achievement

IGERT trainees Cory Rieth and Flavia Filimon (now IGERT alumnus) started a project at the 2007 bootcamp that has continued and has now spawned another project. The newest project seeks to use machine learning to encourage a stronger definition of mirror voxels. The traditional definition views any voxel responding to all reaching conditions more than baseline as a mirror voxel regardless of differences between reaching conditions. Cory and Flavia are using machine learning over patterns of voxels to extract sets of voxels which are both significantly active compared to the baseline condition and which carry no discriminative information between conditions. So, not only is this set of voxels more active than baseline for observed, imagined, and actual reaching, the pattern of information across voxels is similar for all three. Neurons exhibiting these properties would be predicted in any account which describes neurons as extracting generalizable information from their input.