We propose to infer the temporal inter-relationships between (widely distributed) neural systems during a visuomotor task from scalp electrical and magnetic fields. We will integrate and extend a number of recent technological advances: (1) simultaneous recording of 64 EEG channels; (2) recording of 70 channels of magnetic field (MEG) using repeated measurements from two 7-channel SQUID sensors; (3) measurement of locations of electrodes and sensors with respect to brain areas by 3-dimensional digitizers and magnetic resonance image (MRI) scans; and (4) finite-difference deconvolution and source modeling methodologies, taking into account the geometry of each person's brain and skull. We will record EEGs and MEGs from the same participants performing a task which manipulates hand-specific preparation, requires a go/no-go decision based on miscuing of responding hand, requires a precise finger pressure contingent upon a visual numeric stimulus, and gives visual feedback on performance (in each approximately 5- second trial). Then we will develop 3-dimensional models of simultaneously time-varying current sources using the MEG and deconvolved EEG data, both separately and jointly. For each of a number of intervals, a candidate set of sources will be assembled from the MRI and scalp covariance patterns. An exhaustive search will be undertaken for the best subset of those candidate sources which account for the observed potential and field data. The derived models will be tested on independent data sets. The biophysical validity of the source modeling will be checked by 4 tests using intracerebral data obtained from other collaborative studies. Studies to check the results using positron imagin of blood flow will be formulated. Follow-on studies of patients with focal brain lesions will then be proposed.
