Ph.D Thesis


Ph.D StudentBenyamini Miri
SubjectNeural Error Processing in Continuous Reaching Movements and
Experiments with Brain Computer Interface (BCI)
DepartmentDepartment of Mechanical Engineering
Supervisor PROF. Miriam Zacksenhouse
Full Thesis textFull thesis text - English Version


Abstract

Error processing during movements is essential for both on-line error correction and motor learning. Research in error processing contributes to understanding human motor control and learning, with applications to skill acquisition, motor rehabilitation and brain-computer interfaces (BCIs). BCIs may generate more errors than those encountered during normal motor control. Thus, they provide an opportunity to investigate neural correlates of error processing. In our work we characterized neural correlates of error processing in two cases: (1) natural errors during experiments with continuous invasive BCI of finger movements, and (2) imposed errors during continuous reaching movements. Significant characterization and successful detection of neural error processing in the brain may provide a tool for on-line correction of the errors that are made by the interface. 

Neural correlates of error processing during invasive BCI experiments were investigated using data collected in Chestek's Lab in the University of Michigan. In those monkey experiments, the monkey controlled an animated hand on the screen to touch a ball by moving their own fingers. Short movement segments that were consistently toward or away from the target were labeled accordingly and used to train a classifier to differentiate between correct and erroneous movements based on the neural activity. The results indicate that despite the limited number of labeled segments and active neurons in the studied data, the classifier achieved a classification rate of 75% on testing. More details are provided in the Thesis.

BCIs are prone to make errors in the direction of movement. Thus, we investigated Electroencephalography (EEG) potentials during visuo-motor rotations (VMR). We conducted VMR experiments with randomly selected small (±22.5°) or large (±45°)  rotations and investigated ErrPs in response to onset of movement correction (OMC) and saccadic movement onset (SMO). Kinematic analysis indicated that the maximum deviation from a straight line to the target was larger in trials with large compared to small or no rotations, but there was a large overlap. Thus, we also compared ErrPs generated by trials with different maximum deviations.  Our results reveal a significant positive component in trials with large rotations and especially in trials with large maximum deviation. The positive peak appeared 380 msec after SMO and 240 msec after OMC. Furthermore, it was associated with activity in Brodmann area 5, in agreement with the literature.

Investigation of neural activity during Invasive BCIs reveal shifts in channel tuning between training set and brain control set.  Suggesting that adaptation during brain control to BCI errors was required. However, the mechanisms behind those changes, and whether they reflect actual adaptation of the brain to accommodate the BCI, are poorly understood. Using simulation and theoretical analysis, we demonstrate that these changes can be induced by the BMI filter without internal adaptation to errors during brain control.  Since this issue is related to invasive BCI, it is addressed after the investigation of error processing during invasive BCIs.