|Ph.D Thesis||Department of Electrical Engineering|
|Supervisor:||Prof. Emeritus Inbar Gideon|
Movement related potentials (MRPs) are the components of the electroencephalogram related to the generation of voluntary movement. These potentials are known to vary due to the nature of the movement and the environment in which it is performed. Unfortunately, it is difficult to study these variations because MRPs are recorded in highly negative signal-to-noise ratio (SNR) conditions, which require the development of specialized tools for processing such signals. This dissertation investigates several methods for processing such MRPs in order to estimate the brains’ activity related to adaptation of the brain to movement in an unfamiliar environment.
The factors affecting the variability of MRPs during adaptation to a changing force field were investigated using a novel unsupervised learning algorithm and a variant of the matching ursuit algorithm. It was found that contrary to what might have been expected, the process of adaptation is not a gradual process. Instead, after a critical number of iterations large activation can be observed in prefrontal areas of the cortex. The temporal appearance of this activation (i.e. The number of times that the subject has to repeat the task) is dependent on the perceived complexity of the task, as is the power of this activation. This evidence, coupled with other data collected by us and by others, has led us to propose that this pattern of prefrontal activity is a manifestation of task comprehension.