Search

Motor imagery (MI) based brain computer interfaces (BCI) detect changes in brain activity associated with imaginary limb movements, and translate them into device commands. MI based BCIs require training, during which the user gradually learns how to control his or her brain activity with the help of feedback. Additionally, machine learning techniques are frequently used to boost BCI performance and to adapt the decoding algorithm to the user's brain. Thus, both the brain and the machine need to adapt in order to improve performance. To study the utility of co-adaptive training in the BCI paradigm and the time scales involved, we investigated the performance of two groups of subjects, in a 4-day MI experiment using EEG recordings. One group (control, n = 9 subjects) performed the BCI task using a fixed classifier based on MI data from day 1. In the second group (experimental, n = 9 subjects), the classifier was regularly adapted based on brain activity patterns during the experiment days. We found that the experimental group showed a significantly larger change in performance following training compared to the control group. Specifically, although the experimental group exhibited a decrease in performance between days, it showed an increase in performance within each day, which compensated for the decrease. The control group showed decreases both within and between days. A correlation analysis in subjects who had a notable improvement in performance following training showed that performance was mainly associated with modulation of power in the α frequency band. To conclude, continuous updating of the classification algorithm improves the performance of subjects in longitudinal BCI training., (Copyright © 2019 Abu-Rmileh, Zakkay, Shmuelof and Shriki.)

When subjects are intentionally preparing a curved trajectory, they are engaged in a time-consuming trajectory planning process that is separate from target selection. To investigate the construction of such a plan, we examined the effect of artificially shortening preparation time on the performance of intentionally curved trajectories using the Timed Response task that enforces initiation of movements prematurely. Fifteen subjects performed obstacle avoidance movements toward one of four targets that were presented 25 or 350 ms before the "go" signal, imposing short and long preparation time conditions with mean values of 170 ms and 493 ms, respectively. While trajectories with short preparation times showed target specificity at their onset, they were significantly more variable and showed larger angular deviations from the lines connecting their initial position and the target, compared to the trajectories with long preparation times. Importantly, the trajectories of the short preparation time movements still reached their end-point targets accurately, with comparable movement durations. We hypothesize that success in the short preparation time condition is a result of an online control mechanism that allows further refinement of the plan during its execution and study this control mechanism with a novel trajectory analysis approach using minimum jerk optimization and geometrical modeling approaches. Results show a later agreement of the short preparation time trajectories with the optimal minimum jerk trajectory, accompanied by a later initiation of a parabolic segment. Both observations are consistent with the existence of an online trajectory planning process.Our results suggest that when preparation time is not sufficiently long, subjects execute a more variable and less optimally prepared initial trajectory and exploit online control mechanisms to refine their actions on the fly.

Cerebral Palsy (CP) results from an insult to the developing brain and is associated with deficits in locomotor and manual skills and in sensorimotor adaptation. We hypothesized that the poor sensorimotor adaptation in persons with CP is related to their high execution variability and does not reflect a general impairment in adaptation learning. We studied the interaction between performance variability and adaptation deficits using a multi-session locomotor adaptation design in persons with CP. Six adolescents with diplegic CP were exposed, during a period of 15 weeks, to a repeated split-belt treadmill perturbation spread over 30 sessions and were tested again 6 months after the end of training. Compared to age-matched healthy controls, subjects with CP showed poor adaptation and high execution variability in the first exposure to the perturbation. Following training they showed marked reduction in execution variability and an increase in learning rates. The reduction in variability and the improvement in adaptation were highly correlated in the CP group and were retained 6 months after training. Interestingly, despite reducing their variability in the washout phase, subjects with CP did not improve learning rates during washout phases that were introduced only four times during the experiment. Our results suggest that locomotor adaptation in subjects with CP is related to their execution variability. Nevertheless, while variability reduction is generalized to other locomotor contexts, the development of savings requires both reduction in execution variability and multiple exposures to the perturbation.

A methodology for investigating learning is developed using activation distributions, as opposed to standard voxel-level interaction tests. The approach uses tests of dimensionality to consider the ensemble of paired changes in voxel activation. The developed method allows for the investigation of non-focal and non-localized changes due to learning. In exchange for increased power to detect learning-based changes, this procedure sacrifices the localization information gained via voxel-level interaction testing. The test is demonstrated on an arc-pointing motor task for the study of motor learning, which served as the motivation for this methodological development. The proposed framework considers activation distribution, while the specific proposed test investigates linear tests of dimensionality. This paper includes: the development of the framework, a large scale simulation study, and the subsequent application to a study of motor learning in healthy adults. While the performance of the method was excellent when model assumptions held, complications arose in instances of massive numbers of null voxels or varying angles of principal dimension across subjects. Further analysis found that careful masking addressed the former concern, while an angle correction successfully resolved the latter. The simulation results demonstrated that the study of linear dimensionality is able to capture learning effects. The motivating data set used to illustrate the method evaluates two similar arc-pointing tasks, each over two sessions, with training on only one of the tasks in between sessions. The results suggests different activation distribution dimensionality when considering the trained and untrained tasks separately. Specifically, the untrained task evidences greater activation distribution dimensionality than the trained task. However, the direct comparison between the two tasks did not yield a significant result. The nature of the indication for greater dimensionality in the untrained task is explored and found to be non-linear variation in the data.