While many studies used electroencephalography (EEG) to assess mental workload during performance, a fairly limited effort examined this notion in a context of i) practice and ii) team dynamics. Of these limited studies, none employed a combined approach to assess mental workload throughout motor practice. Therefore, the first study aimed to examine whether a combined evaluation of attentional reserve and cognitive-motor effort could collectively assess mental workload dynamics during practice of a novel task. Also, among the limited examination of mental workload and performance in team environments, no work has manipulated interdependence while accounting for psychological factors such as self/collective efficacy, attribution and trust. Thus, the second study aimed to investigate performance, mental workload and these psychological factors while varying interdependence. We employed a human-machine interface where participants learned to perform reaching movements with a virtual robotic arm via limited head motion allowing examination of cognitive-motor practice. Then, once participants learned to control the effector, its controller was manipulated to serve as a synthetic teammate to perform the task with human participants. Three interdependence levels were considered: the human (H-r) and synthetic (h-R) teammate mostly controlled the arm as well as a shared (h-r) control condition. Performance was assessed via the arm kinematics and mental workload via the novelty-P3 and EEG spectral power. Surveys assessed mental workload and psychological factors. A de-noising algorithm allowed removal of artifacts from the EEG signals due to head motion from controlling the effector. The first study suggests that throughout practice the proposed approach could capture an elevation of attentional reserve and an attenuation of cognitive-motor effort, which collectively index a reduction of mental workload along with improved performance. The second study suggests that while the h-r and h-R conditions revealed similar kinematic performance, mental workload and trust, a lower sense of attribution (personal) was observed for the h-R compared to the h-r condition. Additionally, the H-r condition showed the worst performance and highest mental workload. Thus, the h-r interdependence level may provide the best cognitive-psycho-motor performance. Together these studies can improve understanding of cognitive-psycho-motor mechanisms as well as inform the design of assistive technology.