Brain-machine interface (BMI) is a promising technology that can provide accessibility to sensors and actuators using limited physical interaction. This technology can benefit millions of people with physical disabilities, such as Amyotrophic Lateral Sclerosis (ALS) and limb problems. However, its practical application depends critically on the accuracy of interpreting the commands received through BMI. This paper presents two techniques that exploit contextual awareness to improve the accuracy of communication using BMIs. We first present a technique that reduces the false interpretation probability significantly by analyzing the current system state. Then, we quantify the benefits of automating actions with the help of previously learned patterns. Experimental evaluations using a commercial BMI headset and a virtual reality environment show 2.6× decrease in the completion time of a navigation task.