Brain-Computer Interfaces (BCI) allow users to interact with machines without requiring muscular activity. Thus, patients with heavy motor impairment can benefit from these systems. We have implemented an electroencephalography-based BCI which provides four distinct commands. Our system exploits the Motor Imagery of the subject and four different states of mind: imagination of a movement with the left or right arms, both arms simultaneously and no imagination at all. In addition, the BCI exploits specific neurological markers called Steady-State Somatosensory-Evoked Potentials. They are vibrating devices taped on the user's wrists. These markers are measurable on the cortex using electroencephalography. This paper focuses on the Computer Human Interaction aspects. We describe the design and study of two applications controlled by this BCI. The applications differ in two characteristics: their inertia, or rhythm of information flow perceived by the user, and the "punitiveness" of the application in case of mistakes. To study the user experience in perfectly controlled conditions, we used a so-called "sham" feedback in the BCI loop rather than real feedback computed by analysing the user's brain waves. With sham feedback, the BCI provides commands with an a priori defined accuracy. We performed a user experiment of the two applications over a group of ten healthy participants. They tested both applications for different sham accuracies, varying from 45% to 90%. This permits the study and modelling of the relationship between the perceived usability of the system and the performance of the BCI.