Objective. The aim of this work is to assess the usability of two BCI prototypes by measuring interaction with the systems in context, considering the performance, cognitive workload and satisfaction of non-disabled users in order to better understand how the interface affect these parameters. We tested two keyboard- controlled Java BCI prototypes based on the Language Support Program (LSP) and the P300 Speller (P3S).

Main Content. We performed two different evaluations. In the first one, we tested the learnability of BCIs on 6 healthy users through the Thinking Aloud technique. Then, we tested BCI efficiency on 30 participants through the Copy Spelling Task (CST) and we administered the System Usability Scale (SUS) to measure usability and the Survey of Technology Use (SOTU) scale of the Matching Person and Technology (MPT) to measure predisposition to the use of technology. In the second evaluation, we tested again using 61 participants with different computer skills, and administered usability and cognitive workload questionnaires.

Results. The first test showed that all users easily learned how the system worked with the LSP, but failed with P3S. We found that P3S users were more accurate in selecting and recognising letters on the screen. Both SUS and SOTU did not show any significant effects. In the second one, the results showed differences in the number of errors, in user satisfaction and in the cognitive workload.

Conclusions. We found that the Thought Translation Device was more error- resistant, less stressful and more satisfactory for the users compared to the P3S.