Master thesis done at CNBI under the supervision of Prof. Millán and Prof. Chavarriaga.
Error Potentials (ErrP) are one of the most studied paradigms in Brain Computer Interfaces (BCI). They can be detected very accurately, specially in systems with a discrete output where the response is either correct or incorrect. In more recent studies, it has been demonstrated that ErrP also appears in systems with continuous outputs and that it carries more information than just the alert signal. In particular, it has been seen in some studies that the amplitude of the ErrP is modulated by the size of the kinematic error in visuomotor tasks. However, there is still a gap in being able to decode this information in single trials.
In this work we developed an experiment that allowed participants to perform visuomotor tasks. Prediction errors were created by introducing impulse forces in catch trials which participants felt both visually and haptically. These errors elicited the ErrP that was decoded to get the size of the error. The grand averages results confirmed the previous work in literature, showing how the error size modulates the ErrP amplitude. The main achievement of this work is the high correlation accuracy obtained from single trials decoding, which for some subjects exceeded 0.7. These findings suggest the feasibility of introducing ErrP decoders in closed loop, already existing BCI protocols to improve their performance.