Mechanisms in natural and concurrent brain-computer interface motor control

Abstract

A likely future direction for brain-computer interface (BCI) technology is to supplement natural movements of the user, rather than just replacing movement ability lost after injury or illness. The novel control can be viewed as a new motor skill learned by subjects. This thesis investigates basic science questions relating to such a BCI control and for skill learning. We explore the potential for executed movements to occur concurrently to an additional artificial effector by developing an Electrocorticography based BCI that required participants to perform both movement executions and BCI control simultaneously. Crucially the sources of both control signals came from the same cortical site. Subjects were able to gain accurate and independent control and furthermore we observed lasting changes in the neural activity relating to the original movement behaviour specific to the constraints for the concurrent control. We then investigated the capacity of the brain to incorporate independent supernumerary effectors into the body schema. We demonstrated that this was possible by inducing the illusion of ownership and control in a virtual reality third hand using an imitation BCI, without subjects losing a sense of ownership in their existing limbs. Humans show enormous capacity to learn movements of their own limbs and devices controlled by a BCI. To investigate how humans gain a new movement skill we developed a behavioural study to investigate motor skill learning. Subjects were required to track paths in the absence of any external perturbation. We showed that subjects improved their performance and that this generalised to novel paths. We demonstrate that in order to develop the motor skill subjects must increase their movement planning horizon.