The brain and its language in the control of robotic action

Brain Computer Interfaces (BCIs) enable one to control peripheral ICT and robotic devices by processing brain activity on-line. The following epistemological, philosophy of mind, and ethical issues are examined in connection with BCI-controlled actions. (i) According to so-called motor theories of thinking, the unique forms of motor rehabilitation therapy afforded by BCI systems may contrast the decline and extinction of thinking in completely locked-in patients. (ii) Unconscious perceptual processes are used in BCI-enabled, brain-computer cooperative problem solving. There, the functional roles of human “operators” are accounted for at the sub-personal level - without appealing to their intentions, beliefs, and contents of consciousness. Since humans are neither required to act intentionally nor to be aware of their contribution to cooperative problem solving, it is appropriate to ask whether a sub-personal use of human being is lurking here. (iii) Machine learning is crucially involved in human-machine adaptation processes required for BCI operation. The reliability of BCI learning depends on boundary conditions that are difficult to control, such as mental task execution history and overall mental context. For similar reasons, one can hardly deploy the more abstract mathematical framework of statistical learning theory to evaluate the reliability of learning. These epistemological issues distinctively shape ethical issues – notably including autonomy and responsibility problems – in current BCI environments.