A theory of sensorimotor learning for brain-machine interface control

A remarkable demonstration of the flexibility of mammalian motor systems is primates’ ability to learn to control brain-machine interfaces (BMI’s). This constitutes a completely novel and artificial form of motor behavior, yet primates are capable of learning to control BMI’s under a wide range of conditions. BMI’s with carefully calibrated decoders, for example, can be learned with only minutes to hours of practice. With a few weeks of practice, even BMI’s with random decoders can be learned. What are the biological substrates of this learning process? This thesis proposes a simple theory of the computational principles underlying BMI learning. Through comprehensive numerical and formal analysis, we demonstrate that this theory can provide a unifying explanation for various disparate phenomena observed during BMI learning in three different BMI learning tasks. By explicitly modeling the underlying neural circuitry, the theory reveals an interpretation of these phenomena in terms of the biological non-linear dynamics of neural circuits