Training our minds to move matter

We know that our body schema is plastic (see recent post on this), and that our brain's motor routines can learn stable habits of controlling tools and prostheses as if they were our own actual body parts. Ganguly and Carmena have now taken the obvious step of pairing stable recordings from ensembles of primary motor cortex neurons in macaque monkeys with a constant decoder that transforms neural activity to prosthetic movements. Blakeslee points out that this work suggests that learning to move a computer cursor or robotic arm with nothing but thoughts can be no different from learning how to play tennis or ride a bicycle. The brain can form a motor memory to control a disembodied device in a way that mirrors how it controls its own body. Here is their summary, followed by an illustration:

Brain–machine interfaces (BMIs) have the potential to revolutionize the care of neurologically impaired patients. Numerous studies have now shown the feasibility of direct “brain control” of a neuroprosthetic device, yet it remains unclear whether the neural representation for prosthetic control can become consolidated and remain stable over time. This question is especially intriguing given the evidence demonstrating that the neural representation for natural movements can be unstable: BMIs provide a window into the plasticity of cortical circuits in awake-behaving subjects. Here, we show that long-term neuroprosthetic control leads to the formation of a remarkably stable cortical map. Interestingly, this map has the putative attributes of a memory trace, namely, it is stable across time, readily recalled, and resistant to the storage of a second map. The demonstration of such a cortical map for prosthetic control indicates that neuroprosthetic devices could eventually be controlled through the effortless recall of motor memory in a manner that mimics natural skill acquisition and motor control.


Schematics for manual control (MC) and brain control (BC). During MC, the animal physically performs a two-dimensional center-out task using the right upper extremity while the neural activity is recorded. Under BC, the animal performs a similar center-out task using a computer cursor under direct neural control through a decoder trained during MC.