When a novice operator learns to control a novel mechanical system, the brain solves three types of computational problems: optimization of the task performance criteria to arrive at a kinematic plan (i.e., learning how the mechanical system should behave in order to accomplish the goals of the task), learning a model of the forward dynamics of the mechanical system (i.e., acquiring an ability to predict how the mechanical system will behave as a function of current input) and learning a model of the inverse dynamics of the controlled system (i.e., acquiring an ability to predict the inputs that should be provided to the mechanical system for a given desired change in state of that system). For example, consider the case of an operator acting on a joystick that controls the thrust produced by the motors of a remote underwater vehicle. The task involves moving the vehicle from point A to point B. The kinematic plan specifies a smooth vehicle trajectory. The forward model specifies how the vehicle will behave as the operator moves the joystick. The inverse model estimates how the operator should move the joystick so that the vehicle moves along the planned trajectory. An internal model (IM) is a blanket term used to describe the information contained in the solution to these three types of computational problems, and motor memory refers to the representation of IMs in the brain.

People use IMs in nearly every voluntary movement. For example, consider trying to lift an empty bottle of milk that has been painted white on the inside so that it appears full. The motor system will generate muscle activation patterns in the arm that provide the forces appropriate for lifting a full bottle, resulting in a flailing motion. This fact indicates that in programming the motor output to the muscles of the arm, the motor system uses certain visual characteristics of the object to predict and compensate for its mechanical dynamics. Learning IMs has been investigated in experiments in which a subject reaches to a target while holding the handle of a lightweight robotic arm. Disengagement of the robot's motors results in smooth and straight-line movements. Motor learning starts when the investigator programs a pattern of forces for the robotic arm to produce. These forces represent novel dynamics. When a person starts the training process, the computations that the brain performs in programming muscle activations do not take into account the novel forces, resulting in jerky hand movements. To compensate, initially people stiffen the entire limb through general coactivation of the muscles. This results in improved arm stability but serves only as a temporary and relatively ineffective strategy for responding to the perturbations. With training, stiffness returns to normal levels at the same time as the brain builds an IM of the novel dynamics. The motor output changes to specifically account for the additional forces. The development and use of a new IM can be demonstrated by turning off the robot's motors at the onset of movement. The resulting movement is a mirror image of that observed early in the training process. Therefore, the motor command that reaches the muscles includes a prediction by the IM of the forces required to overcome the imposed mechanical dynamics. The motor system retains this skill for months after the training session.



Is there a cause or cure for autism? The Complete Guide To Finally Understanding Autism. Do you have an autistic child or know someone who has autism? Do you understand the special needs of an autistic person?

Get My Free Ebook

Post a comment