There is nothing to be gained by isolating movements to single joints. If you were to set fingers only by using the base knuckle that would imply that the other knuckles are locked in the action, meaning the finger is one large chunk that has to be moved at the base. Rotational inertia grows with the square of the distance from the joint.
With larger units of movements, the absurdity becomes more obvious: does a fast runner flex his legs at the hips or at the knees?
Obviously while different movements have differing involvements of differing parts for differing goals and targets, locking a joint may temporarily project progress because it results in fewer variables you have to consciously control at a given point of time. However, as this control becomes automatic and "muscle memory" takes over most of the kinematic processes, the initial advantage of needing to control only one thing at once for one purpose turns into a mechanical disadvantage.
Movements need to be efficient and organic, and while that mean that certain kinds of movement are mostly focused around the joints most effectively dealing with them, it would be a mistake to try locking everything else down. It has to react just with the right kind of suppleness and support that makes the principal movement not waste any energy on keeping things unmoving that would naturally be involved in the movement.
