I feel that task execution in the brain follows the divide and conquer rule.
Whenever we do a task our brain breaks it down into miniscule smaller tasks.During the learning period each of these smaller isolated tasks are mastered and polished.There actually exists a motivation to do each small task but ultimately it is the motivation for the cumulative task that  drives us to do what we want to do.
   Each of these tasks goes through a process of motivation,performing the task and feedback receptance and processing.Feedback forms an important part of learning, it provides us with  the knowledge of performance and a tendency to correct our errors.With practice our error sensitivity increases,hence our task performance reaches perfection.
 During the process of learning we slowly master every detail and in the end the cumulative task is what we solve.So this is actually a process of recursion, organization and execution.
  Greater the perfect execution of tasks, faster is the rate of adaptation to it.With practice our ability to retain the entire procedure increases.With repetitive learning we slowly acquire a certain degree of automation as it just becomes a process of retrieving the process package and executing it rather than creating the whole package again.
  For example when we tie our shoelace all  the constitutent task packages have already been integrated in our brain and each task becomes independent of the other as time progresses and so when we tie our lace all the tasks like holding our lace,making a knot,tucking the lace into our shoe etc are smoothly executed without any sensory inputs.
   Hence our finetuned ability to do tasks as easily as we actually do is very analogous to solving a jigsaw puzzle blindfolded.