This work is primarily concerned with the problem of how people use their visual perception in the performance of tasks. The tasks with which this work is first and foremost concerned are simple laboratory tasks, in which subjects have to name objects or attributes of objects, or have to read letters or words; that is, tasks in which people report about what they see (Part II), and tasks in which subjects have to make a saccadic eye movement, or have to hit an object at a position; that is, tasks in which subjects do something with regard to what they see (Part III).1 Via an analysis of internal task performance in this type of tasks, I hope to find out what kind of machines human beings basically are.