Programming of action timing cannot be completed until immediately prior to initiation of the response to be controlled.

We consider, in depth, findings from across 6 decades regarding generating a motor response in a simple reaction-time (i.e., delayed response) paradigm. The early results robustly show a response complexity effect whereby the more response elements, the longer the simple reaction time (RT). This effect is puzzling because it indicates that preparation of some aspect of a response cannot be completed prior to the simple RT interval even though a precue had identified the response in advance. Results obtained during the past 10 years indicate that this finding reflects a restriction in advance preparation of the code that controls timing of the onsets for elements making up the response. This code cannot be prepared prior to the simple RT interval even though the parameters to be achieved by the response can be established earlier so that the difficulty of their selection does not influence RT. The general principle that emerges from these studies is that the abstract goal for a response can be determined in advance, but programming of action timing cannot be completed until immediately prior to initiation of the response to be controlled. We suggest that this restriction may arise because limitations in working memory prevent temporary storage of the timing code even though the abstract response goal can be stored until needed. Regardless of its origin, the limitation regarding preparation of response timing represents a fundamental feature in control of motor action.