The Sequencing Layer is responsible for telling the control layer which stimulus-response mapping to do when. It activates and deactivates specific modules in the control layer to accomplish this. It uses the feedback returned by the control layer to monitor progress of desired activities, and to determine when the stimulus-response mapping should change. This feedback also makes it cognizant of failures of the control layer.
Commuication with the control layer is done using ALFA (A Language For Action), a language similar to REX.
The Sequencing Layer requests actions from the Deliberative layer. The returned and pending actions are put on a task queue. The Sequencer, however, is at liberty to ignore the result and perform different actions at different times. In essence, the Sequencer controls the Deliberator.
The Sequencing Layer resembles Firby's RAP (Reactive Action Package) system in that both specify an intermediate level of action control between the high-level goals and the atomic actions. The major difference between the two is that RAPs control atomic actions while the Sequencing Layer has a higher-level view of activities.