Table 1.

Definitions of terms used to describe the case studies under the generative framework. An example application to a simple reactive controller represented with a dynamical system is provided in parentheses.

Term	Definition
Basis model	Describes relatively more concrete aspects of the robot and environment relevant to the target behavior (e.g., equations of motion)
Emergent model	Describes relatively more abstract behavior qualifying as systemic, effective affordance exploitation (e.g., fixed point location)
Generative model	Formal analysis linking features of the basis and emergent models (e.g., stability analysis of fixed point)
Gibsonian affordance	An opportunity for action in an agent-environment system (emergent-level property)
Reactive control	Responsive to robot-environment system's state, with little or no memory
Parallel composition	Controllers operating simultaneously in the same basis level, interacting according to formally described rules
Sequential composition	“Chains” of controllers, with the successful execution of one sub-behavior setting up the next sub-behavior
Hierarchical composition	Controllers operating at different levels of abstraction, e.g., on a single limb, coordination of limbs, center of mass trajectory, or to set a global goal