Analogy Representations.

Knowledge states (both s-images and nv-states) can have analogies between them. Each analogy can have any number of analogical mappings associated with it (determining which mapping is the best is the mapping problem.) Each alignment in a given mapping is called a map.

Similarly knowledge states next to each other in sequences have transform-connections. These are necessary so the agent can track how visual elements in a previous knowledge state change in the next. A difference between analogies and transform connections are that there can be multiple analogical mappings for an analogy, but only one mapping for a transform connection.

Transformations are attached, in fact, to a map between two components of sequential knowledge states. So if a rectangle changes into a circle, the agent knows which rectangle turns into which circle.