CAs are good for Categorisation
- A stimuli is presented to the network
- A CA ignites and the stimuli is categorised as an element of the
category that the CA represents.
- If a stimuli is ambiguous, the CAs should compete and
resolve to one category.
- There are completion effects. If the object is occluded, the
system infers the extra information by activating the entire
CA (including the features represented by the occluded bits).
- Notice that CAs have information about the concept.
- This comes in the form of neurons that represent things
internally, and in the form of connections to other CAs.
