provide persistent neural activity

[xh3b4sd]

To make the cognitive processes within Anna more reflective, additional calculations between core network and knowledge network should be done. E.g. when processing some input, certain routines trigger additional routines in the related contextual scope. The originating routines might be selected randomly or by some bias or preceding evaluations. There might be some inception style injections causing routines to affect. Certain injections could be provided by some dedicated networks around. In case all this causes certain connections to be involved multiple times, due to additional routines visiting connections, weights of such connections should be increased. The following variables should be dynamically evaluated.

• Which routines causing such persistent behaviour within the knowledge network.
• How much weights increase.
• How much additional routines trigger.

[xh3b4sd]

This is from a closed issue for the sake of tracking ideas.

This is currently only a rough idea. The idea network would operate in the background. It takes available information and tries to discover new areas around them. When actively triggered it applies available ideas to the impulse. Based on current models the IdeaNet would sit between the CtxNet and the RespNet.

See #88.