Biological information-processing principles

The basic principles of our design are derived from the biology of the neocortex:

1. Each module stores and processes data of given types characteristic of that module.

2. To form systems, modules are connected in a fixed network with dedicated point-to-point channels.

3. Modules are organized as a perception-action hierarchy.

4. Modules process data received and/or stored locally by them. There is no central manager or controller.

5. Modules are all rule based with a common rule execution process.

6. All modules do similar amounts of processing and run at about the same speed.

7. There is data parallelism in communication, storage and processing. Processing within a module is highly parallel. Parallel coded data is transmitted, stored, and triggers processing. Processing acts on parallel data to produce parallel data.

8. The data items being transmitted, stored and processed can involve a lot of information; they can be complex.

9. The modules act continuously and in parallel; they do not wait for other modules or data.

10. Most of the learning occurs in specialized learning modules, which dynamically create knowledge for modules.