New Ideas for Brain Modelling 7
This work addresses cognitive modeling for neuroscience and AI, but appears incremental as it updates earlier models by moving architecture boundaries and adding complementary views.
The paper tackles the problem of cognitive modeling by creating two systems that represent information at the semantic level and unifying them over a shared structure, resulting in a flexible model that can match patterns across different brain regions and potentially enable knowledge and intelligence through consistent information transposition.
This paper updates the cognitive model, firstly by creating two systems and then unifying them over the same structure. It represents information at the semantic level only, where labelled patterns are aggregated into a 'type-set-match' form. It is described that the aggregations can be used to match across regions with potentially different functionality and therefore give the structure a required amount of flexibility. The theory is that if the model stores information which can be transposed in consistent ways, then that will result in knowledge and some level of intelligence. As part of the design, patterns have to become distinct and that is realised by unique paths through shared aggregated structures. An ensemble-hierarchy relation also helps to define uniqueness through local feedback that may even be an action potential. The earlier models are still consistent in terms of their proposed functionality, but some of the architecture boundaries have been moved to match them up more closely. After pattern optimisation and tree-like aggregations, the two main models differ only in their upper, more intelligent level. One provides a propositional logic for mutually inclusive or exclusive pattern groups and sequences, while the other provides a behaviour script that is constructed from node types. It can be seen that these two views are complimentary and would allow some control over behaviours, as well as memories, that might get selected.