PRIMs Cognitive Architecture

Multi-level architecture

One of the goals of PRIMs is to find a basis in neural networks and neuromorphic hardware. A Cognitive Computer based on neuromorphic hardware requires more than building neural networks out of silicon or other materials. Just like conventional computer architectures, multiple levels of abstraction are needed to support flexible computation, each with its own representation that can be reduced to the underlying level.

A critical difference between computer and cognitive architectures is that the latter is  based on learning. The figure on the right presents a possible framework that is based on what we know about human intelligence and learning.

Many of the components of the framework are based on existing theories: ACT-R, PRIMs, Nengo. The challenge is to combine them productively.

 

For more detail, look at the poster I presented during the 2018 Hannover Cognitive Computing conference.