A Proto Neuro Cognitive Architecture
by Chris Huyck

Our group here at Middlesex has been working on neural simulations and
emulations for over twenty years.  These have largely been based
around cell assemblies (CAs), with the idea that we want to duplicate
human neural processing to eventually move to full strong AI.  A
related community works on cognitive architectures, the core of
cognitive processing.  These architectures are usually based on rule
based systems and associative memories.  At Middlesex, we now have the ability to program rule based systems and associative memories in high level symbolic systems and to cache those systems out to neurons.  At Middlesex, we have also been developing agent technology, based on neurons, including virtual environments and robots.  We are now porting our work to the neurorobotics platform of the HBP so that it can be more broadly used.  Similarly, via the HBP, there is access to large neuromorphic systems capable of running, perhaps, a billion neurons in real time.  I am proposing to combine our neural associative memory work with our neural rule based to develop an proto neuro cognitive architecture.  Working within the HBP, so that agents and large neural systems are available, this architecture can be used to explore neural cognitive architectures.  These explorations can then be used to develop a full-fledged architecture.