Neural Cognitive Architectures
- There are systems that claim to be neural cognitive architectures.
- Eliasmith's Spaun model is one such model.
- Howard similarly claims to have a neural cognitive architecture because
he has a mechanism that makes Taylor Series and these are Turing Complete.
- Just because it's Turing complete, it's not necessarily a cognitive
architecture. So, that's not really accurate.
- Spaun works by approximating vectors and vector operations with
spiking neurons. I'm pretty sure that's Turing complete.
- They're learning is weak.
- I think it's a pretty weak cognitive architecture.
- Moreover, I don't think it's anything like the human (or rat)
neural cognitive architecture.
