- How do we ground symbols?
- I'll assume that we agree that humans:
- use neurons to process information
- symbols are ground in our experience with the environment
- we are genetically predisposed to do certain things (e.g. language),
and this is done by setting up the brains topology.
- This leaves open a lot of possible neural architectures. However,
using current neurologoical evidence we can discount (at the neural
level) feed-forward nets, SOMs, and even ART.
- Cell Assemblies (CAs) remain. The idea was proposed by D.O. Hebb in
1949, and there has been a little bit of neural level computational
- CAs are reverberating circuits of neurons. Neurons tend to
co-activate each other.
- When active they are short-term memories. Since they are learned, they
form long-term memories.