Cell Assemblies and Attractor Dynamics
- This research all derives from Hebb's Cell Assembly (CA)
hypothesis
- The neural foundation of concepts is a group of neurons
that have high mutual synaptic weights. These
form reverberating circuits that persist after input
has ceased.
- So, a CA is an attractor state.
- Consequently, a lot of the attractor network work (Hopfield Nets,
Boltzmann machines etc.) is really relevant.
- A net with CAs is a categoriser. An instance is presented
and is categorised by the CA that ignites.
- We've done a lot of work with categorisation and CAs.
- However, if you stay in one state, you can't do anything else.
- The system must move on to new states!
