Binary Cell Assemblies

We, and others, have modelled CAs with neural simulators.
One common model is the Hopfield model with nets of Integrate and Fire neurons that are well connected bidirectionally.
These models form autoassociative memories that settle into stable states with all neurons in the system maintaining there firing state.
That is a memory is either on or off.
Our models use sets of neurons for CAs, and neurons may participate in more than one CA.
This allows the possibility of a CA being more or less active.
If more neurons are firing, it is more active.
Unfortunately, our simulations to date have shown CAs being more or less binary.
The CAs either have a roughly constant number of neurons firing, or they have none firing.
To some degree there is some flexibility with neurons have sub-threshold activation, or having a few neurons firing, but once it ignites, a CA largely has a roughly constant number of neurons firing.