Implementing Rules with Artificial Neurons
by Chris Huyck and Dainius Kreivenas
- Rules and rule based systems are Turing complete, and widely used
in cognitive science.
- Humans (and all other vertebrates) use neurons to think. Neurons
are the basis of intelligence.
- Spiking compartmental neurons are a good model of neurons. We
use leaky integrate and fire (LIF) neurons.
- Implementing rules in LIF neuron should help general cognitive
scientists write models in neurons.
- We use binary cell assemblies to implement facts, and finite
state automata (FSA) constructed of these to implement rules.
- We have a mechanism to convert an arbitrary rule based system
with initial facts into a neural FSA.