Inhibitory Neurons
- Roughly 20% of neurons in the cortex are inhibitory.
- Inhibitory neurons allow CAs to compete.
- CAs compete to be ignited by Ambiguous stimuli.
- Inhibition requires modification to allow for negative
weighted connections
- Each Neuron is either excitatory or inhibitory (I can't
remember whose Law Gray's Law?)
- Flip the learning rule so that coactive neurons have
weights close to 0, and non-coactive neurons have very
negative weights
- In the most recent simulations, the topology of
inhibitory neurons is different
- Reduces overall activity in the system
- Reduces the likelihood of simulated epilepsy (the whole
net becoming active).
- I think this reduces connection strengths
