Currently No Good Model

• I'll tell you about my binary CA model next week, but it is flawed. In particular, a CA, once ignited, fires persistently indefinitely unless it is explicitly shut down.
• I occasionally am at talks when someone describes there model as having a CA.
• I ask them how long it fires for, and they typically say just once when it's activated.
• You can learn these patterns. It's kind of like word2vec.
• I know Pulvermuller and colleagues do work with learning CAs across areas.
• These generally persist indefinitely.
• Rolls has some nice stuff about the phonetic blackboard, but it doesn't involve learning.
• I'm not aware of a good model of CAs. I'd use it if I was. So, I'm trying to make one myself.