CABot

• The CA work doesn't allow for process.
• That is the system won't move onto a new state.
• We got a grant to build an agent. It finished in 2009.
• We now can implement any process (it's Turing complete).
• The agent, based entirely on fLIF neurons, views the environment, takes commands from the user, and executes them.
• The first version of the system, CABot1, processed the command "Go to the pyramid." This involved planning and interaction with the environment.
• CABot2 added a psychologically plausible parser and rule learning
• CABot3 had improved vision, cognitive map learning, and texture processing for vision.
• It processed the command "Explore" and then one like "Go before the striped pyramid."