Many of the ideas that Hofstadter talks about are very similar to things I have been thinking about in my own research. His research group, however, stays within very restrictive domains which offer insight into the commingling of cognition and high-level perception, while I perceive my own work as bridging between low- and high-level perception. I'm intrigued by the use of "temperature" and probabilistically-run codelets in his systems, which are very closely related to my own blackboard systems.