A few months ago I presented a talk entitled “Reappropriating Ashby’s Ultrastability for the Modelling of Neural Assemblies” at the “Cognition and Consciousness” Retecog Summer School 2012. In the talk I analysed homeostatic models derived from R.W. Ashby’s notion of ultrastability and propose a new model inspired in the ideas of homeokinesis and autopoiesis, where the homeostatic variables are the relations between the components of the system instead of the state of the components themselves. The result is a moving and changing homeostatic area which continuously adapt to the agent’s sensorimotor loop.
I recently obtained new results using just three oscillators as a neural controller of a robot in a phototaxis task, which allows us to watch the homeostatic regulation in 3D. The video shows an agent displaying an auto-regulated preference for two different types of light. This autonomous preference emerges as a result of the coordination of homeostatic regulation and sensorimotor activity.