caetris1

This is because the center point of your models do not expect a foreground/background boundary. By putting a hole at the center, the boundaries seek to stabilize the system by bubbling outward. The best way to put it is that there is no stable state where a doughnut hole exists at the center. An interference pattern ripples outward and destroys the system.

Spectacular. I think most people might be lost on the significance of what has been done here. I suspect that it's because the framing the solution focuses on is based on morphogenesis. That is, in fact, not what is going on here. A more suitable metaphor would be the functional basis of a visual hallucination. The very same mechanics of the demonstrated result are described in Oliver Sack's book on the subject. Learned patterns are procedurally generated through probabilistic inference of adjacent feature vectors. Dreams work in a similar way. To prove that this is what is going on, I suggest training the CA on multiple frames of an animation of the lizard walking.

The research is really promising but the conclusion that the authors come to is that this system can be used to generate self-organizing agents that evolve and replicate. There is no possibility of this result based on the approach. However, they've actually created something much better than the intended result... a real-time auto-complete for visual scenes.