To an outsider like me, this looks very impressive. On looking through the paper giving a 3-page summary of the Denotic Cognitive Event Calculus (http://www.cs.rpi.edu/~govinn/dcec.pdf), I got the impression that this apparent self-awareness was in some way built into the language - for example, 'Self' appears to be a word declared in the syntax. I don't know if this is so, or if it really matters - perhaps someone who does could comment?

I don't know whether this is impressive or not. It's not like they took three robots, told them the puzzle and they started thinking and talking. It depends on how they actually implemented it. With prolog it would take 30 minutes to have something like this, given that you know the puzzle and what you want the robots to do.

No self-awareness was there for sure, though, and I feel whatever they did won't really traspose to anything else easily, so I don't think the result is particularly impressive.

OTOH, the algorithm itself could be new and more performant of whatever Prolog could do, and that would be an achievement. But not really with respect to AI.

Having thought about it some more, what I think is going on here is that the Denotic Cognitive Event Calculus formalizes the deductive steps that a rational and already self-aware agent would perform, and the program is therefore just 'going through the motions'. If so, then this is not like becoming self-aware as a result of experience processed by generalized intelligence, something that several species are capable of.