(1) About the Minimal Representation of Possibility … The way I see it, the minimal representation strategy still has “modality” somewhere lurking in the shadows: in order to select a single simulation, and take it as actual, a space of hypotheses must be constructed somehow, and (to me) it implies that concepts of possibility are implicit to the process. How would they respond to this? Or am I making a conceptual mistake somewhere?
(2) Please, can you go over the logic of how the “gum-ball machine” results can be used as evidence against the minimal representation strategy? I think I get it, but I am not entirely sure I grasp it fully. Why would it be non-graded for it? Also, I am surprised by the similar learning performance in the 50% vs. 2.5%, does it make sense?