I think it's an interesting theory, but I don't find myself very convinced.

Competing hypothesis: People's motivations are less driven by self-interest when the discussion is an abstract one and more driven by self-interest when there are real losses and gains to be had. It still has a self-interest component, but now the claim is that it is those who do not want redistribution acting out of self interest. Self-interest is clearly a component here, but singling it out as a motivation for wanting redistribution makes for certain specific arguments.

I'm also not sure how transferable the results around distribution in a game are for questions with societal implications; I know these sorts of experiments are common in psychology, but I wonder if the limited scope of this game means people are more willing to take everything they've earned since many of the reasons to support redistribution do not exist in a game setting, e.g. real suffering, not wanting violent revolution, etc.