I've been looking for the last ten minutes. Having a hell of a time finding it. The author was Dacher Keltner, it was conducted at Cal (?) in the late nineties (?). Its nickname was the "Cookie Monster" study. I found a YouTube video where he described it [1]. But I can't find the actual study, or any citations of it.
I wonder how successful attempts to replicate it have been. Couldn't find anything on that either.