Abstract
The development of Carnap’s views on knowledge of other minds from 1928 to about 1935 is tracked here in order to clear up a widespread misunderstanding. Early on and well into the 30s their failure is undeniable but it has been badly misdiagnosed. I argue that Carnap was not only not a logical behaviorist but also aimed for a largely non-reductive approach to mental state ascriptions in principle already in Scheinprobleme. The reason for the failure of his anti-reductionist ambition and what it took to overcome it are then investigated in some detail. In the process, it will be shown, Carnap anticipated, in a limited fashion, certain moves familiar from the Wittgensteinian tradition, albeit with the difference that he attempted to naturalize mind talk.