In this study, preschool children were successfully trained in theory of mind tasks, namely false belief inference and appearancereality distinction. Results show that both types of training had a direct effect, as measured by the improvement of performance on the FB and AR delayed post-tests. Both types of training also had an indirect effect, as measured by transfer of the benefits of the training to the task that was not in the training. Finally, both types of training were effective even when the children were trained in a task they had already mastered.

'Theory of mind' development is now an important research field in deaf studies. Past research with the classic false belief task has consistently reported a delay in theory of mind development in deaf children born of hearing parents, while performance of second-generation deaf children is more problematic with some contradictory results. The present paper is aimed at testing the metacognitive abilities of deaf children on two tasks: the appearance-reality paradigm designed by Flavell, Flavell and Green (1983) and the classic false belief inference task (Wimmer & Perner, 1983; Hogrefe, Wimmer & Perner, 1986). Twenty-eight second-generation deaf children, 60 deaf children of hearing parents and 36 hearing children, aged 5 to 7, were tested and compared on three appearance-reality and three false belief items. Results show that early exposure to language, be it signed or oral, facilitates performance on the two theory of mind tasks. In addition, native signers equal hearing children in the appearance-reality task while surpassing them on the false belief one. The differences of performance patterns in the two tasks are discussed in terms of linguistic and metarepresentational development.