This dissertation investigates the hypothesis that acquiring knowledge concerning the syntactic feature Point-of-View (PoV) in CPs of complement clauses (CCs) subcategorized by verbs of belief is a necessary condition for children to achieve a fully developed Theory of Mind (ToM) – the ability to ascribe false beliefs (FBs) to others and to predict their reasoning and action based on it (usually achieved by 4 years of age) (cf. DE VILLIERS AND DE VILLIERS, 2009) (henceforth, PoV-FBHypothesis). In particular, this study aimed to verify the extent to which PoV-FBH can be maintained; to evaluate the possibility of incorporating a PoV feature in an on-line model of syntactic computation grounded in minimalist assumptions (Corrêa & Augusto, 2007; 2011) and to characterize the developmental course of the acquisition of knowledge concerning this feature in the light of a procedural theory of language acquisition (Corrêa, 2009). The possibility of PoV being represented as a formal feature or as a semantic feature was considered in three analyses of CCs. It is argued that no matter how this feature is characterized, the acquisition of the knowledge it represents would inevitably rely on semantic bootstrapping, which presupposes the cognitive ability of distinguishing points of view. As for the empirical investigation, 8 experiments were conducted. Experiments 1-6 aimed at assessing 3;3-3;5 year olds ability to cope with a number of the demands of the traditional FB and Memory for Complements (MC) tasks, which supports PoVFBH, such as the ability to cope with object WH, the factivity of mental verbs and factuality. The production of CCs and childrens ability to cope with a linguistic task that includes a FB situation were also investigated. In experiment 7, childrens ability to assign truth-values to CCs and for both clauses of the complex sentence was evaluated. In Experiment 8 the role of language recursion for first- and second-order FB reasoning was investigated with 6;2 year-old children. In general, the results suggested that i) childrens failure on the traditional FB and MC tasks does not necessarily mean difficulty in dealing with CCs or FBs, insofar as they do not cope with most of the demands these tasks presuppose; ii) 3 year olds correctly assign truth-values to CCs, though assigning the truth-value for the main clause (the whole sentence) independently of the truth-value of the CC is particularly costly; iii) first- and second-order FB reasoning can be undertaken on the basis of non-recursive structures, although recursive structures seem to facilitate it. In sum, both in the light of the possible developmental courses of acquisition considered here and on the basis of the results obtained, it would be hard for the PoV-FBH to be maintained, as it stands. Metacognitive abilities may be required for children to be able to ignore the truthvalue of the CC when assigning a truth-value for the complex sentence. It is argued that the possible role of CCs in the development of ToM is to provide a representation that can be used to facilitate FB reasoning, not to support it.