The current NSM theory regards six mental predicates-- and --as indefinable semantic universals. This paper examines the syntax of their Japanese exponents (omou, sit-te iru, -tai/hosii, miru, kiku and kimoti). Special attention is paid to the syntax and semantics of major complementation types (S no, S koto and S to) found with the majority of these predicates. It is shown that each primitive predicate has a specific set of syntactic frames in which the primitive meaning is expressed, and that the extended meanings which may be expressed in other syntactic environments are specifiable by reductive paraphrase explications.