This paper discusses the development of the Mental zero-Secondary Predicate Construction (Mental zero-SPC) from Old English to Late Modern English (e.g. Apparently they didn't consider her pretty, but I thought her strongly beautiful), paying special attention to the complex interplay of mechanisms underlying the increasing recruitment of internal, mental representation verbs in the zero-SPC. Informed by Traugott's work on semantic change, this pattern of increasing productivity in the Mental zero-SPC is here characterized as a process of internalization. It is argued that the key recruitment mechanisms are polysemization (in Old English), analogical extension (in Middle English), and languageexternal influence from Romance languages (in Middle and Early Modern English). Broadly speaking, the mechanisms of recruitment are seen to become more systematic over time, with focused attractor sets and more large-scale developments only appearing from Middle English onwards. On a more theoretical plane, we address the issue whether the Mental zero-SPC can be considered a multi-source construction, in view of the multiple influences that contribute to its productivity increase. Since these influences do not establish a construction that is entirely new to the language, we argue that the Mental zero-SPC is not a multi-source construction in the strict sense. Rather, the developments characterizing the construction's evolution are defined as cases of 'categorial incursion', as new verb types are added to a construction which was already established in the language.