This paper proposes a gradient redefinition of the notion of factuality, here intended as a dynamic continuum unfolding through several epistemic levels. In this respect, the speaker/writer’s increasing certainty upon the realization of an event or situation is here as factualization. Factualization is a conceptual phenomenon determined by an embodied mechanism (Lakoff & Johnson, 1980, 1999; Lakoff, 1987, 2003; Grush, 2004; Gallese & Lakoff, 2005) of cyclic acquisition and control with respect to a new proposition P. Being a form of subjectification (Traugott, 1989, 1995, 2003, 2010, 2012; Traugott & Dasher, 2002), factualization occurs as the semasiological reanalysis of an epistemic construction. Drawing on Langacker’s (1991, 2008, 2009) notion of the ‘epistemic control cycle’ (see also Kan, Teubner-Rhodes, Drummey, Nutile, Krupa, & Novick, 2013, on cognitive control), I claim and demonstrate that epistemic predicates originally conveying weak certainty towards a proposition P diachronically develop an increasingly factual meaning conveying more and more frequently a subjected form of certainty. This phenomenon is first shown through a qualitative and quantitative corpus analysis from the BNC,1 which provides a measurable account of the various degrees of polysemy of the three epistemic predicates I think, I believe, and I reckon. In addition, I discuss the results of a diachronic corpus survey from the diaCoris on the factualization process of (Io) penso ‘I think’ in Modern Italian during the last 150 years, showing how the contemporary usage of (Io) penso is notably more oriented towards absolute factuality than it was 150 years earlier.