The violation-of-expectation (VOE) paradigm and related methods are the main tools used to study high-level cognition in preverbal infants. Infants' differential looking to conceptually implausible/impossible events has been used as an index of early cognitive competence in many areas, including object knowledge, physics, language, and number. However, an event's plausibility is commonly confounded with its perceptual novelty or familiarity, leading to a variety of interpretations for looking time data (Bogartz, Shinskey & Speaker, 1997). This illustrative study demonstrates the value of factorial designs, in which perceptual (novelty-familiarity) and conceptual (possible-impossible) variables are independently and jointly explored. It also introduces pupil dilation as a viable and complementary dependent measure to study infant cognition. We show that pupil data can assist in the interpretation of otherwise equivocal looking time data. The discussion focuses on methodological considerations in infancy research.