Object permanence has been investigated with a variety of paradigms and measures, yielding heterogeneous findings. The current study employed a novel Violation-of-Expectation paradigm measuring pupil dilation as indicator of cognitive effort and surprise. Across repeated trials, infants watched videos of animated toys either stopping in an open door frame or moving across the open door frame off screen. The door then closed and opened up again to reveal either the toy, or an empty space. In Experiment 1, 18-month-olds's pupils dilated in response to the unexpected empty outcome more than to the expected empty outcome, establishing the paradigm as a suitable measure of violation of object expectation. Using the same paradigm, Experiment 2 revealed an absence of this object expectation effect for 10-month-olds. Results are discussed with regard to paradigmatic aspects and developmental differences. It is suggested that young infants do not automatically represent occluded objects upon perceiving occlusion events, and that occlusion events may initially require relevance in terms of individual activity or social interaction.