Ostensive-referential communication is argued to be pivotal for learning in infancy (Csibra & Gergely, 2009), and even to have a specific effect on what infants learn about objects. Yoon, Johnson & Csibra (2008) found that after viewing an ostensive pointing scene (‘Hey baby’, actress pointing to object, occluder covering object, occluder revealing a change), 9-month-olds detected object identity changes more than object location changes. But after viewing a non-ostensive reaching scene, infants detected object location changes more. These results were interpreted as ostension boosting identity encoding. However, the relative contribution of ostensive and referential signals cannot be concluded from this experiment. In experiment 1, we will conduct a direct replication of the previous study, comparing ostensive pointing and non-ostensive reaching. In experiment 2, we will add two new conditions: ostensive reaching and non-ostensive pointing. Infants will see action videos where an actress performs actions towards a novel object. We will use an eye-tracker to investigate change detection (either of object identity or location), as well as where infants are looking during the action scene. Each infant will see six test scenes: two action conditions (ostensive reaching, non-ostensive pointing) and three different outcomes (no change, identity change, location change). If the effect observed in Yoon’s study is due to ostension boosting object identity encoding, we will observe both ostensive conditions causing longer looking times for the object’s identity change. If the effect is due to referentiality boosting object identity encoding, we will see both pointing conditions causing longer looking times for the identity change. If it is the combination that drives the effect, we will only see longer looking times for the identity change in the ostensive pointing condition. Results from this study will help differentiate the roles of these cues, and how they contribute to infant learning and memory.