Understanding how infants perceive real-world scenes and the type of information they rely on when recognizing different kinds of scenes remains unexplored. In this study, we aimed to investigate the relationship between action and scene information in infants. In a preferential looking paradigm, 18-month-olds were exposed to several trials in which they observed a human performing a given action and a subsequent simultaneous display of two scenes. One of the scenes was congruent with the action, representing the environment where the action is more likely to occur, whereas the other was incongruent. Results revealed a significant preference for looking at the congruent scene, accompanied by a longer first visit duration of that scene. Our findings show that the relation between action and scene information, previously reported for adults, is present already in infancy, suggesting a potential role of action information in shaping the construal of scene representation.

Human adults are extremely good at inferring the overall meaning of scenes in a single glanceÑtheir gistÑsuch as "a park" or "a kitchen". However, little is known about categorical knowledge of scenes and the extraction of scene gist in development. In two preferential-looking experiments, we investigated whether 12- and 18-month-old infants categorize scenes based on gist and the role that verbal input plays in extracting such conceptual commonality. Whereas 12-month-olds showed no evidence of scene categorization, 18-month-olds categorized scenes when the images were presented with a label void of meaning in the familiarization phase. The observed facilitating effect of language on categorization, previously already shown in studies with objects, seems to emerge at a later age for scenes, possibly due to their inherent complexity. Our findings show for the first time that by the age of 18 months, infants can categorize visual scenes based on abstract commonalities.