Sensitivity To natural scene statistics in infancy and its impact on development

Abstract

Infant sensitivities for the statistical regularities found in their visual world are underexplored. From birth, infants are exposed to a complex visual experience with large variations in hues, textures, and forms. The extent of these visual properties means that infants must find a way to filter important information from irrelevant information, parsing visual statistics that aid in their learning of the world around them. Research has shown that infants are sensitive to some of the visual statistics of real and virtual everyday scenes (‘natural scenes’) such as colour distributions and also use these properties to facilitate actions such as target detection. These studies suggest that infants’ sensitivity to these statistics aids in creating a standardized model of their visual environment. Research has also found variations in the timeframes of when infants develop sensitivities to different natural scene statistics, with some sensitivities emerging later than others, prompting questions as to the methods and limitations behind infants’ perceptual learning. However, this area of research remains largely unexplored, and as such many questions remain unanswered. The current review investigates the existing literature from the past 25 years surrounding infant sensitivities to these natural scene statistics, how they use these statistics to find structure in their visual world now and into adulthood, and the effects that different environments’ characteristic visual statistics may have on their subsequent behaviour and development. We also propose directions for future work that will fill the gaps in this underdeveloped, but key, area of infant research.