Building a Large-Scale, Cross-Cultural Dataset to Advance Theory and Methods in Infant Cognition
Research Question: High-profile scientific findings over the past two decades have led to the conclusion that infants – who cannot walk, talk, or interact beyond smiling and cooing – can understand words, estimate quantities, compute probabilities, reason about others' beliefs, and even make moral evaluations. Discoveries about early cognition have radically reshaped our conception of infants, our practices as parents, and our understanding of the origins of human knowledge and learning. Many of these prominent findings across fields including psychology, linguistics, and education rely on a key insight by Fantz (1960): that infants’ attention to familiar versus novel objects and sounds can be used to make inferences about infants’ cognitive processing. Even though virtually all behavioral research methods with infants rely on measures of attention, the conditions under which infants show preferences for familiar and novel stimuli are poorly understood. We plan to build the first large-scale, cross-cultural dataset equipped to answer a key question about infant cognition: what factors influence infants’ attention to familiar and novel stimuli?
Data: We will conduct a behavioral experiment with 2,000+ infants to assess their attention to visual stimuli of various degrees of complexity. Data from each infant will be included in a large dataset, which will enable the development of robust and highly-powered statistical models of infants’ looking patterns and computational models of infants’ information processing, and will facilitate advances in computer science such as improving automated detection and analysis of infants’ looking patterns.
Methods: The experiment will consist of a series of trials in which infants will see a stimulus repeatedly, and then will be presented with this stimulus alongside a novel stimulus. Infants’ looking time to each of the stimuli will be measured. The dominant theoretical model of infants’ attention (proposed by Hunter & Ames, 1988) describes how three key factors shape infants’ attention: the age of the infant, the difficulty of the task, and the amount of time they are exposed to stimuli. These three dimensions will be manipulated in the experiment. In addition to the experimental data, we will collect a range of additional information about participants and their testing contexts: demographic data, cultural background, and information about individual differences in early infant experiences, such as experience with multiple languages or developmental delays.
Challenges: The statistical power associated with infant psychology experiments is often limited by small sample sizes. Large and diverse samples, such as those in our study, will allow us to investigate critical issues of replicability and generalizability that have plagued infant science and the field of psychology more generally. Indeed, much of what we know about human behavior is based on ‘thin slices’ of the population. An equally concerning crisis for psychological science is the reliance on what are known as WEIRD (western, educated, industrialized, rich, democratic) samples. While these samples are often convenient, they do not represent the majority of human beings who live now or who have ever lived. This crisis is especially important for infant research, as the environments in which infants are raised vary dramatically across the globe. Importantly, many of our key assumptions about the methods we use to study infant behavior are rarely tested outside of WEIRD contexts, further calling into question the inferential foundations for studying the developing mind. With 200 collaborators (122 labs, 40 nations, and 6 continents), the scale and diversity of our study is unprecedented, but forming meaningful collaborations with labs from all around the world will be a significant challenge, especially when it comes to deploying the same experiment design at many sites using the same basic technology.
Findings: The results of this study will not only inform our interpretation of prior infant studies, but will also guide future research by generating new predictions, hypotheses, and theories of infant attention.