Proceedings of the Annual Meeting of the Cognitive Science Society

The origin of abstract concepts remains a central question in cognitive science. Empiricists argue that abstract concepts can be learned through a small set of domain-general cognitive mechanisms, whereas nativists argue that most abstract concepts are not learned at all. To explore these opposing views, we examine the abstract concept of number. Drawing on a recent "emergentist" proposal derived from machine learning models, we tested whether stimuli that align with experiences critical for an empiricist view of number perception result in more accurate and less variable numeric estimations. Five- to nine-year-olds were tested on stimuli that conformed to the natural statistics of objects (i.e., number was correlated with features such as cumulative area and spatial position, and the distribution of stimuli was power-law based) or those that did not comply with these statistics. We find little-to-no evidence for predictions of the emergentist theory when presenting children with natural stimuli.