Hysteresis-induced changes in preverbal infants’ approximate number precision

Infants represent the approximate number of items in visual and auditory arrays. These number representations are noisy: for example, whereas 6-month-olds discriminate numerosities that differ by a 1:2 ratio (e.g., 8 vs. 16 dots), they fail to discriminate a 2:3 ratio (e.g., 8 vs. 12 dots) until 9 months old. How should we understand the nature of the representations underlying this performance? One possibility is that the precision of approximate number representations is fixed at a given age; alternatively, precision may be dynamic and context dependent. Here we asked whether one aspect of context—prior numerical experience—influences preverbal approximate number precision. We familiarized 6-month-old infants with pairs of images containing different numerosities. Critically, as trials progressed, the ratio of the two numerosities within each pair also gradually progressed– either from highly discriminable ratios to ratios that became harder to discriminate, or vice versa. After this ordered numerical training, we tested infants’ ability to discriminate numerosities differing by a challenging 2:3 ratio, with which infants of this age typically fail. In three experiments, we found that 6-month-old infants successfully discriminated the 2:3 ratio after starting with easy ratios and progressing to hard ones, but not after starting with hard ratios and progressing to easy ones, despite experiencing identical numerosities across these two conditions. This “numerical hysteresis” effect was feedback-dependent: infants only succeeded with the scaffolded training when they received trial-by-trial feedback. Together, these results provide evidence for a temporary modulation of infants’ number sense.