Abstract
Recent studies with 5-year-old children provide evidence that analog magnitude representations mediate numerical comparison of large sets of visual elements (La Mont, Barth, & Spelke, SRCD 2002). Related studies in adults suggest that these representations are abstract: comparisons across stimulus format or modality are as fast and accurate as comparisons within format or modality (Barth, Kanwisher, & Spelke, Cognition 2003). Here we present evidence that preschool children can also perform across-modality comparisons of large sets of elements that are too numerous and presented too quickly to count. Children were shown animated sequences depicting numerical comparison problems with arrays of 6 to 60 elements. First a pair of rectangular occluders appeared (one blue and one red). An array of blue dots appeared and moved behind the blue occluder. Then a series of tones played; participants had been taught that each tone represented an unseen red dot, concealed behind the red occluder. Finally, the child was asked whether there were more blue dots behind the blue occluder or more red dots behind the red occluder. The numbers of dots differed by ratios of 4:7, 4:6, or 4:5, with red dots more numerous on half the problems. Children succeeded at this task (66% correct overall) and accuracy depended on the difficulty of the comparison ratio (77%, 64%, and 58% correct, respectively): the signature of the analog magnitude system of large number representation. In addition, performance for this crossmodal comparison task was as good as within-modality comparison performance, paralleling previous results with adults and providing evidence that children's large number representations, like adults', are amodal.This ability is present prior to formal math training in school.
This research was supported by a NSF-ROLE grant to E. Spelke.