PROJECT SUMMARY. Arithmetic facts, such as multiplication tables, are learned through verbal rehearsal. In turn, children move from solving problems procedurally (e.g., counting on) to retrieving them from verbal memory. In monolingual children this transition is paralleled by a shift in the supporting brain networks from reliance on the intraparietal sulcus (IPS), involved in numerical calculation, to relying more on language areas, e.g., left middle temporal and superior temporal gyri (MTG/STG). For bilinguals, models of arithmetic suggest that this retrieval process occurs in only one language, or alternatively, that discrete memory stores exist for math facts in each language. Critically, these models are inconsistent with the growing body of research supporting interconnected and overlapping language networks in the bilingual brain, modulated by factors such as language proficiency. This exploratory R21 takes an interdisciplinary perspective, integrating math cognition, cognitive development, and bilingualism research, with the goal of determining if the neural infrastructure for arithmetic in bilingual children is shared or separate across their languages. The work builds on findings that bilingual children show a qualitatively similar language-like brain response during multiplication fact verification in both the language of learning arithmetic (LA+) and the other language (LA-). In two aims, functional magnetic resonance imaging (fMRI) will be used in a cross-sectional sample of 3rd through 5th grade bilinguals to determine 1) the neural networks supporting multiplication fact verification in each language, and 2) if practice in LA- changes these activation patterns. The research focuses on single-digit multiplication given its strong link to language, and uses state-of-the-art methods (e.g., functional localizers and multivariate analysis) to compare activation patterns. Children will judge if the last of three numbers is the correct product of the first two, which will be presented as spoken number words in English (LA+) and Spanish (LA-), separately (e.g., 2 3 6 versus 2 3 8). If arithmetic facts follow bilingual verbal representation patterns, then LA+ and LA- will show overlapping activation patterns in language areas, with recruitment of additional resources for more effortful retrieval. If arithmetic facts are instead language-specific, as models suggest, then separate neural networks should be observed for LA+ and LA-. In Aim 2, a subset of children will return for a second fMRI session after completing training with multiplication facts in LA-. LA- should show increased activation in language areas and decreased activation in additional resources, becoming more similar to LA+ in line with developmental changes. The findings from this research will inform a new bilingual arithmetic model (BAM), which future research can build on to study other types of math concepts across the spectrum of bilinguals. It is essential to consider t...