Electronic Thesis and Dissertation Repository


Doctor of Philosophy




Dr. Daniel Ansari


An arithmetic training study was conducted using a novel paradigm known as Customized Arithmetic Training (CAT). Using the CAT system, self-reports obtained from the participants were used to generate individually tailored problem sets. These problem sets balanced strategy use such that each participant started with an equal amount of problems solved by fact retrieval (e.g., 2 + 2 = 4) and an equal amount of problems solved by procedural calculation (e.g., 34 + 37). Following the training period, participants solved trained and untrained problems from their customized arithmetic sets while undergoing an fMRI scan, after which they again provided self-reported strategy.

Through the use of the CAT paradigm, which tracks (for the first time) arithmetic strategy both pre- and post-training, the neural correlates of arithmetic learning were examined by separating calculated problems which became memorized through training from problems that were rehearsed but did not show a shift in strategy. This analysis produced results consistent with previous studies of arithmetic training, namely a shift from widespread fronto-parietal activation to focal activation of the angular gyrus. However, it also produced several novel findings relating to neural correlates of mental arithmetic, namely an association between right anterior hippocampus in fact retrieval as well as evidence of a temporal gradient which affected brain activity when comparing new vs old arithmetic facts. Furthermore, analysis of training effects on calculated problems (which did not become memorized) revealed a modulation of activity in the putamen, a structure commonly associated with the procedural memory system.