Researchers in numerical cognition usually think that the greatest and most common difficulty in children suffering from dyscalculia is retrieval of arithmetic facts from long-term memory. However, we have recently shown that retrieval might not be the optimum strategy in mental arithmetic. In fact, expert adults would rather solve simple problems such as 3 + 2 by automated and unconscious procedures. Therefore, we hypothesize that children with dyscalculia might not present deficit in retrieval but, instead, in counting procedure automatization. The aim of the current project is to test this challenging position. Through a longitudinal approach, we plan to precisely examine the behavior of children suffering from dyscalculia over a 3-year period. Children will be aged between 8 to 11 years at the beginning of the study and we will precisely observe the evolution of their solution times when they solve simple addition problems involving one-digit numbers. If children with dyscalculia still struggle with simple additions three years, their solution times plotted on the sum of the problems should still follow an exponential function. Indeed, if counting is not automated, difficulties necessarily increase with the progression on the number line or the verbal sequence, hence the exponential function. On the contrary, if counting procedures tend towards automatization, moves along a number line will progressively become as easy at the beginning of the line as at the end, hence the linear function. Importantly, a retrieval model would predict exactly the inverse pattern because, according to this model, the linear function, which is unanimously considered as the hallmark of counting procedures, should progressively be replaced by a non-linear function through practice.

Inclusion Criteria:

• Age between 8 and 11 years old
• Having a dyscalculia as described in the DSM (Diagnostic and Statistical Manual) V

Exclusion Criteria:

• Presenting a global intellectual deficit