Methodology

Principles underlying the design of "The Number Race", an adaptive computer game for remediation of dyscalculia

Anna J Wilson¹ , Stanislas Dehaene^1,2 , Philippe Pinel¹ , Susannah K Revkin¹ , Laurent Cohen^1,3 and David Cohen⁴

¹  INSERM-CEA Unit 562 « Cognitive Neuroimaging », Service Hospitalier Frédéric Joliot, CEA-DRM-DSV, 91401 Orsay, France

²  Collège de France, 11 place Marcelin Berthelot, 75231 Paris Cedex 05, France

³  Service de Neurologie, Hôpital de la Pitié-Salpêtrière, AP-HP, 47 bd de l'Hôpital, 75013, Paris, France

⁴  Department of Child and Adolescent Psychiatry, Université Pierre et Marie Curie, Laboratoire CNRS "Du comportement et de la cognition", Hôpital Pitié-Salpêtrière, AP-HP, 47 bd de l'Hôpital, 75013, Paris, France

author email corresponding author email

Behavioral and Brain Functions 2006, 2:19doi:10.1186/1744-9081-2-19

Published:	30 May 2006

Abstract

Background

Adaptive game software has been successful in remediation of dyslexia. Here we describe the cognitive and algorithmic principles underlying the development of similar software for dyscalculia. Our software is based on current understanding of the cerebral representation of number and the hypotheses that dyscalculia is due to a "core deficit" in number sense or in the link between number sense and symbolic number representations.

Methods

"The Number Race" software trains children on an entertaining numerical comparison task, by presenting problems adapted to the performance level of the individual child. We report full mathematical specifications of the algorithm used, which relies on an internal model of the child's knowledge in a multidimensional "learning space" consisting of three difficulty dimensions: numerical distance, response deadline, and conceptual complexity (from non-symbolic numerosity processing to increasingly complex symbolic operations).

Results

The performance of the software was evaluated both by mathematical simulations and by five weeks of use by nine children with mathematical learning difficulties. The results indicate that the software adapts well to varying levels of initial knowledge and learning speeds. Feedback from children, parents and teachers was positive. A companion article [1] describes the evolution of number sense and arithmetic scores before and after training.

Conclusion

The software, open-source and freely available online, is designed for learning disabled children aged 5–8, and may also be useful for general instruction of normal preschool children. The learning algorithm reported is highly general, and may be applied in other domains.