This study examined the changes in relations between symbolic and non-symbolic representations of numerical quantities subsequent to the implementation of numerical activities in a group of 14 children with Down syndrome (trisomy 21 - T21). Performances of these Down syndrome children were tested three times (at P1 baseline; at P2 after 7 weeks with no special intervention; at P3 after 6 weeks of
systematic numerical activities) both in trained activities and comparative tests of symbolic quantities or not which had not been directly solicited. The results showed a beneficial effect of the activities compared to the period free from solicitations. In the last test (P3), performances significantly improved in numerical knowledge and in comparisons of quantities. In the latter, the change was worthwhile both in terms of symbolic representation âeuro" which was expected insofar as it was solicited primarily during the activities âeuro" as well as non-symbolic representation. This result is discussed in the perspective of a triple-code model : intensification of the practice may have consolidated the relations between the analogue representation, confirmed by the distance effect in taking each test, and the symbolic representation made more accurate as the result of training.