Abstract: Mathematics learning disability (MLD) is an important area of learning disability. Now there emerging a leading paradigm for carrying out research on MLD from the viewpoint of online information processing, and more and more studies focused on discussing the cognitive processing mechanism of MLD. The definition and screening methods of the MLD are still in dispute, therefore it needs further research to explore the different information processing features of different kinds of mathematics. Based on the three-factor model put forward by Baddeley and Hitch (1974), this study designed the experiment tasks of central executive system, visuo-spatial sketchpad and phonological loop, exploring the differences of the three components of working memory between the 55 MLD students and 56 students who are good at math (with 48 male students and 63 female students, average age was 11.97-year-old). Then, according to the criteria of national math curriculum, math learning was further divided into two parts, that is, (a) counting and algebra, (b) space and geometry. The cognitive processing mechanism of various kinds of mathematics study was investigated. In the first study, a serial cognitive-behavior computerized tasks were composed to test the three components of working memory, such as, the stop signal task and Flanker’s tasks testifying the central executive function, N-back and Spatial Figure Position tasks testifying the visuo-spatial sketchpad, and the digit span and sentence span tasks which tested the phonological loop. The cognitive characteristics of the MLD students in some specific math leaning area were analyzed. In the second study, the author chose the specific math knowledge, Axial Symmetry and Centro Symmetry tasks from the space and geometry, which was learned by grade seven students, to discover the cognitive feature between the MLD group and the group who were good at math. The results showed that: (1) the MLD group performed poorer than the normal group in central executive system, visuo-spatial sketchpad and phonological loop, indicating the WM deficits among the MLD students was domain-general. (2) the working memory deficit in the MLD students was domain general, but various kinds of math learning belonged to different cognitive processing mechanism. The tasks of counting and algebra were influenced by the combined role played by the central executive system, visuo-spatial sketchpad and phonological loop; the tasks of space and geometry were influenced by the central executive system, visuo-spatial sketchpad instead of phonological loop. (3) visuo-spatial sketchpad predicted the performance of Axial Symmetry and Centro symmetry tasks, then followed by central executive system, and phonological loop had little effect to this task. The second stud testified the specificity of phonological loop, and the impact to geometry exerted by visuo-spatial sketchpad was obvious. This indicated that during math learning, working memory was not only domain specific but also domain general. The central executive system, visuo-spatial sketchpad were characterized by domain generality and phonological loop was characterized by domain specificity.

Key words: junior school students, mathematics learning disability, central executive system, visuo-spatial sketchpad, phonological loop, counting and algebra learning, space and geometry learning