The present study replicated the investigations of Lemaire, Abdi, and Fayol with some modi®cations: the random time interval generation (RIG) task was used and the stimuli were created in another way. The results provide additional evidence for the crucial role of the central executive in the speed of solving both true and false sums and for the role of the phonological loop in solving false sums. However, the ®ndings concerning the role of this slave system in solving true sums were dierent. Possible explanations and limitations of these results are discussed.

Two experiments were conducted to investigate the role of phonological and executive working-memory components in the carry operation in mental arithmetic. We manipulated the number of carry operations, as previous research had done, but also the value that had to be carried. Results of these experiments show that in addition to the number of carry operations, the value of the carry is also an important variable determining the difficulty of arithmetical sums. Furthermore, both variables (number and value) interacted with each other in such a way that the combination of multiple carries and values of carries larger than one resulted in more difficult problems irrespective of the presence of a working-memory load. The findings with respect to working-memory load suggest that mainly the central executive is important in handling the number of carry operations as well as the value that has to be carried. The implications of the present findings for our views on mental arithmetic and its reliance on working memory are discussed.

In two experiments, we investigated the role of the phonological loop and the central executive in the verification of the complete set of one-digit addition (Experiment 1) and multiplication (Experiment 2) problems. The focus of the present study was on the contradictory results concerning the contribution of the phonological loop in the verification of true problems (e.g., 8 + 4 = 12 or 4 3 6 = 24) reported until now. The results revealed that this slave system is not involved in verifying simple arithmetic problems, in contrast to the central executive. Furthermore, our results indicated that the split effect is due to the use of two different arithmetic strategies.