Currently mathematics difficulties in schools are a major problem due to several factors. Some research suggest that mathematics teaching-learning methodology could be one of the causes. As a result, alternative teaching methods to the traditional approach (ciphers-based closed algorithm [CBC]) have emerged, such as numbers-based open algorithm (ABN) method. Some research about this new approach has emerged, including neuropsychological studies. The current study aims to analyze performance and potential cognitive differences in solving a computerized task linked to eye-tracking device, comparing CBC and ABN approaches. 18 5th & 6th graders participants were evaluated through a computerized mental arithmetic task. Nine participants learned mathematics with CBC, and nine with ABN approach. Participants were distributed according to his/her mathematical performance rate in three sub-groups, three students per sub-group: low, medium, and high. The ABN method group obtained a higher overall score in the computerized task (mean [M][subscript CBC]=16.22; M[subscript ABN]=17.11), but the differences were not statistically significant (p=0.690). However, significant differences have been found in two eye-tracking measures. ABN method group obtained a lower number of fixations average in areas of interest [AOIs]) (M[subscript CBC]=5.01; M[subscript ABN]=3.85; p=0.001), and a lower pupil diameter average in AOIs (M[subscript CBC]=4.07; M[subscript ABN]=3.91; p=0.001). This occurred regardless of the participants' mathematical performance. These results suggest that differences between groups were not in task performance, but in cognitive effort spent in solving the task.