Cancellation tasks have been largely used to evaluate visuospatial function and attention. Cognitive evaluation of low literacy subjects remains a challenge in developing countries, when it becomes necessary to distinguish between what is pathological and what is biased by low education. Performance of river bank dwellers of the Amazon region was studied, in a structured nonverbal cancellation task, verifying their searching strategies (randomized/organized), time of completion, number of correct cancelled targets and number of false-positive targets. A difference was observed in performance and searching strategies between illiterates and literates with only a few years of schooling (mean=0.8, S.D.=1.6 years of education) across all measures. There was a significant difference between literate groups in the searching strategy, as well as between illiterates who had never attended school and those who had, showing that a minimal contact with graphic presentations and organization of writing was able to modify this cognitive function.