The directionality of finger counting (i.e., from left to right or right to left) is supposed to compete with the orientation of the mental number line in determining number mental representations. Indeed, Western individuals who count on their fingers from right to left present a weaker SNARC effect than do individuals for whom the directionality of counting is the same as the mental number line. Observations of natural behavior should be the preferred methodology for classifying individuals according to their counting habits. Yet, to perform such classification, researchers usually rely on questionnaires or reports of imagined behaviors. However, we show in a series of three experiments that, on average, 26% of a sample of adults reported the opposite behavior from the one they actually implemented spontaneously when tested with an original ecological task. In a fourth experiment, this new task proved reliable, using a test-retest method. These results suggest that future studies about counting habits could benefit from the use of more ecological and functional tasks, rather than depending on noncontextualized questionnaires.