Interference of irrelevant salient variables may cause difficulties for students. This study focused on eye tracking during the comparison of perimeters task, in which area is the interfering irrelevant salient variable. There were three trial types: congruent (larger area-larger perimeter), incongruent inverse (larger area-smaller perimeter), and incongruent equal (larger area-equal perimeter). Behavioral findings corroborated previous studies: congruent trials yielded higher accuracy and a shorter reaction time than did incongruent trials. Surprisingly, the area saliency could not be revealed in fixation location and duration measurements in incongruent inverse trials nor in the heat maps for incongruent inverse or incongruent equal trials, suggesting that such processing does not require overt attention; measures of attention shift from one geometrical shape to another were higher for incongruent equal trials, as was pupil dilation, suggesting that greater effort is associated with solving incongruent equal trials.