This study aimed at testing an extension of a theoretical model for the metacognitive monitoring mechanism implied in the detection of inconsistencies when the information provided includes abstract symbols in addition to plain text. Ninety-four postgraduates of STEM specialities were asked to read a worked-out algebra-problem example and to report any incoherence, inconsistency, or error detected in the statement or in the solving procedure. A set of model inspired indexes was defined to describe participants¿ behaviour along the task. The Read & Answer software was used to record online individual processing data and participants¿ reports. Results supported model predictions. Indexes correctly predicted participants¿ outcomes in the task with high accuracy. Specific students¿ behaviours could be associated to observed task outcomes with sufficient reliability within the limitations of the study. In addition, algebra processing was compared with plain text processing.