Our objective in this study was to investigate how the eye-movement behavior and concurrent verbal protocols of students with high-/low-prior-knowledge were reflected in the use of multiple representations for scientific argumentation. We also examined the degree of consistency between eye-fixation data and verbalization to ascertain how and when the eye-mind hypothesis (EMH) applies in this subdomain of scientific argumentation. Our results focused on fixation duration and recorded arguments from 96 college students. The high-prior-knowledge group did not present static patterns in the inspection of multiple representations, which indicates that they tended to select representations according to the contingent demands of the current task, indicating that for them, there was no "most appropriate representation". The high-prior-knowledge group also submitted a greater number of representations and more frequently mentioned multiple representations in their verbal protocols. Finally, the students demonstrated notable discrepancies between eye-movement data and verbal protocols related to representations as well as inconsistencies with previous findings. Thus, the fact that the EMH does not always hold could perhaps be attributed to the scope of interpretation in argumentation tasks and the complexity of information related to some representations, both of which could hinder the instantaneous formation of a gist. Our findings may contribute to reducing the ambiguity and uncertainty involved in the analysis of eye-fixation data when multiple representations are employed for scientific argumentation.