In this eye-tracking experiment, we compared instructional videos with static signals, dynamic signals and no signaling in a within-subject design. We tracked eye movements from 28 undergraduate chemistry students while they were watching instructional videos about reaction mechanisms in the different signaling conditions. Further, we assessed students' cognitive load, as well as retention performance. We employed a Latin square design to control for sequencing and content effects of the instructional videos. Our data showed that dynamic signals helped students to better focus their attention to the relevant features of the representations virtually across the entire time of the video presentation. Furthermore, dynamic signals increased retention performance while they decreased extraneous cognitive load. Overall, our findings show the crucial role of pairing the signaling principle with the temporal contiguity principle in instructional videos to help students navigate through complex symbolic representations and improve their learning success.