While robotics is becoming a popular tool to introduce an integrated STEM curriculum into both, elementary and secondary schools, its impact on students’ learning remains underexplored. This article addresses the complexity of assessing mathematics learning within a robotics-based task by identifying the different types of knowledge and processes related to digital, mathematical and sociocultural contexts inherent in robotics-based learning. We also studied the ways students interpreted the feedback received when performing an assessment task. In the context of the Innovative Learning Agenda implemented in New Brunswick schools, one team of Grade 5–6 students and one team of Grade 6–7 students were asked to perform a robotics-based assessment task. Analysis of students’ performance within this task showed that the interpretation of feedback that students make leads them to think and to act differently when solving different challenges that arise during their work on the assessment task. The identification of this variety may help teachers in dealing with complexity related to this kind of assessment.