Two independent lines of researchmental simulations and thought experimentsprovide strong arguments about the importance of perceptual modalities for the instructional practices in science education. By situating the use of mental simulations in the framework of thought experiments, this study investigated the nature and the role of mental simulations in the context of problem-solving. This study draws on data collected through problem-solving sessions with five physics graduates. Throughout the problem-solving sessions, think-aloud and retrospective questioning were used. Results from this study support some serious concerns put forward by several researchers in the community of science educators about high dependence on descriptive forms of scientific knowledge and exclusion of perceptual modalities from instructional practices. The participants' verbal reports confirmed that they had implicitly or explicitly reached a conclusion that mental simulations were not a legitimate way of reasoning about physics problems, and they consciously avoided the use of mental simulations. This conceptualization seemed to lead participants to compartmentalize mental simulations from formal physics knowledge. Therefore, mental simulations were not refined but kept in a primitive form, which was no more than a retrieval of perceptual representations constructed through observations and experiences of the world. The speculations on the results of the study were based on the interpretations of learning science in terms of the refinement and reorganizations of preinstructional ideas.