The representation of an engineering system can come in different modalities. These modalities may impact how designers’ reason about a given system, which could have implications for how solutions are derived to a given problem. In this paper, function structures, function graphs, and component graphs are compared to understand how each representation modality impacts subsystem clustering behavior. This was achieved through a human-subjects experiment that presented three common products (a hair dryer, a food mixer, and a toilet) represented in one of the three described modalities and a troubleshooting scenario. Results suggest that the effects of the representation modality are consistent between participants at two different universities; however, these effects are not consistent between products. Similarly, the effects of representation modality on the occurrence of element pairs also vary between products, suggesting that the specifics of the system that is represented have a non-negligible effect on the clustering patterns. This analysis can be expanded in future work to include clusters of larger size to take a deeper look at subsystem reasoning. These preliminary results show that representation modality can affect systems thinking. This work takes a step towards understanding how function modeling impacts design cognition at the subsystem level, and how functional decomposition compares to other modalities of system representation.