Navigation based on the formation, storage, and retrieval of spatial information is crucial for the survival of many species. We use a modification of a fully automated T-maze task for mice requiring flexible processing of visual information about the external environment. In our experimental design mice are presented with extra maze visual cues of different levels of complexity - general contexts and salient landmarks. They are conditioned to navigate based on the visual cues, acting against the spontaneous tendency to alternate. The animals are presented with pseudo-random sequences of turns, minimizing the risk of developing strategies independent of visual cues.We show that in this protocol mice learn to use different types of allothetic cues to navigate. We investigated how presenting different cue classes separated from the full visual scene would affect their performance. While the animals successfully navigate using partial cues from different classes achieving similar levels of performance, the brain structures involved in the process differ between groups.Using this system together with iDisco+ optical tissue clearing and light-sheet microscope imaging we identified the locations of individual c-Fos-positive nuclei in the whole brain. That allowed us to analyze which brain structures are involved in landmark and contextual navigation and how changes in functional brain modularity reflect differences between the groups. We noticed increased brain modularity for the mice processing complex contextual cues. Mice exposed to landmark cues had fewer distinct modules, and the network seemed to be dominated by the module containing structures traditionally associated with pattern separation.