Modeling has strongly influenced our understanding of biological navigation, spanning diverse model organisms, neural subsystems, and varying levels of abstraction. Yet to compare models and methods across species requires nontrivial assumptions about how nervous systems of different species are related to each other. Aided by evolutionary considerations and a framing of neural systems according to Marr’s three levels of analysis, the relations between different organisms can be rendered explicit and will be explored here through two case studies. First, head-direction coding in flies, rodents, and potentially humans highlights likely commonalities between phylogenetically distant organisms. Second, different neural implementations of vector navigation in flies and rodents highlight likely differences. Marr’s three levels can also be leveraged to cope with the increasing complexity and fewer constraining data for neural-level models of human navigation. A way forward could lie in the use of large-scale systems-level models, which allow us to explore interactions among multiple neural subsystems during successful navigation and, crucially, idiosyncratic failures. These models may also lead to a mechanistic understanding of how individual differences and the effects of aging across subsystems might manifest in behavior, bringing to the forefront the need for robust, biologically plausible, and gracefully degrading neural models.