A taxonomy of spatial navigation in mammals: Insights from computational modeling

• Published in: Neuroscience and biobehavioral reviews Vol. 176; p. 106282
• Main Authors: Vijayabaskaran, Sandhiya, Zeng, Xiangshuai, Ghazinouri, Behnam, Wiskott, Laurenz, Cheng, Sen
• Format: Journal Article
• Language: English
• Published: United States Elsevier Ltd 01.09.2025
• Subjects: Animals; Artificial agents; Computational models; Computer Simulation; Generalization; Humans; Mammals - physiology; Models, Neurological; Spatial behavior; Spatial navigation; Spatial Navigation - physiology; Spatial representations; Taxonomy; Spatial behavior; Spatial representations; Taxonomy; Computational models; Artificial agents; Spatial navigation; Generalization
• ISSN: 0149-7634; 1873-7528; 1873-7528
• DOI: 10.1016/j.neubiorev.2025.106282

Spatial navigation is a vital cognitive process in nearly all animals, relying on complex neuronal mechanisms to extract, process, and act upon spatial representations. To advance the understanding of spatial navigation and its neural mechanisms, Parra-Barrero et al. (2023) have proposed a taxonomy of spatial navigation processes based on extensive behavioral and neural studies. These processes are hierarchically organized in two levels with navigation strategies at the top and behaviors at the bottom. Building upon this taxonomy, here, we review computational modeling studies on spatial navigation in mammals to provide an overview of the current state of the art and further analyze the navigation processes within the proposed taxonomy. We specifically focus on the representations required by navigation processes, how these representations are extracted, and the computations necessary to execute each strategy and behavior. We propose that the key to understanding what representations and computations are being used by agents lies in testing their ability to generalize to novel situations. We identify three types of generalization relevant for navigation and analyze to what extent current computational models are capable of achieving these types of generalization. Our review shows that while significant progress has been made in modeling navigation, substantial work remains to model and fully understand spatial navigation in mammals. •We build upon and expand a hierarchical, nested taxonomy of spatial navigation.•Each navigation process is characterized by its representations and computations.•The key to pinpointing processes is testing the ability of the agent to generalize.•We review biologically plausible and reinforcement learning models of navigation.•We identify key gaps in the literature on spatial navigation in mammals.