##Towards real-world capable computational cognitive models of spatial memory
How do brains learn where things are, despite sensory noise and spatial complexity? This thesis argues that they use Bayesian mechanisms. It provides evidence that brain cells representing space can perform Bayesian inference and filtering. It shows that human spatial memory structure can be modelled and predicted by clustering in psychological space. It presents a cognitive model of how brains might estimate, correct, and structure representations of space. To demonstrate real-world ability, the accuracy of this model is compared with human data in realistic robotic simulations.