Dissociable systems of spatial representation in mice

Despite the rich existing literature on spatial navigation, the apparent inconsistencies in cue control by various environmental properties (e.g., geometry, featural landmarks) have made it challenging to bring together the evidence on spatial representations at the cellular and behavioural levels. One reason for this difficulty may be that different behavioural tasks are supported by different cognitive processes and underlying neural representations. Therefore, we propose that two distinctions may be important to consider in studies of navigation behaviour: spontaneous vs. highly trained tasks, and cue-specificity. To address these issues, we test the ability of mice to use two types of environmental properties (geometric and featural) across two types of tasks - a spontaneous, working memory task, and a reinforced, reference memory task. While environmental boundary geometry is spontaneously used for reorienting and navigating to a goal, a distinctive featural property is used only as an associative beacon. However, while the use of a featural landmark quickly improves over time, given experience and reinforcement, the use of boundary geometry is less amenable to reinforcement learning. These results support the claim that environmental boundaries and featural landmarks are encoded by separate processes and are dissociable in the ways they influence learning and spatial memory. Preliminary studies show similar behaviour in other species, including rats; by making the above distinctions in future experiments on rats, in combination with simultaneous head-direction and place cell recording, we will explore the neural correlates of these behaviours.