The project concerns both the psychological and neural mechanisms involved in flexible and habitual forms of spatial learning. Traditionally it has been thought that flexible navigation is based upon a map-like representation of a person’s environment – a cognitive map - and that this representation is subserved by the hippocampus. The cognitive map is thought to integrate the relations among stimuli in an environment so that a route through the environment can be planned without having to have direct experience of that route. In addition, influential theories of spatial learning argue that the formation of a cognitive map, based on some parameters around cue type or environmental stability, is incidental – occurring as a person encounters new stimuli, regardless of prior learning experience. Our starting point for our project was to test both the notion of incidental learning in flexible spatial learning, and to determine if a hippocampus-based cognitive map was only one function of the hippocampus in spatial learning, testing if it is also involved in non-mapping functions that involve complex representations of prior and future events and their relations to one another. We conducted a series of spatial learning experiments in which participants control a first-person perspective navigating around and environment. Generally, control was via keys on a computer keyboard and the participant sat a short distance away from a computer screen. In our first output we tested two theories of incidental spatial learning by testing whether certain spatial cues, or environmental stability, created an immunity for spatial learning to be susceptible to a form of cue competition known as blocking. Blocking occurs when prior learning to a cue prevents subsequent learning to an added cue and is the hallmark of associative learning theory. This is because associative learning algorithms rely on prediction error to explain learning, meaning that new learning only occurs when prediction error is high. Incidental learning should occur regardless of prediction error, so some theories of spatial learning predict that blocking will not occur. Our results found no evidence for incidental learning based either on cue type or on environmental stability. Another form of cue competition is overshadowing, when concurrent learning based on two or more cues limits learning based on each, compared to a condition in which learning is based on only one cue. In two sets of studies we examined overshadowing of both boundary cues, which have been argued to be immune to cue competition, and environmental geometry, which has been the subject of a similar claim of modularity. In both studies we found positive evidence for overshadowing, undermining the claim that cognitive mapping is immune to cue competition. Further to our studies of learning based on boundaries and geometry we examined the effects of these parameters on memory for object locations. We found that manipulating these parameters had a profound effect on how participants remembered the locations of objects they had encountered, which may have important implications for theories of episodic memory. The strand of our study examining non-mapping navigation memory was served by two major fMRI studies. Processing of one study was delayed by the Covid-19 pandemic so we are unable to provide those data as yet. The other was designed to observe the activation of the hippocampus and caudate putamen when participants learned their way through a high-sided maze. They could rely on objects they encountered on their way or the sequence of turns, but they could not readily map the environment. We found evidence for caudate putamen activation when behaviour was under the control of landmarks, but evidence for hippocampus activation when behaviour was under control of the sequence. In a subsequent behavioural study we found evidence that sequence learning was still susceptible to blocking. The basis of the second fMRI study was the under-studied effect of reversing a route that was previously learned. In two behavioural studies we examined whether under- or over-training an outward route affected a participant’s ability to reverse their steps, as well we examining the effects of individual differences in spatial wayfinding strategies and behavioural inhibition on reversing a route. Routes could be learned based either on the identity of a landmark at a junction or the sequence of turns taken to reach a goal location. The parameters determined from the behavioural studies formed the basis of the fMRI study, as well as the parameters for a similar behavioural study in which a realistic rendering of a university building was developed. This strand is designed to broaden the scope of our research for more applied work in the future.Problems of navigation (returning to home bases, foraging, efficient route-finding, and selection of escape routes) are universal for all mobile animals, and brain systems have evolved to address these navigation challenges in an optimal and flexible manner. An established division in psychology and neuroscience research has been between a flexible, map-like navigation system, and a habit-like, fixed route-learning system. We tend to become aware of the habit system when it exerts too much control over our behaviour, such as when we find ourselves driving to the office when we meant to take an alternative route, following a lapse of concentration. Currently, two influential models of how these flexible and habit systems function in navigation have been developed, both accounting for important findings in behavioural and brain-systems research. One model focuses on differences between systems at the level of input. In this model the map system codes where things are based on their location with respect to large-scale environmental contours and boundaries, such as room shape in enclosed environments and fences, rivers, and tree-lines in open environments. In contrast, the habit system codes specific actions to be performed with respect to single landmarks e.g. turning left at the junction with the supermarket on the right. A second model focuses instead on whether the type of learning distinguishes flexible and habit-based navigation. In this approach, the flexible system is able to form a "model" of the world, such that if turning left at the junction is appropriate in some situations (such as going to the office), but not others (such as going to the supermarket), this system is able to take account of these contextual associations. The habit system however, accumulates only a history of rewarded responses, so for example if turning left for the office occurs more frequently than turning right for the supermarket, the habit system will exert behavioural pull to turn left, irrespective of one's current goal to go shopping. The aim of the proposed research is to test which of these models is correct, which in turn will answer the fundamental question of how information is processed within and between brain systems. We will gather behavioural and brain scanning data from healthy humans navigating in virtual environments. These data will be used to address three objectives. First, do the brain systems responsible for flexible navigation and fixed route-learning rely on different types of input or different types of learning? That is, do different brain systems process different types of information, or do they process the same kinds of information differently? Second, route-learning has typically been thought not to rely on a brain region known as the hippocampus, yet recent research has shown that the hippocampus is necessary for some forms of route-learning. We hypothesise that this is because under some circumstances route-learning requires a model of the world. We will test the conditions required for hippocampus-based route-learning, and predict that the hippocampus will be required when a sequence of actions must be learned; that is, the action chosen at a particular point in time depends on the actions that came before. The third objective is to apply the factors found to be important in flexible route-learning (objectives 1 and 2) to fire evacuation behaviour in a virtual environment of a real building. This will enable us to optimise training strategies for rehearsing emergency escape behaviour. Some everyday behaviours, such as always entering and leaving one's place of work by the same route, may inhibit flexibility, such as using an unfamiliar exit during an emergency.