Search

We live in compartmentalized, clustered environments and have to deal with spatial information scattered across rooms, streets, neighborhoods, and cities every day of our life. Yet, we are able to piece this information together in our head, for example, in order to find our way from our flat to our workplace, even when faced with construction work and blocked streets. Furthermore, we can point out the direction to the supermarket to a pedestrian without having direct visual access to it. My thesis is concerned with the question of how our memory for spatial relations of places in navigable space (also called survey knowledge) is actually structured. In four consecutive studies, I contrasted two major theo-retical approaches that try to explain how we represent survey knowledge, namely, Euclidean map and enriched graph approaches. Euclidean map approaches assume that spatial locations are represented in a map-like, globally embedded, Euclidean format. Enriched graph approaches propose a partitioned, unitwise representation of places connected in a network. These local units are not required to be globally consistent. In each study, I used different virtual environments, sometimes single rooms, mostly navigable multi-corridor environ-ments, once even an impossible non-Euclidean environment. Participants learned spatial relations between objects spread across these environments and solved survey tasks afterward (e.g., pointing to object locations from memory). Their performance yielded multiple effects. In short, the most prominent effects were: (1) Pointing latency increased with increasing number of places along the route towards the target, (2) facilitated recall along the direction of the initially experienced path walked within the environment, (3) globally incoherent point-ing behavior following the local metrics experienced from place to place, (4) facilitated performance upon alignment with local corridor geometry but also (5) upon alignment with regional geometry and a global main orientation, and (6) decreased pointing latency when pointing beyond regional boundaries. Interpreting these effects jointly implies that human survey knowledge is not repre-sented in the form of a Euclidean mental map embedding all encountered places in a uniform, globally consistent format. Instead, just as the environment we experience, also our memory of it seems to be compartmentalized, consisting of a network of local places connected by directed links that specify how to get from one place to another (rotation and translation) without directly requiring a global calibration. Survey estimates have to be constructed incrementally following this graph structure along the memorized connectivity, thereby relying on the local metrics that enrich the graph entities. These estimates are generally transient but can be retained for a limited amount of time for aiding subsequent estimates. In addition to the local entities of the enriched graph representation, it seems that general reference directions can be acquired during learning a navigable multi-compartment space. Such a reference direction can be understood as a mental “north”, a main direction that is tried to be main-tained and propagated across multiple local places and represented supplemen-tary in memory. It might be limited to only a sub-group of local units, thereby forming regional clusters, or it can cover the entire environment that was encountered. Such a general reference direction can aid the coordination of the local memory units during the construction of survey estimates, however, it does not require a global embedding of all place information into a coherent Euclidean map format. In sum, our representation of navigable space seems to be best described as an impossible puzzle where the memorized pieces and connec-tions do not necessarily match up on a global scale.