Marianne Strickrodt

Qualitative differences in memory for vista and environmental spaces are caused by opaque borders, not movement or successive presentation.

Two classes of space define our everyday experience within our surrounding environment: vista spaces, such as rooms or streets which can be perceived from one vantage point, and environmental spaces, for example, buildings and towns which are grasped from multiple views acquired during locomotion. However, theories of spatial representations often treat both spaces as equal. The present experiments show that this assumption cannot be upheld. Participants learned exactly the same layout of objects either within a single room or spread across multiple corridors. By utilizing a pointing and a placement task we tested the acquired configurational memory. In Experiment 1 retrieving memory of the object layout acquired in environmental space was affected by the distance of the traveled path and the order in which the objects were learned. In contrast, memory retrieval of objects learned in vista space was not bound to distance and relied on different ordering schemes (e.g., along the layout structure). Furthermore, spatial memory of both spaces differed with respect to the employed reference frame orientation. Environmental space memory was organized along the learning experience rather than layout intrinsic structure. In Experiment 2 participants memorized the object layout presented within the vista space room of Experiment 1 while the learning procedure emulated environmental space learning (movement, successive object presentation). Neither factor rendered similar results as found in environmental space learning. This shows that memory differences between vista and environmental space originated mainly from the spatial compartmentalization which was unique to environmental space learning. Our results suggest that transferring conclusions from findings obtained in vista space to environmental spaces and vice versa should be made with caution.

Memory for navigable space is flexible and not restricted to exclusive local or global memory units.

Objects learned within single enclosed spaces (e.g., rooms) can be represented within a single reference frame. Contrarily, the representation of navigable spaces (multiple interconnected enclosed spaces) is less well understood. In this study we examined different levels of integration within memory (local, regional, global), when learning object locations in navigable space. Participants consecutively learned two distinctive regions of a virtual environment that eventually converged at a common transition point and subsequently solved a pointing task. In Experiment 1 pointing latency increased with increasing corridor distance to the target and additionally when pointing into the other region. Further, when pointing within a region alignment with local and regional reference frames, when pointing across regional boundaries alignment with a global reference frame was found to accelerate pointing. Thus, participants memorized local corridors, clustered corridors into regions, and integrated globally across the entire environment. Introducing the transition point at the beginning of learning each region in Experiment 2 caused previous region effects to vanish. Our findings emphasize the importance of locally confined spaces for structuring spatial memory and suggest that the opportunity to integrate novel into existing spatial information early during learning may influence unit formation on the regional level. Further, global representations seem to be consulted only when accessing spatial information beyond regional borders. Our results are inconsistent with conceptions of spatial memory for large scale environments based either exclusively on local reference frames or upon a single reference frame encompassing the whole environment, but rather support hierarchical representation of space.

Spatial Survey Estimation Is Incremental and Relies on Directed Memory Structures

This study examined how navigators of large-scale environmental spaces come up with survey estimates of distant targets. Participants learned a route through a virtual city by walking it multiple times in one direction on an omnidirectional treadmill. After learning, they were teleported to intersections along the route and pointed to multiple other locations. Locations were always queried in chunks of related trials relative to a participant’s current position, either to all locations route forwards or all locations route backwards. For their first pointing, participants took twice as long as for the later pointings and latency correlated with the number of intersections to the target, which was not the case for later pointings. These findings are inconsistent with reading out coordinates from a cognitive map but fit well with constructive theories which suggest that participants integrated locations between their current location and the target along the learned path. Later pointings to adjacent intersections within a chunk of trials continued this process using the previous estimation. Additionally, in first pointings participants’ estimates were quicker and more accurate when targets were located route forwards than route backwards. This route direction effect shows that the long-term memory employed in generating survey estimates must be directed – either in form of a directed graph or a combination of a directed route layer and an undirected survey layer.