When do we integrate spatial information acquired by walking through environmental spaces?

When do we integrate spatial information acquired by walking through environmental spaces? Agnes Henson (agnes.henson@tuebingen.mpg.de) Max Planck Institute for Biological Cybernetics Spemannstr. 38, 72076 Tubingen, Germany Hanspeter A. Mallot (hanspeter.mallot@uni-tuebingen.de) Cognitive Neuroscience, Eberhard-Karls-University Tubingen, Auf der Morgenstelle 28, 72076 Tubingen, Germany Heinrich H. Bulthoff (heinrich.buelthoff@tuebingen.mpg.de) Max Planck Institute for Biological Cybernetics Spemannstr. 38, 72076 Tubingen, Germany Department of Brain and Cognitive Engineering, Korea University, Anam-dong, Seongbuk-gu, Seoul, 136-713 Korea Tobias Meilinger (tobias.meilinger@tuebingen.mpg.de) Max Planck Institute for Biological Cybernetics Spemannstr. 38, 72076 Tubingen, Germany Abstract The present study examined whether spatial information of a novel environment was integrated within a reference frame during initial learning, or only later when required for pointing to other targets. Twenty-two participants repeatedly walked through a multi-corridor virtual environment, presented via a head-mounted display. At several stages within the learning process they were teleported to locations along the route and asked to self-localize and point to other locations. Pointing was faster during later tests as well as for closer targets, both of which might require less integration. Participants tested only after extended exposure (late pointers) took longer than participants who had received testing interspersed throughout the same amount of exposure (early pointers). Pointing latency did not differ between groups when comparing performance on their first pointing test, despite vastly different exposure. These results are inconsistent with the assumption that participants already integrated spatial information within a single reference frame during learning and simply accessed this information during testing. Rather, spatial integration is a time consuming process which is not necessarily undertaken if not required. Keywords: Reference frame; environmental space; spatial integration; survey task; pointing; virtual environment Introduction When exploring a novel environmental space such as a city or building, navigators encounter various views of this space. Each location within the environment is experienced from an egocentric perspective, but for so called survey tasks such as shortcutting, pointing or straight-line distance estimations, these locations must be spatially integrated into a common reference frame (egocentric or allocentric). Spatial integration can be defined here as “the process of combining different spatial representations that have been formed by multiple experiences within a single frame of reference or co-ordinate system” (Meilinger, Berthoz & Wiener, in press). For example, with regard to pointing, navigators must know where their target is relative to their current position, i.e., they must represent the target within the same reference frame as their body. The question asked here is ‘when does this integration happen?’. Many theories concerned with spatial memory assume that when navigating a space, all spatial information is integrated within a single global reference frame, at least eventually, and can then be used for survey tasks (Byrne, Becker & Burgess, 2007; McNamara, Sluzenski & Rump, 2008; O’Keefe, 1991; Poucet, 1993). Some of these positions assume or imply that integration occurs during encoding or consolidation, regardless of whether navigators will use this knowledge for survey tasks or not. Spatial integration is thus independent from accessing this information. Alternatively, multiple representations acquired during navigation might also be kept separate in memory and only integrated when necessary, for example, when conducting a survey task (Meilinger, 2008). These two positions yield different predictions about the time it takes to perform a survey task, such as pointing as a function of (1) the amount of experience with an environment, (2) the amount of prior testing and (3) the distance towards a target. On average, spatial knowledge increases with the amount of learning (Evans, Marrero & Butler, 1981; Garling, Lindberg & Mantyla, 1983; Thorndyke & Hayes-Roth, 1982; but see Ishikawa & Montello, 2006). Repeated testing of navigators’ survey knowledge generally yields an increase in accuracy over learning. If spatial information is integrated during encoding, the acquired knowledge could simply be accessed during survey tasks, independently of