Abstract: Observers can assume two types of frame of reference when process spatial information about their environment. These include egocentric reference system which specifies a location with respect to the observer, and environmental frame of reference which specifies a location relative to other locations independent of the observer. A number of studies suggested that both types of reference systems can be used in the representation of the spatial layout (Wang & Spelke, 2002; Burgess, 2006; Waller & Hodgson, 2006). One special kind of environmental frame of reference has been proposed by McNamara and his colleagues. Since Mou and McNamara’s well-known initial study (Mou & McNamara, 2002), the role of intrinsic axis of spatial layout in spatial memory has been extensively examined. The role of main orientation axis of the environment has also been illustrated. However, it is not clear the relative contribution of intrinsic axis in the object layout and main axis in environment when these two sources of information are not congruent. Moreover, in most of previous studies, all of the objects were positioned in very regular layouts, and had salient intrinsic axes in any testing heading. This might limit how generalizable these results can be applied to spatial learning of every day environment.
In the present research, we manipulated the congruency of the intrinsic reference frames of the object layout and environmental reference frames. The hypothesis was that both environmental frames and intrinsic axes were effective cues in spatial memory, however the use of these frame of references could be interact with the use of egocentric frame of reference. In addition, in our experiment, some of the objects were positioned in an orderly fashion and created the structure of the main intrinsic axes, which were salient in some headings and not salient in other headings, while some other objects are positioned outside those intrinsic axes. The spatial reference systems used in memory to represent the locations of the objects were examined by manipulating the orders of learning directions.
The research was carried out in a rectangular room (5.5m×7m). Eight common objects were placed on a 3m×3m rectangular floor mat. The mat was positioned in the middle of the room so that the outer boundaries of the mat were parallel with the adjacent walls of the rectangular room. Six objects were positioned in a regular layout (two rows and three columns) and which formed the main intrinsic axes of the layout; the other two objects were positioned irregularly in the layout. Two experiments examined the roles of egocentric experience, environmental frames and intrinsic axis to represent the locations of objects in the environment. In Experiment 1, the intrinsic axes of the layout were aligned with the environmental frames, and in Experiment 2, the intrinsic axes were misaligned with the environmental frames. Forty eight undergraduates (aged between 18 and 23) participated in the experiments. Participants learned the locations of eight objects and then made judgments of relative direction based on the memory of the layout (e.g., imagine you are standing at the shoe, facing the clock; point to the bottle). Reaction time and pointing error were recorded and analyzed in mixed-model analyses of variance (ANOVAs) in terms for learning order (0° then 135° vs 135° then 0°), and imagined heading (0° to 315° in 45° increments), position regularity (regularly positioned objects vs irregularly positioned objects). Learning order was between participant variable; position regularity and imagined heading were within participant variables.
The specific findings are presented as follows: first, non-egocentric information had a powerful influence on participants’ representations of the layout. When the main intrinsic axes were congruent with the environmental frames (in Experiment 1), spatial layout were best remembered when participants learned the layout from the same axis (0°). When participants were tested from viewpoints (135°) which were misaligned with the main intrinsic axes of the layout and environmental frames, the spatial layout was represented poorly even when the viewpoint was the first study view, indicating frame of reference from the environment/object array was powerful enough to overshadow the egocentric information. Second, either of intrinsic layout axis and environmental frame are effective cues for spatial representation when paired with egocentric information. When the intrinsic and environmental axis were aligned (Experiment 1) , judgments of relative direction were quicker and more accurate for imagined headings parallel to the main intrinsic axis and environmental frames regardless which view was learned first. However, when the main intrinsic axis were incongruent with environmental frames (Experiment 2), if the first learning view was aligned with the main intrinsic axis, participants performed better in imagined headings of 45°, 135°, 225°, 315° which were aligned with the most salient intrinsic axis, and if the first learning view was aligned with environmental frames, they performed better in imagined headings of 0°, 90°, 180°, 270° which were aligned with environmental frames. This illustrates the use of egocentric information in determining whether the frame of reference from the layout axis or environmental axis would be useful. Third, the intrinsic axes not only facilitated the encoding of objects’ positions during learning phase, but also had an important role in the judgments of object directions during testing if the objects learned were positioned on the intrinsic axis. In both experiment 1 and experiment 2, the judgments of relative directions of objects were more accurate for those target objects positioned on the intrinsic axis than those “irregularly” positioned objects. The judgments for those “regularly” positioned objects were slower, suggesting participants might perform higher-order calculation of the direction of those objects on the axis, rather than simply retrieved directional information from memory.
In conclusion, results in the present study indicated that spatial memories can be sensitive to intrinsic layout axes, environmental frame and also egocentric experiences. If the main intrinsic axes of the layout were congruent with the environmental frames, they would have a powerful influence on spatial representation. However, if the main intrinsic axes of the layout were incongruent with the environmental frames, participants select a reference system according to the learning order (egocentric information in the first learning episode). These results show that both environmental frame and intrinsic axes could be effective reference systems in spatial memory. Egocentric information could encourage the use of either one.

Key words: spatial representation, spatial reference systems, two-system models