Abstract: With visual spatial scene, previous absolute direction judgment (including cardinal direction judgments) studies based on coordinating egocentric and allocentric reference have identified the effect patterns of target position and heading. Also, a variety of ACT-R models have been developed to understand cognitive process of absolute direction judgments. The popular task scenario is asking participant to use a map to determine the absolute direction between two objects in a visual scene, which is presented as 3D or camera picture. In other field of spatial cognition, relative direction/location tasks (i.e., front/back) are used for the purpose of understanding how spatial layout of scene is represented in the mind. This study aimed to investigate how target position and mismatch between two references influence absolute direction judgments based on spatial memory. Based on Zhou et al’s study (2007), the three components of navigation (route specifying, heading identifying, and direction determining) were used to control information present orders in different experiments. The target scene was adapted from Mou (2002) spatial research.
Three 8 × 8 within-subject experiments were conducted. The variables are target position (0o/up, 45o (R45o), 90o (R90o), 135o (R135o), 180o, 225o (R135o), 270o (R90o), and 315o/left-up (R45o)) and heading (N, NE, E, SE, S, SW, W, NW). Participants were asked to remember one scene with regular layout of seven small objects. In another room, with one of the given headings, participants used “verbal report” to give the absolute direction (i.e., the N, NE) between two objects in the memory scene. The information of heading and the two objects were presented on a computer screen. Exp.1 contained all three navigation components. In Exp. 2, participants made the judgments based on heading information after remembering the route information (independent of direction judgments task). Exp. 3 used same tasks as Exp. 2 but the sequence was reversed. Accuracy and response time were collected by E-prime. A total of 60 college students participated in the study and completed three trials (64 problems per trial). Repeated measures statistical analysis was used.
The main finding could be summarized as follows: (1) the North direction advantage effect was found for Exp. 1 and Exp. 2 and the pattern could be described as N < E /S / W < Non-canonical direction (< means shorter response time, and / means no significant difference), and the Canonical direction advantage effect (N = E / S / W < Non-canonical direction) was found for Exp. 3; (2) the response time was shortest at the 0o/up position (except for Exp. 2 where no significant difference was found between 0o and 180o/bottom position), and the advantage effect at 180o/bottom position was more evident for Exp. 1 and Exp. 3 (i.e., R180o < R45o, R90o and R135o) than for Exp. 2 (i.e., R180o < R135o); (3) the Canonical direction-R90o advantage effect which means the left/right position advantage effect is dependent of the canonical direction, was found for all experimental tasks. The results also showed that the impact of target position was greater than that of heading in all experiments.
These findings indicate that conventional effect patterns not only exist in absolute direction judgment, but also apply to spatial memory topics. Compared with absolute direction task, in spatial memory task the effects of position and canonical direction -left/right position increased but the importance of heading lowered. Many strategies have been identified and corresponding ACT-R cognitive models were developed, but there are much difference between visual scene and memory scene. Thus future work should focus on applying and updating these models in spatial memory

Key words: spatial memory scene, absolute direction judgment, egocentric reference frame, allocentric reference frame