ICSC 2006 Dynamics in Spatial Interactions P O S T E R The effect of a reference object’s orientation on the apprehension of spatial terms Takatsugu Kojima  Takashi Kusumi Keywords Spatial cognition x Spatial language x Orientation of a reference object Background The interaction between spatial cognition and language is important to both cognitive science and artificial intelligence. For example, if an interactive humanlike agent could understand spatial terms in natural language in a human manner, simulating human behavior and cognitive processes should be possible. Research in artificial intelligence and linguistics has provided theoretical and computational models of the apprehension of spatial language (Gorniak and Roy 2004; Herskovits 1986; Kreizer 1997; Yamanashi 2000). Cognitive psychology has empirically examined such theoretical and computational works and provided empirical data on the relationship between spatial cognition and language. Imai et al. (1999) examined the assignment of four Japanese spatial terms, mae, ushiro, hidari, and migi (similar to front, behind, left, and right, respectively, in English), in three-dimensional (3-D) computer graphics (CG) space. They found that the four spatial terms did not categorize the given space equally, and that the categorical patterns of space by these spatial terms were influenced by the orientation of the reference; object. When a reference object had no inherent front, mae/ushiro and hidari/migi had symmetrical boundaries in a pair, and the spatial areas of mae/ushiro were larger than those of hidari/migi. When a reference object had an inherent front, the areas of the four terms varied according to the orientation of the reference object. When a reference object was facing toward or away from participants, hidari/migi boundaries were symmetrical T. Kojima ()  T. Kusumi Department of Cognitive Psychology in Education, Graduate school of Education, Kyoto University, Japan e-mail: kojima@cpsy.mbox.media.kyoto-u.ac.jp but mae/ushiro boundaries were asymmetrical. When a reference object was facing sideways, mae/ushiro and hidari/migi had asymmetrical boundaries as pairs. Imai et al. (1999) concluded that the orientation of a reference object influenced the apprehension of the spatial terms dynamically in 3-D space. However, they compared only the boundaries of the four terms and did not examine how the effect influenced the spatial categorical pattern of each spatial term. Their method had some limitations and was not capable of fully examining the effect of a reference object’s orientation. Furthermore, no detailed data were provided for applications to humanlike agents. Thus, we examined the effect of each of the same four terms in 3-D space, with the goal of providing detailed experimental data and general implications for spatial cognition and language understanding. Method Sixty native Japanese graduate and undergraduate students at Kyoto University participated in this study. We divided the participants into three groups of 20 people each, with each group assigned one of the following orientation conditions: rotations of 0°, 90°, and 180°. The experiment was run on a computer with a 17-in. monitor. In the experiment, two windows appeared on the screen. The display window appeared at the top of the screen and was used to present stimuli, whereas the evaluation window appeared below the display window and was used to present statements and receive responses. Four Japanese projective spatial terms were used: mae, ushiro, hidari, and migi (in front of, behind, to the left of, and to the right of in English, respectively). These are basic Japanese spatial terms for 3-D space. The sentences used in this experiment were ‘‘Akai tama-wa midori-no hako-no xx·-ni aru’’ (‘‘The red ball is xx the green box’’). One of the four spatial terms was assigned to xx.