- 東京地学協会 = Tokyo Geographical Society
- 地學雜誌 (ISSN:0022135X)
- vol.103, no.3, pp.221-232, 1994-06-25
Geographical approach to cognitive maps has mainly focused on measurement of their spatial patterns. Recently Lloyd (1989) devised a useful method for analyzing the components of distortions in cognitive maps by means of Euclidean regression. Until now, however, this method has not yet been validated. The purpose of this study was to reexamine his concepts about the distortions in cognitive maps and to test their validity.We pointed out several discrepancies between the concepts of distortions and their operational definitions given by Lloyd (1989). To overcome this problem, the absolute distortion that appears in the configuration before Euclidean regression ought to be distinguished from the systematic distortion that is measured by the parameters of Euclidean regression. While the systematic distortion that indicates a Euclidean property of the distortion can be explained by alignment and rotation heuristics (Tversky, 1981) and implicit scaling model (Holyoak and Mah, 1982), the relative distortion that remains after Euclidean regression cannot be explained by general theories. In addition, these geometrical components of distortion ought to be distinguished from statistical ones, namely, distortion (central tendency) and fuzziness (dispersion) defined by Gale (1982).On the basis of this conceptualization, we carried out an empirical analysis of the distortion in the cognitive map of Kanazawa City. The data used in this study were obtained by a conditional sketch mapping drawn from 113 students of Kanazawa University. Locations to be answered were 21 transportation nodes within the central part of the city known by more than 90 percent of the students. From these locations, two major landmarks of CBD were selected as reference points. Subjects were asked to indicate the remaining 19 locations on the legal-size sheet in which the two reference points were printed.We detected the absolute distortions, overlaying the cognitive maps for all samples on the actual map so as to fit the locations of two reference points into the actual ones. The patterns of the absolute distortion indicated that the amount of errors increased with distance from the reference points, and that the locations in cognitive maps commonly shifted outward from the actual ones. Specifically, these displaced locations in southern or eastern part of the city indicated a counterclockwise shift, which suggested a directional bias in cognitive maps.In order to separate the systematic distortion from the relative one, each of the cognitive configurations was fitted into the actual map by Euclidean regression. Parameter estimates of the scale change averaged 0.593, which suggested that cognitive maps were enlarged about twice the size of the actual map. Mean direction of the rotated angle amounted to-22.2 degrees, which implied that cognitive maps were rotated counterclockwise about 20 degrees from the actual map so as to coordinate the cardinal directions of the cognitive map with the actual one. This tendency can be due to the displacement of two river channels as major reference lines in Kanazawa from cardinal directions.After eliminating the systematic distortions by Euclidean regression, the actual map was overlaid with all the cognitive maps. The overlaid maps indicated that the local patterns of relative distortions reflected hierarchical structure of cognitive maps (Stevens and Coupe, 1978) although the amount of them was smaller than that of the systematic distortions.