Montello (1992) offers a number of interesting comments from the perspective of environmental cognition. In this reply, I would like to clarify one of the assumptions of the original Spatial Representation System (SRS) account (Bryant 1992) and to question Montello's (1992) claim that only egocentric frames of reference are used in working memory.
1.1 Montello (1992) questions Bryant's (1992, section 3.7) claim that people have equal access to inferred and explicitly described or perceived spatial relations. This assumption was originally intended to mean that when the location of an object is encoded in a particular coordinate space, both explicit and implicit relations have equal status. Spatial relations between objects are represented by the relations between their locations in the coordinate space, so that there is no difference between relations that were experienced versus those that were not. All "inferences" are made as a consequence of placing objects in the same frame of reference.
1.2 This is not to imply that people automatically make inferences between different frames of reference or separate spatial models; the studies cited by Montello (1992) demonstrate this. The study by Hanley and Levine (1983) is especially illustrative. In their experiments, subjects learned two different paths by touch, one with their left hand and the other with their right. After a study period, subjects were informed that there was a common point on each path and that they were to integrate the two paths into a single configuration. Then subjects were tested on their ability to move from one point to another in the combined paths. Hanley and Levine found that the speed and accuracy of subjects' responses were equivalent for originally learned and inferred movements within a path they had learned. On the other hand, subjects were slower and less accurate when attempting to infer the relation between points from different paths. In other words, for each path, subjects built a spatial representation in which inferred spatial relations were just accessible as experienced relations. When they attempted to combine two distinct spatial representations, however, inferred relations did not have equal status with experienced relations within the paths.
1.3 The results of Hanley and Levine (1983) imply that integrating existing spatial models requires more complicated computations than originally locating objects in a spatial model. What determines the relative accuracy of inferred relations in this case is the complexity of the computations required to translate from one spatial model or frame of reference to another. Along this line, Rieser (1989) reported differences in people's ability to locate objects after rotating or moving relative to the objects. In his experiments, subjects sat in the center of a room surrounded by objects. After observing the objects, subjects were blindfolded and asked to point to objects to provide baseline accuracy and response time measures. In the rotation condition, subjects either physically rotated to face a new direction or imagined such a rotation. In the translation condition, subjects either physically moved to the location of one of the objects or asked to imagine such a translation. Rieser found that subjects were less accurate and took longer to point to objects when they had rotated or imagined a rotation but were as fast and accurate in pointing after a translation (providing the subject did not also rotate). Rieser suggested that subjects encoded the object to object relations and were able to retrieve them directly in the translation condition but that additional processing was required to compute new object-to-self relations after a rotation. Consistent with this idea was the finding that the magnitude of error depended on the amount of rotation.
1.4 Thus, there is evidence (see also Bryant, Tversky, & Franklin 1992; Byrne & Johnson-Laird 1989) that people have equal access to experienced and inferred spatial relations within a single frame of reference. As a person locates objects with the coordinate space formed by a reference frame, the SRS account predicts that there should be no difference between spatial relations that are perceived or heard in a description and those which are not. Montello (1992), however, is certainly right that inferred spatial relations are not equivalent to experienced relations in the larger sense when we discuss inferences between separate spatial models or different frames of reference.
2.1 Montello (1992) equates allocentric (survey) representations with long-term memory (LTM) and egocentric (route) representations with working memory (WM). In this scheme, one stores a survey map of the environment in LTM, but to use that spatial knowledge to guide behavior in space one must convert the survey map to an egocentric route representation in WM. Thus, Montello proposes not only that people use two different frames of reference for representing space, but that they are limited in which frame can be used in long- and short-term memory.
2.2 In the SRS account, the distinction between LTM and WM was not explicitly made. The account assumes that the SRS creates mental models based on allocentric and egocentric reference frames. Mental models interact with WM during retrieval (Johnson-Laird 1989), so I must assume that if people use allocentric models in LTM they can instantiate these models in WM as well. This does not mean that people cannot use an allocentric cognitive map in LTM to generate an egocentric route map or to imagine a survey map. These are undoubtedly common strategies for dealing with the task of navigating through the environment. However, it seems an unnecessary restriction to say that WM can only accommodate egocentric representations.
2.3 Recently, Hirtle and Mascolo (1991) examined the strategies and heuristics people use to estimate distances. Some of the strategies they discovered involve egocentric or route representations such as estimating the distance between two places by calculating the time it would take to travel from one to the other. Other strategies, however, involved computations performed directly on an allocentric cognitive map. These included the use of analogies (A is to B as C is to D) and triangulation, in which two sides of a triangle are used to calculate the hypotenuse. Hirtle and Mascolo (1991) found evidence of up to 21 distinct heuristics for calculating distance, many of which were performed on allocentric representations. It is likely that many other spatial tasks can likewise be performed by a number of different strategies using egocentric or allocentric representations.
2.4 Another example in which people seem to use allocentric representations in WM is the phenomenon of spatial priming. In a classic study, McNamara (1986) found that when subjects learn the layout of objects in a room, their recognition latencies for the objects are influenced by the objects' spatial proximity in the layout. Thus, an item facilitated recognition of another more if the two had been close together in the environment rather than far apart. This priming effect was based on an allocentric representation of the layout rather than an egocentric one because priming was determined by location in the room independent of the route or egocentric perspective subjects used during study. Spatial priming seems to depend on having the allocentric representation in WM because the priming effects are not obligatory and are found only when subjects are cued to the spatial relations at the time of the recognition test (Clayton & Chattin 1989; McNamara, Altarriba, Bendele, Johnson, & Clayton 1989). Although subjects must retrieve the spatial representation to generate priming, the recognition test does not require mental imagery and there is no evidence that subjects scanned a mental image.
Bryant, D. J. (1992) A Spatial Representation System in Humans. PSYCOLOQUY 3(16) space.1.
Bryant, D. J., Tversky, B., & Franklin, N. (1992) Internal and External Spatial Frameworks for Representing Described Scenes. Journal of Memory and Language 31: 74-98.
Byrne, R. M. J., & Johnson-Laird, P. N. (1989) Spatial reasoning. Journal of Memory and Language 28: 564-575.
Clayton, K., & Chattin, D. (1989) Spatial and Semantic Priming Effects in Tests of Spatial Knowledge. Journal of Experimental Psychology: Learning, Memory, and Cognition 15: 495-506.
Hanley, G. L., & Levine, M. (1983) Spatial Problem Solving: The Integration of Independently Learned Cognitive Maps. Memory & Cognition 11: 415-422.
Hirtle, S. C., & Mascolo, M. F. (1991) The Heuristics of Spatial Cognition. In the Proceedings of the 13th Annual Conference of the Cognitive Science Society (pp. 629-634). Hillsdale, NJ: Erlbaum.
Johnson-Laird, P. N. (1989) Mental Models. In M. I. Posner (Ed.), Foundations of Cognitive Science. Cambridge, MA: MIT Press.
McNamara, T. P. (1986) Mental Representations of Spatial Relations. Cognitive Psychology 18:87-121.
McNamara, T. P., Altarriba, J., Bendele, M., Johnson, S. C., & Clayton, K. N. (1989). Constraints on Priming in Spatial Memory: Naturally Learned Versus Experimentally Learned Environments. Memory & Cognition 17: 444-453.
Montello, D. R. (1992) Characteristics of Environmental Spatial Cognition. PSYCOLOQUY 3(52) space.10.
Rieser, J. J. (1989) Access to Knowledge of Spatial Structure at Novel Points of Observation. Journal of Experimental Psychology: Learning, Memory and Cognition 15: 1157-1165.