Is There a Geometric Module for Spatial Orientation? Squaring Theory and Evidence

Article excerpt

There is evidence, beginning with Cheng (1986), that mobile animals may use the geometry of surrounding areas to reorient following disorientation. Gallistel (1990) proposed that geometry is used to compute the major or minor axes of space and suggested that such information might form an encapsulated cognitive module. Research reviewed here, conducted on a wide variety of species since the initial discovery of the use of geometry and the formulation of the modularity claim, has supported some aspects of the approach, while casting doubt on others. Three possible processing models are presented that vary in the way in which (and the extent to which) they instantiate the modularity claim. The extant data do not permit us to discriminate among them. We propose a modified concept of modularity for which an empirical program of research is more tractable.

To navigate in the world, an animal usually needs to figure out which direction is which, a problem also known as determining the heading (Gallistel, 1990). That is, it needs to establish which way it is facing with respect to some frame that specifies directions. Such a frame may be given by cues external to the animal, such as the pattern formed by the sun and polarized light in the sky or the pattern formed by distant landmarks. But the "frame" may also be given by internal cues-for instance, when nothing external to the animal is available. An animal may compare its current heading with the direction in which it started its journey. In some cases of navigation, the heading needs to be continuously computed. Path integration is such a mechanism (for reviews, see Biegler, 2000, Collett & Collett, 2000, 2002, Etienne, Berlie, Georgakopoulos, & Maurer, 1998, Gallistel, 1990, Newcombe & Huttenlocher, 2000, Wehner, Michel, & Antonsen, 1996, and Wehner & Srinivasan, 2003). In path integration, an animal keeps track of the straight-line distance and direction to its starting point as it travels. In outdoor environments, many animals, especially insects and birds, use a sun compass (e.g., Wehner & Wehner, 1990; Wiltschko & Balda, 1989) and large-scale landmarks (Dyer & Gould, 1983; Gagliardo, Ioalé, & Bingman, 1999; von Frisch & Lindauer, 1954) to establish heading. Vertebrates tested in indoor environments, however, do not have cues from the sky. They typically use the surrounding landmark cues to tell which direction is which. One kind of cue that has received marked attention over the past 2 decades is the overall geometric shape of the environment, called geometric information or geometric cues. Evidence that we will review below shows that geometric information is frequently used in relocating desired targets in a range of vertebrate species.

The idea that vertebrate animals use the geometry of the surrounding environment to locate places started with the work of Cheng (1986). Since then, similar paradigms of research have been conducted on a range of species, including human children and adults, monkeys, birds, and fish. Gallistel (1990) formulated a theoretical mechanism by which environmental geometry is extracted by computing the major and minor axes of a space and proposed that a geometric module is responsible for these computations. The idea of a geometric module forms a cornerstone of a recent proposal that human spatial representation is typically limited in nature-momentary, egocentric, and limited in informational content (Wang & Spelke, 2002, 2003)-as well as of proposals that initial modularity in human infants is overcome in adults through the acquisition of spatial language (Hermer & Spelke, 1996; Hermer-Vazquez, Moffet, & Munkholm, 2001). In terms of underlying neural mechanisms, Epstein and Kanwisher (1998) have proposed that a part of the human brain, in the parahippocampal area, is dedicated to parsing and encoding the geometry of the environment, although others have disagreed (e.g., Maguire, Burgess, et al. …