Spatial representation in body coordinates: Evidence from errors in remembering positions of visual and auditory targets after active eye, head, and body movements

Canadian Journal of Experimental Psychology, Mar 2003 by Kopinska, A, Harris, L R

Abstract Eight participants were presented with auditory or visual targets and then indicated the target's remembered positions relative to their head eight seconds after actively moving their eyes, head or body to pull apart head, retinal, body, and external space reference frames. Remembered target position was indicated by repositioning sounds or lights. Localization errors were found related to head-on-body position but not of eye-in-head or body-in-space for both auditory (0.023 dB/deg in the direction of head displacement) and visual targets (0.068 deg/deg in the direction opposite to head displacement). The results indicate that both auditory and visual localization use head-on-body information, suggesting a common coding into body coordinates - the only conversion that requires this information.

Determining the location of real-world objects requires the integration of sensory information concerning not only the location of the stimulus relative to the sense organs but also of the sense organs relative to some reference system common to all the senses. The retina, head, body, or external space could theoretically provide such a reference frame. A series of conversions is then needed to move information from its initial coding in the frame of a particular sense organ into such a frame. Since eye, head, body, and space reference frames move relative to each other every time the eye, head, or body move, knowledge of the position of each frame relative to the others is needed when converting information from one to another (Harris, 1997; Poppel, 1973). Any errors in coding the relative position of the frames could therefore lead to corresponding errors in the stored location.

Our experiments looked for reference frame conversion errors that would betray which conversions had been done and thus reveal which frame was used to code the information. After viewing a target, participants were asked to reproduce its position relative to the head after changing eye, head, or body position. If the location of a sound were stored in head coordinates, for example, then no conversions would be necessary to solve the reproduction task and no errors would be expected related to movements of the eyes in the head or head on the body. If location were stored in visual coordinates then eye-in-head information would be required both to store and retrieve the location, and systematic errors might be found related to the position of the eye in the head. Similarly, if sound location were stored in body coordinates then head-on-body information would be required for storing and retrieving, and errors might be found related to head position. Visual location is originally in eye coordinates and therefore eye-in-head information is always required for a head-based task. But if visual information were stored in body coordinates then head-on-body information would also be required, and errors might be found related to head position.

If no systematic errors are found related to the relative positioning of these frames it does not necessarily indicate that a particular conversion is not done. It could be that it was done perfectly and left no errors. However, if systematic errors are found they are strong indications of particular conversions taking place.

Possible neural substrates exist that could support any of these frames. Reports of auditory receptive fields moving in the superior colliculus tend to remain aligned with their visual counterparts during eye movements (Jay & Sparks, 1984), support the visual frame as a feasible candidate. Modulation of visual fields in the parietal cortex in response to eye position could represent a neural basis for a head-centred coding system (Bremmer, Distler, & Hoffman, 1997). Head position related cells in the parietal (Brotchie, Andersen, Snyder, & Goodman, 1995) could support body-centred localization. Using spatial coordinates is initially appealing in its directness (Mergner, Nasios, Maurer, & Becker, 2001) but since such coding is not related to any part of the organism, the representation would be unaffected by any movements of any extent or direction. Hippocampal cells show some of the required properties (Georgesfrancois, Rolls, & Robertson, 1999).

Effects of eye position on the perception of sound direction have been studied extensively (Bohlander, 1984; Goldstein & Rosenthal-Viet, 1926; Lackner, 1973a; Lewald, 1997, 1998; Lewald & Ehrenstein, 1996, 1998a; Pierce, 1901; Ryan & Schehr, 1941; Weerts & Thurlow, 1971) with inconclusive results. We have investigated eye position effects using a novel approach: using intracranial auditory targets and measuring their remembered positions always with respect to the head. With our design, participants could use a clearly defined head frame of reference for localization: The auditory targets were heard "inside the head" (lateralized) and the task was to move another sound heard "inside the head" to the remembered location. With this design, an effect of eye position on the remembered position of a sound would suggest auditory information being transferred into retinal frame of reference.