Reference systems for coding spatial information in ... - CiteSeerX

Exp Brain Res (1993) 93:324-331

Experimental Brain Research © Springer-Verlag 1993

Reference systems for coding spatial information in normal subjects and a deafferented patient J. Blouin1, C. Bard1, N. Teasdale1, J. Paillard2, M. Fleury1, R. Forget1,4, Y. Lamarre3 1

Université Laval, Laboratoire de Performance Motrice Humaine, PEPS Ste-Foy, Québec GIK 7P4, Canada CNRS, Laboratoire de Neurosciences Fonctionnelles, Marseille, France 3 Université de Montréal, Centre de Recherche en Sciences Neurologiques, Montréal, Canada 4 Hôpital Hôtel Dieu de Montréal, Montréal, Canada 2

Received: 1 October 1991/Accepted: 26 March 1992

Abstract To produce accurate goal-directed arm movements, subjects must determine the precise location of target object. Position of extracorporeal objects can be determined using: (a) an egocentric frame of reference, in which the target is localized in relation to the position of the body; and/or (b) an allocentric system, in which target position is determined in relation to stable visual landmarks surrounding the target (Bridgeman 1989; Paillard 1991). The present experiment was based oh the premise that (a) the presence of a structured visual environment enables the use of an allocentric frame of reference, and (b) the sole presence of a visual target within a homogeneous background forces the registration of the target location by an egocentric system. Normal subjects and a deafferented patient (i.e., with an impaired egocentric system) pointed to visual targets presented in both visual environments to evaluate the efficiency of the two reference systems. For normals, the visual environment conditions did not affect pointing accuracy. However, kinematic parameters were affected by the presence or absence of a structured visual surrounding. For the deafferented patient, the presence of a structured visual environment permitted a decrease in spatial errors when compared with the unstructured surrounding condition (for movements with or without visual feedback of the trajectory). Overall, results support the existence of an egocentric and an allocentric reference system capable of organizing extracorporeal space during arm movements directed toward visual targets. Key words: Reference systems - Reaching movements Deafferented human - Spatial processing - Human

Introduction The encoding of spatial information is a critical step in the programming and guiding of spatially oriented movements.

Correspondence to: J. Blouin

Encoding a target location implies the existence of a frame of reference in which space coordinates can be defined. Two main reference systems are usually considered, egocentric and allocentric (Paillard 1987, 1991; Paillard et al. 1981; Ventre et al. 1984). The egocentric frame stems from the continuously updated static and dynamic proprioceptive signals that tune the motor apparatus with the registration of gaze direction for directing actions in the extracorporeal space (Jeannerod 1991a; Paillard 1987). The egocentric system corresponds to a sensorimotor mode of processing spatial information and benefits from a direct dialogue with the physical world (Paillard 1987, 1991). For Bridgeman (Bridgeman 1989; Bridgeman et al. 1979), this body-centered mapping represents the motor-oriented mode of organizing extrapersonal space. Depending on experimental conditions, authors have localized the origin or center of the egocentric system at different areas within the body architecture [e.g., head-centered (Bard et al. 1990; Biguer et al. 1982; Grossberg and Kuperstein 1989; Jeannerod 1988; Roll et al. 1986), shoulder-centered (Caminiti et al. 1990; Flanders et al. 1991; Soechting et al. 1990), and trunk-centered (Yardley 1990)]. The allocentric system derives from a memory-based internal construct built from extracting the stable covariant features of the visual environment that remain stable when the body moves. Here, both the body and the object are located through the retinal coordinate system in a visual world perceptually stabilized (Paillard 1987). It corresponds to the representative (Paillard 1987) or the cognitive (Bridgeman et al. 1979; Bridgeman 1989) mode of processing spatial information. Egocentric and allocentric reference systems are believed to have separate neural substrates. The posterior parietal region has been identified for providing an allocentric representation of the extracorporeal world that remains stable when the body changes position (Kesner et al. 1989; Paillard 1987). Cells of the hippocampus might have an important contribution by storing spatial information of the physical environment necessary to construct an

325

allocentric system (Rolls 1991). On the other hand, the fronto-striatal system presumably plays a major role in the elaboration of the egocentric reference system (Paillard 1982). Therefore, both the egocentric and allocentric reference systems are candidates for coding the position of objects in space. Pioneer works on these systems come from developmental research and concern their emergence during infancy. The first oriented movements probably unfold within an egocentric system whereas the cognitive allocentric system is believed to emerge later during ontogeny (Acredolo 1977, 1978; Rebok 1987). More recently, several researchers have focused on the dialogue between the two reference systems in the coding of target position during pointing movements and tasks involving judgement of objects' positions. When the position of an object is altered either by moving its surrounding environment or by changing its absolute position, subjects are more accurate in performing motor acts (hand of eye movements) than in producing perceptual judgements of actual target position (Andersen 1987; Bridgeman 1989; Bridgeman et al. 1979; Honda 1985, 1990). Stark and Bridgeman (1983) have also produced a discrepancy between the egocentric and allocentric maps by pressing the eye while subjects maintained gaze fixation at an illuminated target presented within a structured environment Although subjects correctly perceived the target location (because of the stable visual landmarks present within the environment), they made systematic pointing errors due to the perturbed internal motor mapping. These results are in agreement with the existence of two processing modes for mapping extracorporeal space and suggest that their contribution might be a function of the task to be performed. The present experiment was based on the premise that (a) the presence of a structured visual environment enables the use of an allocentric frame of reference, and (b) the sole presence of a visual target within a homogeneous background forces the registration of the target location by an egocentric system. We were interested in assessing whether, in an oculomanual task, normal subjects can use either their sensorimotor or cognitive map or a combination of both to locate the position of the target without affecting their movement accuracy. We also tested the prediction that a deafferented subject, having an impaired egocentric frame of reference, should be unable to reach accurately a visual target presented with an unstructured visual environment but should be capable of doing so when a structured environment allows the use of an allocentric frame of reference.

published elsewhere (Cooke et al. 1985; Forget and Lamarre 1987). Briefly, the patient permanently lost her large myelinated fibres of the four limbs subsequent to two episodes of sensory polyneuropathy. Continuous investigations of the patient during the last 14 years have shown a definitive loss of all somato-sensory modalities with preservation of pain and temperature sensitivity and have shown normal efferent motor pathways (Bard et al. 1991; Cooke et al. 1985; Forget and Lamarre 1987, 1990; Teasdale et al. 1991). All subjects went through each experimental procedure.

Apparatus Figure 1 is a schematic representation of the experimental setting. Subjects sat comfortably on an adjustable seat, with their chest leaning against a vertical restraint A 1.5-m hand-held pointer extended from the floor between the subjects' legs. The pointer was mounted onto a universal joint attached to the floor and could be moved by extending the arm. Horizontal tracks were fixed on both sides of the pointer to remove the directional component of the movement and to restrain it to the sagittal plane. A small lightemitting diode (LED) was fixed on the tip of the pointer and could provide visual feedback of the movements when they were performed in a dark environment. Two LED targets were fixed at eye level on a panel above the area accessible by the pointer. The targets were located in the subjects’ sagittal plane at, respectively, 28 and 36 cm from a constant starting position. About 3 cm separated the tip of the pointer and the targets when they were aligned. The use of two targets was intended to prevent the subjects from using a similar amplitude programming strategy across trials, and only movements directed toward the 28-cm target were analyzed. In the initial position, the subjects held the pointer with their preferred hand at chin level, and the tip of the pointer was at nose level (without touching the face). For movements performed in an unstructured visual environment, the room lights were turned off, leaving the subjects and the targets in total darkness. To structure the visual environment, the lights of the room remained on and the subjects could see the apparatus environment (mainly composed of a series of targets not used in the present experiment, the apparatus setup, and the wall facing the subjects). As shown in Fig. 1, for movements executed in a structured visual environment without visual feedback of the limb, panels were fixed below subjects’ eye level and under the targets to obstruct vision of both arm and pointer. A light piece of cloth was suspended to the panel fixed under the targets to preclude vision of the pointer after movement initiation. The position of the tip of the pointer was obtained through two linear potentiometers fixed on a steel frame facing the subject; small-gauge wires were attached to these potentiometers 5 cm below the tip of the pointer. The signals from the potentiometers were digitized at 1000 Hz, and the cartesian position of the pointer was obtained through trigonometric transformations. Velocity-time and acceleration-time curves were obtained through optimized regularized Fourier series (Hatze 1981). The first moment that the velocity increased above 10 mm/s was defined as movement onset and the first velocity drop under 50 mm/s was defined as the end of the movement, and movement accuracy was measured at that time.

Materials and methods Procedure Subjects Ten normal subjects (eight right- and two left-handed) and one right-handed deafferented subject (the patient) participated in the experiment. The subjects used their preferred hand to perform the task. Normal subjects’ age ranged from 23 to 35 years and the patient was 42 years old. The patient was familiar with the laboratory environment and had participated in previous experiments with this apparatus. A detailed clinical description of this patient has been

Within 2 s following a verbal preparatory signal given by the experimenter, the LED fixed at the tip of the pointer came on for 500 ms. For normal subjects, this initial visual information served to calibrate the sensory information from body proprioceptors in the visual map (Prablanc et al. 1979); for the deafferented subject, it served to calibrate the starting position (Ghez et al. 1990). Then, one LED target turned on and indicated to the subjects to move the pointer rapidly toward the target and to stop precisely under i t .

326

Fig. 1. Schematic representation of the experimental setting for the unstructured environment/open-loop condition. See the text for details. Additional targets were used to structure the environment and are not represented here. GWs, gauge wires; Ps, panels; C, light piece of cloth; POTs, potentiometers; VR, vertical restraint; HTs, horizontal tracks; J, universal joint; LED, light-emitting diode on tip of the pointer.

Subjects were not required to initiate their movements as quickly as possible. All visual information was withdrawn when the pointer came to rest, to eliminate visual feedback of movement endpoint and target position, which is known to provide valuable information for programming subsequent movements in deafferented patients (Ghez et al. 1990) and normal subjects (Beaubaton and Hay 1986). For security and technical reasons, special procedures were used with the deafferented subject before and during the experiment. A large belt secured the patient to the seat and prevented her from falling. This was necessary because, in total darkness conditions, the subject lost all indication about the relative position of the different parts of her body. Also, the patient's right hand was attached to the pointer to maintain a stable hand position on the lever without visual feedback. An assistant stood beside the patient during the testing session, to secure the subject and to help in returning the pointer to the same starting position. Subjects performed ten practice trials in the structured environment/closed-loop condition before starting the experiment. For both visual environments, 10 trials were performed by the patient and 15 by normals to each target with and without vision of the trajectory (closed- and open-loop conditions, respectively). The closed-loop/structured environment condition was always presented first. The remaining vision and environment conditions were randomly presented across subjects.

Fig. 2A, B. Constant errors and standard deviations for normals (A) and the deafferented subject (B). The presence of a structured surrounding did not affect movement accuracy of normals but largely enhanced spatial processing for the deafferented subject. Symbols without error bars have a standard deviation smaller than the symbol size

Results Data obtained from normal subjects were all submitted to a 2 x 2 (environment conditions x visual condition analysis of variance (ANOVA), with repeated measured on both factors. For the deafferented patient, t-tests of the differences between two means (Sokal and Rohlf 1982) were used to assess the difference between the various experimental conditions. Spatial accuracy Constant errors for the normal subjects and the deafferented subject are presented in Fig. 2. Constant errors were defined as the distance between the target and the pointer endpoint (i.e., amplitude errors). Negative errors indicate that movements were shorter than the target distance. For normals, the ANOVA showed a mean

327

effect of vision [F (1,9) = 10.87; P0.05]. The interaction environment x vision was also nonsignificant [F (1,9) = 0.61; P>0.05]. Normals were more accurate when movements were controlled under visual/proprioceptive feedback than under proprioceptive information only (0.6 and 3.1 cm for the closed-loop and open-loop conditions, respectively). The deafferented patient systematically undershot the targets during movements performed without vision of the limb and generally overshot those performed with visual feedback of the arm. The structured environment increased movement accuracy for both closed-loop conditions [0.86 vs 2.27 cm for the structured and unstructured visual environment, respectively; t(18) = 2.21, P