Influence of visual constraints in the trajectory formation of grasping movements

Transcription

1 Neuroscience Letters 401 (2006) Influence of visual constraints in the trajectory formation of grasping movements R. Palluel-Germain a,, F. Boy a, J.P. Orliaguet a, Y. Coello b a Laboratoire de Psychologie et NeuroCognition (LPNC), CNRS UMR 5105, Université Pierre Mendès France, Grenoble II, 1251, Avenue Centrale BP 47, Grenoble Cedex 9, France b Laboratoire URECA, Université Charles de Gaulle, Lille, France Received 18 November 2005; received in revised form 8 February 2006; accepted 24 February 2006 Abstract The main objective of the present study is to show that the visual context can influence the trajectory formation of grasping movements. We asked participants to reach and grasp a cylinder disposed at three different positions: 20, 0 and 20 of eccentricity with respect to the midsagittal axis. Grasping movements were performed in a direct and in an indirect visual feedback condition (i.e., controlled through a vertical video display). Results revealed that for grasping movements directed toward objects located at 20 and 0, path curvatures of the wrist, the thumb and the index finger were significantly straighter in the indirect visual feedback condition. However, no significant difference concerning hand path curvature was observed when the movement was directed toward the object located at 20. This suggests that grasping movements controlled through a remote visual feedback tend to be planned in extrinsic space and that the effect of the visual context on movement planning appears to be not isotropic over the workspace Elsevier Ireland Ltd. All rights reserved. Keywords: Natural grasp; Motor planning; Visual feedback; Video-controlled movements; Context effect The paths taken by the hand in goal-directed movements are generally viewed as containing information about how movements are planned and controlled (for a review see [4]). For instance, considering pointing movements, it has been shown that hand paths were roughly straight whatever the direction of the movement [16]. This extrinsic stability suggests that manual goal-directed movements are coded as a displacement along a straight line between the initial position of the hand and the goal of the movement (extrinsic space planning) and that the appropriate covariation of the joint angles is specified at some later stage [7,23,26,27]. Alternatively, some other studies observed that the shape of the hand movement varied as a function of the movement direction, thus implying that the movement is not programmed to follow a straight line in the extrinsic space [1,5,14,17]. This last result is compatible with the idea that the motor system plans movement in joint coordinates (intrinsic space planning). Movements in extrinsic space are therefore curved simply as a result of the non-linear relation between cartesian coordinates and joint angle coordinates [1]. Corresponding author. Tel.: ; fax: address: richard.palluel@upmf-grenoble.fr (R. Palluel-Germain). These controversial data were reconciled by studies showing that the features of a trajectory could depend on the execution constraints imposed to the movement. Desmurget et al. [3], for instance, observed that planar (2D) pointings evidenced straight hand paths whereas trajectory curvature of three-dimensional (3D) pointings depended on movement direction. This suggests that planar movements are planned in extrinsic space and that differently, 3D movements are planned in intrinsic space. Interestingly, Palluel-Germain et al. [18] showed recently that not only physical constraints but also the visual context can affect trajectory formation. Indeed, three-dimensional pointing movements executed in a situation of remote visual control (i.e., with the use of a video-system) exhibited roughly straight paths whatever the movement direction. The authors then concluded that in this particular situation subjects estimated the desired trajectory by using hand and target position displayed on the screen, with the consequence that reaching movements are planned and controlled in extrinsic space. Therefore, it seems that in a remote control situation, the constraints arising from the visual feedback surpass the biomechanical constraints thereby leading to modifications in the shape of the hand displacement. The present study was designed to test this hypothesis by analysing grasping movements i.e., /$ see front matter 2006 Elsevier Ireland Ltd. All rights reserved. doi: /j.neulet

2 98 R. Palluel-Germain et al. / Neuroscience Letters 401 (2006) movements in which the hand path is strongly compelled by both task and biomechanical constraints. Indeed, several studies [5,6,10,19,22] showed that the trajectories of grasping movements are influenced, at least in part, by the final posture reached by the arm with respect to the properties of the object (localisation, orientation, shape). For example, when subjects were asked to reach and grasp cylindrical objects placed at different locations, Paulignan et al. [19] reported that the configuration reached by the arm was stereotyped for a given position of the object. More specifically the hand trajectory exhibited curved path that changed significantly as a function of the object position. Such results suggest that the hand path, in grasping movements, is widely determined by the constraints of the task [5,10,19,22,24]. Therefore, if goal-directed movements carried out in an indirect visual feedback situation tend to be planned and controlled in extrinsic space, we expect that even grasping movements should exhibit roughly straight hand paths whatever the object s position. Moreover, as the visual feedback is available during the entire movement, it is thus difficult to disentangle whether the observed modifications of the path shape result from the influence of either planning or feedback control processes. Consequently, to determine the part played by planning processes, the shape of the hand trajectory was assessed right at the start of the movement. In order to test these hypotheses, we asked participants to carry out grasping movements in two conditions of visual feedback: either under an indirect visual feedback (i.e., controlled through a vertical video-screen) or in a direct visual feedback condition. Moreover, movements were executed toward objects placed at different locations. After providing informed consent, eight subjects (18 25 years) participated in this experiment. All subjects were right handed, and naïve as to the purpose of the experiment. None of them reported any sensory or motor deficits, and all had normal or corrected-to-normal vision. The apparatus is shown in Fig. 1. Subjects sat in front of an untextured white table positioned slightly above the navel and maintained their trunk straight. Each trial started by placing the tip of the thumb and index finger in contact on the top of a support (1 cm diameter and 5 cm in height) aligned with the midsagittal plane. Subjects were instructed to hold the remaining fingers clenched into the palm and to keep the lower arm and the hand on the table surface. In this configuration, the thumb and the index finger were approximately at 15 cm from the body. After a go instruction from the experimenter, subjects were asked to reach and grasp a cylinder (diameter 2.5 cm, height 1.5 cm) disposed one at a time at three fixed locations: 20 on the left of subject s sagittal axis ( 20 ), 0 (along the midsagittal axis) and 20 on the right (see Fig. 1). The distance between the starting position and the object locations was 25 cm. Subjects were instructed to pick up the cylinder with the thumb and the index finger. No instruction was given as concern the movement speed, the movement path and the contact positions of the digits on the cylinder. Having grasped the cylinder, subjects lifted it a few centimetres off the table. For every session the subject s left hand remained on his/her left thigh. Each subject performed grasping movements in an indirect visual feedback condition (IVF) and Fig. 1. Experimental conditions. (A) Indirect visual feedback condition. Cylinder, hand position and movement were visible on the video screen. An opaque cloth suspended between the screen and the subject prevented the direct vision of the hand. (B) Direct visual feedback condition. The cylinder was directly visible in the workspace. Subjects began each trial by positioning the tip of the thumb and index finger in contact on the top of a support (1 cm diameter and 5 cm in height) aligned with the midsagittal plane. Subjects were asked to reach and grasp a cylinder (diameter 2.5 cm, height 1.5 cm) disposed at 25 cm from the starting position. Three objects locations were used: 0 (plain circles), 20 and 20 (dotted circles) centred on the subject s body axis. in a direct visual feedback condition (DVF). In the IVF condition participants were seated in front of a colour video screen (0.55 m 0.40 m) located approximately 0.50 m from the head. A Sony video camera placed 1.10 m above the table recorded arm displacement (1:1 spatial relationship) and transmitted in real time movement images on the screen. The starting position of the hand, the cylinder location and the displacement of the hand were continuously available on the video screen. On the video display the hand starting position and the cylinder located at 0 were in alignment with the subject s midsagittal axis. An opaque cloth placed between the screen and the subject prevented the direct vision of the hand. In the DVF condition participants performed the same task in the absence of the occluder

3 R. Palluel-Germain et al. / Neuroscience Letters 401 (2006) (see Fig. 1B). The IVF and the DVF conditions were performed in two independent sessions that were counterbalanced across participants. In each experimental condition the 30 movements (3 object positions 10 trials) were carried out in a random order. The positional data of the movements were recorded using the CMS movement registration system (manufacturer: ZEBRIS, Isny, Germany). In this measuring system the positions of miniature ultrasonic transmitters were sampled at a frequency of 100 Hz at a spatial resolution of about 0.1 mm. Two markers were stuck on the tips of the thumb and index finger, respectively. Another one was placed on the radius process of the wrist. Positional data were filtered using a second-order Butterworth dual-pass filter (cut-off frequency: 10 Hz). Movement onset was defined as the time when the thumb speed first exceeded 3 cm/s. The movement endpoint was determined visually on the curve of the thumb and index finger distance, as the point where this distance stopped decreasing (i.e., the time when the fingers came into contact with the object). Therefore, the period following the contact with the object, during which the cylinder was lifted, was not considered. In order to assess movement curvature we first calculated the Path Curvature Index (PCI) [1]. For each trial, the maximum distance between the trajectory and the straight line joining the starting- and the end-point was divided by the amplitude of the movement. Therefore, this measure increases with curvature (a straight line would have a null PCI value). To calculate average paths for each marker position and each condition, the starting position of each trajectory was first realigned with the mean starting position. The trajectories were then resampled at 50 evenly spaced points over the duration of the movement. Finally, averaging these resampled trajectories allowed to compute mean trajectories. We also calculated the maximum distance (MD) between the trajectory and the straight line joining the startingand the end-point (negative value for deviations to the left of the straight line). In addition, in order to determine whether a difference in curvature was present at the beginning of the movement, we measured the distance between the path and the straight line 150 ms after movement onset. Finally, movement time was analysed. For the movement time (MT) we performed a two-way analysis of variance (ANOVA) with repeated measures on the visual feedback (IVF and DVF) and the object position ( 20, 0 and 20 ). For the other parameters a two within-subject (visual feedback conditions) 3 (marker positions) 3 (object positions) ANOVA was conducted. For each analysis a significance level of 0.05 was chosen. For comparison of each marker a Tukey HSD test was used as a post hoc test. Fig. 2B shows the average hand paths for each condition. Each trajectory represents the average of 10 replications for all 8 subjects (80 paths). We can observe that hand-paths tend to be straighter in the IVF than in the DVF condition only for movements executed toward the 20 and the 0 directions. The statistical analyses confirm this observation. PCI was less pronounced in IVF than in DVF condition, F(1, 7) = 11.38, p < There was also a significant main effect of the object position, F(2, 14) = 17.01, p < 0.001, and a significant interaction between the object position and the visual feedback situation, F(2, 14) = 16.91, p < To investigate this interaction further, we then used simple effects analysis to compare the two visual feedback conditions for each object position and each marker (see Fig. 2A). When the object was at 20, a significant difference between the two visual feedback conditions was observed for the three marker positions: trajectories were straighter in the IVF condition (wrist: F(1, 7) = 39.81, p < 0.001; thumb: F(1, 7) = 24.47, p < 0.01; index finger: F(1, 7) = 25.98, p < 0.01). The same effects were found for movements directed at the object located at 0 (wrist: F(1, 7) = 10.49, p < 0.02; thumb: F(1, 7) = 8.29, p < 0.03; index finger: F(1, 7) = 6.35, p < 0.04). Conversely, when the object was positioned at 20, differences between IVF and DVF conditions were not significant whatever the marker position (wrist: F(1, 7) = 1.90, p > 0.05; thumb: F(1, 7) = 2.80, p > 0.05; index finger: F(1, 7) = 1.78, p > 0.05). The post hoc analysis revealed that in each condition of visual feedback, trajectories of thumb and index finger were straighter than the trajectory of the wrist. No significant difference was noted between trajectory curvatures of thumb and index finger. We can notice that, when the object was positioned at 20, the maximum distance (MD) between the trajectory and the straight line joining the starting and the end point was on average 10 and 39 mm for the IVF and DVF conditions, respectively, F(1, 7) = 17.20, p < For movements directed at the object located at 0, MD was on average 7 mm for the IVF condition and 28 mm for the DVF condition, F(1, 7) = 8.71, p < Conversely, no significant difference was noted between IVF and DVF condition when the object was positioned at 20 : MD was on average 19 mm for the IVF condition and 28 mm for the DVF condition, F(1, 7) = 1.43, p > Therefore, as observed in Fig. 2A, in the IVF condition trajectories tended to be curved leftward with respect to the sagittal axis when the object was positioned at 20, whereas they tended to be curved rightward when the object was positioned at 0 and at 20. The effect of the visual feedback on trajectory curvature was present from the beginning of the movement. The perpendicular distance of paths from a straight line connecting start- and endpoints 150 ms after movement onset was indeed less important in the IVF condition when the object was positioned at 20 (wrist: F(1, 7) = 9.25, p < 0.02; thumb: F(1, 7) = 9.87, p < 0.02; index finger: F(1, 7) = 10.44, p < 0.02) and at 0 (wrist: F(1, 7) = 7.30, p < 0.05; thumb: F(1, 7) = 7.18, p < 0.05; index finger: F(1, 7) = 6.43, p < 0.01). No significant difference was observed for the object located at 20 (all ps > 0.05). As concerned MT no significant difference was observed between the IVF (1171 ms) and the DVF conditions (892 ms), F(1, 7) = 3.38, p > 0.05 and between the object positions, F(2, 14) = 1.23, p > A significant interaction was noted between the object position and the visual feedback, F(2, 14) = 4.29, p < The analysis of the simple effects showed that MT was longer in the IVF situation for the 20 position (IVF: 1241 ms, DVF: 876 ms, F(1, 7) = 6.49, p < 0.05). No significant difference was observed for the 0 position (IVF: 1180 ms, DVF: 882 ms, F(1, 7) = 2.47, p > 0.05) and for the 20 position (IVF: 1091 ms, DVF: 918 ms, F(1, 7) = 1.77, p > 0.05). However, we can notice that concerning the 0 position, the standard deviation was 450 ms in the IVF condition: for two subjects the mean

4 100 R. Palluel-Germain et al. / Neuroscience Letters 401 (2006) Fig. 2. Hand path curvature. (A) Mean hand path curvature and standard deviations according to the visual feedback, the object position and the marker position. (B) Mean normalised trajectories according to the visual feedback (IVF open circles and DVF thin lines) for the wrist, the thumb and the index finger. Dotted circles figure the three different objects positions. MT was very long (2218 and 1493 ms). For the other subjects mean MT was 955 ms (MTs ranging from 766 to 1100 ms). The goal of the present experiment was to evaluate whether changing the visual context could affect the processes involved in planning grasping movements. To this aim we compared the spatial temporal characteristics of grasping movements performed either under a remote visual control condition, or in a direct visual feedback condition. Results show that the curvature of the hand path performed by the hand was significantly straighter in IVF than in DVF condition for grasping movements directed at the objects located at 20 on the left and at 0 with respect to the subject s midsagittal axis. Such contrasted performances could result from the fact that controlling movements through a videoscreen constrains subjects to produce slow movements. In this case, the increase of on-line adjustments in slower movements might cause hand paths to be less curved. However, this hypothesis seems hardly plausible since we observed comparable MTs (for most of the participants) in both IVF and DVF conditions for the object located at 0 whereas path curvatures differed, the hand path being straighter in the IVF condition. Then if one assumes that hand trajectories can give insight into how visually directed movements are planned [3,4,16,17], we rather suppose that the differences observed in path curvature provide evidence that these two situations of visual feedback involve different control strategies. Furthermore, this assumption is backed up by the fact that differences in path curvature exist even right at the beginning of the movement (i.e., 150 ms after movement onset). Therefore, observing straighter trajectories under indirect visual feedback suggests that, in this situation, movements tend to be planned in extrinsic space so that the resulting path is roughly straight. These observations confirm previous results obtained with pointing movements [18]. In addition, as we can consider that in grasping movements the hand path is widely determined by the demands of the task [5,6,10,19,22,24], the present data show that the visual feedback constraint appears to dominate biomechanical constraints and has a massive influence on the trajectory formation. An interesting consequence of the present results is that they can reconcile controversial data reported in the literature. Indeed, as pointed in the introduction a large set of data showed that when subjects carried out reaching movements, hand paths were consistently straight [9,16,26,27]. By contrast, other studies observed that the shape of the movement carried out by the hand varied as a function of the movement direction. This outcome was interpreted as implying that reaching movements are not planned to follow a straight line in the extrinsic space

5 R. Palluel-Germain et al. / Neuroscience Letters 401 (2006) [1,5,14,17]. It is worth noting that a large set of studies describing extrinsic space planning used an indirect visual control situation in order to study motor planning processes [8,9,11,13,15]. Thus, divergent results obtained in the literature could be explained by the nature of the visual feedback provided when executing hand movement. A similar conclusion may be inferred from results obtained by Messier and Kalaska [15]. They indeed show the type of visual feedback provided (either IVF or DVF conditions) has an influence on the pattern of variable errors evidenced in a pointing task. Our results extend this finding by showing that, unlike under direct visual feedback situation, movements performed under indirect visual feedback tend to be planned in extrinsic space. Movements in the IVF condition being controlled though a video-system, it might be argued that the contrasted performances observed between IVF and DVF conditions for the 20 and 0 object positions originated from the fact that the visual feedback was in two dimensions in the IVF condition. Thus, one may think that this pattern of results could result from the processing of the monocular cues dominant in the IVF condition. However, it has been shown that the shapes of movement trajectories in a reaching task are similar under binocular and monocular viewing [12]. In addition, as the cylinder was viewed from the top in the IVF condition, we may suppose that the use of depth cues specifying the height of the object and thus the size of the grasping surface was limited in the IVF condition. However, this hypothesis is unlikely because (1) no significant difference in the height of the hand trajectory was observed between IVF and DVF conditions (z-axis, results not presented), and (2) once participants completed the first movement, they had access to the vertical dimension of the target object. We rather think that a visual constraint can straighten the hand path because in a remote control situation subjects estimate the desired trajectory by using hand and target positions displayed on the video screen. The idea of a visual hand-centred perception of target position is in agreement with studies concerned with frames of reference used in adaptation to directional bias in a video-controlled reaching task [2,20,21]. Boy et al. [2], for example, showed that in a remote control situation involving a non-congruent visual scene according to the actual work space, the initial orientation of hand trajectory in the first trial corresponded to the directional bias introduced. Therefore, in a remote visual control condition, the movement is initially coded as a vector between the initial hand position and the target position both determined from the visual display. However, this conclusion has to be moderated. Indeed, when the object was located at 20 with respect to the sagittal axis, movements carried out under the two conditions of visual feedback were quite similar, as regard to the hand path and to the movement duration. This result stresses the fact that the effect of the visual context on movement planning is not isotropic: when movements are controlled through a video screen, the effect of the visual context seems to differ according to the visual hemispace considered. Some previous reports are in agreement with the fact that the features of the hand trajectory in goaldirected movements depends on the workspace region in which the movement is executed. For example, in the horizontal plane, transverse movements were found to be outwardly convex, but radial movements were relatively straight [25]. In the vertical plane up and down movements were found to be outwardly convex, whereas back and forth movements were relatively straight [1]. In the same vein, the present data show that controlling one s own movements through a video display could also induce a modification in path curvature to an extent that depends on movement direction. Acknowledgement We want to thank Matthieu Bain to help running this experiment. References [1] C.G. Atkeson, J.M. Hollerbach, Kinematic features of unrestrained vertical arm movements, J. Neurosci. 5 (1985) [2] F. Boy, R. Palluel-Germain, J.P. Orliaguet, Y. Coello, Dissociation between where and how judgements of one s own motor performance in a video-controlled reaching task, Neurosci. Lett. 386 (2005) [3] M. Desmurget, M. Jordan, C. Prablanc, M. Jeannerod, Constrained and unconstrained movements involve different control strategies, J. Neurophysiol. 77 (1997) [4] M. Desmurget, D. Pelisson, Y. Rossetti, C. Prablanc, From eye to hand: planning goal-directed movements, Neurosci. Biobehav. Rev. 22 (1998) [5] M. Desmurget, C. Prablanc, Postural control of three-dimensional prehension movements, J. Neurophysiol. 77 (1997) [6] M. Desmurget, C. Prablanc, Y. Rossetti, M. Arzi, Y. Paulignan, C. Urquizar, J.C. Mignot, Postural and synergic control for threedimensional movements of reaching and grasping, J. Neurophysiol. 74 (1995) [7] T. Flash, N. Hogan, The coordination of arm movements: an experimentally confirmed mathematical model, J. Neurosci. 5 (1985) [8] J. Gordon, M.F. Ghilardi, C. Ghez, Accuracy of planar reaching movements. I. Independence of direction and extent variability, Exp. Brain Res. 99 (1994) [9] J. Gordon, M.F. Ghilardi, C. Ghez, Impairments of reaching movements in patients without proprioception. I. Spatial errors, J. Neurophysiol. 73 (1995) [10] H. Grea, M. Desmurget, C. Prablanc, Postural invariance in threedimensional reaching and grasping movements, Exp. Brain Res. 134 (2000) [11] H. Imamizu, S. Shimojo, The locus of visual-motor learning at the task or manipulator level: implications from intermanual transfer, J. Exp. Psychol. Hum. Percept. Perform. 21 (1995) [12] D.C. Knill, D. Kersten, Visuomotor sensitivity to visual information about surface orientation, J. Neurophysiol. 91 (2004) [13] J.W. Krakauer, Z.M. Pine, M.F. Ghilardi, C. Ghez, Learning of visuomotor transformations for vectorial planning of reaching trajectories, J. Neurosci. 20 (2000) [14] F. Lacquaniti, J.F. Soechting, S.A. Terzuolo, Path constraints on pointto-point arm movements in three-dimensional space, Neuroscience 17 (1986) [15] J. Messier, J.F. Kalaska, Differential effect of task conditions on errors of direction and extent of reaching movements, Exp. Brain Res. 115 (1997) [16] P. Morasso, Spatial control of arm movements, Exp. Brain Res. 42 (1981) [17] R. Osu, Y. Uno, Y. Koike, M. Kawato, Possible explanations for trajectory curvature in multijoint arm movements, J. Exp. Psychol. Hum. Percept. Perform. 23 (1997)

6 102 R. Palluel-Germain et al. / Neuroscience Letters 401 (2006) [18] R. Palluel-Germain, F. Boy, J.P. Orliaguet, Y. Coello, Visual and motor constraints on trajectory planning in pointing movements, Neurosci. Lett. 372 (2004) [19] Y. Paulignan, V.G. Frak, I. Toni, M. Jeannerod, Influence of object position and size on human prehension movements, Exp. Brain Res. 114 (1997) [20] I. Pennel, Y. Coello, J.P. Orliaguet, Frame of reference and adaptation to directional bias in a video-controlled reaching task, Ergonomics 45 (2002) [21] I. Pennel, Y. Coello, J.P. Orliaguet, Visuokinesthetic realignment in a video-controlled reaching task, J. Mot. Behav. 35 (2003) [22] D.A. Rosenbaum, R.J. Meulenbroek, J. Vaughan, C. Jansen, Posturebased motion planning: applications to grasping, Psychol. Rev. 108 (2001) [23] R.L. Sainburg, J.E. Lateiner, M.L. Latash, L.B. Bagesteiro, Effects of altering initial position on movement direction and extent, J. Neurophysiol. 89 (2003) [24] J.B. Smeets, E. Brenner, A new view on grasping, Motor Control 3 (1999) [25] Y. Uno, M. Kawato, R. Suzuki, Formation and control of optimal trajectory in human multijoint arm movement. Minimum torque-change model, Biol. Cybern. 61 (1989) [26] P. Vindras, P. Viviani, Frames of reference and control parameters in visuomanual pointing, J. Exp. Psychol. Hum. Percept. Perform. 24 (1998) [27] P. Vindras, P. Viviani, Altering the visuomotor gain. Evidence that motor plans deal with vector quantities, Exp. Brain Res. 147 (2002)