Representation of hand position prior to movement

Representation of hand position prior to movement and motor variabilityq

Can. J. Physiol. Pharmacol. Downloaded from www.nrcresearchpress.com by 199.201.121.12 on 06/05/13 For personal use only.

Miehe! Desmuaget, Yves Rossetti, Claude Prablane, George EmStelmaeh, and Mare Jeannerod

Abstract: Pointing accurary of six h u m n subjects was measured in two blocked conditions where the hand was either never visible (T: target only) or only visible in static position prior to movement onset (H+T: hand target). It was shown in condition H + T that, viewing the hand prior to movement greatly decreased end-point variability compared with condition T. This effect was associated with a significant modification of the movement kinematics: the H + T condition induced a shortened acceleration phase with a corresponding lengthened deceleration phase, compared with the T condition. These results led us to the hypothesis that viewing the hand prior to movement onset allowed a decrease of pointing variability through a feedback process. This hypothesis was further tested by turning the target off during the deceleration phase of the movement at half peak velocity- Ht was shown that turning the target off had no effect upon the T condition but induced a significant increase of pointing variability in the H + T condition. This result suggests that vision of the static hand enhances the proprioceptive localization of the limb and allows for a better visual to kinesthesic feedback.

+

Key words: proprioception, vision, motor control, pointing, feedback, human.

WCsurnC : La prkcision de six sujets humains, sournis a une tiche de pointage, a Ctk rnesurke sous deux conditions blocpCes pour lesquelles la main n'ktait, soit jamais visible (T: cible seule), soit visible uniquement en position statique avant le debut du mouvement (H+T: main cible). Les rksultats ont rnontrC que la vision de la main avam le debut du geste (H+T) entrainait une diminution importante de la variabilite finale des pointages, par rapport a la situation noir coknplet (T). Cet effet s'accompagnait de modifications significatives de la cinkmatique du geste. En comparaison de la condition T, H + T induisait ainsi une diminution de la phase d'accClCraton, et usne augmentation de celle de dCcC1Cration. Ces modifications nous ont amen6 a postuler que la vision initiale de la m i n permettait de rCduire la variabilite finale du geste 8 travers un prscessus rktroactif. Cette hypothhse fklt testCe en Cteignant la cible durant la phase de dCctlkration du mouvement, 8 la rnoiti6 du pic de vitesse. I1 a CtC constatt que l'extinction de la cible Ctait sans effet pour la condition T, abrs qu'elle ianpliquait une augmentation de variabilitk sous H+T. Ces rCsaaltats suggkont donc que la vision initide de la main agit en amkliorant la localisation de l'effecteur, ce qui perrnet une optimisation des: boucles de rktrocontrole visuo-kinesthesiques.

+

wahats ck& : proprioception, vision, control moteur, pointage, rCtroaction, hunaain.

The determination of the accuracy of an aimed movernent such as grasping an object is a very complex procedure not only because of the high number of necessary parameters but SO because of the large set, if not infinite, of correct responses. Indeed there is theoretically no strict correspondence between an object and the final configuration of the effectors used to grasp it. The use of a simple movement such as pointing with the index fingertip on a target is an easier Received January 25, 1994.

M. Dsrnurgt, YoRossetti," C. RabBmc, and M. Jewerasd. Vision et motricitC, Institut national de la santk et de la recherche rnCdicale, Unite 94, F-$9500, Bron, France. G.E. Stehach. Department of Exercise Science and Physical Education, State University, Tempe, AZ, U.S.A. This paper has undergone the Journal's usual peer review. Author for correspsndence.

approach, more reductisnist, and nonetheless ecological if we consider for instance in life every day the pointing behaviour required for man -machine interactions. However, it has the advantage of having a strict matching between stimulus and response, although the coincidence between the index Gngertip and the target may be obtained by several joint configurations. For a target-pointing task, the error for each trial can be defined in an external frame of reference by the threedimensional vector joining the target to the final hit point- For most pointing operations, performed on a planar board, the error Is even ;educed to two dimensions. Fo; a series of pointi n g ~onto a given target, the error is defined by its two main components: the constant error (CE) is the two-dimensional error vector joining the target to the average final hit point. The variable error (VE) or scatter along the two dimensions x and y can be defined by the surface of the ellipse where the probability of finding a pointing movement exceeds 66%. Movement variability is classically considered as a product of the motor command variabilitymArguments for this view

Can. J. Physiol. Pharmaeol. 73: 262 -272 (1995). Printed in Canada / Imprim6 au Canada


have been supprted by experimental data (e.g ., Schmidt et al. 1979, 1985; Jeannerod 1988; Kawato 1991; Meyer et al. 1990). Basically, it has been shown that force variability produced by a rapid a m movement is proportional to the mean force intensity and to the movement amplitude. However, further studies have shown that movement variability does not reach its maximum toward movement end, but rather around peak velocity (e. g ., Paulignan et al. 1991; Rossetti et d. 1991). This suggests that some mechanisms are able to reduce movement variability that arises from the initial motor impulse. This idea was explored by Darling and Cooke (1987a, H987b, 1 9 8 7 ~who ) ~ showed a decrease in the spatiotemporal variability of pointing trajectories during learning. Surprisingly, they observed an associated increase of electromyographic (EMG) variability. Their interpretation was that learning results in a coupling between agonist and antagonist bursts, so that the variability of the initial maximum agonist activity can be further compensated by the following antagonist activity. The mechanism responsible for this coupling may be based on efference copy or on muscle afferents. In this recent conception, the general idea remains that movement variability is produced either by the efferent or by the afferent signals from the initid agonist impulse. The aim of this work was to expand this classical view by identifying other factors possibly involved in the production of variability, such as localization of the goal and of the effector before (static factors) and during (dynamic factors) the movement. In the next sections, the so-called "open loop situation" will define pointing movements performed at small Hightemitting diode (LED) targets in an otherwise dark room in which the hand is never seen even when it is on the target. Conversely, the so-called "closed loop situation9' will correspond to the same pointings at small LED targets, but with the hand continuously seen.

Dynamic factors related to movement variability Some experiments have demonstrated that the alteration, after movement onset, of the arm information induced an increase in movement variability. Thus Prablanc et d.(1979a) have described a considerable decrease of VE in the open loop situation compared with the closed loop situation (Conti and Beaubaton 1976; Beaubaton and Hay 1986; Velay and Beaubaton 1986; for a review see also Jeannerod 1988). Interestingly, the same phenomenon occurs if the visual target is suppressed at the onset or during the pointing movement (Elliott et d. 1991; Prablanc et al. 1979a, 1986). These results could suggest that movement variability also depends on a ' 'visual -proprioceptive feedback loop' ' that allows for the comparison, either continuous or discrete, between the visual position of the target and the proprioceptive position of the moving arm. In other words, VE could be related to the errors that occur in the sensory transformation of the internal representation of the visual target location into a kinematic representation of the arm movement (Soechting and Flanders 1989a, 1989b). The influence of the visual proprioceptive loop on movement variability will be studied in the second experiment presented here. Static factors rehted to mc~vementvariability It is likely, at least for multijoint movement, that the nervous system needs information about the initial position of both

the goal and the effector in order to program the movement (see, among others, Bock and Eckmiller 1986; Bock and Arnold 1993; Ghez et diT. 1990; Larish et al. 1984). Thus, the accuracy of the internal target and arm representation.may be a parameter of movement variability. Reconstmction of the target position in a body-centered space should require integration of the target's retinal position with respect to the line of sight, the extraretind signal related to the position of the eyes in the orbit, and the neck signal related to the head position on the body. Whereas increasing retinal eccentricity when pointing to targets under peripheral vision may not affect VE (Bock 1986), increasing the amplitude of the extraretinal (Prablanc et al. 1979b; Biguer et al. 1984) or of the neck (Roll et al. 1986) signals increases pointing variability. In addition the increased variability related to eccentric eye positions is much larger than that related to eccentric head positions (Rossetti et al. 1994a). Another example suggesting that the acuity of target representation may affect pointing variability is provided by a task of blind pointing toward the contralateral hand. When position signals encoding the target arm are degraded by imposing extreme joint positions for that arm, pointing with the other m toward the target finger located on the s m e p i n t h o m e s much more variable (Rossetti et al. 1994b). It is interesting to note that a linkage between the acuity of target representation and pointing variability was suggested by Bowditch and Southard as early as 1880 (Bowditch and Southard 1880). These authors performed an experiment to compare the spatial acuity of the visual and the proprioceptive localizations. The subject with eyes closed was pointing toward a target that had been previously viewed or touched. Bowditch and Southard measured the number of pointings obtained in concentric circles around the target, which is a good measure of end-point variability. Their conclusion was that visual acuity was better than proprioceptive acuity. The representation of initid arm position also seems to contribute to motor variability. Prablanc et d.(1979b) compared the pinting performance toward visual targets when subjects could see their hand prior to pointing in the dark ("static closed loop situation9') to the performance when they were pointing in full darkness. They showed that pointing variability was reduced by about one third when the visual information about hand position was available prior to movement. This result suggests that the quality of the internally represented arm position prior to movement contributes to pointing variability. This result was then confirmed by Elliott et al. (1991), Ghez et d. (1990), and Rossetti et al. (1994~).These latter authors showed in deafferented patients that viewing the hand at rest prior to movement could substantially decrease movement variability. However, Ghez et d. (1990) failed to replicate this effect with normal subjects (see also Proteau and Materniuk 1993). Whereas previous experiments have only focused on the end-point errors, the present work also considered the kinematic characteristics in order to andyse the time course of the movement in the static closed loop situation. Precisely, the current study consisted of two distinct experiments. The first one investigated the role of ' "static factors9 on movement variability by determining how initial vision of the limb prior to movement improves its accuracy. To answer to this question we tested whether the decreased pointing variability was

Can. J. Physiol. Pharmacol. VoI. 73, 1995 reflected in early or late kinematic aspects of the movement. The purpose of the second experiment was to go h e a d a step further in analyzing the influence of the "dynamic factors" on movement accuracy.

Matmials and methods

Procedure The subject's task was to point with his fingertip as accurately and quickly as possible (i.e., to find his or her own best trade-off between reaction time plus movement time and accuracy) once a target was lit and then to return back to the starting position located close to the body within the sagittd plane. Positioning the finger on the starting position automatically triggered the beginning of the next trial. The target presentation order was randomly selected by the control program to reduce motor and spatial memory acquisition (Georgopoulos et al. 1981). Two experimental conditions were considered (see Fig. 1, upper part). (i) The first was the "target-only" (T) condition, in which the room was kept in h l l darkness throughout the experiment. (is') The second was the "view sf hand and target" (H+T) condition, in which the hand was only visible at rest, starting from 300 ms prior to target presentation to movement onset, and then remained hidden until the next trid. Six right-handed subjects (5 males, 1 female) from 22 to 47 years of age participated in this experiment. Each condition was blocked, and all subjects were submitted to both experimental conditions. There were 78 trials per session (7 targets x 10 repetitions), and the order of presentation between the two experimental conditions was randomized (Snedecor and Cochran 1984).

Apparatus The pointing apparatus was a horizontal table made of a flat black surface (1.58 m deep by 2-00m wide). LEDs were positioned on a circular ring above the working surface. A semireflecting mirror was d s o placed between the table and the diode plane. The mirror and the LEDs were arranged on the ring so that the virtual images of the LEDs appeared on the horizontal table on a half circle (595-mm radius centred on the subject body axis) and, at 18" increments, from 20" to the left 4-20) s f the sagittal plane passing by the circle centre, to 40" to the right (640). With this apparatus the subject could see, by means of the mirror, the virtual image of the target on the pointing surface without seeing his hand (for a more detailed description of this apparatus see Prablanc et al. B979a). Thus, this setup did not allow howledge of pointing errors. In both conditions, subjects sat comfortably in a dentist chair facing the table. The chair was placed so that the vertical body axis of the subjects coincided with the centre of the target circle, with the sagittal axis aligned with the 8" target. Chair height was adjusted for each subject such that the elbow would not touch the table during the pointing task. The initial starting point of the hand was composed of a tactile mark on the table (10-mm diameter), Hocatd 235 mrn h e a d of the body axis. Movements of the tmnk were prevented by using a seat belt. This experiment was hlly controlled on-line by a program run on a PDP 11/73 computer. The x and y horizontal cornponents of the hand movement were recorded through a small infrared-emitting diode (3 x 2 mm) placed on the nail sf the index finger and coupled with a Selspot 2 system. The Selspot camera, positioned 3 m above the working surface, provided the coordinates of the fingertip with an accuracy of

Data analysis Position data were filtered and derived through a frequency impulse response (FIR) filter with a cutoff frequency of 20 Hz and with a window of f9 frames. The main kinematic landmarks were computed. The onset and end of the finger movements were determined automatically from position, veBocity, and acceleration thresholds, which provided reaction time (RT) and movement time (MT). Time to peak acceleration QTPA), time to peak velocity (TPV), and corresponding values of tangential velocity and acceleration were also cornputed, as well as deceleration time (MT - TPV). The x and y coordinates of trajectory end points were used to compute spatial errors for each trial. The constant radial error defined the distance from the mean reached point to the pointed target, measured along the radius from the b d y axis. This error sign was defined as positive when the subjects overshot the target, and negative when they undershot it. The constant angular error described the angle between the target and the reached point, the angle vertex being the subject's vertical axis. This error was defined as positive when the subjects pointed (with their right hand) to the right sf the target and negative when they pointed to the left. The variable error, also called the 66ellipseerror,9' was calculated for each target and condition by the formula ar x SD, x SD,. A two-way analysis of variance tested the influence sf target eccentricity and visual conditions T and H +T on the different parameters. As no parameter exhibited any interaction between target eccentricity and visual conditions (lowest p > 8.85), all eccentricities were pooled; there were thus seven averaged measures per subject. The t test comparison between H + T and T conditions was performed on seven

Experiment 1


f8.7 mm. All the data were collected at a sampling rate of 200 Hz.

As mentioned in the Introduction, some experimental results have suggested that viewing the static hand prior to initiating a god-directed movement could improve pointing accuracy. However, they are partly contradictory and it is not clear whether the view of the static hand decreases only variable errors (Elliott et d.1921; Rosetti et d- 1994~)or dso constant errors (Prablanc et al. H979b). m e first part of this experiment deals with this issue. Another important question was to determine through which process vision improves movement accuracy. Indeed, this improvement can be related either to a feedforward or to a feedback mechanism. It is interesting to note that these two assumptions have different implications. The first one implies that improvement of accuracy will be associated with early changes in the kinematic aspects of the movement cornpared with those produced in total darkness. Conversely the second one implies that changes will occur later, during the deceleration phase of the movement. The second part of this experiment was designed to disentangle these hypotheses.

Desrnurget et al.


Fig. 8 . Schematic representation of the experimental conditions US& in the two experiments. Upper part: setup B,.In the target only condition (T),the room was kept in frall darkness throughout. In the vision of hand and target condition (H+T), the hand was visible at rest only from 300 m before target presentation until movement onset. Lower part: setup D,. The T and H + T conditions are the same as in setup D, . In contrast to B, the target is turned off during the deceleration phase of the hand movement, at half peak velocity.

FEATION POINT

ON TARGET

B

EYE

I

Position

r-----------

HAND

om TARGET

I

EYE

I Position !-----------I Position

I

I B

Can. J. Physiol. Pharrnacol. Vol. 73, 1995

Table I. Mean and standard deviation for spatial and tenraporal parameters in the six subjects from experiment 1. Constant error is represented by angular error and radial error, whereas variable error is represented by the ellipse error containing 66% of the pointings. -

T

H+T


Mean Accuracy Angular error (") Radial error (mm) Ellipse error dm2) Temporal and kinematic parameters (ms) Reaction time Movement time Time to peak acceleration Time to peak velocity Deceleration time

SD

Mean

SD

Fig. 3. Movement duration is unaffected by experimental conditions, either under vision of the hand prior to the movement (W+T) or no vision at all of the hand (T). However, the kinematic profile is slightly but significantly affected: the H+T condition corresponds to a shortened acceleration phase, with a corresponding lengthened deceleration phase, compared with the T condition. Deceleration Time TPV TPA

0.18 23 275

2.7 30 142

332

147 95 35

547

$0 227 32 1

* p < 0.05

45

83

Fig. 2. Error parameters as a function of experimental conditions. Vision of the hand prior to movement initiation (H+T) strongly decreases variable errors compared with the no vision of hand condition (T). These experimental conditions have no effect on radial and angular consfant errors.

""1

Angular emor (0.1")

Experimental condition (targets) x six (subjects) for all parameters. Data displayed in the Results section are means (SD) sf pointings over subjects. The threshold for statistical significance was set at 0.05.

Results Results are summarized in Table 1 and Figs. 2 and 3. Pointing accuracy Constant error: Angular error ranged from 0.18" (2.7) for H + T to 8,14" (4.2) for T. This difference did not reach statistical significance (t = 0.13, p > 0.85). Radial error ranged from 23 aaarrm (30) for H +% to 26 mm (57) for T. As previously, this difference did not reach statistical significance ( t = -0.61, p > 0.5). Variable error: Ellipse error ranged from 275 mm2 (24%)

T HsT Experimental condition

for H + T to 492 mm2 (215) for T. En contrast with constant error, this difference was very significant (t = 5, p < 0.0001) and showed that seeing the hand prior to the movement permitted a 56% reduction of pointing variability. Temporal and kinematics parameters Reaction time was significantly shorter in condition H + T (332 ms (147)) than in condition T (367 ms (122)) (t = -3.1, p < 0.002). Movement time was almost the same in the two conditions (547 was (95) for H + T versus 546 ms (105) for 9). Time to peak acceleration (%PA)was attained significantly sooner in condition H + T (88 ms (35)) compared with condition T (98 ms (45)) ( t = -2.0, p < 0.05). Time to peak velocity (TPV), d s o called acceleration time, showed a similar evolution to TPA. Indeed TPV was attained significantly sooner in condition H f T (227 ms (45)) compared with condition T (240 ms (5 1)) (t = -3.2, p < 0.002) Deceleration time (i.e., time following peak velocity) was longer in condition H +T (32 1 ms (83)) than in condition T (306 ms (76)). Interestingly, this difference was statistically significant (t = 2.0, p < 0.85). Thus, kinematic landmarks showed a slight modification of the structure of the pointing in the condition H+T. Statistical analysis when data are scaled in percentage of movement confirm this result: when subjects were allowed to see their static hand prior to movement onset, they exhibited an increase of deceleration time, whereas acceleration time decreased, within the same movement time. As can be seen in Fig. 3, the asymmetry between acceleration and deceleration times increased from condition T to condition H T.

+

Discussion Error analysis This experiment showed that viewing the hand prior to


initiating a pointing movement greatly reduces the end-point variability. This result on hand pointing variability is in agreement with the data of previous studies by Prablanc et al. (1979b) and Elliott et al. (1991). However, when instead of variable error, constant error is considered, it appears that our data (and those from Elliott et al. 1991) depart from those of Prablanc et al. (1979b). Indeed, we failed to confirm a positive effect of the static closed loop situation, i.e., viewing the hand prior to movement, on the constant error. This discrepancy seems important in explaining the effective process by which the view of the hand improves movement accuracy. This assumption will be considered below. Paillard and Brouchon (1974) showed in a finger to finger pointing task that the introduction of a delay between the positioning of the target hand and the pointing movement induced a systematic pointing error after a delay of about 8 s. This result was confirmed by Wann and Ibrahim (1992), who showed moreover that the drift of the proprioceptive information was stopped if brief glimpses of the target limb were allowed. An hypothesis for explaining the effect of viewing the (static) hand prior to movement initiation on pointing accuracy is then the calibration of proprioception by vision (Prablanc et al. 1979b; Jeannerod and Prablanc 1983; Jeannerod 1988, 1989, 1991; Rossetti et al. 1994~).A corollary of this assumption is that the accuracy of recalibration is dependent on the acuity of the visual information a b u t hand position. Accordingly, we can assume that the difference between Prablanc9s and our results originates in the type of vision of the initial position of the limb. Indeed, foveal vision used by Prablanc et al. (1979b) is ?mown to allow a better encoding of the spatial position of static cues compared with peripheral vision (Paillard 1980, 1982; Paillard and Amblard 1985). This could imply that the same mechanism is responsible for the decrease of the variable and constant error in the static closed loop situation and that the absence of decrease of constant error in the present experiment is likely due to the poorer visual location of the hand in the peripheral visual field than in the fovea. However, some data indicate that the above hypothesis is incorrect and that a process of recalibration of proprioception by vision does not explain the decrease in variable error observed in the present study. Indeed, since the inter-trial latency (about 2 s) in our experiment is much less than the visual drift threshold (about 8 s), the proprioceptive map does not have enough time to shift with respect to the extrapersonal space. Moreover, Velay (1984) showed in a proprioceptive pointing task that the pointing delay that caused an increase of constant error did not affect variable error. Taken together, these data lead us to the hypothesis that in fact, the combination of proprioceptive and visual information before movement onset improves pointing accuracy by decreasing the noise bound to the internal representation of limb position relative to the body. Such a mechanism was observed at the neural level by Meredith and Stein (1986) in the selectivity of a visual and auditory cell encoding the same location of space. Moreover, it is possible that the constant error decrease observed by Prablanc et al. (1979b) in the static closed loop situation is due to factors not directly related to a reset of the proprioceptive map with respect to the visual one. These factors are the efference copy of the ocular saccade and the allocentric frame of reference used to

encode the movement. They are discussed below. The eye saccade is known to guide the limb motion toward the target (Festinger and Canon 1965; Prablanc et al. 1979b; Abrams and Landgraf 1990; Monda 1984). In particular, the "outflow" theory proposes that the arm motor command system uses the information derived from the eye saccadic command to program the movement (Hansen and Skavenski 1977; PClisson et al. 1986; Pklisson 1987). Thus we can postulate that the full compatibility between the arm and eyes displacement in the Prablanc et d.(19793) experiment explains the constant error decrease noted by these authors. Second, the existence of a dual processing mode for locating a target in space has often been suggested (Matin et al. 1982; Abrams and Landgraf 1990; Paillard 1991; Blouin et al. 1993). Thus, position of extrapersonal objects can be determined using (b) an dlocentric frame of reference, in which target position is determined essentially through a retinal signal in relation to stable external visual landmarks, (ki) an egocentric frame of reference, in which target position is determined only through an extraretinal signal in relation to the position of the b d y . In the first case, information about location of head relative to the tmnk has little importance, whereas in the second case it is critical. Indeed, the "headfixed" situation leads to an increase of pointing errors with respect to the "head-free" situation (Marteniuk 1978; Roll et al. 1981, 1986; Biguer et 1984; Jeannerod 1988). On this basis, it is possible to conjecture that the experimental setup with head fixed used by Prablanc et d. (1979a, 197%) resulted in an artificial increase of constant error in their open loop situation, and that there was no real decrease of angular and radial errors as a result of the vision of the static hand before movement onset. Indeed, in the static closed loop condition the subjects were able to use an dlocentric frame of reference to encode target position with respect to the hand, and fixation of the head had likely no influence on movement accuracy. Conversely, the open loop situation compels the subject to use an egocentric frame of reference. Determination of head position with respect to the tmnk is then critical to encode target position with respect to the body. Thus fixation of the head in this situation may have degraded target lmdkation and consequently hmd movement. Temporal and kinemtic analysis Our kinematic data showed a modification of the structure of the pointing movement in the condition H f T: when subjects were allowed to see their static hand prior to movement onsets, they exhibited a decrease of acceleration time, with a corresponding increase of deceleration time. Both early and late kinematic parameters were then affected in this static closed loop situation; and the observed improved accuracy might have been related to both feedback and feedfornard processes. Qi) Decrease of movemend varMbibbty related to a feedfornard process Some experimental data support the view that knowing the initial position of the limb is critical for movement encoding. Bock and colleagues r e p r t d results that showed that amplitude and direction rather than final position are the controlled variables for pointing movement (Bock and Echiller 1986; Bock and Arnold 1993). These results fit well with the


Can. J. Physiol. Pharrnacol. VoB. 73, 1995

"vectorial coding hypothesis" (Georgopoulos et al. 1983). Indeed the existence of a directional coding of the movement by cell populations has been confirmed at the electrophysiological level (GeorgopouHos % 986; Georgopoulos et al. 1986). Moreover, Ghez et al. (1998) showed for normal subjects that the initial acceleration of the arm was proportional to the pointing distance, and that the direction of the movement was already accurate at the beginning of the limb motion. When considering a single joint positioning task following the laws of the ' 'equilibrium point hypothesis, ' the directly controlled variable is not the joint angle but the difference between agonist and antagonist muscle activity. As a consequence, changing the location of the rotation axis s f the elbow results in an error of the previously learned angle (Polit and Bizzi 19791, which stresses the necessity for an either implicit or explicit knowledge sf initial limb position. As we showed in the previous section, position sense can rapidly Hose its exact relationship with extrapersonal space when vision of the arm is prevented. Taken together, all these data suggest that the static closed loop situation can act on movement accuracy through the improvement of the knowledge of initial position of the limb, The accuracy improvement noticed in the H +T condition can also be due to both a better egocentric encoding of target position, as the hand and the target within its environment are visible, and (or) a better specification of the vector of movement, because the effector and the goal are visible simultaneously. Regarding the first assumption, it is interesting to note that Velay and Beaubaton (1986) failed to show a significant improvement in the accuracy in a situation where contextual information was available before and after the movement (without vision s f the limb). This result suggests that vision of the environment alone before movement onset is insufficient to improve both the internal representation of target location and the pointing accuracy. According to the second assumption, the simultaneous presentation of the hand and the target would allow a better encoding of the vector movement by vision. This single (visual) modality encoding could improve movement accuracy by suppressing the visual -- proprioceptive sensorimotor transformation that occurs in the dark situation. Indeed some studies have shown that unimodal tasks are better achieved than multimodal tasks. For instance, when a subject is asked to match his right arm position with his left arm, he is more accurate when the stimulus and the movement are presented in the same modality (visual -visual or proprioceptive proprioceptive) than when they are presented in different modalities (proprioceptive -visual or visud -proprioceptive) (OrHiaguet 1985). Qii) Decrease of movement variability through a feedback

process Some experiments have suggested the existence of fine on-line motor control adjustment based on an error signal arising from the comparison between the prsprioceptive position of the finger and the visual position sf the target (Prablanc et al. 1986; Prablmc and Pklisson 1998; Prablanc md Martin 1992; Cord0 1990; Redon et al. 1991). Logically, the reliability of this corrective process is dependent on the reliability of the encoding hand psition itself. Then, a better specification of

the initial position of the limb relative to the extrapersonal space can enhance movement control during the deceleration phase by improving the computation of hand position relative to the target (Rossetti et al. 1994.c). This corrective process will be addressed more specifically in the second experiment.

Experiment 2 Closed loop models propose that movement accuracy depends on the comparison during arm displacement between the visual position of the target and the visud and (or) proprioceptive position of the limb (Carlton 1979; Howarth and Bowden 1971; Crossman and Goodewe 1983; Prablanc et al. 1986). According to this view, experimental results showed that the impairment of the signals arising from either the arm (Prablanc et al. 19'99~;Redon et al. 1991) or the target (Prablanc et al. 1986; Elliott et d. 1991; Rossetti et al. 1994~) could dramatically decrease movement accuracy. For instance, transient vibration of the biceps during the movement, which is known t s alter proprioceptive signals, induces a shift of paintings toward a visual target (Redon et d. 1991). The present experiment was designed to test the hypothesis that viewing the static hand prior to movement onset would decrease pointing variability through the enhancement of the visual -psoprioceptive feedback loop. The prediction of this hypothesis is that extinguishing the target during the find part of the movement after peak velocity must be more dramatic for pointing accuracy in the H + T condition than in the T condition.

Materials and methods Apparatus The apparatus used in the first experiment was altered in the following ways for the second experiment. The LEDs were positioned underneath the table at the same place as in the first experiment. Furthermore, to prevent the subject from using the visaad information about target location to guide his unseen hand toward the god9the target LED was extinguished during the deceleration phase of the movement, when instantaneous velocity of the hand pointing reached half its maximum value (see Fig. 1, lower part).

Procedure Procedure and conditions were otherwise the same as in the first experiment. Data 6k~mlysis Data analysis was identical with that in the previous experiment. Variable error and kinematics data obtained in the first experiment were compared with those obtained in this second experiment. A two-way ANOVA (setup X vision condition) was computed to test the differences between the setup Dl (target on during all the movement) and D2 (target off before the end of the msvement) for conditions T and H + T . The threshold for statistical significance was set at 0.85, and the t test was used for post hoc analysis.

Results Results are summarized in Table 2. The constant errors of hand pointing are not shown because, as in experiment 1,


none was dependent upon any of the investigated factors. Pointing accuracy Ellipse error analysis revealed a main effect for vision condition ( p < 0.0001), a main effect for experimental setup ( p < 0.05), and no interaction between these two factors ( p > 0.5). The difference between experimental setups was significant only in the H + T condition, as demonstrated by a post hoc analysis. In the T condition, where ellipse error for D2 to 492 mm2 (215) for ranged from 558 mma"276) Dl, the DllD2 difference did not reach statistical significance (t = -0.055, p > 0.9). Conversely, in the H +T condition, where ellipse error ranged from 360 m2(177) for D2 to 275 mm2 (142) for Dl, the D1/D2difference was statistically significant (t = 2.1, p < 0.05). This result indicates that the subject benefits from the vision of the target during the last part of the movement (Dl) only when he is dlowed to see his hand before pointing onset (H T) . The difference between T and H T vision conditions was significant for both setups Dl and D2, indicating that the vision of the hand prior to movement strongly reduced movement variability, despite target extinction during the deceleration phase. With the D2 setup, ellipse error decreased from 558 mm2 (276) in the T condition to 368 mm2 (177) in the H + T condition (t = 3, p < 0.01).

+ +

Temporal and kinematics parameters Movement time analysis revealed a main effect only for experimental setup ( p < 0.8001). Subjects moved more slower in Dl (547 ms (181)) than in D2 (403 ms (31)). This increase of movement time was significant in all visual conditions. In the T condition, i.e., with the hand in the dark, MT ranged from 398 ms (28) for D2 to 546 ms (105) for Dl (t = -5.7, p < 0.0001). In the H + T condition, i.e., with the hand seen prior to the movement, MT ranged from 487 ms (33) for D2 to 547 ms (95) for Dl (t = -6.2, p < 0.0801). The decrease of movement time observed for D2 (target off during the last part of the pointing) was entirely due to a decrease of deceleration time. Indeed, there was no clear effect of the experimental setups on the time to peak acceleration ( p > 0.05) or on the time to peak velocity ( p > 0.05). Statistical analysis revealed for these two parameters only a significant effect of vision condition ( p < 0.05), as observed in the first experiment. Deceleration time analysis revealed a main effect for vision condition ( p < 0.05), a main effect for experimental setup ( p < 0.0001), and no interaction between these two factors ( p > 0.5). Deceleration time was longer in Dl (312 ms (79)) than in D2 (187 ms (23)), for the two visual conditions. In the T condition, this parameter ranged from 173 ms (24) for D2 to 3% ms (76) for Dl (t = -7.8, p < 0.0001). In the H +T condition, it ranged from 200 ms (28) for D2 to 321 rns (83) for Dl (t = -6.5, p < 0.0001). All these kinematic features appear clearly in Table 2.

Discussion This experiment showed that the extinction of the target during the movement acted differently upon the T and H +T conditions. When subjects were dlowed to see their hand prior to the movement (H T), the vision of the target until

+

Table 2. Mean and standard deviation for all parameters in setups D, and D2 for the six subjects from experiments B and 2.

Ellipse error (mm2) MT (ms) TPA (ms) TPV (ms)

Deceleration time (ms)

492 (215) 546 (105) 98 (45) 248 (51) 306 (76)

275 (142) 547 (95) 80 (35) 227 (45) 32 1 ($3)

558 (276) 398 (28) 187 (24) 226 (2) 173 (24)

the end of the pointing had a significant effect on movement variability. Interestingly the velocity profiles in setups Dl and D2 were very similar up to peak velocity. If a different strategy was to be used in Dl and D2, it could apply only to the duration of the deceleration phase. Indeed the increase in deceleration time, observed under setup Dl compared with setup D2, indicated that subjects used the visual target for guiding the pointing through a feedback process. Conversely when the hand was never visible (T), extinguishing the target during the movement had no effect on pointing variability, notwithstanding an increase in deceleration time comparable with what was observed in the first situation. This result indicated the involvement of a feedback process, which was, however, not accurate enough to decrease pointing variability. Taken together, the above-mentioned data strongly suggest that view of the hand prior to movement onset can improve pointing accuracy through an enhancement of the proprioceptive localization of the limb. This allows a further optimization of the comparison between the visual target position and the felt arm position during the deceleration phase of the movement. However, this terminal feedback process is unable to explain by itself the whole effect of seeing the hand prior to movement onset. Indeed, accuracy was better achieved in condition H +T than in condition T, even when the target was switched off during the movement (D2). This suggests that the initial view of the hand also acts through a feedfornard process (see also the discussion of the first experiment).

General discussion The main issue of this study was to have some insight into the cooperation between vision and proprioception and to understand how it improves hand motor response. A first explanation is that vision of the hand allows a resetting and a sharpening of the proprioceptive map by aligning it with the visual map. Another explanation relies on the existence of two factors, controlling on one side the constant error biases and on the other side the end-point variability. The constant error could depend on a full spatial compatibility between initial and final positions of the oculomotor and the hand squelettomotor systems. (A simple example of fuK spatial compatiblity is the saccadic eye movement between two lit points within prehension space, followed quickly by the


Can. J. Physiol. Pharmaeol. Mi. 73, 1995

hand pointing between the same lit points). This factor is supposed to dlow an updating by gaze fixation and saccade, specifying both the location of the target and the exact amount of hand displacement. This cmcid factor was present in the exprirnent by Prablanc et al. (1979b) and absent in ours (present study; Rossetti et al. 1994~)and those of Elliott et d. (1991) and Ghez et al. (1990), as the eye movement vector was quite different from the hand movement vector. Indeed if the final position for the eye md hand were common, the initid p i n t s were sensibly different. The second factor acting upon the variability is supposed to be an allocentric definition of the movement common to the two (spatially compatible or not) types of experiments, and having the same effect in both peripheral and central vision of the hand, thus explaining the similarity of results on variable error. The second main aspect of this study was the respective role of feedforward and feedback processes in achieving a better hand pointing response with the initial view of the hand compared with pointing in the dark. The first experiment ation of the accelerationldecelerationratio with a shorter acceleration time and a longer deceleration time associated with a decrease of pointing error variability in the static closed loop situation compared with pointing in the dark. In the second experiment where vision of the target was turned off during the last part of the deceleration, deceleration time no longer increased under the static closed loop situation compared with pointing in the dark. As expected, the nearly unchanged deceleration time was conversely associated with a nearly unchanged pointing error variability. Taken together, thew two experimental results strongly suggest that some feedback processing occurs during the deceleration phase. This feedback comparing both the visual location of the target and the current dynamic hand position is responsible for the final error correction. When one branch of the cornparator is either removed or altered, it results in an increased variability. The results of our experiments are consistent with two other experiments. In the first one, under both initial view of the hand and full spatial compatibility between saccade and hand movement, duration of target presentation was shown to have an effect on pointing accuracy (Prablmc et al. 1986). In the second one (Wedon et d. 1991) with biceps vibration, the largest errors were observed when the vibration was applied during the end part of the movement, suggesting also that kinesthesic feedback was more influential during the last phase of the movement.

Acknowledgements The authors are grateful to Mohammad Arzi for statistical advice, Christian Urquizar for technical assistance, and Marcia Riley for useful comments. This work was supported by the FCdkration des Aveugles de France.

References Abrams, R.A., and Landgraf, J.Z. 1990. Differential use of distance and location information for spatial localization. Percept. Psychophys . 47: 349 -359. Beaubaton, D., and Hay, L. 1986. Contribution s f visual information to feedforward and feedback processes in rapid pointing movements. Hum. Mov. Sci. 5: 19 -34.

Biguer, B., Prablanc, C., and Jeannerod, M. 1984. The contribution of coordinated eye and head movements in hand pointing accuracy. Exp. Brain Wes. 55: 462 -469. Blouin, J., Bard, C., Teasdale, N . , Paillard, J., Fleury, M . , Forget, R., and Lamarre, Y. 1993. Reference systems for coding spatial information in normal subjects and a deafkrented patient. Exp. Brain Wes. 93: 324-331. Bock, 0. 1986. Contribution of retinal versus extraretinal signals towards visual localization in goal-directed movements. Exp. Brain Res. 64: 476-482. Bock, O., and Arnold, K. 1993. Error accumulation and error correction in sequential pointing movements. Exp. Brain Res. 95: 111-117. Bock, O . , and Eckmiller, R. 1986. Goal-directed arm movements in absence of visual guidance: evidence for amplitude rather than position control. Exp. Brain Wes. 62: 451-458. Bowditch, H.P., and Soubard, W.F. 1880. A comparison of sight and touch. J. Physiol. (London), 3: 232 -245. Carlton, L.G. 1979. Control processes in the production of discrete aiming responses. J. Hum. Mov. 5 : 115 - 124. Conti, P., and Beaubaton, D. 1976. Utilisation des informations visuelles dans le contr8le du mouvement : Ctude de la prCcision des pointages chez l'homme. Trav. Hum. 38: 19-32. Cordo, P.J. 1990. Kinesthetic control of a multijoint movement sequence. J. Neurophysiol. 63: 161 - 172. Crossman, E.R.F.W., and Goodewe, P.J. 1993. Feedback control of hand-movement and Fitts9 law. Q . J. Exp. Psychol. 3588: 25 1-278. Darling, W.G., and Cooke, J.D. 1987a. Changes in the variability of movement trajectories with practice. J. Mot. Behav. 19: 291 -309. Darling, W.G., and Coske, J.D. 1987b. Movement related EMGs become more variable during learning of fast accurate movements. J. Mot. Behav. 19: 31 1-331. Darling, W.G., and Cooke, J.D. 1987~.A linked muscular activation model for movement generation and control. J. Mot. Behav. 19: 333 -354. Elliott, D., Carson, R.G., Godman, D., and Chua, R. 1991. Discrete vs. continuous visual control of manual aiming. Hum. Mov. Sci. 10: 393 -418. Festinger, L., and Canon, L.K. 1965. Information about spatial location based on howledge about efference. Psychol. Rev. 72: 373 -384. Georgopoulos, A. P. 1986. On reaching. Annu. Rev. Neurosci. 9: 147 - 170. Georgopodos, A.P., Caminiti, R., Kdaska, J.F., and Massey, J.T. 1983. Spatial coding of movement: a hypothesis concerning the coding of movement direction by motor cortical populations. Exp. Brain Wes. Suppl. 7: 327-336. Georgopoulos, A.P., Schwartz, A.B., and Kettner, R.E. 1986. Neuronid population coding of movement direction. Science (FVashington, D. C .), 233: 1416- 1419. Ghez, C., Gordon, J., Ghilardi, M.F., Christakos, C.N., and Cooper, S.E. 1990. Roles of proprioceptive input


in the programming of arm trajectories. Cold Spring Harbor Symp. Quant. Biol. EV: 837 -847. Harasen, R.M., and Skavenski, A. A. 1977. Accuracy of eye position information for motor control. Vision Res. 17: 919-926. Honda, H. 11984. Functional between hand differences and outflow eye position information. Q. J. Exp. Psychol. 70: 189-192. Howarth, C.I., and Bowden, 3.M. 1971. The relationship between speed and accuracy of movement aimed at a target. Acta Psychol. 35: 207 -2 18. Jeannerod, M. 2988. The neural and behavioural organization of goal-directed movements. Clarendon Press, Oxford. Jeannerod, M. 1989. The representation of the goal of an action and its role in the control of goal-directed movements. In Computional neurosciences. Edited by E.L. Schwartz. MIT Press, Cambridge, Mass. Jeannerod, M. 1991 . The interaction of visual and proprioceptive cues in controlling reaching movements. In Motor control: concept and issues. Edited by D.R. Humphrey and H. J. Freund. Wiley and Sons Etd., New York. Jeannerod, M., and Prablanc, C. 1983. The visual control of reaching movements. In Motor control mechanisms in man. Edited by 3.E. Desmedt. Raven Press, New York. pp. 13-29. Kawato, M. 1991 . Optimization and learning in neurd networks for formation and control of coordinated movement. Hn Attention and performance. Vol. 14. Edited by D.E. Meyer. Lawrence Erlbaum, Hillsdde, N .J . Larisk, D.D., Volp, C.M., and Wallace, S.A. 1984. An empirical note of attaining a spatial target after distorting the initial conditions of movement via muscle vibration. J. Mot. Behav. 15: 15 -83. Marteniuk, R.G. 1978. The role of the eye and head positions in slow movement execution. IH Information processing in motor control and learning. Edited by G. Stelmach. Academic Press, London, New York. pp. 267-288. Matin, L., Picoult, E., Stevens, J.K., Edwards, M.W., Young, D., and MacArthur, R. 1982. Bculoparalytic illusion: visual-field dependant spatial mislocalizations by humans partially paralysed with curare. Science (Washington, D.C.), 216: 198 -281. Meredith, M.A., and Stein, B.E. 1986. Visual auditory and somatosensory convergence on cells in superior colliculus results in multisensory integrations. J . Neurophy siol . 56: 640 -662. Meyer, D.E., Keith-Smith, J.E., Kornblum, S., Abrams, R., and Wright, C .E. 1990. Speed - accuracy tradeoffs in aimed movements: toward a theory of rapid voluntary action. In Attention and performance. Vol. 13. Motor representation and control. Edited by M. Jeannerod. pp. 173 -226. Orliaguet, J .P. 1985. Dominance visuelle ou proprioceptive lors de la perception de la position d'un bras chez l'enfant. C A . Psychol. Cognit. 5: 609 -618. Paillard, J. 1980. The multichanneling of visual cues and the organization of a visually guided response. In

Tutorials in motor behavior. Edited by G.E. Stelmch and 3. Recgaain. North-Holland Publishing Company, Amsterdam. pp. 259 -279. Paillard, J. 1982. The contribution of peripheral and central vision to visudly guided reaching. In Analysis of visual behavior. Edited by D.J. Ingle, M. A. Goodale, and R.J. W. Mansfield. MPT Press, Cambridge, Mass. pp. 367-382. Paillard, J . 1991. Motor and representational framing of space. In Brain and space. Edited by J. Paillard. Oxford University Press, Oxford. pp . 163 - 182. Paillard, J., and Amblard, B. 1985. Static versus kinetic visual cues for the processing of spatial relationships. In Brain mechanisms and spatial vision. Edited by M. Jeannerod and D.N. Lee. Martinius Nijhoff, Amsterdam. pp. 299 - 330. Paillard, J., and Brouchon, M. 1974. A proprioceptive contribution to the spatial encoding of position cues for balistic movements. Brain Res . 71: 273 -284. Paulignan, Y., MacKenzie, C., Marteniuk, R., and Jeannerod, M. 199 1 . Selective perturbation of visual input during prehension movements. 1 . The effects of changing object position. Exp. Brain Res. $3: 502-512. PClisson, D. 1987. ktude du geste et de l'orientation saccadique du regard vers une cible visuelle: r8le des boucles internes de rCtrocontr8le. Th2se de doctorat de 32me cycle, UniversitC Claude Bernard, Lyon I. Pklisson, D,, Prablanc, C., Goodale, M.A., and Jeannerod, M. 1986. Visual control of reaching movements without vision of the limb. 11. Evidence of fast unconscious processes correcting the trajectory of the hand to the final position of a double-step stimulus. Exp. Brain Res. 62: 303 -31 1. Polit, A., and Bizzi, E. 1979. Characteristics of motor programs underlying arm movements in monkeys. J. Neurophysiol. 42: 183 - 194. Prablanc, C., and Martin, 0. 1992. Automatic control during hand reaching at undetected two-dimensional target displacements. J . Neurophysiol . 67: 455 -469. Prablanc, C., and PClisson, D. 1990. Gaze saccade orienting and hand pointing are locked to their goal by quick internd loops. Hn Attention and performance. Vol. 13. Motor representation and control. Edited by M. Jeannerod. Erlbaum, Hillsdale, N. J. pp. 653 - 676. Prablanc, C., Echallier, J.E., Jeannerod, M., and Komilis, E. 1979a. Optimal response of eye and hand motor systems in pointing at a visual target. I. Spatio-temporal characteristics of eye -hand movements and their relationships when varying the amount of visual information. Biol. Cybern. 35: 9 13 - 124. Prablanc, C., Echallier, J.E., Jeannerod, M., and Komilis, E. 1979b. Optimal response of eye and hand motor systems in pointing at a visual target. 11. Static and dynamic visual cues in the control of hand movement. Biol. Cybern. 35: 183 - 187. Prablanc, G., PClisson, D., and Goodde, M.A. 1986. Visual control of reaching movements without vision of the limb. I. Role of retinal feedback of target position in guiding the hand. Exp. Brain Res. 62: 293 - 302.


Can. J. Physiol. Pharrnacol. VoI. 73, 1995

Proteau, L., and Materniuk, R.G. 1993. Static visual information and the learning and control of a manual aiming movement. Hum. Mov. Sci. 12: 515 -536. Redon, C., Hay, L., and Velay, J.-L. 1991. Proprioceptive control of goal-directed movements in man, studied by means of vibratory muscle tendon stimulation. J. Mot. Behav. 23: 101 - 108. Roll, R., Bard, C., and Paillard, %.1981. R8le des mouvements ~Cphalogyressur la prCcision d'un ajustement visuo-moteur chez l ' h o m e . J. Phy siol. (Paris), 77: MA. Roll, R., Bard, C., and Paillard, J. 1986. Head orienting contributes to the directional accuracy of aiming at distant targets. Hum. Mov. Sci. 5 : 359 -371. Rossetti, Y., Paulignm, Y., Boisson, D., Rode, G., and Prablanc , D . 1991. Pointing movements in hemiparetic patients. Innovation and technology in biology and medicine, 12: 22 -29. Rossetti, Y., Tadary, B., and Prablanc, C. 1994s. Optimal contribution of head m d eye positions to spatial accuracy in man tested by visually directed pointing. Exp. Brain Res. 97: 487 -496. Rossetti, Y., Meclder, C., and Prablanc, C. 1994b. Is there an optimal arm posture? Deterioration of finger locdization precision and comfort sensation in extreme arm-point postures. Exp. Brain Res. 99: 131- 136. Rossetti, Y.,Stehack, G . , Desmurget, M., Prablanc, C., and Jeannerod, M. 1 9 9 4 ~ The . effect of viewing the static hand prior to movement onset on pointing kinematics and variability. Exp. Brain Res. 181: 323 -330.

Schmidt, R.A., Zelznik, H., Hawkins, B., Franks, J.S., and Quinn, J.T. 1979. Motor output variability: a theory for the accuracy of rapid motor acts. Psychol. Rev. 86: 415-451. Schmidt, R.A., Sherwmd, D.E., Zelaznik, H.W., and Leikind, B.J. 1985. Speed -accuracy trade-offs in motor behavior: theories of impulse variability. In Motor behavior, programming, control and acquisition. Edited by H. Heuer, U. Keinbeck, and K.H. Schmidt. pp. 79-123. Snedecor, G .W., and Cochran, W .G. 1984. Statistical methods. 6th ed. French translation by HH. Boelle and E. C d a j i . ACTA (Association de Coordination Technique Agricole) , Paris. p. 649. (Original edition: Iowa State University Press, Ames, Iowa. 195'7.) Soechting, J.F., and Flanders, M. 1989s. Sensorimotor representations for pointing to targets in three-dimensional space. J . Neurophysiol. 62: 582 -594. Soechting, J.F., and Flanders, M. 1989b. Errors in pointing are due to approximations in sensorimotor transformations. 9. Neurophysiol. 62: 595 -608. Velay , J.-E. 1984. Sens de la position: les traitements sensori-moteurs des informations proprioceptives. Th8se de Doctorat de 3kme cycle, UniversitC de Provence, Aix-Marseille II. Velay , J. -L., and Beaubaton, D. 1986. Influence of visual context on pointing movement accuracy. Cah . Psychol . Cognit. 6: 447 -456. Wann, J.P., and Ibrhim, S.F. 1992. Does limb proprioception drift? Exp. Brain Res. 91 : 162- 966.