Perceptualand Motor Skills, 1992, 75, 375-384. O Perceptual and Motor Skills 1992
CHILDREN'S DEVELOPMENT O F REACHING: TEMPORAL AND
SPATIAL ASPECTS OF AIMED WHOLE-ARM MOVEMENTS ' CECILE FAYT, NICOLE SCHEPENS, AND MARC MINET Loboratory of Neurophysiology Universio of Louvain
Summary.-Age-related changes in movement time and spatial distribution of pointing errors were investigated using a whole-arm target-aimed task. 60 children (6 to 11 yr.) and 10 young adults were required to reach towards targets located on a vertical screen in four conditions: target bt for 4 sec. in light (I), in darkness (2), target lit for 200 msec. in light (3), in darkness (4). Accuracy was perfect in Conditions 1 and 2, with a significant increase of MT in Condition 2 between the ages 7 and 10. Accuracy decreased slightly in Condition 3 and strongly in Condition 4 , despite a similar shortening of MT. In Condition 4 the subjects undershot the mrget position: horizontally from age 8 and vertically in all age groups, with an increase of vertical bias by girls at age 9. These results suggest age-related changes in computation of arm movement towards the target.
The reaching behavior has been extensively investigated in human adults; see Jeannerod (1988) for a review. Two successive phases are usually described in the execution of aimed movements: an essentially ballistic approach phase, performed on the basis of an initial programming of the movement, and a terminal visually guided phase, allowing adjustments of the trajectory. From a developmental viewpoint, the ontogenetic priority of the approach phase was observed in babies (von Hofsten, 1982). Wishart, Bower, and Dunkeld (1978) observed the emergence of visual guidance from 5 months of age. However, adult-like reaching behavior is not achieved during infancy, and childhood seems to be an important period of maturation for movement control. Indeed, the results of Hay (1978, 1979) suggested a nonmonotonous development of motor-control components during childhood: In a visual open-loop pointing task, Hay (1978) observed a degradation of accuracy around 7 years of age. Kinematic data of the movement showed a majority of ballistic-like velocity profiles at age 5 , an important braking activity at age 7, and a majority of asymmetrical velocity profiles with a prolonged terminal deceleration phase from age 10 (Hay, 1979). Taken together (Hay, 1978, 1979, 1981), these results suggested "a predominance of the programming system at 5 yr. and guiding system at 7 yr., followed at 9 and 11 yr. by integration of both systems" (Hay, 1979, p. 189). 'We thank Roland Maire for help in data collection and analysis and Christian Stoquart for technical assistance. Cicile Fayt is a Research Assistant of the National Fund for Scientific Research of Belgium. Requests for reprints should be sent to CCcile Fayt, Laboratory of Neur~ph~siology, University of Louvain, 54 Avenue Hippocrate, 1200 Brussels, Belgium.
376
C. FAYT, ETAL
Evidence for a discontinuity in the development of processes underlying motor control was also reported by von Hofsten and Rosblad (1988). In their study, children (4 to 12 yr.) had to reach for dots located on a horizontal table with a pin, moving the hand underneath the table. When the child placed the index finger of the other hand on the dot and laced the pin underneath it with the eyes closed (proprioceptive condition), the performance improved between the ages of 5 and 8 years. The authors related this result to those of Hay (1978): The development of a precise control of hand movements between the ages of 5 and 8 years implies probably that during this period the child focuses on the corrective phase of reaching. Nevertheless, in children who pointed to visual targets located in the vertical plane, Brown, Sepehr, Ettlinger, and Skreczek (1986, 1987) found a roughly linear increase in accuracy between the ages of 1.5 and 8 years, whatever the availability of visual information; the authors concluded that the development of the programming and of guiding phases of aimed movements were continuous and parallel. In fact, the latter authors graded the pointings as a function of their position in predetermined concentric areas around the target, but did not consider localization of error in oriented space. However, for the same error defined as above, the subject can overshoot or undershoot the target position. I n our experiment, we studied whether there was a systematic bias in pointing positions and whether it changed with age. We assumed that changes with age in spatial distribution of error could reflect changes in the relative importance of approach and adjustment components in motor control. This hypothesis was tested in a whole-arm target-aimed task, similar to that of Brown, et al. (1986) with the exception that the subject pointed with a stylus instead of the index finger. Spatial errors were studied in their horizontal and vertical components on the frontal plane where the targets were presented. The task was performed in different visual and nonvisual conditions by 6- to 11-year-old children and a reference group of young adults.
METHOD Subjects Sixty healthy children from 6 to 11 years old and ten healthy young adults participated in the experiment. They were assigned by age in seven groups of ten subjects each, the two sexes being equally represented in each group. The mean ages were (in years and months) 6.1, 7.0, 8.1, 8.11, 10.1, 11.1, and 22.0. All the subjects were right-handed for handwriting and drawing and had normal or corrected to normal vision. Experimental Setup The subject sat in front of a vertical frame consisting of a transparent digitizing surface (Calcomp 640) joined with a panel supporting light-emit-
AIMED WHOLE-ARM MOVEMENTS IN CHILDREN
377
ting diodes (LEDs), 3 mm in diameter. Three of the LEDs-positioned horizontally at eye level-were used as targets: one in front of the subject, and two at left, each one separated from the other by 10 cm. The useful part of the digitizing surface was delimited by a thin paper border. The distance between the subject and the surface was adapted for each subject to have the arm half-stretched when the tip of the stylus touched the surface. The experiment was controlled by an IBM AT13 computer. The subject pointed with the right hand, holding a stylus as one holds a pencil. The armrest at the right of the subject was equipped with a microswitch for recording time parameters (see below).
Procedure The subject was instructed to maintain a stable posture of the trunk. No restrictions were imposed on head and eye movements. At the beginning of a trial, the right hand rested o n the armrest, depressing the microswitch. One of the experimenters, seated near the device, warned the subject that a target would soon be Illuminated. The subject was instructed to point the stylus as accurately as possible to the target, without any speed constraint. Each target was randomly presented three times by condition. The four conditions were (1)target lit for 4 sec. with the room lights on (visual control condition), (2) target lit for 4 sec. in darkness (the tip of the stylus became visible in the immediate proximity of the target, making this a terminal visual feedback condition), (3) target lit for 200 msec. with the room lights on (vision of movement but not of target), (4) target lit for 200 msec. in darkness (nonvisual condition). O n the basis of preliminary observations with a group of 6-year-old children, it appeared necessary to run the experimental conditions in an increasing order of predictable difficulties, i.e., 1, 2, 3, 4. There was a small rest after the second condition (a few minutes with the room lights turned on). Parameters of the Movement Reaction time (RT) was the time from presentation of the target to the release of the microswitch. This parameter was only used to eliminate trials on which KT was shorter than 200 msec. (less than 5% of trials by a subject), ensuring the validity of Conditions 3 and 4. Movement time (MT) was the time from the release of the microswitch to the contact of the stylus on the digitizing surface. This contact allowed the recording of the rectangular coordinates (x,y) of the stylus position. Analysis From these rectangular coordinates, the distance between the stylus irnpact point and the target location was calculated as the square root of x2 + y 2 and named distance error (d). O n the other hand, x and y coordinates allowed distinguishing a horizontal bias from a vertical bias in pointing posi-
378
C. FAYT, ETAL.
tions: positive x corresponded to a leftward bias (horizontal overshoot) and negative x to a rightward bias (horizontal undershoot); positive y corresponded to a bias above the target (vertical overshoot) and negative y to a bias under the target (vertical undershoot). The results related to the three targets were confounded. Constant errors (CEd, CEx, CEy) were calculated, respectively, as the mean of distance errors, x and y coordinates by condition and by subject. Variable errors (VEd, VEX, VEy) measured the intraindividual variability of constant errors and were calculated, respectively, as the square root of the variance of the differences between d and CEd, x and CEx, y and CEy (Schmidt, 1988, p. 58). MT mean was calculated as the mean of movement times by condition and by subject. These dependent variables were subjected to separate 7 . 2 . 4 (age . sex . condition) analyses of variance with repeated measures on condition. Follow-up comparisons were conducted with the Duncan's test using p = 0.05.
Constant Errors Distance error (CEd).-A significant main effect was found for age (F,,,, = 2.96, p c 0 . 0 5 ) ; CEd decreased with age and was significantly smaller in lo-, 11-year-old children and adults than in 6-, 7-, 8-, and 9-year-old children (Duncan's test). Condition had a significant effect (F,,,,, = 445.69, p
CHILDREN'S DEVELOPMENT O F REACHING: TEMPORAL AND
SPATIAL ASPECTS OF AIMED WHOLE-ARM MOVEMENTS ' CECILE FAYT, NICOLE SCHEPENS, AND MARC MINET Loboratory of Neurophysiology Universio of Louvain
Summary.-Age-related changes in movement time and spatial distribution of pointing errors were investigated using a whole-arm target-aimed task. 60 children (6 to 11 yr.) and 10 young adults were required to reach towards targets located on a vertical screen in four conditions: target bt for 4 sec. in light (I), in darkness (2), target lit for 200 msec. in light (3), in darkness (4). Accuracy was perfect in Conditions 1 and 2, with a significant increase of MT in Condition 2 between the ages 7 and 10. Accuracy decreased slightly in Condition 3 and strongly in Condition 4 , despite a similar shortening of MT. In Condition 4 the subjects undershot the mrget position: horizontally from age 8 and vertically in all age groups, with an increase of vertical bias by girls at age 9. These results suggest age-related changes in computation of arm movement towards the target.
The reaching behavior has been extensively investigated in human adults; see Jeannerod (1988) for a review. Two successive phases are usually described in the execution of aimed movements: an essentially ballistic approach phase, performed on the basis of an initial programming of the movement, and a terminal visually guided phase, allowing adjustments of the trajectory. From a developmental viewpoint, the ontogenetic priority of the approach phase was observed in babies (von Hofsten, 1982). Wishart, Bower, and Dunkeld (1978) observed the emergence of visual guidance from 5 months of age. However, adult-like reaching behavior is not achieved during infancy, and childhood seems to be an important period of maturation for movement control. Indeed, the results of Hay (1978, 1979) suggested a nonmonotonous development of motor-control components during childhood: In a visual open-loop pointing task, Hay (1978) observed a degradation of accuracy around 7 years of age. Kinematic data of the movement showed a majority of ballistic-like velocity profiles at age 5 , an important braking activity at age 7, and a majority of asymmetrical velocity profiles with a prolonged terminal deceleration phase from age 10 (Hay, 1979). Taken together (Hay, 1978, 1979, 1981), these results suggested "a predominance of the programming system at 5 yr. and guiding system at 7 yr., followed at 9 and 11 yr. by integration of both systems" (Hay, 1979, p. 189). 'We thank Roland Maire for help in data collection and analysis and Christian Stoquart for technical assistance. Cicile Fayt is a Research Assistant of the National Fund for Scientific Research of Belgium. Requests for reprints should be sent to CCcile Fayt, Laboratory of Neur~ph~siology, University of Louvain, 54 Avenue Hippocrate, 1200 Brussels, Belgium.
376
C. FAYT, ETAL
Evidence for a discontinuity in the development of processes underlying motor control was also reported by von Hofsten and Rosblad (1988). In their study, children (4 to 12 yr.) had to reach for dots located on a horizontal table with a pin, moving the hand underneath the table. When the child placed the index finger of the other hand on the dot and laced the pin underneath it with the eyes closed (proprioceptive condition), the performance improved between the ages of 5 and 8 years. The authors related this result to those of Hay (1978): The development of a precise control of hand movements between the ages of 5 and 8 years implies probably that during this period the child focuses on the corrective phase of reaching. Nevertheless, in children who pointed to visual targets located in the vertical plane, Brown, Sepehr, Ettlinger, and Skreczek (1986, 1987) found a roughly linear increase in accuracy between the ages of 1.5 and 8 years, whatever the availability of visual information; the authors concluded that the development of the programming and of guiding phases of aimed movements were continuous and parallel. In fact, the latter authors graded the pointings as a function of their position in predetermined concentric areas around the target, but did not consider localization of error in oriented space. However, for the same error defined as above, the subject can overshoot or undershoot the target position. I n our experiment, we studied whether there was a systematic bias in pointing positions and whether it changed with age. We assumed that changes with age in spatial distribution of error could reflect changes in the relative importance of approach and adjustment components in motor control. This hypothesis was tested in a whole-arm target-aimed task, similar to that of Brown, et al. (1986) with the exception that the subject pointed with a stylus instead of the index finger. Spatial errors were studied in their horizontal and vertical components on the frontal plane where the targets were presented. The task was performed in different visual and nonvisual conditions by 6- to 11-year-old children and a reference group of young adults.
METHOD Subjects Sixty healthy children from 6 to 11 years old and ten healthy young adults participated in the experiment. They were assigned by age in seven groups of ten subjects each, the two sexes being equally represented in each group. The mean ages were (in years and months) 6.1, 7.0, 8.1, 8.11, 10.1, 11.1, and 22.0. All the subjects were right-handed for handwriting and drawing and had normal or corrected to normal vision. Experimental Setup The subject sat in front of a vertical frame consisting of a transparent digitizing surface (Calcomp 640) joined with a panel supporting light-emit-
AIMED WHOLE-ARM MOVEMENTS IN CHILDREN
377
ting diodes (LEDs), 3 mm in diameter. Three of the LEDs-positioned horizontally at eye level-were used as targets: one in front of the subject, and two at left, each one separated from the other by 10 cm. The useful part of the digitizing surface was delimited by a thin paper border. The distance between the subject and the surface was adapted for each subject to have the arm half-stretched when the tip of the stylus touched the surface. The experiment was controlled by an IBM AT13 computer. The subject pointed with the right hand, holding a stylus as one holds a pencil. The armrest at the right of the subject was equipped with a microswitch for recording time parameters (see below).
Procedure The subject was instructed to maintain a stable posture of the trunk. No restrictions were imposed on head and eye movements. At the beginning of a trial, the right hand rested o n the armrest, depressing the microswitch. One of the experimenters, seated near the device, warned the subject that a target would soon be Illuminated. The subject was instructed to point the stylus as accurately as possible to the target, without any speed constraint. Each target was randomly presented three times by condition. The four conditions were (1)target lit for 4 sec. with the room lights on (visual control condition), (2) target lit for 4 sec. in darkness (the tip of the stylus became visible in the immediate proximity of the target, making this a terminal visual feedback condition), (3) target lit for 200 msec. with the room lights on (vision of movement but not of target), (4) target lit for 200 msec. in darkness (nonvisual condition). O n the basis of preliminary observations with a group of 6-year-old children, it appeared necessary to run the experimental conditions in an increasing order of predictable difficulties, i.e., 1, 2, 3, 4. There was a small rest after the second condition (a few minutes with the room lights turned on). Parameters of the Movement Reaction time (RT) was the time from presentation of the target to the release of the microswitch. This parameter was only used to eliminate trials on which KT was shorter than 200 msec. (less than 5% of trials by a subject), ensuring the validity of Conditions 3 and 4. Movement time (MT) was the time from the release of the microswitch to the contact of the stylus on the digitizing surface. This contact allowed the recording of the rectangular coordinates (x,y) of the stylus position. Analysis From these rectangular coordinates, the distance between the stylus irnpact point and the target location was calculated as the square root of x2 + y 2 and named distance error (d). O n the other hand, x and y coordinates allowed distinguishing a horizontal bias from a vertical bias in pointing posi-
378
C. FAYT, ETAL.
tions: positive x corresponded to a leftward bias (horizontal overshoot) and negative x to a rightward bias (horizontal undershoot); positive y corresponded to a bias above the target (vertical overshoot) and negative y to a bias under the target (vertical undershoot). The results related to the three targets were confounded. Constant errors (CEd, CEx, CEy) were calculated, respectively, as the mean of distance errors, x and y coordinates by condition and by subject. Variable errors (VEd, VEX, VEy) measured the intraindividual variability of constant errors and were calculated, respectively, as the square root of the variance of the differences between d and CEd, x and CEx, y and CEy (Schmidt, 1988, p. 58). MT mean was calculated as the mean of movement times by condition and by subject. These dependent variables were subjected to separate 7 . 2 . 4 (age . sex . condition) analyses of variance with repeated measures on condition. Follow-up comparisons were conducted with the Duncan's test using p = 0.05.
Constant Errors Distance error (CEd).-A significant main effect was found for age (F,,,, = 2.96, p c 0 . 0 5 ) ; CEd decreased with age and was significantly smaller in lo-, 11-year-old children and adults than in 6-, 7-, 8-, and 9-year-old children (Duncan's test). Condition had a significant effect (F,,,,, = 445.69, p
