Development of visually guided hand orientation in reaching

JOURNAL

OF EXPERIMENTAL

CHILD

PSYCHOLOGY

38, 208-219 (1984)

Development of Visually Guided Hand Orientation in Reaching CLAES VON HOFSTEN AND SHIRIN FAZEL-ZANDY University

of Uppsala

The development of an ability to use vision in adjusting the hand and the fingers to the orientation of an object to be grasped was studied in a group of 15 infants. They were 18 weeks at the first session and were seen at 4-week intervals until 34 weeks old. At each session they were presented with horizontal and vertical rods. The orientation of the hand of the infant when reaching for these rods was measured at each 60-msec interval during the last 540 msec of the approach. It was found that even at the youngest age there were signs of adjustment of the hand to the orientation of the object. However, at that age the adjustments were rather incomplete. During the months that followed there was a rapid improvement in the skill studied. The findings were in accordance with the idea that information about object orientation is accessible to the manual system when infants start reaching for objects but that the system has yet to be tuned and calibrated before functioning adequately.

Prehension is a highly visual skill. Vision enters not only in determining where in space the approach should be directed but also in adjusting the hand and the digits to the size, form, and orientation of the object before any contact is made. Such adjustments will secure a smooth and efficient grasp. When infants at about 4 months of age reliably start to contact objects reached for, they cannot yet grasp them very well. Contact is often made with the back of the hand, and grasping, if any, is slow and awkward. In a longitudinal study of reaching for moving objects, von Hofsten and Lindhagen (1979) found that although the object was frequently contacted at 15 weeks of age, it always slipped out of the hand and was lost during or before the infant’s attempts to grasp it. At this age, the arm transport is also rather unstable and awkward (von Hofsten, 1979). Von Hofsten (1979) found that smooth grasping does not appear until after the arm transport has been stabilized at 5 to 6 months of age. These observations fit well with anatomical findings by Kuypers (1962, 1964) on the primate motor system. He found that the mechanisms conRequests for reprints should be sent to Dr. Claes von Hofsten, Department of Psychology RMhusesplanaden 2, S-90247 Umei, Sweden. 0022-0965/84 $3 .OO Copyright 0 1984 by Academic Press. Inc. All rights of reproduction in any form reserved.

208

HAND

ORIENTATION

IN

REACHING

209

trolling the movements of the arm are different from those that control the fine movements of the hand. Distinct changes take place in these systems after birth (Kuypers, 1962). The “proximal” motor system of the arm which is mainly organized at brainstem level generally matures ahead of the “distal” motor system of the hand which is cortically organized. Lawrence and Hopkins (1972) studied infant rhesus monkeys and found directed arm movements to locations of objects but no independent hand or finger movements before 3 months of age. If the ability to control independent hand and finger movement develops only after the infant has started to contact objects successfully the question arises as to how early grasping of objects is controlled. The grasp reflex that can already be observed in the neonate is tactually elicited (Twitchell, 1970). It has therefore been argued that the early grasping of objects reached for is tactually controlled too (White, Castle, & Held, 1964). However, White et al. (1964) made no attempt to test this hypothesis. At the moment, there exist no clear data that could answer the questions of how early grasping is controlled or when infants start to be able to use vision in adjusting the hand and the fingers to the properties of the object to be grasped. The only serious attempt to address these questions was made by Lockman, Ashmead and Bushnell (1984). They measured the adjustment of hand orientation to a vertical and a horizontal object during reaching in a group of Smonth-olds and a group of 9-month-olds. To ensure a smooth grasp, it is important that the orientation of the hand is adjusted to the orientation of the object. Ideally, at the moment of contact, the hand should be aligned with the object in such a way that the fingers are perpendicular to the long axis of the object. Lockman, Ashmead, and Bushnell (1984) found clear evidence of anticipatory hand adjustment only in the 9-month-olds. The Smonth-olds showed a trend in the right direction but this was not found to be statistically significant. Thus, it is still possible that Smonth-olds are able to make visually guided adjustments of hand orientation but not in such a clear way as to be revealed by the very crude measurements made by Lockman et al. (1984). Judgments of hand orientation from vertical to horizontal were made directly from the videoscreen on a 4-unit scale. The purpose of the present study was to pursue the questions of visual guidance of hand orientation further. More refined measurements were used to make it possible to identify also less distinct adjustments of hand orientation. A longitudinal approach was chosen to make it possible to study the development of this ability too.

METHOD Subjects. Eighteen full-term, healthy infants were tested. They were all born in October 1981. Three of the subjects were eliminated from the study due to moving away from Uppsala (two cases) and fussing at more

210

VON HOFSTEN

AND FAZEL-ZANDY

than one session (one case). Thus, seven females and eight males completed the longitudinal program. The infants were 18 weeks old at their first visit and were seen every fourth week until 34 weeks of age (all ages + 1 week). One subject (J.S.) missed the second session and one (C.L.) missed the first and the third (see Table 1). The subjects came mostly from middle class families. Parents were paid 40 SwCr (about $7) for each session. Apparatus and stimuli. The infant sat comfortably in a semireclining seat (an ordinary baby chair) where he or she could freely move his or her arms. The seat was placed on a table to which the stimulus device was attached. To allow three-dimensional analysis of hand movements, the situation was recorded by two Sony AVC-3250 CE television cameras placed at right angles to one another, one above and one in front of the infant. The two pictures were fed into a Sony 3760 CE video recorder via a mixer and a digital clock giving the time on TV screen in milliseconds. The resulting picture is shown in Fig. 1. The object was a rod 12 cm long and 5 mm in diameter. It was made of aluminum and covered with brightly colored woolen cloth, red in the middle section (6 cm) and blue on each side of that (3 cm). The stimulus rod was perpendicularly attached at its center to a horizontally oriented metal rod around which it could be turned either to a horizontal or a vertical position. The whole device could further be moved around a vertical axis 70 cm from the stimulus rod. Procedure. The infant was brought into the laboratory by a parent (or parents), who stayed there during testing out of sight of the child. When alert and quiet, the infant was placed by the parent in the infant seat, and the height and distance of the object were adjusted. For each infant

J

FIG.

1. The two views of the infant as seen from the video.

HAND

ORIENTATION

IN

REACHING

211

the center of the rod was placed at nose height at a distance comfortable for reaching (about 12 cm). The rod was presented 12 times to each infant, 4 times on the subject’s midline, 4 times 10 cm to the right, and 4 times 10 cm to the left of the midline. At each position the rod was horizontal on two trials and vertical on two trials. The order of presentation was random. Each presentation started when the infant’s arms rested in his or her lap. If the infant did not reach for the object when presented to him or her, the experimenter tried to attract the infant’s attention to the object by shaking it or by moving it across the infant’s field of view. When the infant had reached for the rod or the rod had remained in front of the infant for more than 2 min, the next condition was presented. If the infant cried or fussed a pause was made and the parent was asked to try to soothe the infant. Pacifiers were discouraged, but were allowed if nothing else would keep the infant quiet in the seat. However, most infants were interested in the task and reached for the object as soon as it was placed in front of them. The whole session usually took less than 5 min. Measurement of hand orientation. The purpose of measuring hand orientation was to find out whether the orientation of the hand was adjusted toward the orientation of the rod before contacting it. Reaches have different durations. Sometimes, especially if the infant is moving much, it may even be difficult to determine when a reach begins. However, the final approach toward the target is usually rather regular and takes around half a second. Measurements were therefore confined to this part of the reach. It was done in the following way: First, for each reach to be evaluated, the time of contact with the rod was determined from inspection of the upper as well as the frontal videoview of the reach. The videotape was then run backward to a point 540 msec earlier in time where measurement of hand orientation was to start. Hand orientation was always measured from the frontal video view (see Fig. 1). The position on the screen of the base joint of the index finger as well as the base joint of the little finger was read into a computer with the aid of an XY-reader. The XY-reader had a precision of about 0.6 mm in the horizontal and 0.5 mm in the vertical dimension. From the two points in the frontal plane so determined, hand orientation was calculated. The procedure was repeated at every 60-msec interval up to the time of contact between hand and rod making altogether 10 readings of hand orientation for each reach. If a reach had a shorter duration than 540 msec or if the hand was occluded during the first part of the reach, fewer than 10 readings were secured for that reach. RESULTS

As each condition was presented to the infant twice, there were at most 12 reaches to be analyzed for each infant and visit. However,

212

VON

HOFSTEN

AND

FAZEL-ZANDY

sometimes infants did not reach for the object or made incomplete movements toward it. Only reaches where the hand finally contacted the object were considered. There were altogether 745 such cases. Out of those, all reaches were omitted where the subject used the hand contralateral to the side to which the object was placed. It turned out to be difficult to analyze such reaches as the hand then did not move toward the video camera ahead but rather across the field of view. There were 79 such cases, or 22, 17, 9, 15, and 16 at 18, 22, 26, 30, and 34 weeks of age, respectively. Finally 13 other cases of nondecodable reaches where the reaching hand was occluded by the other arm or by the apparatus itself were omitted as were 3 cases where the subject just touched the end of the rod with thumb and index finger. Thus, a total of 650 reaches were analyzed, distributed over subjects, ages, and object orientation as shown in Table 1. Ideally, 10 readings were secured per reach. This was true in 590 cases. In the remaining 60 cases, the number of readings per reach were fewer or 9, 8, 7, 6, 5, 4, 3, and 2 in 12, 7, 11, 13, 10, 5, 1, and 1 cases, respectively. The reliability of measurement was evaluated by a split-half productmoment correlation method on all subjects at all ages. This was done by correlating the even readings of each reach with the uneven ones, i.e., the first reading of a reach was correlated with the second, the third with the fourth, etc. This correlation was found to be .93.

THE DISTRIBUTION

TABLE I OF MEASURED REACHES OVER SUBJECTS, AND TARGET ORIENTATIONS Age

18

AGES,

(weeks)

22

26

30

34

Subject

V

H

V

H

V

H

V

H

V

H

E.A. I.L. A.T. D.F. L.A. C.S. D.B. A.B. A.H. E.M. CL. K.B. D.M. G.B. J.S.

0 2 4 2 2 3 6 4 0 0

0 1 3 3 3 3 3 4 0 2

6 6 6 5 5 5 5 6 4 5

3 4 0 3

2 6 5 6 4 5 4 4 3 4 4 5 6 3

6 6 6 6 5 4 5 5 3 4

2 4 0 4

3 5 6 5 5 5 6 5 2 4 4 5 5 5

4 6 6 5

4 5 6 4

6 6 6 6 5 6 6 4 6 5 6 5 4 6 6

5 3 6 5 2 5 6 6 6 4 6 6 5 6 5

4 6 5 5 6 6 5 6 6 5 4 6 4 5 6

4 5 6 6 6 6 4 5 5 5 6 6 4 5 4

Nore.

--

V, vertical;

H,

-horizontal.

--

HAND

ORIENTATION

213

IN REACHING

In Fig, 2 the progression of hand orientation over time for reaches aimed at vertical as well as horizontal targets is shown for each age level of the study. It can be seen from Fig. 2 that the difference in hand orientation at the end of the approach is in the direction of target orientation for all the age levels of the study. Figure 2 also shows that, for all age levels except at 22 weeks, hand orientation is on the average in the direction of target orientation for the whole measured part of the reach. Finally, it can be seen from Fig. 2 that the mean adjustment of hand orientation to target orientation improves during the measured part of the reach for all age levels except at 18 weeks. An analysis of variance was performed for each age level of the study. As the number of reaches per subject were not always the same, these analyses were based on subject means rather than individual reaches. Further, in the analyses, the sequence of 10 readings per reach was

18 W&I

-

22

270

380

310

30

WeeLI

iso’

270

380

510

TIME*I.~Ec)

IO

150

270

390

510

TIMEWSEC)

FIG. 2. The progression of hand orientation during the last 540 msec of the approach for reaches aimed at vertical and horizontal targets for each age level of the study. Two consecutive readings make up each data point. H denotes reaches for a horizontal object and V reaches for a vertical object (0” is vertical, 90” is horizontal).

214

VON HOFSTEN

AND FAZEL-ZANDY

divided into subgroups of two consecutive readings each, treating the two readings as replications. The main effect of target orientation was found to be significant on the 1% level for all age levels except at 22 weeks. The interaction between target orientation and time was found to be significant on the 1% level at 22, 30, and 34 weeks and significant on the 5% level at 26 weeks (see Table 2). Separate analyses of variance were also performed on the data from the last two readings of each reach. In these ANOVAs, the main effect of target orientation was found to be significant on the 1% level for all age levels of the study (see Table 2).

Individual results are shown in Fig. 3. The dependent variable in these graphs is mean difference in hand orientation between reaches for horizontal and vertical objects at a certain time during reaching. A positive score denotes a difference in the direction predicted by object orientation. A score increasing over time means that the hand rotates toward the orientation of the object during the approach. Two consecutive readings make up each data point in Fig. 3. Curves with unfilled circles are the pooled result from the 18- and 22-week sessions and curves with filled circles are the pooled results from the 30- and 34-week sessions. As can be seen from Fig. 3, all infants showed at least some signs of adapting hand orientation to object orientation. At the younger age levels 11 of the 15 subjects had a difference in hand orientation in the direction of object orientation toward the end of the approach and at the older age levels this was true for all subjects. Differences in hand orientation in the direction of object orientation were found for the whole measured part of the reach in 4 cases at the younger ages and in 13 cases at the older age levels. Analyses of variance were carried out on each subject on the study. In 11 cases the predicted main effect of target orientation was found to be significant on the 1% level and in one case on the 5% level (see Table 3). TABLE F SCORES FOR THE ANALYSES

4s (weeks)

Orientation (whole sequence)

18 22 26 30 34

89.5 (1,80)** Reverse effect 778.0 (1,140)** 147.0 (1,150)** 838.9 (1,150)**

2

OF VARIANCE

BY

Orientation X time (whole sequence) 1.29 19.1 2.61 26.7 13.0

(4,gO)n.s. (4,140)** (4,140)* (4,150)** (4,150)**

AGE LEVEL Orientation (last 120 msec) 42.2 44.2 299.9 800.3 639.6

(1,16)** (1,28)** (1,28)** (1,30)** (1,30)**

Note. Analyses were carried out for the whole recorded movement sequence as well as for the last two readings; n.s., not significant; df in parentheses. * Significant on 5% level. ** Significant on 1% level.

HAND ORIENTATION

215

IN REACHING

____-___._______ L- A.T. 1, , , ,y.-

30

270

610

30

270

E A.

30’

510

TIMEWSEC)

+30 +20 “-J+ t10 0_._._.__ ___ ___.-.-10 b-i+ 0.F. 40 L-

.d.

_ _ _ _ __ _ _ __ _ _ _ __ _ _

E.M.

0.0. e

E

30

270

510

30

270

510

30

270

610

30

270

510 TIMEOASEC)

t30 y +20 +I E +10 c-+---y P 0I_.._.. .... . ._ ______

: 5 -10 E -20 -Y kYL 0 30

//-i

ILL.

0.M D

270

30

610

270

510

E

30

270

AH. e

610

TIME(MSEC)

0.0.

30

270

610

30

270

510

30

270

640 TIMEWSEC)

FIG. 3. Mean difference in hand orientation (degrees) between reaches for horizontal and vertical objects at different times during the last 540 msec. of the approach. A positive score denotes a difference in the direction predicted by object orientation. Curves with unfilled circles are the pooled result from the 18- and 22-week sessions and curves with tilled circles are the pooled results from the 30- and 34-week sessions. Each diagram depicts the results from one subject. Two consecutive readings make up each data point.

216

VON HOFSTEN

AND FAZEL-ZANDY TABLE

F SCORESFOR

THE ANALYSES

3 OF VARIANCE

BY SUBJECT ---

Subject A.B. A.T. C.S. E.A. D.F. E.M. J.S. G.B. I.L. L.A. D.M. A.H. K.B. C.L. D.B.

Orientation 1000.6 (1,50)** 943.8 (1,50)** 173.4 (1,50)** 3.6 (1,40)n.s. 199.4 (1,50)** 298.5 (1,40)** 194.9 (1,40)** 10.6 (1,40)** 4.3 (1,40)* 83.9 (1,50)** Reverse effect 85.3 (1,40)** 114.6 (1,40)** Reverse effect 145.7 (1,50)**

Orientation

X time

3.2 (4,50)* 3.1 (4,50)* 1.3 (4,50)n.s. 1.1 (4,40)n.s. 13.8 (4,50)** Reverse effect 27.4 (4,40)** 5.4 (4,40)** 0.4 (4,40)n.s. 10.1 (4,50)** 13.8 (4,50)** 5.2 (4,40)** 4.1 (4,40)** 5.3 (4,30)** 5.3 (4,50)**

Orientation

-

X age

189.8 (4,50)** 8.9 (4,50)** 12.8 (4,50)** 23.9 (3,40)** 86.4 (4,50)** 99.4 (3,40)** 6.2 (3,40)** Reverse effect 152.2 (3,40)** 50.6 (4,50)** Reverse effect Reverse effect 85.2 (3,40)** 52.3 (2,30)** 65.5 (4,50)**

Note. ns., not significant; df in parentheses. * Significant on 5% level. ** Significant on 1% level.

As indicated above, the results improved with age. In 8 cases, the results for the older age level is at every data point above the results for the younger age level and in 12 cases this is at least true for the last data point, just before touch. The analyses of variance showed that the interaction between target orientation and age in the direction of a closer fit between hand and target orientation was significant on the 1% level in 12 cases (see Table 3). Finally, Fig. 3 shows that, at least for some subjects like J.S., the correspondence between hand and target orientation improved considerably during the measured part of the reach. The last data point of the approach is above the first one in 11 cases at the younger age level and in 13 cases at the older age level. The analyses of variance showed that the interactions between target orientation and time in the direction of a closer fit between hand and target orientation toward the end of the approach were significant on the 1% level in 9 cases and on the 5% level in 2 cases (see Table 3). It was not clear from the present study whether the ability to carry out adjustments of hand orientation to target orientation during the approach improveswithage.Insixcasestheolderagelevelshowsagreateradjustment during the measured part of the reach and in nine cases it is the younger age level that adjusts more. However, in several cases, like for A.B., the hand is already well adjusted to target orientation where measurement

HAND

ORIENTATION

IN

REACHING

217

starts and there is then obviously less need for further adjustment during that part of the approach. The adjustment of hand orientation to object orientation is necessary but not sufficient for securing a smooth grasp. It is also important that the object is encountered with the palm side of the hand. In the present study it was frequently observed in the younger infants that even when they had turned the hand in the correct direction the object was encountered with the back of the hand. The percentage of instances where the object was first touched with the back of the hand or back of the fingers was 26, 33, 21, 11, and 4 in the 18, 22, 26, 30, and 34-week-old infants, respectively. DISCUSSION The results show that visual information about object orientation is already accessible to the manual system when infants start grasping objects. When the 18-week-old infants reached for the rod they adjusted their hand orientation, although crudely, to the orientation of the rod before touching it. The results do not lend support to the hypothesis that tactual guidance of grasping precedes visual guidance in the development. This does not mean that tactual guidance is unimportant. At this age, most adjustments are done after the object has been contacted and, undoubtedly, the haptic system is an important determiner of those adjustments. The adjustments of the hand to the orientation of the object became more precise with age. This development reflects the maturing distal motor system of the hand. During the same period as that of the present study, the whole grasping act becomes much better controlled and under increasing influence of visual guidance. Whereas at 18 weeks of age infants will only grasp objects in a crude and awkward way with their whole hand, they will at 34 weeks start to be able to pick up small pellets with their thumb and index finger (Gesell, 1928). However, it is not only the grasping act itself that becomes better controlled during the period under study. During the earlier half of the period there is a dramatic stabilization of the motor system of the arm, resulting in smooth straight approaches toward the object (von Hofsten, 1979). This must also affect the infant’s ability to control the movement of the hand during the approach. One important source of error was revealed through the examination of individual reaches. The task did not always turn out to be quite as simple and straightforward as the authors had intended. The rod was supposed to evoke a grasping response in which the fingers would close around the rod, securing it in the hand. This was also the behavior observed in the great majority of cases. However, the rod also seemed to elicit other kinds of behavior like swiping for it, or tapping the end of it with the whole hand, or reaching for the end of the rod and securing

218

VON

HOFSTEN

AND

FAZEL-ZANDY

it with a pincette grasp. The last type of act was especially apparent in a few of the older infants. Three reaches were omitted for that reason; but in most cases the subject changed strategy during the encounter with the object and finally grasped it with the whole hand. In those cases it was impossible to know whether the first behavior was a mistake or if the intention was different. Thus, such reaches were included in the data treatment and constituted a source of error, not so important as to conceal the major trends, but large enough to distort some of the individual results. Adjustment of the hand before or during the early part of the reach seems to be at least as important as adjustments done during the measured later part of it. Indeed, it could often clearly be observed from the videoscreen how subjects rotated their hand in the appropriate direction before starting the approach. This appears to be a good strategy in the present situation with a stationary rod to be reached for. The motor control task during the approach then becomes simpler as there will be one parameter less to keep track of. This advantage will, of course, be lost if the object to be reached for is moving or animate. In such a situation, the subject has to be prepared for changes in the orientation of the object to be reached for during the approach of the hand. Therefore, to test the subject’s ability to adjust hand orientation to object orientation during the approach requires a situation where the orientation of the object may change during the subject’s approach toward it. The obtained results do not reflect any systematic change in strategy of hand adjustment over age. The strategy of adjusting the hand to object orientation early in the reach or before it starts and the strategy of doing the adjustments during the later part of the reach are both represented in individual results at the older as well as at the younger age levels. Rather the two different strategies seemed to be preferred by different subjects who even retained them over age sometimes. For instance, J.S. made most of his adjustments during the measured part of the reach at both age levels shown in Fig. 2, while A.B. and E.M. made their adjustments predominantly before that part of the reach. Infants’ acquisition of the ability to visually control the delicate movements of the hand is an important development. It marks the first step toward mastery over objects and tools. With the exception of Halverson’s (1931, 1933, 1937) classical descriptive studies very little is known about this development. We need more precise and quantitative data on when and how vision takes over the control of the hand. The present study is just the first step toward such knowledge. The next question to be answered is how vision starts to guide the adjustments of the hand to the form and size of the object to be grasped. We also need to know how experience enters into this development. Infants spend considerable time looking at their hands. Even if some of the basic aspects of eye-

VON HOFSTEN

219

AND FAZEL-ZANDY

hand coordination are preformed, experiencemay play a crucial role for the differentiation of manual motor skill. REFERENCES Gesell, A. (1928). Infancy and human growth. New York, McMillan. Halverson, H. M. (1931). Study of prehension in infants. Genetic Psychology

Monographs,

10, 107-285. Halverson, H. M. (1933). The acquisition of ski11in infancy. Journal ofGenetic Psychology, 43, 3-48. Halverson, H. M. (1937). Studies on the grasping responses of early infancy. I. Journal of Genetic

51, 393-424.

Psychology,

Hofsten, C. von. (1979). Development Journal

of Human

Movement

of visually directed reaching: The approach phase.

Studies,

5, 160-178.

Hofsten, C. von, & Lindhagen, K. (1979). Observations on the development of reaching for moving objects. Journal of Experimental Psychology, 28, 158-173. Kuypers, H. G. J. M. (1962). Corticospinal connections: Postnatal development in the rhesus monkey. Science, 138, 678-680. Kuypers, H. G. J. M. (1964). The descending pathways to the spinal cord, their anatomy and function. In J. C. Eccles (Ed.), Organization of the spinal cord. Amsterdam: Elsevier. Lawrence, D. G., & Hopkins, D. A. (1972). Developmental aspects of pyramidal motor control in the rhesus monkey. Brain Research, 40, 117-118. Lockman, J. J., Ashmead, D. H., 8z Bushnell, E. W. (1984). The development of anticipatory hand orientation during infancy. Journal of Experimental Child Psychology, 37, 176186. Twitchell, T. E. (1970). Reflex mechanisms and the development of prehension. In K. Connolly (Ed.), Mechanisms of motor skill development. New York/London: Academic Press. White, B. L., Castle, P., & Held, R. (1964). Observations on the development of visually directed reaching. Child Development, 35, 349-364. RECEIVED:

February 4, 1983;

REVISED:

June 17, 1983; July 25, 1983.