the development of prospective grasping control between 5 and 7 months: a longitudinal study

INFANCY, 7(2), 143-161 Copyright 0 2005, Lawrence Erlbaum Associates, Inc.

The Development of Prospective Grasping Control Between 5 and 7

Months: A Longitudinal Study

David C . Witherington Department of Psychology lJ?iiversily ($New Mexico

By 7 months, infants, when reaching for an object, visually guide their grasp by preorienting their hands to match the object’s orientation. Evidence at earlier ages, however, for prospective grasp control via anticipatory hand orientation is mixed. This study examined longitudinally the development of anticipatory hand orientation in 15 infants, seen every 3 weeks between 5 and 7.5 months. On each visit, infants were given 8 trials of reaching for an object oriented vertically and horizontally. Hand orientation at the first point of contact, prior to any tactile feedback, indexed infant prospective grasp. Between 5 and 7 months, infants showed evidence for qualitative transition in prospective control of grasp, supporting the contention that control of grasp shifts from being based on tactual feedback to being visually and therefore prospectively based. Implications for how prospective grasp emerges developmen- tally are discussed.

A cornerstone of skilled action is its future orientation. To successfully interact with the environment, organisms must prepare-through the coordination of per- ception and action-for forces, both intra- and extraorganismic, that constantly threaten to destabilize the balance and rhythm of action (Lee, 1993; von Hofsten, 1993). Maintaining an upright position while walking, for example, requires antic- ipating imbalance by enacting compensatory postural adjustments before imbalance has a chance to occur; similarly, catching a moving object requires anticipat- ing the trajectory of the object by moving the arm to a space the object does not yet occupy. The development of efficient, successful action thus depends on prospective control, both to counteract destabilization from gravitational and reactive

Requests for reprints should be sent to David C. Witherington, Department of Psychology, Logan Hall, University of New Mexico. Albuquerque. NM 87131-1 161. E-mail: [email protected]

144 WITHERINGTON

forces and to implement adjustments to object and environmental properties (von Hofsten, 1993, 1997).

In infant development, one of the best studied forms of prospective control is anticipatory hand orientation during reaching, in which grasping becomes future oriented. Skilled reaches for objects involve adjustments of the grasping hand to match object properties in advance of tactile feedback from actual contact. By 7 to 9 months, infants, when reaching for vertically and horizontally oriented objects, demonstrate prospective control of grasping by preorienting their hand in accor- dance with the objects’ orientation (Lockman, Ashmead, & Bushnell, 1984; McCarty, Clifton, Ashmead, Lee, & Goubet, 2001; Morrongiello & Rocca, 1989; Newell, Scully, McDonald, & Baillargeon, 1990; von Hofsten & Fazel-Zandy, 1984; Wentworth, Benson, & Haith, 2000). By 1 1 months, infants show clear signs of extending anticipatory hand orientation to diagonally oriented objects (Went- worth et al., 2000). Prior to 7 months, however, the extent to which infants exercise prospective control over their grasping is subject to a degree of debate.

Twenty years ago, for example, von Hofsten and Fazel-Zandy (1 984) reported evidence for anticipatory hand orientation in the reaching of infants as young as 4.5 months, when infants are just beginning to visually guide their reaching and to coordinate reaching with grasping (e.g., McDonnell, 1979; Piaget, 1952; White, Castle, & Held, 1964). Lockman et al. (1984), in contrast, found no evidence for anticipatory hand orientation in their 5-month-old sample: Only after grasping objects did 5-month-olds’ hand configurations differ significantly from previous points in their reaches. Uniformly, both of these studies reveal progression between 4.5 and 7 months in the precision with which infants grasp objects before tactile feedback. Nonetheless, the nature of this progression in prospective grasp control remains at issue, especially in light of two recent studies. Consistent with Lockman et al., McCarty et al. (2001) found correspondence between hand alignment and object orientation at the first point of contact (before tactile feedback) for infants at 7 months but not at 5 months. Consistent with von Hofsten and Fazel-Zandy, Wentworth et al. (2000) found such differences in infants as young as 5.5 months. Is the period between 5 and 7 months one during which prospective control of grasp emerges in qualitative, transitional fashion? Or are the changes evident between 5 and 7 months more quantitative in nature, the result of refinement in a skill that already accompanies infants’ earliest visually guided reaching efforts? McCarty et al. specifically framed these questions in terms of whether early grasping is under tactual feedback control or whether early grasping already dem- onstrates signs of prospective control.

Whereas von Hofsten and Fazel-Zandy (1 984) examined group differences by means of contrasting infant hand orientation on vertical and horizontal trials, Lockman et al. (1984) employed a difference score to chart the extent of match between hand orientation and dowel orientation. Yet, such methodological differences are unlikely to account for discrepant results. Both Wentworth et al. (2000)

PROSPECTIVE GRASPING CONTROL 145

and McCarty et al. (2001) used comparable indexes for assessing group differences in hand orientation accuracy but generated discrepant results regarding the availability of prospective control in 5-month-olds. Such discrepancies more likely stem from the overly distant ages typically contrasted and from the preponderance of cross-sectional work. Notwithstanding von Hofsten and Fa~el-Zandy, who studied infants at 4-week intervals from 18 to 34 weeks of age, the literature in development of prospective control is largely characterized by contrasts of infants at 5 months with infants 7 months or older. As a result, studies have not systematically targeted developmental change within the period of 5 to 7 months. Consider the possibility that the 5th month is typically a month of considerable transition in infant prospective control. The nature of 5-month-old sampling would then become critically important in charting developmental change. McCarty et al. sampled 5-month-olds at the beginning of the 5th month (M = 153 days. SD = 6.9), whereas Wentworth et al. sampled infants at the middle of their 5th month ( M = 166 days, SD = 2.04). Given the increased inter- (and intra-) individual variability that marks periods of developmental transition (e.g., Thelen & Smith, 1994), different sam- ples of 5-month-olds could vary substantially in the extent to which they exhibit anticipatory hand orientation. Studying infants at both the beginning and end of their 5th month would allow for more systematic examination of the developmental course of prospective grasp control.

More important, averaged, cross-sectional data are inadequate for studying developmental transitions because individual infant trajectories cannot be explored and the timing of developmental transitions cannot be individualized. Develop- mental transitions are necessarily individualized phenomena, and interindividual variability in the timing of transitions can be considerable. Without longitudinal data to chart individual trajectories, transitional change may be “averaged out” as similarly aged infants at different transitional points are grouped to yield 5- or 7-month data. Only von Hofsten and Fazel-Zandy ( 1 984) offered a systematic longitudinal account of development in infant prospective control, and their results suggest that grasping is to a degree prospectively controlled even at 4.5 months. However, in their analysis of individual infants, each infant’s data were pooled across 18 and 22 weeks and were contrasted with pooled data across 30 and 34 weeks. The possibility remains that this pooling of data could potentially mask transition points. Furthermore, von Hofsten and Fazel-Zandy did not specifically address intraindividual variability in their longitudinal work. Both Wentworth et al. (2000) and McCarty et al. (2001) described large trial-by-trial variability in infants’ accuracy of hand orientation and acknowledged the extent to which group data obscure such variability. However, without longitudinal data, we cannot de- velopmentally extend such analysis of intraindividual variability.

Given continued questions regarding the nature of developmental change in prospective grasp control, this study examined longitudinally its developmental trajectory between 5 and 7 months. This study also extends previous work by at-

146 WITHERINGTON

tending to the potentially transitional nature of the 5th month, by examining individualized developmental transitions, and by charting intraindividual variability. Infants were tested at the beginning and at the end of their 5th month, in addition to being tested in the middle of the 6th month and 2 weeks into the 7th month. If prospective control of grasp emerges between 5 and 7 months, then evidence for a qualitative, transitional shift from tactually controlled reactive grasping to visually controlled prospective grasping should be present. If, on the other hand, prospective control of grasp as a skill is already present by 5 months, changes that occur between 5 and 7 months should be in the form of quantitative increases in skill level.

METHOD

Participants

Fifteen infants accompanied by their mothers were studied longitudinally approxi- mately every 3 weeks from 5 months to 7.5 months. Each of the 15 infants com- pleted four visits, the first of which occurred in the first or second week of the infant’s 5th month. At the first visit (5 months 1 week [5;1]), mean age for infants was 155 days (SD = 5.25); at the second visit (5;4), mean age was 177 days (SD = 4.78); at the third visit (6;3), mean age was 197 days (SD = 4.76), and at the fourth visit (7;2), mean age was 218 days (SD = 5.44). Of the 20 infants originally sampled, 5 did not complete more than two visits due to scheduling conflicts. Only healthy, full-term infants (2 38 weeks gestational age) who had already begun to visually direct their reaching, as reported by their parents, participated. The final sample came from predominantly middle-class families. Eight of the 15 infants were female. Thirteen infants were White. and 2 were African American.

Apparatus

Each infant sat on his or her mother’s lap facing a 2 m x 1.22 m surface wall, covered with a uniform blue cloth. An orange and white target object, 26 cm long and 1.5 cm in diameter, was presented within reaching distance of the infant by means of a 1.23 m long support dowel (1.5 cm in diameter) that projected from behind the surface wall. The target object was presented midline at the level of each infant’s chest and at a distance of about three-quarter arm’s length from the infant to ensure its graspability. During reaching trials, Experimenter 2, by rotating the support dowel, could adjust the orientation of the target object to present it vertically o r horizontally to the infant. Two video cameras, positioned at right angles to one another with one above and one directly in front of the infant, provided a three-di- mensional analysis of infants’ reaching behavior. Signals from both cameras were transmitted through a video mixer (Videonics MX-1 Digital Video Mixer) to gen-


erate a time-locked, split-screen display of both views that was recorded for subsequent analysis.

Procedure

Before testing began, Experimenter 1 occluded the target object with a 46 x 67 cm cardboard screen while Experimenter 2, seated out of view behind the surface wall, made necessary changes in both the height and the distance of the target object relative to the infant. To prepare for testing, Experimenter 1 ensured that the infant sat squarely in front of the target object hidden from view by the screen. Experimenter I then engaged the infant’s attention to the screen by tapping it. As soon as the infant directed visual attention to the screen, Experimenter 1 instructed the mother to lightly restrain her infant’s hands at the waist. Experimenter 1 then quickly removed the screen to reveal the target object in either the horizontal or vertical orientation. Once the infant fixated the target, the mother released her infant’s arms and repositioned her own hands around the infant’s waist to provide postural support for the infant. If the infant was not initially inclined to reach for the target object, Experimenter 1 encouraged the infant to reach by lightly tapping the middle of the target object. Mothers also often verbally encouraged their infants to reach for the target.

An individual reaching trial ended (a) when the infant’s grasp on the target object was fully established; (b) if no grasp occurred, after the infant at least touched the target: or (c) after 30 sec elapsed. After each trial, Experimenter 1 repositioned the screen to occlude the infant’s view of the target object while Experimenter 2 adjusted the target object to its proper orientation for the next trial. Infants received eight reaching trials, with the dowel oriented vertically on four of the trials and horizontally on four trials. The order in which the horizontal and vertical trials were presented was randomly determined, but infants never received more than two consecutive trials of the same orientation. Of 480 total trials across all 15 participants, only 7 trials, or 1.4%, involved no reach within 30 sec. A small subset of infants received additional, replacement trials for those on which they reached to the target object without first looking at it. Both exper- imenters monitored infants’ looking activity during the procedure to ensure that visually guided reaching occurred, and the procedure was designed to minimize reaching without looking. Only 22 trials, or 5% of the total sample, involved reaching without an initial look to the target object, and these trials were replaced during testing with additional trials.

Data Reduction

Infant hand orientation relative to target object orientation at the point of first contact, just prior to tactual feedback, served as the primary variable of interest. Point of first contact was defined as one video frame prior to the frame in which infants

148 WITHEKINGTON

first touched the object. Timing for the inception of the reach (the point of the reach at which the infant first moves his or her hand forward) was also coded. Interrater reliability for the temporal identification of both points in the reach, derived from Pearson correlation coefficients on a random 20% of the participants, was above .99. Mean difference in rater timing was two video frames (SD = 2 frames). To quantify hand orientation, we adopted the scoring system used by Lockman et al. (1984) and Morrongiello and Rocca (1989): A 16-equal-interval scale (0"-360", with each interval comprising 22.5') captured the angle of the infant's hand relative to gravitational vertical. The first hand to make any contact with the target object was scored using this scale, both at the time of first contact and at the inception of the reach itself. Figure 1 depicts the scale with bird's-eye view illustrations of hand orientations corresponding to points 1,5,9, and 13. On the scale, a hand vertically oriented like the orientation used to shake hands (thumb up) was assigned a 1. A hand horizontally oriented like the orientation of a hand with the palm down lying on a flat surface was assigned a 5. A horizontally oriented hand with the palm up lying on a flat surface scored a 13. Hand orientations corresponding to points on the scale from 8 to 12 are highly unlikely to occur, owing to the difficulty of rotating the arm and wrist to those points. Two coders independently scored hand orientation for every session. Both coders achieved perfect matches on 82% of the trials and scored within one interval of each other on 98% of the trials. A third coder re- viewed the 18% of trials on which disagreement occurred to resolve differences with the other two coders.

Two indexes of anticipatory hand orientation were employed: (a) unconverted hand orientation scores (the raw 16 equal-interval scale mentioned previously),

k

FIGURE 1 view illustrations of hand configurations accompanying points I , 5,9. and 13 of the scale.

Diagram of 16-interval. 360" scale for scoring hand orientation. with bird's-eye


comparable to the indexes used by von Hofsten and Fazel-Zandy (1984), Wen- tworth et al. (2000), and McCarty et al. (2001); and (b) orientation difference scores, comparable to the index used by Lockman et al. (1 984) and Morrongiello and Rocca (1989). Unconverted hand orientation scores occupied the range of 1 to 5 for the overwhelming majority of trials, both for reach inception and point of first contact. For only 8 trials out of 473 codeable trials at reach inception (1.6%), and for only 5 trials out of 473 codeable trials at first contact (1.1 %), hand orientation scores moved beyond the 1-to-5 range to include scores of 15, 16, 6, and 7. No more than one of these scores characterized any given infant session, with the ex- ception of 1 infant's session that featured two outer-range scores. Scores of 1 and 5 represented the full range from vertical to horizontal orientation, rendering infant hand orientation scores within this range readily interpretable. Scores beyond this range (e.g., 16, 6), however, complicate averaging procedures as they artificially inflate overall scores. An infant who scores a 16 in a vertical orientation trial has a hand orientation only 22.5" different from a perfect score of 1, yet averaging a score of 16 with scores of 1 s and 2s produces an unrepresentatively high estimate. Similarly, a score of 6 in a horizontal orientation trial, which is 22.5" removed from a perfect score of 5 , will move an average score closer to 5, even though a 6 is comparable to a 4 in terms of its relation to 5. Given how few scores fell outside the range of 1 to 5 , and the interpretational ambiguity that inclusion of these scores would cause when computing mean performance estimates, we excluded the 8 reach inception scores and the 5 first contact scores that fell outside the 1-to-5 range for all analyses involving unconverted hand orientation scores. For orientation difference scores, these few trials remained in the sample, as computation of difference scores established a common metric for all points on the 16-point scale.

Orientation difference scores reflected the degree to which each hand orientation score corresponded to the orientation of the target object on any given trial and were computed as the difference between dowel orientation and hand orientation score (Lockman et al., 1984; Morrongiello & Rocca, 1989). For example, vertical dowel trials scored a 1 or 9 for the target orientation, and horizontal dowel trials scored a 5 or 13. If, on a vertical orientation trial, infant hand orientation scores a 3, then the target object orientation, being vertical, would be assigned a 1 because this score corresponds to the infant hand orientation more readily than a 9 would. As a result, the orientation difference score for the trial would equal 3 - 1, or 2. Orientation difference scores ranged from 0 to 4, with a 4 corresponding to a hand orientation, either forehand or backhand, that deviated

'The 13 trials involving hand orientations outside the 1-to-5 range could also be transformed to make them interpretable within the 1-to-5 range. For example. a score of 16, one interval removed from a perfect I , could be recoded as a 2 within the 1-to-5 range. We transformed the 13 excluded trials in this fashion and reran all hand orientation analyses with these 13 trials included. The pattern of results was exactly the same.

150 WITHERINGTON

90" from the Orientation of the target object, and a 0 corresponding to perfect hand-dowel orientation correspondence. For each infant session, a mean orientation difference score and mean hand orientation scores (one for vertical dowel trials, one for horizontal dowel trials) were the primary indexes of performance.

RESULTS

Analysis of Mean Orientation Difference Scores

Figure 2 presents infant mean orientation difference data at first contact across age as a function of dowel orientation. A 4 (age) x 2 (dowel orientation) repeated mea- sures analysis of variance (ANOVA) on mean orientation difference scores at first contact yielded a main effect for age, F(3,42) = 6.12, p < .01, and a main effect for dowel orientation, F( 1, 14) = 20.80, p < .001. No significant Age x Dowel Orienta- tion interaction emerged, F(3,42) = 0.46, p > .lo. Infant mean orientation difference scores at first contact, collapsed across dowel orientation, progressively and steadily improved across age, from 1.55 (SD = 0.5 1 ) at 5;l months to 1.22 (SD = 0.54) at 5;4 months, 1.03 (SD = 0.61) at 6;3 months, and 0.83 (SD = 0.44) at 7;2 months. These scores produced a clear and significant linear function across age, F( 1, 14) = 12.98, p < .01, and strongly coincide at 5; 1 and at 7;2 months, respectively, with Lockman et al.'s (1984) 5- and 9-month-old data as well as with Morrongiello and Rocca's (1989) 5- and 7-month-old data. Post hoc comparisons for the main effect of dowel orientation-involving paired-sample 1 tests with cor- rections for familywise Type I error through Dunn's test at an alpha of .05 (see Howell, 1987)-revealed significantly higher mean orientation difference scores

1.2 . E 2 . 1 1.8 .

1.2 .

3 0.8 . 0

B 0.4

0.2 . " -

$1 S:4 6;3 7;2

Age (months; weeks)

FIGURE 2 as a function of age (with standard error bars).

Mean orientation difference data at first contact for vertical and horimntal trials


for vertical compared to horizontal trials at 5;4 and 7;2 months, reflecting greater mismatch between hand and dowel orientation for vertical trials, ts( 14) > 3.52, ps < .01. At 6;3 months, the difference between vertical and horizontal trial scores achieved precorrection significance, t( 14) = 2.64, p < .05, and approached postcorrection significance, which required a t value of 2.86 (Howell, 1987, p. 590). These findings again closely correspond with Lockman et al.'s (1984) data and are discussed later in the context of mean hand orientation scores.

At the group level, mean orientation difference scores at first contact improved linearly between 5 and 7 months, which argues for a quantitative model of skill refinement. Figure 3 presents individual trajectories, collapsed across dowel orientation, for all 15 participants.* Only 3 of the infants' individual trajectories significantly mapped to a linear function: S1, S4, and S7, all Fs( 1, 25) > 5.60, ps < .05. Another 2 infants' trajectories approached significance for a linear function at an alpha of .05: S9, F(1,25) = 4 . 1 2 , ~ < .lo; and S12, F( I , 25) = 3 . 2 6 , ~ < .lo. The remaining 10 infants' individual trajectories, however, showed no significant evidence of linearity. Thus, group means obscure the relative heterogeneity with which development proceeds in mean orientation difference scores at first contact. In fact, examining the proportion of infants across time who demonstrated mean orientation difference scores of 1 .OO or below at first contact offers a view of dis- continuous change in prospective grasp control, with the 5th month serving as a fo- cal point for qualitative transition. At 5;l months, only 1 infant in the sample averaged 1.00 or below, in contrast to 7,9, and 10 infants at 5;4,6;3, and 7;2 months, respectively. Planned comparisons, using the Wilcoxon matched-pairs signed ranks test, revealed significant difference between 5; 1 and 5;4 months ( Z = -2 .45 ,~ < .05), but no significant differences either between 5;4 and 6;3 or between 6;3 and 7;2 months.

Thus, orientation difference scores at first contact provide evidence between 5 and 7 months for linear, quantitative refinement at the group level but developmental discontinuity in the skill of prospective grasp control at the individual level. Furthermore, a sudden jump between 5; 1 and 5;4 months in the number of infants whose mean orientation difference scores fell within 22.5" of perfect alignment suggests that the 5th month may serve as a transition point for the

?Given that infants generally exhibited greater match between hand and dowel orientation on horizontal trials relative to vertical trials, the question arises as to whether performance on vertical trials across age is largely responsible for individual differences in mean orientation difference score trajectories. For 7 of the 15 infants, hand orientation difference scores on horizontal trials fell within 22.5" of perfect match through the age range studied, whereas scores on vertical trials began with greater mismatch and over time assumed greater degrees of match. However, for the remaining 8 infants, no such pattern held. Instead, some infants showed a reverse pattern. with near-perfect performance on vertical trials and improvement from greater degrees of mismatch between dowel and hand orientation in horizontal trials across age. Other infants showed marked improvements in match in both horizontal and vertical trials.

WITHERINGTON

1 0.5 lpji, 0 5;l 5;4 63 72

3.5 ss 3 1

3: 3 s2

2.5

0.5

5;l $4 63 73

S6 ’.: ‘1 2.5

0.5 0 1

5;l 5;4 63 72

s9 ’.: ‘1

0.5 YkL 0

%I 5;4 6;) 72

3.5 3 1

0 1 5;l 5;4 63 72

1 0.5 i!LL 0

s3 2.5

5;l 5;4 6 3 72 5;I 5;4 6;) 72

S8 ”.: ‘1

$1 5;4 4 3 72 - . . . . .

5;l 5;4 6;3 73

3.5 s 12 3 1

1 0.5 e1“4-1 0

5;l 5;4 6;) 72

SIS

2.5 1 T 2.5 4

5;l 5;4 63 72 5;l 5;4 w 72 5:l 5;4 63 72

Age in Months

FIGURE 3 age (with intertrial standard error bars).

Individual infant mean orientation difference data at first contact as a function of

consolidation of prospective grasp control. However, these data do not in them- selves address the particular nature of developmental progress either within the 5th month specifically or between 5 and 7 months in general. During this potential period of transition from 5;l to 5;4 months, are infants undergoing a shift from tactually controlled to visually controlled grasping, or are infants simply in the process of consolidating a skill of prospective grasping control already evident at 5; l months? Turning to hand orientation scores offers further insight into this question.


Analysis of Mean Hand Orientation Scores

Figure 4 plots infants’ mean hand orientation at reach inception and at the point of first contact for vertical and horizontal dowel trials. Hand orientation at reach inception serves as a control basis of comparison for hand orientation at first contact, without which differences in hand orientation scores at first contact are potentially spurious. A 4 (age) x 2 (dowel orientation) x 2 (time during reach) repeated mea- sures ANOVA yielded a main effect for dowel orientation, F( l , 14) = 8 l .64, p < .001, and interaction effects for Age x Dowel Orientation, F(3,42) = 3 . 3 3 , ~ < .05; Dowel Orientation x Time During Reach, F( 1, 14) = 60.56, p < .001; and, most critically, Age x Dowel Orientation x Time During Reach, F(3.42) = 5 . 1 4 , ~ < 4 1 . As the three-way interaction of age, dowel orientation, and time during reach reveals, differences between vertical and horizontal trials in mean hand orientation varied as a function of point in reach (inception vs. first contact), but this interaction itself varied as a function of age.

Corrected paired-sample t tests elucidate the nature of this three-way interaction. Post hoc comparisons revealed significant differences in infants’ mean hand orientations at first contact as a function of dowel orientation at every age sampled, all rs(14) > 3.44, all ps < .05. Although infants’ prospective grasp was highly imprecise at 5 months, often requiring considerable postcontact correction in grasp configuration for successful grasp to occur, anticipatory hand orientation was seemingly evident in infant reaching at every age sampled in this study. This finding is consistent with von Hofsten and Fazel-Zandy (1984) and Wentworth et al. (2000) and supports a quantitative skill refinement model of

HORIZ 5 . 4.5 .

e 4 .

f 3.5 . i

f 3 .

3 2.5

5 E 1 .

1.5 .

FIGURE 4 and horizontal trials as a function of age (with standard error bars).

Mean hand orientation scores at both reach inception and tir\t contact for vertical

154 WITHERINGTON

prospective grasp control between 5 and 7 months. However, at 5;l months, infants’ hand orientation at first contact did not significantly differ from orientation at the inception of reach for either vertical, t( 14) = 0.45, p > .lo, or horizontal, r(14) = -0.68, p > .lo, trials. In contrast, at 5;4, 6;3, and 7;2 months, infants’ hand orientation at first contact significantly differed from orientation at reach inception for horizontal trials, all rs( 14) > -3.78, all ps < .05. For vertical trials, first contact orientation significantly differed from reach inception orientation at 7;2 months, t( 14) = 3.69, p < .05. At 6;3 months, the difference for vertical trials achieved precorrection significance, t(14) = 2.27, p < .05, and approached postcorrection significance, which required a t value of 3.43 (Howell, 1987, p. 590). No statistically significant difference emerged for vertical trials at 5;4 months, however.

Thus, with mean hand orientation scores at first contact juxtaposed against scores at reach inception, infants from 5;4 months on-but not reliably at 5;l months-demonstrated evidence for prospective grasp control at first contact, especially in the context of horizontally oriented objects. Recall from infants’ mean orientation difference scores the greater mismatch between hand and dowel orientation for vertical trials. Infants clearly mastered an anticipatory horizontal hand orientation at an earlier age than an anticipatory vertical hand orientation. This may derive from the tendency of infants at all ages in our sample to assume a hand orientation closer to horizontal than to vertical at the start of their reaches, consistent again with Lockman et al.’s (1984) sample. Mean hand orientation scores at reach inception ranged from 3.10 to 3.67. Generally, infants, when reaching for the vertically positioned dowel, had in effect more ground to cover than for a horizontally positioned dowel, an argument first offered by Lockman et al. (1984). However, infants’ greater success in the context of horizontally oriented objects did not hold at 5; 1 months, when infants’ hand orientations at reach inception and first contact were virtually identical for vertical and horizontal trials. These results accord with McCarty et al.’s (2001) conclusion that 5-month-olds do not preorient their hands to an object’s orientation prior to tactual contact. These results also lend some support to the transitional nature of the 5th month, from primarily tactually controlled (tactual feedback-based) to visually controlled (prospective) grasping.

Evidence for developmental discontinuity in general must be framed in the context of individual developmental trajectories, given individual variability in the timing of developmental transitions. With a longitudinal design, times of transition can be uniquely defined for each infant. Furthermore, at points of transition, intraindividual variability should rise considerably relative to periods of developmental stability (e.g., Hartelman, van der Maas, & Molenaar, 1998; Thelen & Smith, 1994). In the following section we individualize transition points for each infant to chart potential qualitative transition in infants’ prospective grasp control. We also chart developmental change in infants’ intertrial variability.


Analysis of Individualized Transitions in Mean Orientation Scores and lntraindividual Variability

To target individual points of transition in prospective grasp control, we referred to our earlier proportional analysis of mean orientation difference scores at first contact. For each infant, the first session in which his or her mean orientation difference score equaled or fell below 1 .OO was identified and compared with the session immediately preceding it, thus establishing a “before” and “after” transition comparison (see Figure 3). Mean hand orientation scores at both reach inception and first contact for each of the two sessions were then tabulated for each infant. Group “before” and “after” hand orientation means could thus be compared, taking into account individual differences in the timing of transition. One infant (S 10) already scored well below 1 .OO in mean orientation difference at 5; 1 months, and 2 other infants (S 1 1 and S 15) never scored at or below 1 .OO for any of the ages sampled. An additional infant (S5) initially fell below 1 .OO but then in the following sessions rose well above 1 .OO again, rendering ambiguous this infant’s transition point. Of the remaining 1 I infants for whom transition sessions were readily identifiable, 5 infants (S2, S3, S6, S9, and S14) transitioned between 5;l and 5;4 months, 3 infants (S 1, S4, and S 12) transitioned between 5;4 and 6;3 months, and 3 infants (S7, S8, and S13) transitioned between 6;3 and 7;2 months.

Table 1 provides mean hand orientation scores at both reach inception and first contact for before and after transition, individualized for the 11 infants with clear transition points. Before transition, infants’ mean hand orientations did not significantly differ from reach inception to first contact, either for vertical trials, t( 10) = 0.41, p > .lo, or for horizontal trials, t( 10) = -1.48, p > .lo. After transition, simply 3 weeks later, infants’ mean hand orientations significantly differed from reach inception to first contact, both for vertical trials, t( 10) = 11.98 ,~ < .OOl, and for horizontal trials, t( 10) = -5.01, p < .01. Thus, when we individualize transition points for mean orientation difference scores at first contact, infants demonstrate a dra-

TABLE 1 Mean Hand Orientation Scores at Reach Inception and First Contact Before and After Developmental Transition in Orientation Difference

Scores

Vertical Trials Horizontal Trials

Reurli Inception First Contact Reach Incepfion Firsf Contact

Transition M SE M SE M SE M SE ~~

Before 3.20 0.16 3.07 0.31 3.56 0.19 4.08 0.26 After 3.50 0.12 1.93 0.17 3.41 0.22 4.55 0.09

Note. N = I I .

156 WITHERINGTON

matic shift across the 3-week transition, from no significant evidence for prospective grasp control to consistent evidence for prospective grasp control. These results support discontinuity in the development of prospective grasp control and argue for a qualitative transition in such control between 5 and 7 months.

As Table 1 illustrates, however, the difference in mean hand orientation from reach inception to first contact, although not statistically significant, is in the appropriate direction for both vertical (3.20 to 3.07) and horizontal (3.56 to 4.08) trials before transition. Lockman et al. (1984) and McCarty et al. (2001) found directionally appropriate orientation scores prior to tactual feedback for infants at 5 months, although such preorientation was marginal and failed to achieve statistical significance in both studies. Nonetheless, the directionally appropriate nature of our results before transition could call into question the interpretation applied previously to these data. Perhaps before transition results failed to achieve statistical significance simply due to relatively small sample size. An inspection of individual infants’ before and after scores, however, underscores both the qualitative nature of transition in prospective grasp control between 5 and 7 months and the extent to which statistically nonsignificant but directionally appropriate group means can be illusory in the story they tell. Of the 11 analyzed, only 3 infants before transition exhibited directionally appropriate change in mean hand orientation from reach inception to first contact, with the remaining 8 infants exhibiting directionally inappropriate change. In fact, of these 8 infants, half erred in directional change from reach inception to first contact on horizontal trials only (e.g., from a score of 3.67 to 2.67), and the other half erred on vertical trials only (e.g., from a score of 3.00 to 3.75). Thus, each of these 8 infants, before transition, consistently changed his or her hand orientation in a single direction (either always toward a vertical or always toward a horizontal orientation) from reach inception to first contact, irrespective of the orientation of the dowel for which he or she was reaching. In contrast, all 1 1 infants after transition showed directionally appropriate change in mean hand orientation from reach inception to first contact, for both vertical and horizontal trials.

Analysis of intraindividual variability across individualized transitions offers further support for a qualitative transition in infant prospective control between 5 and 7 months. To index within-session variability of infant grasping, we employed the coefficient of variation, which normalized the standard deviation for each infant session relative to the infant’s mean performance. The coefficient of variation of mean orientation difference scores at first contact was computed for each of the 1 1 infants before and after transition in mean orientation difference scores. Mean coefficient of variation fell from 1.35 (SD = 0.42) before transition to 0.85 (SD = 0.38) after transition, just 3 weeks later, t( 10) = 3.68, p < .01. An examination of each infant before and after transition revealed that 10 of the 11 infants showed de- creases in coefficient of variation, consistent with group differences, and 7 of these 10 infants showed their largest decrease in coefficient of variation across the tran-


sition itself. At the group level, mean difference in coefficient of variation across the transition covered and in fact extended the full range of age-based mean variability in the coefficient of variation, which ran from a high of 1.21 (SD = 0.47) at 5;l months to a low of 0.93 (SD = 0.59) at 7;2 months. It is also important to note that when differences in mean coefficient of variation were analyzed simply as a function of age, no significant differences emerged, highlighting the importance of individualizing developmental transitions. With individualized transitions in place, infant intraindividual variability shows a dramatic decrease in the span of just 3 weeks, again underscoring the qualitative nature of this transition, especially in light of our mean hand orientation results.

In the context of infants’ within-session variability, are infants showing signs of learning across trials? Could the reduced intraindividual variability infants demonstrated after transition reflect an increase across trials in infants’ accuracy of grasping? We compared infants’ hand orientation difference scores on their first vertical and horizontal trial with scores on their last vertical and horizontal trial both before and after transition. No significant differences emerged between first and last vertical trials or between first and last horizontal trials, either before or after transition. Thus, infants showed no systematic evidence of within-session learning in their prospective grasping, which was also borne out for the full 15-infant sample considered as a group and individually.

DISCUSSION

The purpose of this study was to examine the development of anticipatory hand orientation from 5 to 7.5 months with a longitudinal sample and to extend previous work by individualizing pQtential qualitative transition in prospective control, charting intraindividual variability over time, and by more systematically investi- gating the 5th month for its potential transitional character. Recent findings by Wentworth et al. (2000) and McCarty et al. (2001) in the infant literature on prospective grasp control have revived a debate first crystallized in papers by von Hofsten and Fazel-Zandy (1 984) on the one hand and Lockman et al. (1984) on the other. Namely, does prospective grasp control emerge in transitional fashion between 5 and 7 months, or is it already present, albeit in nascent form by 5 months, with developmental change between 5 and 7 months reflecting progres- sive refinement of the skill? This study’s results, taken as a whole, converge to support a qualitative transition model for prospective grasp control between 5 and 7 months, consistent with Lockman et al.’s (1984) and McCarty et al’s (2001) conclusions. AIthough infants in general demonstrated linear progression in anticipatory hand orientation between 5 and 7 months and showed evidence for differentiated anticipatory hand orientation at first contact as a function of dowel orientation from the first age of testing (5; 1 months), an examination of the data on a

158 WITHERINGTON

more individualized basis, with controls in place, provides consistent evidence for a qualitative, transitional characterization of prospective grasp control development between 5 and 7 months, from tactually controlled to increasingly visually controlled grasping.

Evidence for anticipatory hand orientations as early as 4.5 to 5 months from the work of von Hofsten and Fazel-Zandy (1984) and Wentworth et al. (2000) suggests that some degree of prospective control in grasping, however rudimentary in form, is present at the earliest stages of visually guided reaching. Our results fail to support such a characterization. Both at the group level and the individual level, infants in our study between 5 and 7 months typically started out showing no difference in hand orientation from the inception of reach to first contact with the object. In fact, a majority of infants in this study showed no evidence for directionally appropriate hand orientation at some point between 5 and 7 months, only to transition within 3 weeks to directionally appropriate, differentiated hand orientation at first contact with objects both vertical and horizontal. Furthermore, these infants showed a marked reduction in intertrial variability of hand orientation scores across this transition. Before transition, intertrial variability was significantly higher than after transition variability, suggesting that for the majority of infants in this study, we tapped into a distinct, qualitative shift in prospective grasp control between 5 and 7 months.

For many skilled actions, such as maintaining upright posture while walking, the absence of prospective control has potentially serious consequences, such as the organism’s losing balance and falling. Failure to control grasp prospectively, however, is much less likely to result in serious consequences; outside of initial mishandling resulting in the dropping of the object, reshaping one’s hand after contact with an object usually means little more than a slight delay in successful grasping. Our results speak to the potential utility of couching prospective control in terms of its functional consequences for the organism’s well-being. Perhaps the developmental course for prospective control in any given skill coincides to a considerable degree with the functional consequences lack of prospective control would generate for the organism. Whether or not such considerations provide a useful framework for understanding development in prospective control, our results at least underscore the need to consider the developmental course of prospective control on a skill-by-skill basis.

With respect to the skill of grasping, this study provides evidence for qualitative transition in the typical development of prospective grasp control between 5 and 7 months, both through age-based group data and through more individualized analyses. This study also reveals considerable individual variability in the developmental timing of such qualitative transition, which itself highlights the inadequacy of age-based investigations (recall that 1 infant of the 15 studied was already showing sophisticated prospective control of grasp at 5; 1 months). If, as this study and others (e.g., McCarty et al., 2001) suggest, infants undergo a qualitative transition


from tactually controlled to increasingly visually controlled grasping after visually guided reaching emerges, the next step for investigation should target what processes might underlie the transition. One obvious candidate is simply the experience that derives from coordinating reaching and grasping in the context of visually guided reaching.

The developmental course for reaching and grasping has been systematically charted across numerous studies. Neonates engage in a form of “prereaching” behavior consisting of spontaneous forward arm extensions directed toward objects on which visual attention is fixated (Bower, Broughton, & Moore, 1970; McDonnell, 1979; Trevarthen, 1975; von Hofsten, 1982). Prior to 2 months, infants’ prereaching activity is typically accompanied by an opening of the hand, with fanlike extension of the fingers, but contact with objects is highly unreliable, with no attempt to grasp the objects (von Hofsten, 1982, 1984). Furthermore, hand opening acts as an extension of the arm movement synergy, not as an adaptive action in its own right; hand opening accompanies forward extensions irrespective of whether the infant visually fixates an object. At 2 to 3 months, in contrast, there ensues a period of reaching accompanied by a fisted hand and characterized as swip- ing. After this period, around 4 to 5 months, infants begin again to open their hands as they reach for objects, but now in the context of attempts to grasp objects that are visually fixated (Field, 1977; von Hofsten, 1984; White et al., 1964). This visually guided or goal-directed reaching involves coordinated reaching and grasping activity and is viewed as qualitatively distinct from the prereaching efforts that pre- cede it. Specifically, between 15 and 18 weeks, a transitional period between prereaching and visually guided reaching typically ensues. At 15 weeks, reaching is openhanded but terminates exclusively in a touch of the object, with no ensuing grasp. By 18 weeks, this reaching without grasping has declined in frequency dra- matically, to be replaced by reaching coordinated with grasping activity (von Hofsten & Lindhagen, 1979). Wimmers, Savelsbergh, Beek, and Hopkins (1998) underscored the qualitative nature of this transition through an application of ca- tastrophe detection procedures to their own sample of infants studied weekly from 2 to 6 months.

Once visually guided reaching emerges, does it set in motion the experiential conditions that typically help establish prospective grasp control? Is visually guided reaching experience per se an important formative influence in this regard? Perhaps developmental changes in manual exploration, independent of coordination in reaching and grasping, give rise to prospective grasping. With the emer- gence of rhythmical stereotypies and examining behavior in manual exploration around 4 to 5 months, infants embark on a new means of actively engaging objects at the haptic level, one increasingly integrated with visual inspection (Bushnell & Boudreau, 1993; Ruff, 1986; Thelen, 1979). Whatever processes might be involved, the issue of how prospective grasp control emerges, how infants transition from tactual to visual control of grasping, is in clear need of investigation.

160 WITHERINGTON

In addition, our results call for more systematic investigation of the developmental course in prospective grasp control beyond 7 months. Although this study yields clear evidence for qualitative transition in prospective grasp control, testing did not continue beyond 7.5 months. For some infants in the sample, reduced intraindividual variability and lowered orientation difference scores were not maintained across time, suggesting that the transition they underwent in prospective control did not yield stability of accurate responding. Clearly, the transitional change documented in this study does not suggest that after transition infants have now acquired prospective control; prospective control continues to develop beyond 7 months, both quantitatively, in terms of skill consolidation, and also qualitatively. Subsequent development in prospective control has been investigated, most recently in work by Wentworth et al. (2000) and McCarty et al. (2001), but the lack of longitudinal work past 8 months of age limits the conclusions drawn in regard to the nature of its specific developmental course.

In conclusion, this study, through a longitudinal sampling of infants, helps resolve the current debate over the origins of prospective grasping by more systematically charting its developmental course between 5 and 7.5 months. Our results suggest that this time period typically involves a qualitative transition in infant grasping from tactile to visual control, consistent with the recent work of McCarty et al. (2001).

ACKNOWLEDGMENTS

This research is based in part on a dissertation that was submitted in partial fulfill- ment of the requirements for the PhD at the University of California, Berkeley, and was supported by a Graduate Student Research Grant from the Institute of Human Development. I thank Joseph J. Campos and John S. Watson, my dissertation com- mittee cochairs; Martin Banks; Margaret Boucher, Lisa Casman, Julie Hwang, Hannah Lohman, Tim Martin, Hyun Joo Park, Christine Sun, Sarah Yi, and Peter Yurista, for their help in data collection and coding; and the mothers and infants who participated in this study.

REFERENCES

Bower, T. G. R., Broughton, J. M.. & Moore, M. K. (1970). Demonstration of intention in the reaching behavior of neonate humans. Nature, 228, 6 7 9 4 8 I .

Bushnell, E. W., & Boudreau, J. P. (1993). Motor development and the mind: The potential role of motor abilitiesasadeterminantofaspectsofperceptual development. ChildDevelopment, 64. 1005-1021.

Field, J. (1977). Coordination of vision and prehension in young infants. Child Development, 48, 97-103.


Hartelman, P. A. I., van der M

Howell, D. C. (1987). Statistical methods for psychology. Boston: Duxbury. Lee, D. N. ( 1993). Bodyyenvironment coupling. In U. Neisser (Ed.), 7heperreivedse@ Ecologicaland

interpersonal sources ($self-knowledge (Emory Symposia in Cognition, Vol. 5, pp. 4347) . New York: Cambridge University Press.

Lockman, J . L., Ashmead, D. H., & Rushnell, E. W. (1984). The development of anticipatory hand orientation during infancy. Journal of Experimental Child Psychology, 37, 176-1 86.

McCarty, M. E., Clifton, R. K., Ashmead, D. H., Lee, P., & Goubet, N. (2001 ). How infants use vision for grasping objeets. Child Development, 72, 973-987.

McDonnell, P. M. (1979). Patterns of eye-hand coordination in the first year of life. Canadian Journal

Morrongiello, B. A.. & Rocca, P. T. (1989). Visual feedback and anticipatory hand orientation during

Newell, K. M., Scully, D. M., McDonald, P. V., & Raillargeon. R. ( 1990). Fdsk constraints and infant

Piaget, J. (1952). The origins v f intelligence in children. New York: Norton. Ruff, H. A. ( 1986). Components of attention during infants’ manipulative exploration. Child Develop-

Thelen, E. (1979). Rhythmical stereotypies in normal human infants. Animal Behavior, 27, 699-7 IS. Thelen, E., & Smith, L. B. ( 1994). A dynamic systems approach to the deivlvpmenf of cognition and ac-

Trevarthen, C. ( I 975). Growth of visuomotor coordination in infants. Journal of Human Movement

von Hofsten, c. (1982). Eye-hand coordination in the newborn. Developmental Psychology, 18,

von Hofsten, C. ( 1984). Developmental changes in the organization of prereaching movements. Ikvel-

von Hofsten, C. (1993). Prospective control: A basic aspect of action development. Human Develop- ment, 36. 253-270.

von Hofsten, C. (1997). On the early development of predictive abilities. In C. Dent-Read & P. Zukow-Goldring (Eds.), Evolving explanations of development: Ecological approaches fo orgati- ism-environment systems (pp. 163-194). Washington, DC: American Psychological Association.

von Hofsten, C., & Fax-Zandy, S. (1984). Development of visually guided hand orientation in reaching. Journal of Experimenfal Child Psychology, 38, 208-2 19.

von Hofsten, C.. & Lindhagen, K. (1979). Observations on the development of reaching for moving objects. Journal of Experimental Child Psychology, 28, 158-1 73.

Wentworth, N., Benson, J . B., & Haith. M. M. (2000). The development ofinfants’ reaches to stationary and moving targets. Child Development. 71, 576601.

White, B. L., Castle, P.. & Held, R. (1964). Observations on the development of visually-directed reaching. Child Development, 35, 349-364.

Wimmers, R. H., Savelsbergh, G. J. P., Beek, P. J.. & Hopkins, B. (1998). Evidence for a phase transition in the early development of prehension. Developmentrtl Pqchvbivlvgy, 32, 235-248.

. H. L. J., & Molenaar, P. C. M. (1998). Detecting and modeling developmental transitions. British Journal of Ilevelvpmental Psychology, 16, 97-122.

of Psvchology, 33, 253-267.

infants’ reaching. Perceptual & Motor Skills, 69, 787-802.

grip configurations. I~evelvpmental PsychohiokJgy, 22. 8 17-832.

ment, 57, 105-1 14.

tion. Cambridge, MA: MIT Press.

Studies, I , 57.

450-46 1 .

opmentul f.S.Vcho/ogy, 20, 378-388.

the development of prospective grasping control between 5 and 7 months: a longitudinal study

Documents