Keeping track of locations during movement in 8- to 10-month-old infants

Keeping track of locations during movement in 8- to 10-month-old infants

JOURNAL OF EXPERIMENTAL CHILD PSYCHOLOGY 32, 48-64 (1981) Keeping Track of Locations during Movement in 8- to 1OMonth-Old Infants EUGENE C. GOLD...

1MB Sizes 0 Downloads 9 Views

JOURNAL

OF EXPERIMENTAL

CHILD

PSYCHOLOGY

32, 48-64 (1981)

Keeping Track of Locations during Movement in 8- to 1OMonth-Old Infants EUGENE

C. GOLDFIELD AND DONALD J. DICKERSON University

qf Connecticut

Infants of 84 and 94 mo of age were tested for the ability to “keep track,” i.e., to determine the location of an object hidden in one of two covered containers before their left-right positions were reversed. Infants in both age groups for whom the covers were the same color and younger infants for whom the covers were different colors were generally unable to keep track. Only the older infants provided with different colored covers were able to do so. An analysis which separated keeping track from the sensorimotor stage 4 error indicated that (a) there was no contingency between the two and (b) there were developmental differences in the nature of the error.

Piaget (1954) theorizes that the infant’s conception of space changes dramatically from the beginning of sensorimotor stage 4 (approximately 8 mo) to the middle of sensorimotor stage 5 (approximately 15 mo). At the outset of this period the infant has an egocentric, or subjective, conception of space in which the location of an object or event is specified only by its position relative to the infant’s own body. By the middle of stage 5, however, the infant develops an objective conception of space in which the location of an object or event is specified by its position relative to other objects. The subjective space thus has a single reference point that moves as the infant moves, while the objective space has multiple reference points most of which remain fixed as the infant moves. Piaget (1954) also holds that a close relation obtains between the development of an objective conception of space and the devleopment of This article is based on a dissertation submitted by the first author to the Graduate School of the University of Connecticut in partial fulfillment of the requirements for the Ph.D. degree. Thanks are extended to Sam Witryol and Michael Turvey for their critical comments and to Patricia Singleton for assisting in data collection. The research was supported in part by the Research Foundation of the University of Connecticut. Portions of this paper were presented at the annual meeting of the Jean Piaget Society, Philadelphia, June, 1979. Requests for reprints should be sent to the first author at the Language and Cognitive Development Center, 11 Wyman Street, Boston. MA 02130.

48 0022-0965/81/040048-17$02.00/O Copyright 0 1981 by Academic Press, Inc. All rights of reproduction in any form reserved.

KEEPING

TRACK

IN INFANTS

49

the concept of object permanence, i.e., the understanding that objects continue to exist independently of actions upon them. Object permanence is usually assessed with some version of the hidden object test. At the outset of sensorimotor stage 4 infants will search for a hidden object, but they also will make an unusual error. If the infant is allowed to find the object in one location, A, and the object then is hidden in a different location, B, the infant will search where the object was previously found, at A, rather than where it was last seen, at B. This phenomenon is well documented (Gratch, 1975; Harris, 1975) and is termed the AB (read A not B) error. By stage 5 (12 mo) infants no longer make the error. The An error is interpreted as indicating a limited conception of object permanence. The stage 4 infant conceives of the object as existing at a particular location. The specification of this location, furthermore, is in terms of its position relative to the infant’s body. Piaget thus ties the An error to an egocentric conception of space. The relating of the AB error to an egocentric space clearly is not required. Infants could have an objective conception of space and still view the object as existing at a particular location. Harris (1977), has, in fact, suggested that infants who make the AB error perseverate in looking in the specific place where the object was previously found rather than, as Piaget suggests, in the direction where the object was previously found. Harris’s position is tantamount to saying that infants have an objective, or partially objective, conception of space during sensorimotor stage 4. The available evidence suggests that Piaget’s theory is correct in general but also provides some support for Harris’s position. The development of the infant’s conception of space has been studied by testing the ability to “keep track” of the location where an event, e.g., the appearance of a person or the hiding of an object, had previously occurred. Infants with an objective space should be able to keep track, while infants with an egocentric space should not. Acredolo (1978) used a procedure in which infants first were trained to anticipate the appearance of a person in one of two windows and then were tested for anticipatory looking to the correct window after movement. She found that 6-mo-old infants were unable to keep track; 16-mo-old infants were able to keep track, and 1 I-mo-old infants were able to keep track consistently only when the location was marked by a distinctive visual stimulus (a landmark). Bremner (1978b) used a one-trial procedure in which infants first watched while an object was hidden in one of two containers and then were allowed to search for the object after movement occurred that reversed the left-right position of the containers. He reported findings for 9-mo-old infants that showed them to behave similarly to the 1l-moold infants in Acredolo’s study. The evidence thus indicates that 6-mo-

50

GOLDFIELD

AND

DICKERSON

old infants have an egocentric conception of space and that 16-mo-old infants have an objective conception of space, as Piaget (1954) suggests. Infants at 9 and 11 mo of age, toward the middle of sensorimotor stage 4, seem, however, to have a partially objective conception of space. The evidence bearing on the relation between the infants’ conceptions of space and the permanent object is consistent with that just reviewed. In making the AB error 9-mo-old infants perseverate in searching in the previously-rewarded direction when the two hiding places have identical covers (Bremner & Bryant, 1977) and perseverate in searching in the previously-rewarded place when the two hiding places have differentcolored covers (Bremner, 1978a). Infants 9-mo-old thus seem to have a partially objective conception of space since they keep track and respond to specific places when landmarks are provided, but fail to keep track and respond directionally when landmarks are not provided. The making of the AB error does not, additionally, seem to be tied to a completely egocentric conception of space. The present study further explores the development of the infant’s conception of space, as indicated by the ability to keep track of the location of an event during movement, and the relation between the infant’s conception of space and the AB error. Both 8& and 9&-mo-old infants are studied. The former are closer to the beginning of sensorimotor stage 4 than in previous work, and the latter are approximately the same age as in the previous work with 9-mo-old (actually 9f-mo-old) infants (Bremner, 1978a,b; Bremner & Bryant, 1977). The procedure involved administering the hidden object test in a fashion similar to that used by Bremner (1978b). The object (a toy) first was hidden repeatedly in one of two containers until the infant searched correctly on two successive trials. On the test trial, which followed immediately, the object was hidden again, but before the infant was allowed to search she was reoriented by one of three types of movement. Two types of reorientation, (a) rotating the platform holding the containers by 180” (move platform) and (b) moving the infant to the opposite side of the platform (move infant), resulted in a reversal of the left-right positions of the containers, while one type of reorientation, (c) moving the infant 90” around the platform and then back to the original position (control), did not. Each infant received six training-test sequences, two with each type of reorientation. In one of the two sequences the object was hidden in the same container on the test trial as on the preceding training trials and in the other sequence the object was hidden in the other (different) container. For different groups at each age the covers were identical or differently colored. The infant’s ability to keep track of a location during movement was assessed by the level of correct responding on the same container test trials. Different container test

KEEPING

TRACK

IN INFANTS

51

trials were not used in the assessment of keeping track because the possible occurrence of the AB error makes the container that the infant is tracking ambiguous. The relation between the infant’s conceptions of space (keeping track) and object permanence (the AB error) was assessed by analyzing the pattern of correct responses over all six test trials. The study also involved an attempt to monitor the visual activity of the infants during movement. Bremner (1978b) reported unsystematic observations that suggested that keeping track might be related to visual tracking of the correct location.’ METHOD Subjects

Forty-eight infants from the Hartford and Storrs, Connecticut areas served as subjects. The infants were all white and from middle or uppermiddle class homes. The two age groups contained equal numbers of males and females, and they had mean ages of 259 (8i mo) and 289 (94 mo) days and age ranges of 244 to 270 days and 271 to 304 days, respectively. Potential subjects were obtained from hospital birth lists. Parents were contacted by letter and their consent was obtained by telephone. Testing took place with the parent present in either a room in the Human Development Center at the University of Connecticut or a room in the Department of Psychology at Trinity College. Apparatus

A 25 x 30 cm white wooden platform holding two white wooden containers, 12 cm on a side, was used. The rotating platform was atop a pedestal 75 cm from the floor. Two sets of 15 x 15 cm wooden container covers were used; a nondistinctive set, in which both covers were white, and a distinctive set, in which one cover was yellow and the other was green. The objects that were hidden were a string of keys, a string of balls, a toy kitten, and a toy giraffe. All were colorful and made noise when squeezed or shaken. Design and General Procedure

Age and distinctiveness of the covers were between-subjects variables. There were, therefore, four major groups: younger-nondistinctive, younger-distinctive, older-nondistinctive, and older-distinctive. Each group contained equal numbers of male and female infants. Type of reorientation (move platform, move infant, or control) and test container (same or different) were within subjects variables. ’ Note the distinction between keeping track as measured by manual search and tracking as measured by visual activity.

52

GOLDFIELD

AND

DICKERSON

Each infant received six training-test sequences. The two sequences with the same type of reorientation were blocked together in the presentation order. The order of the three blocks of two training-test sequences was counterbalanced over subjects, and the order within a block of the two sequences ending in the same and different container test trials were counterbalanced over blocks and subjects. An observer recorded on each test trial where the infant searched and where the infant looked following the halfway point of reorientation. The observer was hidden behind screens that were situated around three sides of the platform, an arrangement that allowed the observer to move in order to keep the infant’s face in view at all times. Procedure

Each infant was brought individually by the parent into the testing room. The parent was seated in a swivel chair with the infant on her lap at the platform containing the apparatus. The experimenter was seated on the opposite side of the platform. Infants were first allowed to familiarize themselves with the experimenter, the apparatus, and the toys. Care was taken to ensure that infants were capable of lifting the covers and were interested in the toys. A partial hiding procedure, which has been shown to increase search for the hidden object (Miller, Cohen, & Hill, 1970), was used on two warmup trials. With the platform pulled out of reach of the infant, a toy was lowered into one of the two containers and then lifted out twice in succession. It was then lowered into the container a third time and was left with part of it hanging over the side. The covers were simultaneously placed on the two containers, the toy remaining partially visible. After a 3-set delay the platform was moved forward and the infant was allowed to search for the toy. The procedure was then repeated with the toy hidden in the other container. There were six training-test sequences in the experimental period. On the training trials in each sequence the toy was hidden in the same container. After a criterion of two consecutive correct responses was reached the toy was hidden for the test trial and one of the three types of reorientation was performed. On move platform trials the parent and infant remained in position while the experimenter slowly rotated the platform so that the correct container passed directly in front of the infant. On move infant and control trials the table remained stationary and the seated parent slowly wheeled herself and the infant around the table in a manner such that the infant faced the platform and the correct container was on the side by which the infant moved. On move infant trials the experimenter simultaneously moved in his chair around the other side of the platform to the position previously occupied by the parent.

KEEPINGTRACKININFANTS

53

If the infant initially searched in the wrong container on a training or test trial, she was allowed to continue searching until the toy was found or 1 min elapsed. If the toy was not found within a minute, the cover was removed and the toy was raised to the top of the container for the infant to grasp. The infant was allowed to play with the toy briefly before the next trial. If the infant appeared to be getting bored with any toy, a different one was introduced during a series of training trials but not between the last training trial and the test trial. RESULTS Manual

Search

Before evaluating the effect of the major variables, an assessment was made of the effect of the order of the six test trials on correct manual search. The proportions of correct responses on the six successive test trials were 52, .46, 58, 54, .56, and .50, respectively. In an age x sex x distinctiveness x trials analysis of variance all interactions involving trials plus the main effect of trials were nonsignificant. Correct search, therefore, did not change appreciably as a function of trials. The correct searches on all three same container trials and on the different container control trials are summarized below. The former measure the infant’s ability to keep track of a specific location during movement, and the latter measure the incidence of the An error. The different container trials cannot be used to measure the ability to keep track since, once the object is moved, the specific container being tracked cannot be determined. An infant making the AB error would be trying to keep track of the container in which the object was hidden on the training trials, while an infant not making the error would be trying to keep track of the container in which the object was hidden on the test trial. The proportion of correct responses on move platform and move infant test trials with the different container ranged only from .50 to .58 among the four age x distinctiveness groups. The responses on these trials will figure in the subsequent analysis of the pattern of responses over all six test trials. Table 1 presents the proportion of correct responses on same container trials. The two younger groups tend to show more correct responses on control trials than on move platform and move infant trials. In the two older groups, however, correct responding on move platform trials approaches that on control trials and surpasses that on move infant trials. Overall, the performance of the first three groups in the table are comparable, while that of the older-distinctive group is superior. An age x sex x distinctiveness x reorientation analysis of variance of the correct response scores yielded an interaction of age and distinctiveness that approached significance, F(1, 40) = 3.35, p = .075, and

54

GOLDFIELD

AND DICKERSON TABLE

PROPORTION

OF CORRECT

1

RESFQNSES ON SAME CONTAINER

Reorientation Group Youngernondistinctive Youngerdistinctive Oldernondistinctive Olderdistinctive Combined

TRIALS

trials

Move platform __--~

Move infant

Control

Combined

.25

.25

35

.42

.42

.25

.58

.42

.58

.I7

.67

.47

.83 .52

.50 .29

.83 .71

.?2 .51

significant main effects of age, F(1, 40) = 6.98, p = .012, and reorientation, F(2, 88) = 8.20, p = .0006. The significant age effect indicates that older infants searched correctly more frequently than younger infants. The significant reorientation effect is due to greater accuracy on control trials than on move platform and move infant trials and greater accuracy on move platform trials than on move infant trials. Two correlated t tests showed that the difference between control and move platform trials only approached significance, t(47) = 1.84, p = .072, while the difference between move platform and move infant trials was significant, t(47) = 2.41, p = .020. The infants, therefore, had difficulty following displacements that reversed the left-right positions of the containers and they had more difficulty when they were moved than when the platform was moved. The age x distinctiveness interaction was also evaluated by t tests. While the younger-nondistinctive and older-nondistinctive groups did not differ significantly, t(22) = .93, p = .363, the younger-distinctive and older-distinctive groups did, t(22) = 4.19, p = .0004. In addition, there was a significant difference between the older-nondistinctive and older-distinctive groups, t(22) = 3.35, p = .003. The performance of the older-distinctive group thus was superior to the performance of the other three groups. The use of distinctive covers seems to enhance the ability of 94, but not 8Bmo-old infants, to keep track of a location. The proportion of correct responses on different container control trials were .50, .67, .75, and .25 for the younger-nondistinctive, younger-distinctive, older-nondistinctive, and older-distinctive groups, respectively. Only the older-distinctive group is showing an error rate that suggests the AB error. Since each infant received one different container control trial, the results can be evaluated by chi-square comThe tests yielded three nonsignificant differences; parisons.

KEEPING

55

TRACK IN INFANTS

younger-nondistinctive vs younger-distinctive, x2(l) = 0.17, p = .068, younger-nondistinctive vs older-nondistinctive, x’(1) = 0.71, p = .399, and younger-distinctive vs older-distinctive, x’(1) = 2.69, p = .lOl; and one significant difference; older-nondistinctive vs older-distinctive, x’(l) = 4.17, p = .041. The somewhat surprising conclusions are (a) that the An error is not occurring with high frequency and (b) that older infants are more likely to make the AB error when distinctive covers are used. Visual Tracking Table 2 presents the proportion of some container trials on which infants visually tracked the correct location of the hidden object, i.e., looked at the cover of the correct container during reorientation. Infants were scored as visually tracking the correct container if they kept their eyes on it during the period between the halfway point and the completion of reorientation. Infants who looked back and forth or looked away were not scored as visually tracking. The control trials were eliminated since, when the infant is reoriented on these trials, there is no left-right reversal of the array and visual tracking would be less important for correct responding. An age x sex x distinctiveness x reorientation analysis of variance indicated significant interactions of sex, distinctiveness, and reorientation, F(1, 40) = 5.44, p = .025, and distinctiveness and reorientation, F(l,40) = 5.44, p = .025, and a significant main effect of distinctiveness, F(1, 40) = 23.81, p = .00002. These effects occurred because, except for male infants on move platform trials, visual tracking of the correct container occurred more frequently when distinctive covers were used. Female infants performed identically on move platform and move infant trials; they tracked on .25 of the trials with nondistinctive covers and ~OWRTION

OF SAME CONTAINER

TRIALS DURING

TABLE

2

THAT

CORRECT

Reorientation Group Youngernondistinctive Youngerdistinctive Oldernondistinctive Olderdistinctive Combined

LOCATION

WAS

VISUALLY

TRACKED

REORIENTATION

trials

Move platform

Move infant

.42

.33

.38

.7.5

.92

.83

.50

.I7

.33

.58 .56

.83 .54

.71 .55

___-

Combined

--

____.

56

GOLDFIELD

AND

DICKERSON

.83 of the trials with distinctive covers. Similarly, male infants tracked on .25 of the move infant trials with nondistinctive covers and .91 of these trials with distinctive covers. On move platform trials, however, male infants tracked on .67 of the trials with nondistinctive covers and SO of the trials with distinctive covers. Table 3 presents the contingency between visual tracking and manual search. In all but one instance the proportion of correct responses is higher on trials where visual tracking occurred. Two chi-square tests, one for move platform and one for move infant trials, were conducted in order to determine whether the relation was significant. The tests compared correct and incorrect responding on trials where visual tracking did and did not occur for all 48 infants. The results indicated that the contingency was significant for move infant trials, x2 (1) = 4.01, p = .045, and nonsignificant for move platform trials, x’ (1) = 1.28, p = .258. Response Patterns

Each infant received six test trials, one same container and one different container, with each of the three types of reorientation. The pattern of responses over the six trials are used here to index the determinants of an infant’s responses. The analysis considers the role of two determinants, (a) whether infants search in the location where the object was hidden on the preceding training trials or in the location where the object was hidden on the test trial, i.e., whether or not they make the AI? error, and (b) whether infants keep track of the container during reorientation PROFQRTION

TABLE 3 OF CORRECT RESPONSES ON SAME CONTAINER TRIALS AS A FUNCTION TRACKING OF CORRECT LOCATION DURING REORIENTATION Reorientation

Group Youngernondistinctive Youngerdistinctive Oldernondistinctive Olderdistinctive Combined

Tracking

Move

platform

OF VISUAL

trials Move

infant

Yes No

.40 .I4

.25 .25

Yes No

.44 .33

.27 .oo

Yes No

.67 .50

.50 .I0

Yes No YCS No

.86 .80 .59 .43

.60 .oo .41 .14

KEEPING

TRACK

57

IN INFANTS

in which they would have searched without reorientation. The combining of these two two-level determinants produces four types of “response systems.” Which of the four response systems determines an infant’s responses is indicated by the pattern of responses over test trials. In a Type 1 response system the infant both makes the AB error and fails to keep track. She, therefore, selects the position (left or right) in which the object was hidden on the training trials. In a Type 2 response system the infant does not make the An error but still is unable to keep track. She, therefore, responds to the position in which the object was hidden on the test trial. In a Type 3 response system the infant still makes the AB error but is able to keep track. She, therefore, responds to the container in which the object was hidden on the training trials. Finally, in a Type 4 response system the infant does not make the AB error and is able to keep track. She, therefore, responds to the container in which the object was hidden on the test trial. This last response system produces consistently correct responses. The patterns of correct and incorrect responses predicted by the different response systems are shown in Table 4. There are 64 (26) possible patterns of correct and incorrect responses, only four of which are perfectly consistent with the response systems. Only six infants showed these ideal patterns. Twenty-eight of the remaining 42 infants, however, showed patterns that differed from an ideal (6-O) pattern on only a single trial. Since each ideal pattern differs from the other three on at least three trials, it is possible that the patterns that differed on a single trial (5-l patterns) also were indicative of response systems. With this assumption invoked, 34 of 48 infants were scored as showing patterns consistent with response systems. The question arises as to whether the inference of a response system from a response pattern is justified, especially when the imperfect 5-l patterns are included. One justification is derived by comparing the obtained number of response patterns that correspond to response systems TABLE RESPONSE PATTERN

4

FOR THE FOUR SYSTEMS

Test trials

Response system

Move platform Same container

Move platform Different container

Move infant Same container

Move infant Different container

Control Same container

Control Different container

Type 1

-

+

-

+

+

-

Type 2 Type 3 Type 4

+ +

+ + +

+ +

+

+ +

+

58

GOLDFIELD

AND DICKERSON

with the number expected under an assumption of chance responding. Since 28 of the 64 possible response patterns index response systems, chance responding would lead to an expectation that 21, i.e., 48 (28/64), infants would show response systems. The obtained value of 34 differs significantly from the expected value of 21, x’(l) = 14.31, p = .0002. Infants are thus showing patterns that correspond to response systems at a significantly greater than chance frequency. A second method of justifying the analysis is to construct a simple model of infant behavior that would predict the frequency of 6-0, 5- 1, and residual response patterns. To do this two assumptions are required; (a) each infant’s behavior is determined by one of the response systems and (b) on a given trial the infant with a certain probability makes a “mistake,” perhaps because of inattention or distraction, and selects the wrong container, i.e., the one that is opposite to the one dictated by her response system. Since all four response systems predict a correct response on same container control trials, these trials can be used to estimate the probability (p,) of such a “mistake.” Given these assumptions, it is possible to predict the number of 6-0,5-l, and residual response patterns. These computations were made separately for the four age x distinctiveness groups.’ Table 5 gives the results of the computations, i.e., the expected number of 6-0, 5-1, and residual patterns, along with the obtained number of such patterns. The fit between the expected and obtained values appears generally good. Two chi-square tests were conducted in order to test the goodness of fit. The first test compared the expected and obtained numbers of response systems (6-O plus 5-l) and residual patterns over the four groups. The 6-O and 5-l patterns were combined because of the problem of small expected frequencies. This 4 x 2 (group x pattern) analysis showed that the difference between the obtained and expected frequencies would occur more than 80% of the time by chance, x2(6) = 2.29, p = .891. The second test compared the expected and obtained numbers of 6-0, 5-1, and residual patterns for the four groups combined. This analysis showed that the difference between the expected and obtained frequencies would occur more than 30% of the time by chance, x’(2) = 2.32, p = .314. The statistical tests thus confirm that the fit between the predictions of the model and the obtained values is relatively good. Since the model assumes that each infant’s behavior is determined by a response system, the response system analysis derives support. Table 6 presents the proportion of infants showing different response systems as a function of age and distinctiveness. The incidence of the ability to keep track of a location can be determined by comparing the ’ An appendix describing the steps for computing the expected frequencies of the four response systems is available upon request.

KEEPING

TRACK TABLE

PREDICTED

AND OBTAINED

NUMBER

OF SUBJECTS

59

IN INFANTS 5 EXHIBITING

DIFFERENT

RESFQNSE PATTERNS

Group’

P,

Pattern

Predicted no.

Obtained no.

Youngernondistinctive

.25

6-O 5-t

2.32 6.18 3.50

7.00 3.00

0.93 5.05 6.02

1.00 7.00 4.00

1.41 5.75 4.84

1.00 7.00 4.00

4.00 5.99 2.01 8.66 22.97

2.00 7.00 3.00 6.00 28.00 14.00

Residual Youngerdistinctive

.42

6-O 5-1

Residual Oldernondistinctive

.33

6-O 5-1

Residual Olderdistinctive

.I7

Total

.29

6-O 5-1

Residual 6-O 5-1

Residual

2.00

16.37

-

-

a n = 12 for each group.

frequency of Types 1 and 2 response systems, where keeping track is absent, with the frequency of Types 3 and 4 response systems, where keeping track is present. The results are consistent with the earlier manual search results. Only a small proportion of the younger infants are able to keep track in both distinctiveness conditions, and while older infants have similar difficulties in the nondistinctive condition, the presence of distinctive covers facilitates their keeping track. Fifty percent TABLE PROFQRTION

OF INFANTS

SHOWING

6

THE DIFFERENT

RESPONSE SYSTEMS

Response systems Group Youngernondistinctive Youngerdistinctive Oldernondistinctive Olderdistinctive Total

Residual

Type 1

Type 2

.25

.33

.08

.08

.25

.17

.25

.08

.I7

.33

.25

.25

.oo

.I7

.33

.08 .I9

.I7 .25

.25 .I0

.25 .I7

.25 .29

Type

3

Type

4

60

GOLDFIELD

AND

DICKERSON

of the older infants (67% of the infants with categorizable patterns) in the distinctive condition showed response systems consistent with keeping track. ‘The incidence of the AB error can be determined by comparing the frequency of Types 1 and 3 response systems, where it is present, with the frequency of Types 2 and 4 response systems, where it is absent. A surprisingly low incidence of the AB error is indicated again. Only 29% of the infants in the total group (41% of the infants with categorizable patterns) showed response systems indicating its presence. Furthermore, there clearly is no significant difference in the incidence of the error as a function of age or distinctiveness. These results are generally consistent with those reported earlier on correct responses on different container control trials. One difference is that the earlier data indicated a higher incidence of the An error for the older-distinctive group. The results also are pertinent to the issue of whether infants making the AB error perseverate in responding to the previously-rewarded direction or to the previously-rewarded place (container). In order to examine developmental differences two Fisher’s exact tests were conducted, one for each age group, comparing the incidence of Type 1 (perseverative directional error) and Type 3 (perseverative place error) response systems in the distinctive and nondistinctive conditions. The test for the younger infants showed that the type of error was unrelated to distinctiveness (p = .286). Younger infants showed more Type 1 than Type 3 response systems regardless of the distinctiveness of the covers. The test for the older infants, however, indicated that type of error was influenced by distinctiveness (p = .029). The older infants showed more Type 1 than Type 3 response systems with nondistinctive covers, and more Type 3 than Type 1 response systems with distinctive covers. The results in Table 6 speak also to the contingency between keeping track of a location and making the An error. It appears that there is no contingency worth noting. If the infants who do not keep track of a location (Types 1 and 2) are compared with those who do keep track (Types 3 and 4) then slightly more do not make the An error (Types 2 and 4) than make it (Types 1 and 3). Similarly, when infants who make the AB error (Types 1 and 3) are compared with those who do not (Types 2 and 4), then slightly more infants do not keep track (Types 1 and 2) than do keep track (Types 3 and 4). These patterns tend to hold for all four age by distinctiveness groups. An exception is the older-distinctive group where the pattern for the latter comparison is different. No contingency appears in this case either, however. In order to show the absence of the contingency more clearly the totals at the bottom of Table 6 have been reorganized in Table 7. This table gives the proportion of total categorizable response patterns, i.e., all but residual patterns, that fit each cell of a 2 x 2 matrix formed by

KEEPING

TRACK TABLE

CONTINGENCY

BETWEEN

KEEPING

7 TRACK

AND THE

ti Keeping track Absent Present

61

IN INFANTS

AB ERROR Error

Present

Absent

.26 .I5

.35 .24

combining the presence and absence of the AB error with the absence and presence of the ability to keep track. The Type 1, Type 2, Type 3, and Type 4 response systems fit the upper left, upper right, lower left, and lower right quadrants of the matrix, respectively. A contingency chisquare test of the frequencies upon which the proportions are based was nonsignificant, x2 < 1. DISCUSSION The 88 mo (younger) infants in the present study, like the 6-mo-olds in Acredolo’s (1978) study, seemed to specify the location of an event only in terms of its position relative to their own bodies. These infants were generally unable to keep track of the location of the hidden object when movement reversed the left-right positions of the two containers, even in the distinctive cover condition. The 94 mo (older) infants, however, performed much like the 9-mo-olds in Bremner’s (1978b) study. With nondistinctive covers they were generally unable to keep track but they were generally able to do so with distinctive covers. This pattern of results supports Piaget’s (1954) claim that infants have an egocentric conception of space at the beginning of sensorimotor stage 4. By the middle of stage 4, however, the infant’s conception of space is no longer completely egocentric. The visual tracking of the correct location during reorientation was not related to age but was related to the sex of the infant and to the distinctiveness of the covers. Infants of both sexes were more likely to track the correct container on move infant trials when distinctive covers were used than when nondistinctive covers were used. The results were in the same direction for female infants on move platform trials but not for male infants. Distinctive covers seem, therefore, to attract the attention of infants more than nondistinctive covers in most cases. Bremner’s (1978b) suggestion that keeping track of a location might be related to the visual tracking of that location during movement was at least weakly supported by the results. The proposed contingency between correct responding and visual tracking was statistically significant for move infant trials but not for move platform trials. It appears

62

GOLDFIELD

AND

DICKERSON

that while visual tracking may facilitate keeping track when the infant moves, its absence does not preclude keeping track. The developmental change in the ability to keep track that was observed in this study may be explained by learning that takes place toward the end of the first year. Most infants begin to crawl at around 8 mo of age (Kopp, 1979) and, thus, they are in a position for the first time to discover that relations among stationary objects remain invariant during movement (i.e., Gibson, 1969; Gibson, 1979). By attending to these invariant relations during movement they develop a conception of space which has multiple reference points. Of course, the relations among the objects are more apparent when the covers are distinctively different, which explains why keeping track of landmarks develops first, Infants were more likely to search correctly on move platform trials than on move infant trials. Bremner (1978b), on the other hand, reported higher accuracy on move infant trials. The discrepancy may be due to a difference in the move infant procedures. In both studies the experimenter maintained a position opposite the infant as the infant was moved around the platform. The location of the infant’s mother during reorientation differed, however. In Bremner’s study the mother stood behind the experimenter before movement and remained in position during and after reorientation. It is possible that with this procedure the infant used the mother as an additional landmark and thus was better able to keep track of the location. In the present study the infant was seated on the mother’s lap and both moved on move infant trials. The mother was not, as a consequence, available as a stable landmark. Piaget’s assertion of a close relation between spatial development and the development of object permanence was not supported by the findings of the present study. The response system analysis indicated no contingency between the ability to keep track (spatial development) and the absence of the AB error (object permanence). The results tend to confirm Piaget’s contention that only stage 4 infants specify the location of an event in terms of its position relative to their own bodies. Infants who are only a bit older, however, can specify the location of an event in terms of its position relative to distinctive landmarks. Both younger and older infants, additionally, showed about the same frequency of the AB error and, more importantly, the expected contingency between the ability to keep track and the absence of the An error was not observed. Piaget’s assertion of a close tie between these phenomena must be called into question by the results. The response system data showed developmental differences in the type of perseveration occurring when infants make the AB error. Types 1 and 3 response systems both indicate the presence of the error, but Type 1 reflects a perseverative directional response while Type 3 reflects

KEEPING

63

TRACK IN INFANTS

a perseverative place response. The 8&mo-old infants showed more Type 1 than Type 3 response systems regardless of the distinctiveness of the covers. The 9&mo-old infants also showed more Type 1 than Type 3 response systems with nondistinctive covers, but showed more Type 3 than Type 1 response systems with distinctive covers. The results, thus, indicate that 8&mo-old infants are primarily directional perseverators, while 9&mo-old infants are directional perseverators when landmarks are not available and place perseverators when they are available. The 9fmo-old infants in this study thus seem to perform like the 9-mo-old infants in previous studies (Bremner, 1978a; Bremner & Bryant, 1977). The low incidence of the An error in this study was surprising. While it is possible that the 8kmo-olds in the present study had developed beyond the point where the AI? error occurs, no one else has reported the absence of the error at this age. It seems more likely that the procedures used in this study reduced the frequency of the An error. The critical factor may be what was done when the infant first searched incorrectly for the hidden object. Most studies have used a noncorrection procedure where the infant is prevented from searching further. This study used a correction procedure where the infant was allowed to continue search after an error. It may be that with this procedure the infants learned during the early training trials that the object did not exist at a particular place but, instead, existed where it was last seen. This interpretation is supported by the results of a study by Cornell (1979) in which a correction procedure was also used. Infants 9-mo-old became increasingly likely to search correctly over a series of 12 training trials in which the hiding container and its position varied. They apparently were learning to search where they last saw the object. REFERENCES Acredolo, 1978,

L. P. Development

of spatial orientation in infancy. Developmental

Psychology,

14, 224-234.

Bremner, .I. G. Spatial errors made by infants: Inadequate spatial cues or evidence of egocentrism? British Journal of Psychology, 1978, 69, 77-84. (a) Bremner, J. G. Egocentric versus allocentric spatial coding in nine-month-old infants: Factors influencing the choice of code. Developmental Psychology, 1978, 14, 346-355. (b) Bremner, .I. G., & Bryant, P. E. Place versus response as the basis of spatial errors made by young infants. Journal of Experimental Child Psychology, 1977, 23, 162-171. Cornell, E. H. The effects of cue reliability on infants’ manual search. Journal of Experimental

Child

Psychology,

Gibson, E. J. Principles Appleton-Century-Crofts, Gibson, J. J. The ecological

of

1979. 26, 81-91. perceptual learning

and

New

York:

Boston: Houghton

Mifflin,

development.

1969. approach

to visual

perception.

1979.

Gratch, G. Recent studies based on Piaget’s view of object concept development. In L. B. Cohen & P. Salapatek (Eds.), Infant perception: From sensation to cognition. New York: Academic Press. 1975.

64

GOLDFIELD

AND DICKERSON

Harris, P. L. Development of search and object permanence during infancy. Psychological Bulletin, 1975, 82, 332-344. Harris, P. L. The child’s representation of space. In G. Butterworth (Ed.). The child’s representation of the world. New York: Plenum, 1977. Kopp, C. B. Perspectives on infant motor system development. In M. Bornstein & W. Kessen (Eds.), Psychological development from infancy. Hillsdale, New Jersey: Erlbaum, 1979. Miller, D. J., Cohen, L. B., & Hill, K. T. A methodological investigation of Piaget’s theory of object concept development in the sensorimotor period. Journal of Experimental Child Psychology. 1970, 9, 59-85. Piaget, J. The construction of reality in the child. New York: Basic Books, 1954. RECEIVED:

December 3, 1979;

REVISED:

October 7, 1980.