Appearance-Reality Distinction: Children's

Copyright 1993 by the American Psychological Association, Inc. 0012-1649/93/S3.00

Developmental Psychology 1993, Vol. 29, No. 5,907-914

Appearance-Reality Distinction: Children's Understanding of the Physical and Affective Domains Margaret Friend and Teresa L. Davis The relation between children's performance on physical and affective appearance-reality (AR) tasks was investigated with a within-subjects design. Children participated in a standard color physical task (a color cutout behind afilter)and an affective measure that used stories and pictures depicting characters who were hiding their emotions. The performance of 147 4- to 8-year-olds was examined under conditions of varying task similarity. As expected, older children performed better than younger children on both tasks. More important, children's physical and affective AR performance differed significantly only when the affective task required a greater use of inference than the physical task. Results are discussed in terms of the general representational ability underlying AR performance across domains and the ecological validity of these findings.

The ability to distinguish "what is" from "what appears to be" is critical in many domains of human cognition. Recent work on children's theory of mind has focused on the importance of the appearance-reality (AR) distinction as one indicator of children's metarepresentational ability (Flavell, Miller, & Miller, 1993). The earliest estimates place the emergence of the AR distinction at approximately 3 to 4 years of age (Gross, 1989; Harris, Donnelly, Guz, & Pitt-Watson, 1986), although many studies do notfindevidence of this ability until age 5 to 6 (Flavell, 1988; Flavell, Green, & Flavell, 1986; Flavell, Green, Wahl,& Flavell 1987). Although the AR task has been extended to a number of physical domains (i.e., visual, auditory, olfactory, and tactile), the standard task is visual in nature (Flavell et al., 1986). In one version, children are shown cardboard cutouts of animal shapes. The cutouts are placed behind afilterthat changes their color, and the children are asked what color the animal looks when it is behind thefilterand what color it is "really and truly" Six- and 7-year-olds are quite capable of responding with the "apparent" color of an object when asked the appearance question and with the "real" color of the object when asked the

Margaret Friend (now at Department of Cognitive Science, University of California, Irvine) and Teresa L. Davis (now at Department of Psychology, Middle Tennessee State University), Department of Psychology, University of Florida. This research was supported by a National Institute of Child Health and Human Development traineeship to us, awarded to the Department of Psychology, University of Florida (National Research Service Award No. T32HD07318). We wish to acknowledge Scott A. Miller, Patricia H. Miller, M. Jeffrey Farrar, and our reviewers for their valuable comments on earlier versions of this article and our research assistants, Pamela Goldsmith, Lori Driscoll, and Anne Townsend, for data collection and coding. We are also grateful to the children of Baby Gator, Tot Learning Center, and St. Michael's Episcopal Church preschools and P. K. Ybnge Elementary School for their participation in this research. Correspondence concerning this article should be addressed to Margaret Friend, Department of Cognitive Science, School of Social Sciences, University of California, Irvine, California 92717.

reality question, but 3- and 4-year-olds do not show consistently correct performance (Flavell, 1986). However, children are able to give accurate responses to the AR questions earlier (approximately 3 years of age) when the task is tactile instead of visual (Flavell, Green, & Flavell, 1989). Flavell et al. (1989) suggested that the discrepancy between the acquisition of visual and tactile AR distinctions may be due, in part, to increased perceptual availability of tactile reality over visual reality. Whereas both the real (e.g., the ice cube is really cold) and apparent (e.g., the ice cube does not feel cold to a finger wearing an insulated glove) states of objects remain perceptually available in the tactile task, in the visual task, the child must generate and maintain one (real) mental representation of an object while simultaneously considering a second (apparent) representation. Variations in perceptual availability may also contribute to differences in the acquisition of the AR distinction across additional domains. For example, recent research on children's understanding of emotion has focused on the acquisition of the AR distinction in the affective domain in which real and apparent states are never perceptually available in the same, objective sense as object properties such as color. As a result, the ability to infer emotion from relevant cues may be an important component of AR performance in the affective domain. At issue is whether an understanding of AR distinctions is evident across physical and affective domains at about the same point in development. Harris et al. (1986) suggested that, to the extent that both affective and physical AR tap the same basic cognitive processes, some correspondence between performance in these domains can be expected. Children's understanding of the AR distinction in the affective domain has been assessed in a display rules paradigm. This involves telling children a story about a situation in which the protagonist would probably feel one emotion but would have reason to appear to be feeling something else (Gnepp, 1983; Gnepp & Hess, 1986; Harris et al., 1986; Harris & Gross, 1988; Saarni, 1979). The children are asked how the protagonist feels and looks, and why he feels and looks that way. Children's correct responses and justifications provide a measure of their understanding of af-

907

908

MARGARET FRIEND AND TERESA L. DAVIS

fective display rules and of their ability to make affective AR distinctions. Several studies using this paradigm have indicated that children 10 years old and older have a more sophisticated understanding of affective AR than younger school-age children (Gnepp, 1983; Gnepp & Hess, 1986; Saarni, 1979). However, Harris et al. (1986) found that even 4-year-olds demonstrate some understanding of affective AR when they are oriented to relevant cues in the stories during questioning. Data on within-subjects comparisons of physical and affective AR are limited to a single study. Gross (1989) examined the performance of 3- and 4-year-olds on modified versions of physical and affective AR tasks. In the physical task, children were shown pens whose caps did not match the actual color of the pen. Each pen appeared to be one color (the color of the cap) but was really a different color. Instead of hearing stories in the affective task, children were shown dolls that wore masks (apparent emotion) over their faces (real emotion). Both 3- and 4-year-olds had some ability to make physical and affective AR distinctions, suggesting that the AR distinction is present early in development and that it is domain-general. However, understanding affective AR requires more than detecting discrepancies between successive facial expressions. Almost by definition, facial expression is apparent emotion and, while it is an important cue to emotion, it does not occur in isolation. Children also have access to situational information (e.g., Billy was opening a disappointing birthday present when he smiled). To discern Billy's real and apparent emotions, the child must infer his real emotion from the situational context and identify his apparent emotion from his facial expression. The identification of facial expression in the absence of situational context may not reflect a recognition of simultaneous, yet discrepant emotions (one apparent, one real) that is central to the AR distinction. Although some general level of cognitive development is sufficient for the acquisition of the basic AR distinction, further development may be required before it is generalized to more complex tasks. This may be because of differences in the perceptual availability of real and apparent states across domains and of the corresponding need to infer those states that are not perceptually available. Furthermore, the ability to infer affective states from other cues may be a critical factor in making affective AR distinctions. Children are frequently confronted with situations (e.g., receiving an unwanted birthday present) in which real emotion must be inferred from situational cues while the facial expression reflects apparent emotion. To the extent that affective AR requires cognitive skills beyond the dual representation of appearance and reality, it is reasonable to expect children's performance across physical and affective tasks to appear domain-specific. It has not been possible to address the issue of the domain specificity/generality of the AR distinction on the basis of previous research for two reasons. First, with the exception of a single study, there were no within-subjects comparisons of physical and affective AR. The ability of individual children who perform well on physical tasks to perform well on affective tasks, or vice versa, has not been assessed. Second, the task demands for physical and affective AR have traditionally been quite different. In physical tasks, children are exposed to the real state of the stimulus (i.e., its real color) as part of the proce-

dure. However, children are not exposed to a visual representation of the real emotion in affective tasks. Similarly, children are shown the apparent state of the stimulus in physical tasks but have to infer the apparent state in affective tasks. The purpose of this study is to inform both methodological and conceptual issues surrounding the assessment of the domain-specificity/generality of the AR distinction. Of methodological interest is whether the perceptual availability of real and apparent states influences children's AR performance across domains. Of conceptual interest is the extent to which physical and affective AR reflect general cognitive processes as opposed to domain-specific knowledge. We used a within-subjects design and systematically manipulated the perceptual availability of real and apparent affective states. When the perceptual availability of affective states is high, physical and affective AR task demands are the most similar. When perceptual availability is low, the affective task approximates more naturalistic problems and requires greater inference than the physical task. We made three predictions. First, when the perceptual availability of real and apparent states is similar across tasks, children should perform equally well on physical and affective AR. Second, as the level of inference in the affective task approaches what the child might encounter in a naturalistic setting, physical and affective AR performance should diverge. Finally, differences in performance across tasks should be due to the extent to which children are able to identify real physical and affective states. Apparent states, which remain perceptually available throughout the procedure, should be easily identified in either task.

Method Subjects Subjects were recruited from an elementary school associated with the university and several area preschools. One hundred and fortyseven children ranging in age from 4 years to 8 years 4 months (M= 6 years 1 month) participated. Seventy-four of the children were boys and 73 were girls. Four groups were formed: 4-year-olds (4 to 5 years; M = 4 years 6 months; A^= 33; 15 boys and 18 girls), 5-year-olds (5 years to 5 years 10 months; M= 5 years 6 months; iV= 39; 20 boys and 19 girls), 6-year-olds (6 to 7 years; M = 6 years 6 months; N = 34; 19 boys and 15 girls), and 7- to 8-year-olds (7 years to 8 years 5 months; M=l years 8 months; N = 41; 20 boys and 21 girls). Forty-nine children, an approximately equal number of boys and girls from each age group, were assigned to each of three conditions.

Materials Physical task. Stimuli for the physical AR task were cardboard cutouts of animal figures (butterfly, duck, elephant, and rabbit) varying in size from 5 to 7 cm in height and 4 to 9 cm in length. There was one white, one green, one yellow, and one red cardboardfigure.Filters for altering the apparent color of thefigureswere two 10x25 cm sheets of transparent plastic (one red and one green), which were housed in clear plastic frames. Affective task. In the affective task, we used a story-telling paradigm (Gnepp & Hess, 1986; Harris et al., 1986; Harris & Gross, 1988; Saarni, 1979). The stimuli were four stories in which a target child (Sally or Billy) was involved in situations designed to elicit positive or

909

APPEARANCE-REALITY DISTINCTION negative emotions (see the Appendix). In each story, the child was given a display rule (e.g., Billy didn't want to hurt his uncle's feelings) as a reason for Sally's (Billy's) real and apparent emotions to differ. Two children acted out each story and were photographed. Both of the children were female. In the version shown to female subjects, the children (Sally and Anne) were photographed with their hair worn down, and one child wore a dress. In the version shown to male subjects, both children (Billy and Sam) were dressed in T-shirts and pants and wore their hair up in baseball caps. For each scene within a story, a photograph depicting that scene and a typescript for the experimenter to read were glued onto 18 X 19-cm pieces of cardboard.

Procedure Before beginning either task, children received a color pretest consisting of five 7.5 X 7.5-cm colored squares of cardboard (black, orange, red, green, and white) whose colors they were asked to name. All of the children successfully completed the color pretest. The order of tasks (physical and affective) was counterbalanced across subjects. The tasks were administered by two experimenters who each tested an equal number of children from the preschools and elementary school. Physical task. The physical task was the standard color AR task taken from Flavell et al. (1987) with the exception that the children did not receive pretraining because it was not possible to give equivalent pretraining on affective AR. In addition, children were asked to justify their choices of real and apparent color. Children were shown a cardboard cutout and were asked, "What animal is this? That's right. Watch while I put the (butterfly) behind here." The experimenter placed the object approximately 1 cm behind the filter and asked, "When you look at the (butterfly) right now, does it look (red) or does it look (black)? Why does it look (black)? What color is the (butterfly) really and truly? Is it really and truly (red) or is it really and truly (black)? Why is it really and truly (red)?" The order of appearance and reality questions and the order of real and apparent states (red or black) within each question were counterbalanced across subjects. There were four object-filter combinations: white object, green filter (apparent color is green); green object, red filter (apparent color is black); red object, green filter (apparent color is black); and yellow object, red filter (apparent color is orange). The order ofstimulus presentation was randomized across subjects. Affective task. The children were told that they would hear stories about Sally (Billy) and would be asked questions about how Sally (Billy) looked and felt. The experimenter placed two full-face photographs of the target child on the table and labeled them (happy and sad). The children were asked to point to the face corresponding to each label. All of the children easily distinguished the faces on the basis of these labels. These photographs remained on the table during the rest of the affective AR procedure. The order of story presentation was randomized across subjects. As the experimenter read each scene, she placed the photograph corresponding to that scene on the table. The similarity of the affective task to the physical task was systematically manipulated. In Condition 1 (high task similarity), children were shown a picture of Sally's (Billy's) real emotion immediately following the situational information that elicited it. The experimenter removed this picture, provided a display rule as the motivation for disguising the real emotion, and showed a picture of the apparent emotion. In Condition 2 (moderate similarity), children were told how Sally (Billy) felt but were not shown a picture of the real emotion. As in Condition 1, they were given a display rule and shown a picture of the apparent emotion. In Condition 3 (low similarity), children were given a display rule and shown a picture of the apparent emotion but were not given any explicit information about the real emotion. The real emotion had to be inferred from the situational cues in the story. In each condition, the photographs portraying the situational infor-

mation and the apparent emotion remained on the table during questioning. The experimenter asked, "How does Sally really and truly feel, does she feel happy or does she feel sad?" (reality question) and "When you look at Sally right now, does she look happy or does she look sad?" (appearance question). Children pointed to one of the full-face photographs in response to each question. Next the experimenter asked for justifications, "Why does Sally feel that way?" (following the reality question) and "Why does Sally look that way?" (following the appearance question). The order of questions (reality and appearance) and of real and apparent states (happy and sad) within each question were counterbalanced across subjects. The form of the questions was chosen to be similar to the questions in the physical task.

Scoring Each child received a total of eight trials (four physical and four affective). Six summary scores ranging from 0 to 4 were calculated: physical appearance, physical reality, affective appearance, affective reality, physical justifications, and affective justifications. A score of 1 was given for each question (appearance or reality) correctly answered in each trial. Children also received a score of 1 for each AR trial in which they gave correct justifications for both the appearance and reality questions. Following Harris et al. (1986), we classified children's justifications into one of six categories: 1. Both correct: The response to the appearance question refers to the presence of thefilter(in the physical task) or to a display rule (in the affective task). The response to the reality question refers to the color of the paper (in the physical task) or to precipitating events in the story (in the affective task). 2. Reality correct; appearance = reality: A correct reply is given to the reality question but the same reply is also given to the appearance question. 3. Reality correct; appearance incorrect: A correct reply is given to the reality question and a different, but incorrect, reply is given to the appearance question. 4. Appearance correct; reality = appearance: A correct reply is given to the appearance question but the same reply is also given to the reality question. 5. Appearance correct; reality incorrect: A correct reply is given to the appearance question and a different, but incorrect, reply is given to the reality question. 6. Both incorrect: Responses to both the appearance and reality questions were incorrect or were uninformative (e.g., "I don't know"). Justifications for the physical and affective tasks were coded by two experimenters at interrater agreements of .83 and .81, respectively. Disagreements were resolved by discussion.

Results Choices of Real and Apparent States We analyzed the data using a 2 x 2 X 4 x 3 x 2 (Task X Question Type X Age X Condition X Sex) repeated measures multivariate analysis of variance (MANOVA) with task (physical or affective) and question type (reality or appearance) as repeated factors. Children performed significantly better on physical (M = 3.47) than on affective AR (M= 3.13), F(l, 123) = 23.46, p < .01, and gave more correct responses to appearance questions (M = 3.84) than to reality questions (M= 2.77), F(l, 123) = 140.40, p < .01. There was a main effect of age, F(3,123) = 12.5, p < .01, indicating that performance across tasks improved with age. Bonferroni comparisons revealed that 6-year-

910


olds, /(65) = 4.65, p < .01, and 7-year-olds, /(72) = 5.71, p < .01, performed better than 4-year-olds. The 5-year-olds did not differ significantly from any other groups. Of primary interest were the interactions of Task X Condition, F(2,123) = 5.92, p < .01; Task X Question, F{\, 123) = 55.34, p < .01; Question X Age, F(3,123) = 6.23, p < .01, and Task X Question X Age, F(3, 123) = 3.94, p < .01 (see Table 1 for means). Condition effects. Planned comparisons of the Task X Condition interaction indicated that children's affective (M = 2.92) and physical {M = 3.55) AR performance differed significantly only in Condition 3, Bonferroni r(96) = 5.05, p < .01. As predicted, physical and affective AR performance diverged when the inference required in the affective task approximated what children could be expected to encounter in naturalistic settings. When the affective task required no more inference than the physical task, differences in performance were not obtained. The effect of condition on the correspondence of children's physical and affective AR scores was also evident at the level of individual performance. A cross-classification of individual AR performance on both tasks as a function of condition is presented in Table 2. The proportion of children performing at 3 to 4 trials correct on the affective task decreased markedly from Conditions 1 and 2 (57% and 61 %, respectively) to Condition 3 (28%). In contrast, the proportion of children performing at 3 to 4 trials correct on the physical task remained relatively stable across Conditions 1 to 3 (61%, 84%, and 76%, respectively). This further corroborates thefindingthat, as the level of inference in the affective task increases, children's physical AR performance generalizes less readily to the affective domain. Question effects. Both the Task X Question and Question X Age interactions are subsumed by the higher order Task X Question X Age interaction. This interaction was due to 4-year-olds' Bonferroni /(32) = 4.50, p < .01, and 6-year-olds' Bonferroni ^(33) = 4.83, p < .01, correctly identifying real physical states

Table 1 Mean Performance and Standard Deviations on Appearance and Reality Questions as a Function of Task and Age M age Task

4 yr

5yr

6 yr

7yr

Overall

3.33 1.11

3.77 0.54

3.94 0.24

3.88 0.46

3.74 0.68

2.64 1.47

3.03 1.42

3.68 0.84

3.51 1.22

3.21 1.32

3.93 0.24

3.97 0.16

3.94 0.34

3.93 0.35

3.94 0.28

1.39 1.46

2.31 1.41

2.56 1.46

3.07 1.17

2.33 1.49

2.82 0.71

3.27 0.63

3.53 0.52

3.60 0.45

Physical Appearance M SD

Reality M SD

Affective Appearance M SD

Reality M SD

Overall M SD

Note. The maximum possible score was 4.

more than real affective states. Neither 5-year-olds nor 7-yearolds showed this pattern. The lack of a difference between the identification of real physical and affective states for 5-year-olds is attributable to their improvement over 4-year-olds in answers to questions about real affective states, r(70) = 2.65, p< .01, but not in their answers to questions about real physical states. In contrast, the lack of difference between the identification of real physical and affective states for the 7-year-olds appears to reflect a generalized understanding of the appearance-reality problem that is reflected in high scores on both questions. Analysis of Individual Stories The stories in the affective task differed from each other in terms of the valence of real and apparent states and the motivation for disguising the real affective state. In the birthday and fight stories, the situation elicited a negative (real) emotion that was to be disguised with a positive (apparent) emotion. In the airplane and drawing stories, the situation elicited a positive (real) emotion that was to be disguised with a negative (apparent) emotion. Harris et al. (1986) found that children show better performance when the emotion to be hidden is negative than when it is positive. To determine whether children responded differently to the stories, we conducted a 4 X 4 (Age X Story) repeated measures chi-square analysis with story as the repeated factor and frequency of correct responses to the reality question as the dependent measure. Responses to the appearance question were at ceiling (at least 97% of children correct) for all of the stories and so were not included. The analysis revealed main effects of age, x2(3, N= 147) = 29.36, p< .01, and story, x2(3, N = 147) = 35.17, p < .01. Contrasts controlling familywise error rate revealed that 7-year-olds (77% of children correct) performed at a higher level than 4-year-olds (35% correct), x 2 0, N=1A)= 28.60, p < .01, across levels of story. The story main effect was due to children's performing better on the birthday story (69% correct), which involved hiding a negative emotion, than on the airplane story (50% correct), x2(l, N= 147) = 19.28, p< .01, or drawing story (47% correct), x2(l, N = 147) = 26.23, p < .01, both of which involved hiding a positive emotion. Similarly, children performed better on thefightstory (70% correct) than on the airplane story, x2(l, N= 147) = 19.43, p < .01, or drawing story, x2(l, N= 147) = 27.51, p < .01. When the valence of the emotion to be hidden was positive, children had more difficulty than when the valence was negative. However, this difficulty did not seem to influence relative performance on affective and physical AR because, when the availability of real and apparent states was similar, children performed equally well across tasks. Interestingly, the two negative valence stories (birthday and fight) differ in the motivation for implementing a display rule. In the birthday story the motivation is prosocial (i.e., not hurting Uncle Jeff's feelings), whereas in thefightstory the motivation is to continue playing with the other child. Although the present study was not designed to address this issue, motivation for display rule usage does not appear to have influenced children's affective AR performance. Analyses of Justifications A 2 X 3 X 4 X 2 (Task X Condition X Age X Sex) repeated measures M ANOVA with task as the repeated factor and num-

911

APPEARANCE-REALITY DISTINCTION Table 2 Cross-Classification of Individual Appearance-Reality Performance as a Function ojCondition Affective trials correct Condition 3

Condition 2

Condition 1 Physical trials correct

0

1

2

3

4

0

1

2

3

4

0

1

2

3

4

0 1 2 3 4

10 8 2 2 0

0 4

0 2 2 0 8

4 ? 0 4 6

0 2 2 4 33

2 2 0 4 4

2 0 0 4 2

0 0 0 2 16

0 2 0 0 18

6 0 2 4 28

6 2 2 2 10

2 2 0 2 10

4 2 2 0 24

0 0 0 0 10

0 2 0 0 16

n ? 2

Note. Values represent the percentage of children (out of 49) in each cell.

ber of trials on which both justifications were correct as the dependent measure was conducted. There was a main effect of age, F(3,123) = 19.09, p < .01, indicating an increase with age in the number of trials on which correct justifications were given. Bonferroni comparisons revealed that 7-year-olds {M = 4.73) gave correct justifications on more trials than 4-year-olds (M = 1.27), f(72) = 6.94, p < .01, or 5-year-olds (M= 2.74), t(78) = 4.05, p < .01, and that 6-year-olds (M = 3.74) were correct on more trials than 4-year-olds, /(65) = 4.52, p < .01. The 6-yearolds did not differ significantly from either the 5-year-olds or the 7-year-olds. The omnibus test also revealed a Task x Condition interaction, F(2,123) = 7.26, p < .01, but Bonferroni tests did not detect any significant differences between cells. The lack of a task effect suggests that the number of correct justifications was comparable across domains. Justifications for the physical task as a function of age are presented in Table 3. This table indicates that children's physical justification errors were very systematic. When they were partially correct, children typically gave a correct justification for the apparent color but gave the same, or an incorrect, justification for the real color (see Table 3). For example, children would often explain both the real and apparent color of an object by reference to thefilter(e.g., "because this thing is in the way"). This conflation of appearance and reality, known as the phenomenism error (Flavell, Flavell, & Green, 1983), accounted for 40% to 50% of children's justifications on the physical task. Table 3 Percentage of Children's Justifications on the Physical Task as a Function of Age M age Response category

4yr

5yr

6 yr

7yr

Appearance and reality correct Reality correct; appearance = reality Reality correct; appearance incorrect Appearance correct; reality = appearance Appearance correct; reality incorrect Appearance and reality incorrect

18 0 2

35 0 1

45 0 2

54 0 2

23 29 79

26 25 13

15 30 7

16 24 2

Note. Percentages may not total 100 because of rounding.

Justifications for the affective task as a function of age and story valence are presented in Table 4. Justification errors in the affective task were less systematic than those in the physical task. When children provided a correct justification for one of their emotion choices, they often provided a correct justification for the other emotion choice as well. Rarely were children's affective justifications partially correct. To determine whether story valence influenced children's affective justifications, we conducted a 4 X 2 (Age X Story Valence) repeated measures MANOVA with the number of trials on which both justifications were correct (out of 4) as the dependent measure. The analysis revealed main effects of age, F{3,143) = 17.47, p < .01, and story valence, F(l, 143) = 66.05, p < .01. The age main effect duplicated the pattern obtained across tasks. The main effect of story valence was due to children providing correct justifications on more trials (out of 2) when the protagonist disguised a negative emotion (M = 1.05) than when the protagonist disguised a positive emotion (M = 0.54). In summary, there was an increase with age in accurate performance across AR tasks, and as predicted, when the perceptual availability of real and apparent states was similar across tasks, children performed equally well on physical and affective AR. When the level of inference required in the affective task increased (and the level of inference in the physical task remained the same), children performed better on physical than on affective AR. This pattern of performance was obtained at the level of individual data as well. Children were equally able to justify their responses across tasks but were more likely to be partially correct in the physical than in the affective task. Finally, in both choices of emotion and justifications, children performed better on affective stories that involved hiding a negative emotion than on those that involved hiding a positive emotion. Discussion Two crucial findings derive from this experiment. First, the more similar the perceptual availability of real and apparent states, the more similarly children perform across domains. Second, the basic AR distinction appears to be domain-general, although domain-specific knowledge may be necessary for good performance on more naturalistic tasks. Importantly, these patterns are obtained for both group and individual per-

912


Table 4 Percentage of Children's Justifications on the Affective Task as a Function of Age and Story Valence (Positive or Negative) Mean age 4yr

5 yr

6 yr

7yr

Response category

Positive

Negative

Positive

Negative

Positive

Negative

Positive

Negative

Appearance and reality correct Reality correct; appearance = reality Reality correct; appearance incorrect Appearance correct; reality = appearance Appearance correct; reality incorrect Appearance and reality incorrect

10 6 3 3 3 75

18 4 6 12 5 54

19 6 12 10 8 45

49 6 14 6 6 18

31 8 13 10 4 32

64 1 7 6 2 18

49 1 12 4 9 24

79 1 7 5 4 4

Note. Percentages may not total 100 because of rounding.

formance. When task demands are similar, children who perform well on the physical task also perform well on the affective task. The general age-related improvement across tasks most likely reflects developmental changes in representational ability. Flavell (1986) argued that physical AR tasks tap the ability to simultaneously consider alternative representations of an object. Given similar task demands, application of this skill to the affective domain should result in good performance. For instance, Sally could be represented as both appearing to feel happy but really feeling sad. By age 5 or 6, children are acquiring a representational competence that facilitates performance on a variety of tasks. Convergent evidence for this position is provided by the fact that the performance of 3- to 5-year-olds on false belief, appearance-reality, and representational change tasks (all of which require the dual representation of objects and events) is positively correlated and improves dramatically during this period (Gopnik & Astington, 1988). In the present study, children recognized that "what something is" may differ from "what it appears to be" across fairly complex versions of physical and affective AR tasks by approximately 5 to 6 years of age. Making physical and affective tasks more similar provides an assessment of the domain-general nature of the AR distinction. However, this account may overestimate competence in naturalistic settings in which generalization of the AR distinction is impeded by domain-specific constraints. Naturalistic affective discrepancies are more cognitively demanding than the tasks presented in Conditions 1 and 2. When children observe a person's affective appearance, access to that person's affective reality is available only through inference. In addition, the child must think recursively (e.g., Sally doesn't want her uncle to know how she feels) to recognize the motivational bases for displaying a false emotion (Harris et al., 1986). The coordination of dual representations of affective events may be necessary, but insufficient, for affective competence in complex, naturalistic tasks. Because inference may be critical to the extension of affective AR to naturalistic settings, the information provided to children in Condition 3 simulated what might be available outside the laboratory for the affective task, but not for the physical task. Children were given situational information and were shown an affective display. It was in this condition that we observed a delay in the acquisition of affective AR relative to physical AR.

This delay can be attributed to the inference required when real and apparent emotions are not perceptually available. But it also has broader implications: The acquisition of cognitive skills such as the coordination of alternative representations does not ensure competence in social-cognitive domains. Other factors such as environmental support in inferring others' emotions and in disguising one's own emotions (Fischer & Pipp, 1984) and recursive thinking (Harris et al., 1986) may precede the application of representational competence to the domain of emotions, particularly in naturalistic settings. Also, the recent work on affective display rules involving the use of dolls as stimuli has implications for the role of pretend play in facilitating performance (Banerjee, 1992). Even physical AR tasks do not necessarily tap the ability to make naturalistic AR distinctions. Complex perceptual illusions such as the moon illusion (the moon appears closer at the horizon than when it is straight overhead) and motion parallax are naturalistic analogs of standard laboratory AR problems that rely on specific sorts of domain knowledge beyond the basic AR distinction. For both physical AR and affective AR, children's performance may be an overestimation of performance outside the laboratory. Similarly, good performance in the laboratory cannot be equated with an accurate articulation of the AR problem. Even 6- and 7-year-olds provided correct justifications only about half of the time. In the physical task, children tended to focus on the reason for the apparent color (the presence of the filter) and give this same reason or a different, but equally incorrect, reason for the real color. This response tendency seems to reflect a reliance on the most salient aspect of the procedure (placing the object behind the filter). Children were less likely to be partially correct in the affective task than in the physical task. As a result, the proportion of incorrect justifications was higher in the affective task. One interpretation of this finding is that children had difficulty remembering situational and display rule information in the stories. However, the evidence does not support this argument. First, the proportion of correct justifications across tasks was comparable, suggesting that children had no more difficulty justifying affective responses than physical responses. Second, subjects in the Harris et al. (1986) study (which included memory probes) made very few memory errors, and the proportion of correct justifications in the present study mirrors that obtained by Harris et al. Therefore, memory limitations cannot account for the pattern of errors obtained in the present study.

APPEARANCE-REALITY DISTINCTION

913

Annual Meeting of the Jean Piaget Society, Montreal, Quebec, Canada. Fischer, K. W, & Pipp, S. L. (1984). Processes of cognitive development: Optimal level and skill acquisition. In R. J. Sternberg (Ed.), Mechanisms of cognitive development (pp. 46-80). New \foik: Freeman. Flavell, J. H. (1986). The development of children's knowledge about the appearance-reality distinction. American Psychologist, 41,418425. Flavell, J. H. (1988). The development of children's knowledge about the mind: From cognitive connections to mental representations. In J. W Astington, P. L. Harris, & D. R. Olson (Eds.), Developing theories of mind (pp. 244-267). Cambridge, England: Cambridge University Press. Flavell, J. H., Flavell, E. R., & Green, F. L. (1983). Development of the appearance-reality distinction. Cognitive Psychology, 15, 95-120. Flavell, J. H., Green, F. L., & Flavell, E. R. (1986). The development of knowledge about the appearance-reality distinction. Monographs of the Society for Research in Child Development, 57(1, Serial No. 212). Flavell, J. H., Green, F. L., & Flavell, E. R. (1989). Young children's ability to differentiate appearance-reality and level 2 perspectives in the tactile modality. Child Development, 60, 201-213. Flavell, J. H., Green, F. L., Wahl, K. E., & Flavell, E. R. (1987). The effects of question clarification and memory aids on young children's performance on appearance-reality tasks. Cognitive Development, 2,127-144. Flavell, J. H., Miller, P. H., & Miller, S. A. (1993). Cognitive development (3rd ed.). Englewood Cliffs, NJ: Prentice Hall. Gnepp, J. (1983). Inferring emotions from conflicting cues. Developmental Psychology, 19, 805-814. Gnepp, J., & Hess, D. L. R. (1986). Children's understanding of verbal and facial display rules. Developmental Psychology, 22,103-108. Gopnik, A., & Astington, J. W (1988). Children's understanding of representational change and its relation to the understanding of false belief and the appearance-reality distinction. Child Development, 59, 26-37. Gross, D. (1989, April). Young children's understanding of misleading emotional displays. Paper presented at the Biennial Meeting of the Society for Research in Child Development, Kansas City, MO. Harris, P. L., Donnelly, K., Guz, G. R., & Pitt-Watson, R. (1986). Children's understanding of the distinction between real and apparent emotion. Child Development, 57, 895-909. Harris, P. L., & Gross, D. (1988). Children's understanding of real and apparent emotion. In J. W Astington, P. L. Harris, & D. R. Olson (Eds.), Developing theories of mind (pp. 295-314). Cambridge, EnReferences gland: Cambridge University Press. Banerjee, M. (1992, May). Preschooler's knowledge of emotion regula- Saarni, C. (1979). Children's understanding of display rules for exprestion: Display rules and regulation strategies. Paper presented at the sive behavior. Developmental Psychology, 15, 424-429.

Instead, children who have a fragile grasp of the AR distinction do not show the same pattern of partially correct responses (i.e., a focus on the cause for the apparent state) in the affective task as they do in the physical task, because the stories do not yield perceptually salient information about the causes for apparent states. It could also be argued that the inclusion of stories that involved hiding a positive emotion resulted in an underestimate of children's AR knowledge. Children performed less well on these stories than on stories that involved hiding a negative emotion. However, the stories represent a sample of situations that children could be expected to encounter. Decrements in performance due to this variability may reflect true limitations in the generalization of AR competence to the affective domain. Also, despite differences across stories, children's physical and affective AR performance diverged only in Condition 3. These condition effects cannot be attributed to the stories themselves, but rather reflect the influence of domain-specific constraints on the basic AR distinction. In summary, careful task analysis is critical to addressing the issue of the domain-specificity/generality of the AR distinction. In previous studies, physical AR competence has been equated with maintaining a representation of the real state of an object, whereas affective competence has been equated with generating and maintaining a representation of an apparent state. The two tasks elicit different competencies, and domain differences have been confounded with procedural variation. By manipulating the availability of real and apparent states in the affective task, we have addressed these confounds and demonstrated that both physical and affective AR rely on a general underlying representational competence. However, this competence may be only a prerequisite for performance in naturalistic settings. The work on the domain specificity/generality of the AR distinction is in its earliest stages and, as careful task analyses of AR problems proceed, researchers will be in a position to estimate developmental changes in representational competence with greater precision. An exploration of across-domain AR competence in naturalistic contexts would be particularly fruitful in this regard.

(Appendix follows on next page)

914


Appendix Stories Used in the Affective Task Birthday Story Sally's Uncle Jeff gave her a birthday present. Her uncle said to go ahead and open the birthday present. When Sally opened the box, she saw the present was just a baby rattle. Condition 1: This is how Sally felt when she saw the baby rattle, (sad photo). But Sally didn't want her uncle to know how she felt because she didn't want to hurt her uncle's feelings. Condition 2: Sally was sad that the present was a baby rattle, but Sally didn't want her uncle to know how she felt because she didn't want to hurt her uncle's feelings. Condition 3: Sally didn't want her uncle to know how she felt because she didn't want to hurt her uncle's feelings. AH conditions: So she looked like this when she saw the rattle (happy photo).

Airplane Story Sally and her friend Anne were playing with Sally's new toy airplane. Every time Sally wanted to play, Anne would take the airplane away. Anne's mom saw her take the airplane and scolded Anne for being mean. Condition 1: This is how Sally felt when Anne got in trouble for being mean (happy photo). But she didn't want Anne to know how she felt because then Anne might get mad at her. Condition 2: Sally was happy that Anne got in trouble for being mean, but she didn't want Anne to know how she felt because then Anne might get mad at her. Condition 3: Sally didn't want Anne to know how she felt because then Anne might get mad at her. All conditions: So she looked like this when Anne got in trouble (sad photo).

Fight Story Sally's mother said that her friend could spend the night as long as there was no fighting. But Sally's friend called her a mean name. Just

then, Sally's mother came back to see if the girls were getting along okay. Condition 1: This is how Sally felt when her friend called her a mean name (sad photo). But she didn't want her mother to know how she felt because then her friend might have to go home. Condition 2: Sally felt sad that her friend called her a mean name, but she didn't want her mother to know how she felt because then her friend might have to go home. Condition 3: Sally didn't want her mother to know how she felt because then her friend might have to go home. All conditions: So Sally looked like this when her mother came over (happy photo).

Drawing Story The teacher told Sally and Anne to draw some pictures. Sally wanted to have the best picture so she scribbled all over Anne's drawing. When the teacher saw it, she said that Anne's drawing was very messy. Condition 1: This is how Sally felt when her teacher did not like Anne's drawing (happy photo). But Sally didn't want the teacher to know how she felt because then she might get in trouble. Condition 2: Sally was happy that the teacher didn't like Anne's drawing, but she didn't want the teacher to know how she felt because then she might get in trouble. Condition 3: Sally didn't want the teacher to know how she felt because then she might get in trouble. All conditions: So she looked like this when the teacher did not like Anne's drawing (sad photo).

Received January 30,1992 Revision received July 22,1992 Accepted December 31,1992