Abstract
The present study explored children’s developing understanding that the same stimulus or event can support multiple legitimate interpretations (here called interpretive diversity). Nineteen boys and girls ranging in age from five through eight years were given 3 sets of tasks meant to assess their understanding of interpretive diversity. The first set of tasks involved cases in which the stimuli were inherently ambiguous (e.g., a line drawing that could be seen as either a duck or a rabbit). The second set of tasks determined participants’ sensitivity to differences in interpretation that can arise as a result of differences in story characters situational history, while the third involved differences in story characters personality characteristics. Scores on the three different types of tasks were found to be reliably different, with situational tasks being the easiest, and the other two types of tasks having approximately the same level of difficulty. It is argued that this pattern of results may be due to differences in the ease with which children can simulate the mental states of other people in these different types of situations.

Children are not born with the knowledge that people have private mental lives, filled with unique beliefs, desires, and intentions. Young children typically expect everyone to know, think, and feel the same thing. Over the course of the preschool years, they acquire the ability to attribute mental states to others ("he thinks there are cookies in the jar, but it’s really empty"), which in turn enables them to predict and explain the behavior of those around them ("he looked in the cookie jar because he wanted a cookie, now he's sad because it is empty"). The arrival of these skills has come to be called the development of a "theory of mind." Theory of mind is one of the most widely researched topics in the area of developmental psychology, and is the source of a very lively debate. Researchers agree that children begin their journey with the realization that there is a relationship between knowledge and perception (if you see it, you know it). What is still a matter of contention, however, is whether this discovery is made separately from the realization that it is possible for people to have different, yet legitimate beliefs about exactly the same thing. The importance of this insight becomes apparent when we notice that it must come with an awareness that mental content is not simply the result of perception or experience–beliefs are created in the mind rather than simply copied from the world. This discovery marks the advent of an "interpretive theory of mind." The present research was conducted in an effort to discover whether the children’s understanding of the nature of interpretation develops in a step-like fashion with different forms or sources of interpretive diversity being easier to comprehend than others.

The inability of young children to attribute mental states to others may initially come as a surprise to some adults, but is easily demonstrated by a "false-belief" task. In the classic form of this test, a child is shown a Smarties box and asked to take a guess at the contents–the obvious answer being of course, Smarties. The box is opened to reveal pencils. The surprised child is then asked to anticipate what some third party, who was not present when the box was opened, would suppose the box to contain. A child of the age of two or three, would typically answer "pencils"–expecting the other person to somehow be aware of the true contents of the box. This type of error is known as a "reality error." Children at this stage of development are not yet cognizant of the relationship between knowledge and perception; they have no idea that people will not be aware of facts that they have not been exposed to. A child of three or four years will have made this leap, being able to correctly attribute a false belief to anyone who had been deprived of a peek into the box ("they’ll think Smarties").

Children who have only just become capable of passing false-belief tasks are still limited in their understanding of the mind to what Chandler and Boyes (1982) have dubbed "copy theory." They are only aware of the mind as a recording device–an entity that copies the world, without having any interpretive influence over its own content. This is obviously a somewhat impoverished view of mental life relative to an adult conception in which experiences are altered as we assimilate them, attaching meaning and significance to them as we go along. What is lacking in children at this stage of development is an understanding of the mind’s influence on mental content. Minds not only "fit themselves to an independent reality, [but are] also capable of altering or deforming that reality in such a way as to make it better conform to the sorts of intelligence seeking to understand it" (Chandler & Lalonde, 1996, p. 112). Children who realize that ignorance can lead to a false belief understand the relationship between perception and knowledge, and as such understand beliefs as a representations of the world. They have not, however, begun to understand that the mind is active in interpreting the world and that beliefs are constructed by the mind as much as by the reality the mind seeks to represent. The fact that is it possible for people to have different, legitimate interpretations of the same thing is therefore beyond their grasp.

Deciding just when it is, exactly, that children attain what can be called an interpretive theory of mind is a very contentious issue. There are those who believe that the ability to pass false-belief tests is indicative of an interpretive theory of mind (Wimmer & Perner, 1983; Gopnik & Astington, 1988; Perner, 1991; Ruffman, Olson, & Astington, 1991). According to Perner, the beginning of an understanding of knowledge as representation carries with it a number of other characteristic abilities, one of which being an understanding that the mind is actively interpretive (Perner, 1991). While it is generally admitted that a child’s understanding of interpretation at this stage is limited, supporters of this view contend that no further conceptual advances take place: the only thing that stands between the child’s early view and an adult conception is practice.

In contrast to this view, a second group of researchers (e.g., Carpendale & Chandler, 1996; Chandler & Lalonde, 1996; Pillow 1991, 1993; Taylor, 1988) hold that false-belief tasks measure only an appreciation of ignorance rather than interpretation. Chandler and Lalonde (1996), for example, have found that children as young as three years of age are able to pass false-belief tests, but that it is not until age 5-7 that children begin to show some understanding of interpretation. False-belief understanding appears to be a necessary prerequisite for an interpretive view of knowledge acquisition, but is by no means sufficient. The results of a study by Carpendale and Chandler (1996) are also consistent with this hypothesis. The present study, which falls in this second camp, attempts to uncover more milestones along the route toward the development of a fully interpretive theory of mind. One possibility is that children’s ability to see the potential for interpretive diversity depends on the kinds of stimuli used.

Research by Gnepp and Gould (1985), and Pillow (1991, 1993) has examined the child’s developing capacity to factor in personal characteristics and immediate personal history when making inferences about another person’s reaction to an event. Results indicate that younger children tend to be more likely to make use of situational variables than personality characteristics in their explanations of other people’s mental states. There is also an age related increase in their tendency to make use of personality characteristics when attributing mental states to others.

Such findings have led us to hypothesize that the developing child finds some kinds of stimuli easier to understand as sources of interpretive diversity than others. The thesis set forward here is that children will first be able to understand that one stimulus/event can support multiple legitimate interpretations when there is something ambiguous in the nature of that stimulus. They will then come to appreciate that a person’s immediate situational history can have an impact on their reactions. There will finally be a recognition of the influence that personality characteristics can have on the way we interpret our experiences. For the purpose of this experiment, these different kinds of stimuli will be known as stimulus, situational, and personologically based sources of interpretive diversity. We predict that children will begin to see how these different kinds of stimuli can lead to multiple legitimate interpretations over the ages of five through eight, and that their understanding of them will appear in a particular developmental order. The nature of these different sources of interpretive diversity make it reasonable to expect these results: they can be placed on a continuum, from concrete to abstract. Ambiguities in a stimulus are directly observable in the real world. Differences in people’s experiential history can be seen as well, but are made distant by time. Personality characteristics however, are abstract constructs, and are therefore never directly observable. They can only be inferred after observing a person’s behavior over a period of time. It seems natural that children would take longer to understand, and realize the relevance of some of these more abstract concepts their relations to the way in which we interpret our experiences.

Method

Twenty boys and girls ranging in age from five to eight years and attending the University of Victoria Childcare Complex participated in the study. Our intention to test a larger sample of children was thwarted by a school strike that led parents of children originally enrolled in the study to seek alternative care. Time constraints imposed on the study prevented us from extending the study long enough to obtain the number of participants we originally wanted.

Subjects were given one test of false-belief understanding, and nine different tests of their ability to appreciate interpretive diversity. The 9 interpretive tests were composed of three different kinds of stimuli (three for each different source).

False-Belief Test.

The false belief measure used was part of a "droodle task" developed by Chandler and Helm (1984). Because this same task also served as the basis for assessing participant’s interpretive understanding of ambiguous stimuli, the procedure and stimuli are described in full in the following section.

Stimulus-based Sources of Interpretive Diversity.

Three different tasks were used to assess participant’s understanding of the fact that ambiguous stimuli can result in interpretive diversity.

1. "Droodle" task (Chandler and Helm, 1984):

In this task, children are shown a cartoon drawing depicting an elephant eating a peanut (see Figure 1), and asked to describe the picture. A cover is then placed over the drawing, leaving only a small portion of the original drawing visible. Participants are then introduced to a doll (Michelle), who has "seen" only this small portion of the original picture. They are asked what Michelle might think this is a picture of. A child who has is unable to attribute false beliefs to others will typically respond that Michelle will think this is a picture of an elephant eating a peanut. Because it would be extremely unlikely for anyone who has seen only the limited view of the picture to interpret the picture in this way, such a response can be considered a "reality error" and indicates that the child is inappropriately attributing her own privileged knowledge to the puppet. Children who understand the role of perception in belief formation will attribute a false-belief to the doll; they will say the doll thinks this is a picture of a chimney, or a hockey stick, or some other imagined thing. This first portion of the task served as the measure of false belief understanding. Children who attributed a false belief to Michelle (that is, said that she would think the picture was something other than what the child herself knew it to be), were credited with an understanding of false belief and a copy theory of mind.

Figure 1: Elephant eating a peanut (Droodle task)




In order to assess whether a subject is aware of the interpretive nature of belief formation, we went on to ask the child how a second doll (Jim) would interpret the restricted view of the cartoon. Because the restricted view is so impoverished–consisting of just a few intersecting lines–it affords a large set of different, yet legitimate, interpretations as to its identity or content. A child who (like adults) is aware that one could rightly "see" almost anything in this tiny picture, will claim that, in contrast to what they said Michelle thinks, Jim will think the picture contains something new and different. Michelle is free to entertain one interpretation (one false belief), while Jim can entertain another. Thus, the child is said to have an interpretive theory of mind if they attribute a false belief to Michelle, and then go on to attribute a different false belief to Jim.

In scoring this task, a child who claims that Michelle and Jim will hold the same false belief (or a child who commits two reality errors) is not considered to have an understanding of interpretation and is awarded a score of 0. Children who give two different false beliefs are said to have passed the task and are given a score of 1.

2. Ambiguous Referential Communication (Carpendale and Chandler, 1996):

In this task, a penny is hidden under one of three cards, the backs of which are marked by a large red block, a large blue block, or a small red block. The participant is told that there is a penny "under the card with the red block." Two puppets are introduced and made to endorse two different versions of this ambiguous statement (one puppet claims the penny is under the large red block, while the other claims it is under the small red block). The child is then asked whether or not it is "OK" for the puppets to have two different interpretations of the drawings, and then to explain their answer.

To be scored as passing this task, children must say that it is "OK" for the puppets to have two different (yet correct) interpretations of the ambiguous statement. They must also be able to point to the ambiguous nature of the stimulus as an explanation for this difference of opinion. Children who are unable to state that the stimulus is the source of interpretive diversity in this task, or who say that it is not "OK" for the puppets to have different ideas, will not receive any points for this exercise (scoring ranges from 0-1).

3. Ambiguous Figures:

Subjects are shown one of two ambiguous line drawings (see Figure 2): Jastrow’s "duck-rabbit" (Jastrow, 1990), or Bugelski’s "rat-man" (Bugelski, 1990). Two puppets were once again made to endorse one of the two possible interpretations of the drawings. Participants are asked if this is "OK" and to explain their answers.

This test is scored in the same manner as the Ambiguous Referential Communication task–to receive a score of 1, participants must acknowledge that two interpretations are possible and that this results from the ambiguous nature of the pictures.

Figure 2: The "duck-rabbit" and "rat-man" (Stimulus task)

Situational Sources of Interpretive Diversity.

The multiple levels of interpretation present in the narrative stimulus materials that make up this set of tasks derive from events that transpire in the experience of story characters immediately prior to their being exposed to the new story event. These tasks will all be in the form of a short script based on those found in a study by Gnepp and Gould (1985). For example:

This is a story about a boy (girl) named Pat. One day Pat picked up his (her) gerbil, and the gerbil bit him (her), and it hurt. The next day in school, Pat’s teacher said, "Pat, it’s your turn to feed the gerbil."

How do you think Pat will feel about feeding the gerbil that day?

A child who is capable of taking situational variables into account when making inferences about another person’s mental state will appreciate that being bitten by a gerbil will likely cause Pat to be none too excited about feeding the classroom gerbil that day. A child who is not capable of such reasoning, will expect Pat to exhibit the usual warm response of her classmates when they are given the chance to feed the class pet.

Children who make use of situational information to explain the mental state the character in the script will be awarded one point; while those who make no use of such information will not be awarded any points for this exercise (range: 0-1).

Personological Sources of Interpretive Diversity.

Children who have reached this level of Interpretive Theory of Mind sophistication will be aware that personality characteristics can influence people’s interpretation of a stimulus or event. Tests of this insight were also given in the form of short vignettes. For example:

Jane and Sally are twin sisters. Like all sisters, they are different in many ways. Jane is very adventurous. Sally, unlike her sister, is very careful and always thinks before she does something. Today is their birthday. When they arrive at school, they find that they have been left presents in the cloak room. The cards are signed from Johnny, and they both know that Johnny likes to play mean tricks on people.

What do you think Jane will want to do with her present?

What do you think Sally will want to do with hers?

A child who understands the influence of personality characteristics will appreciate that these girls may view their "presents" in different ways. Jane may be excited, and want to open the present without carefully having considered the consequences, while Sally may be skeptical. A child without these insights will make inferences about the girl’s mental states based on the usually reliable dictum that everyone likes to get presents.

Scoring of these tasks is carried out in the same manner described above for situational tasks; children making use of personality information when inferring the mental state of a character in the vignette will receive one point, while children who do not make use of this information will receive a score of zero.

Results

Of twenty children who participated in this study, data from one was excluded from the final analysis due to inattentiveness during the procedure.

On the Droodles task, all children showed an understanding of false-belief. That is, all participants claimed that the first puppet would ‘think’ the picture contained something other than what the child knew it to be.

Scores on the Droodle, Ambiguous Referential Communication and Ambiguous Figures tasks were combined to form a total "Stumulus score" that indexed the child’s understanding of Stimulus sources of interpretive diversity. Scores on tests of an understanding of Situational and Personological sources of interpretive diversity were similarly combined to form "Situational," and "Personological" scores. Scores on each of these three combined measures could range from zero to three. To create an overall Pass/Fail classification on these combined measures, children scoring either 0 or 1 on any of these combined scores were classified as having failed that set of tasks (for a new score of 0), and children receiving scores of two or three were treated as having passed (a score of 1).

The pattern of mean scores for the three different types of tasks was not what we had predicted: the average score on Stimulus tasks was .47, .58 for Personological, and .84 for Situational tasks (see Figure 3). However, a repeated measures ANOVA revealed a significant effect of TASK, F (36,2) = 4.226, p < .05. Within-subjects contrasts indicated that scores on Situational tasks were reliably higher than scores on Stimulus and Personological tasks, F(18,1) = 4.800, p<.05, and F(18,1) = 9.460, p<.01. Scores on Stimulus and Personological tasks were not reliably different. A chi-square analysis of performance on Stimulus and Personological tasks indicates that there may be a relationship between these two types of tasks in subject performance–the statistic was significant at the .10 level (X² = .281, p<10). The pattern of cell frequencies indicates that Personological tasks are easier than Stimulus tasks. Chi-square analyses did not indicate any other significant associations, although the overall pattern of cell populations indicates that Situational tasks are by far the easiest, Personological tasks intermediate, followed very closely by Stimulus tasks.

This relationship between the three different sources of interpretive diversity can be seen when one looks at individual participants’ performance on the three different task types. When Stimulus and Situational tasks are compared, nine out of nineteen participants exhibited the same level of performance on both tasks. This pattern of results reveals little about the order in which these skills are acquired. That is, children who perform at ceiling by passing all of the tasks are "developmentally uninteresting". Of greater interest, at least for the present purpose, are the remaining ten participants whose performance showed some degree of variability across tasks. Nine of these 10 participants performed better on the Situational tasks than they did on the Stimulus tasks. This suggests that Stimulus tasks are the more difficult of the two (see Figure 4). When Situational and Personological tasks were examined, 12 participants showed the same level of performance on the two types of tasks. Of the remaining 7 participants, six performed better on Situational tasks. Situational tasks appear to be the easier of the two tasks once again (see Figure 5). Finally, ten participants showed equal performance on Stimulus and Personological tasks. Of the remaining 9, four performed better on Stimulus tasks, and 5 performed better on the Personological tasks. These two tasks appear to be equally difficult (see Figure 6).

Discussion

The pattern of results obtained in this study was not as what was expected. We had predicted that Stimulus tasks would be the easiest for this group of children, followed by Situational and then Personological tasks. Instead, it appears that Situational tasks are the easiest of those presented here, with little apparent difference between Stimulus and Personological tasks. While these results are unexpected, they are not unexplainable. There are at least two possible explanations for this turn of events. First, it might be the case that the pattern of performance we observed represents the true nature of children’s developmental progress. In that case, it would seem that we were simply wrong in the way we had conceptualized the developmental process as lying along a continuum from concrete to abstract forms of representation. It is possible that differences in difficulty reside instead in how easy it is for children to simulate the mental state of the characters in these tasks. In tests of an understanding of Situational sources of interpretive diversity, for example, children could easily place themselves in any one of these situations, and imagine how they themselves would respond; "what would I do in that situation." For the other two types of tasks, children must accomplish something quite different, and apparently more difficult: "If I were not me, how would I feel/react in this situation." If this rather different representational ability is, in fact, required for these tasks, then it would make sense that children pass the stimulus and personological tasks at roughly the same time.

It may, however, be too early to throw out our original thesis. There are alternative explanations for these results. As has been stated, our sample size was unexpectedly reduced by a public school strike. The low sample size alone is reason to believe that this study was an inadequate test of our hypothesis. In addition, there are a number of procedural problems with that may have affected our results.

When children were brought in for testing, they were read a short script to inform them of what they were about to experience. This script was included as part of our efforts to obtain informed consent from the children prior to testing. One section of this standard script may have been particularly problematic given our hypothesis about interpretive diversity:

I’m trying to understand how children figure out what other people are thinking. Sometimes it’s hard to tell, because different people think different things. I want to know why children think different people can think different things.

There is some concern that this passage may have unintentionally affected children’s answers in the Stimulus portions of the tests. They were initially asked whether or not it was "OK" for the two puppets to endorse alternate interpretations. They were then asked to provide an explanation for their answer. Many children who said "yes" went on to explain the legitimacy of these different interpretations as though they were merely matters of opinion. Children explained these interpretations as being legitimate because "people have the right to pick," or "different people have different feelings." There is the possibility that hearing the statement "different people think different things," while being introduced to the study increased the likelihood that children would interpret these situations as matters of personal opinion or taste, thinking that this is what the experimenter expected or wanted to hear. It is known that even preschool children understand that people are free to have different likes and dislikes (you like chocolate, I like vanilla). If our pre-test consent procedure led participants to form the impression that we were asking about matters of taste ("Is it OK to say different things"), then the sensitivity of our tests was clearly compromised.

In addition, our tests themselves may have been overly complicated for our own purpose. The Situational and Personological sections of the test, for example, included two vignettes that dealt with more than one character. This seemed to confuse many of the younger participants, who had difficulty remembering which character was which. Children who experienced such difficulties may have had an increased likelihood to claim not to know how to interpret these particular vignettes, even if they were perfectly capable of passing this general kind of task. In the future, then, it would be wise to include only one target character per vignette.

There were vignettes in the Personological group of tests in which children exhibited a strong tendency to interpret the characters traits as global. In one script, for example, one character was described as cheerful and friendly, and the other as grumpy and solitary. Children were then asked how these characters would feel as a result of events that occurred over the course of the story. Many of the children responded that one character would feel happy, and the other grumpy, simply because "they’re always that way." In these Personological tasks, it might have been better to concentrate children’s attention on how a character might act or behave rather than on feelings or emotions.

If these changes to testing procedures were made, and if a larger sample were obtained, it is possible that our original thesis would be supported; Stimulus sources of interpretive diversity may indeed be the easiest, followed by Situational and Personological sources. We had originally expected this order of acquisition because we conceived of these different sources as occupying different locations on a continuum, from concrete to abstract. Ambiguities in stimuli are directly observable, as are differences in experiential history, although made distant by time. Differences in personality characteristics are never directly observable, and must be inferred over a long period of observation. However, the developmental order that was observed here does make sense theoretically. If children do attribute mental states to others by mentally simulating their experience, then children may have an easier time simulating the mental states of others in examples of Situational interpretive diversity where it is arguably easier to imagine oneself in someone else’s shoes. All that needs to be added to one’s own experience of the event is the proviso "but what if ____ had just happened to me before I saw this"? For Stimulus and Personological tasks, by contrast, such simulation is more complicated because it is more difficult to divorce oneself from one’s own current experience of the event.

Were this study to be repeated, with the procedural modifications noted above, the results might indeed be different. Deciding just how different, however, remains a matter of interpretation.

References

Chander, M.J., & Boyes M. (1982). Social Cognitive Development. In B. B. Wolman (Ed), Handbook of developmental psychology (pp. 387-402). Englewood Cliffs, NJ: Prentice Hall.

Chandler, M.J., & Lalonde, C.E. (1996). Shifting to an interpretive theory of mind: 5- to 7-year olds’ changing conceptions of mental life. In A. Sameroff & M. Haith. The five to seven year shift: The age of reason and responsibility. (pp. 111-139). Chicago, IL: The University of Chicago Press.

Gnepp, J. & Gould, M. E., (1985). The development of depersonalized inferences: understanding of other people’s emotional reactions in the light of their prior experiences. Child Development, 56, 1455-1464.

Gopik, 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.

Perner, J. (1991). Understanding the representational mind. Cambridge, MA: MIT Press.

Pillow, B. (1991). Children’s understanding of biased social cognition. Developmental Psychology, 27, 539-551

Pillow, B. (1993, March). Children’s understanding of biased interpretation. Poster session presented at the biennial meeting of the society for Research in Child Development, New Orleans, LA.

Taylor, M. (1988). The development of children’s ability to distinguish what they know from what they see. Child Development, 59, 703-718.

Wimmer, H. and Perner, J. (1983). Beliefs about beliefs: Representation and constraining of wrong beliefs in young children’s understanding of deception. Cognition, 13, 103-128.