the influence of nonverbal cues associated with looking behavior on young children's...

21
THE INFLUENCE OF NONVERBAL CUES ASSOCIATED WITH LOOKING BEHAVIOR ON YOUNG CHILDREN'S MENTALISTIC ATTRIBUTIONS Derek E. Montgomery, Christy Moran, and Leslie M. Bach ABSTRACT: The ability of children to use looking behavior as a cue to guide their mentalistic attributions was assessed. In Experiment 1 video displays were pre- sented in which a protagonist faced one of two potential goals, half of the time standing equidistant from both targets and in the remaining trials standing closer to the target not being faced. Preschoolers consistently based their inferences of an- other's attention and goal on the direction in which the protagonist was facing. However, in Experiments 2 and 3 preschoolers experienced difficulty in correctly inferring an actor's desired goal when the protagonist's body was oriented in a direction opposite of where she was looking. Under these conditions, only young elementary school children consistently inferred that the protagonist's goal was the target being looked at (Experiment 2). The results from these three studies suggest that an important development occurring in childhood is the ability to consistently distinguish gaze from body orientation when inferring the goals of another. The implications of this development for children's understanding of the mind are dis- cussed. One of the most active research areas in recent years in the field of social cognitive development concerns children's developing theory of mind (Flavell & Miller, in press). A theory of mind is the set of implicit assumptions one makes about a hypothesized causal interrelation of mind (e.g., desires, beliefs) and behavior. The syllogism "If an agent desires X (mental state), and sees that it exists, she will do things to get X (behavior)" is a core assumption of an adult-like theory of mind (Gopnik & Wellman, The authors gratefully acknowledge Scott Miller and two anonymous reviewers for very helpful comments on an earlier version of the manuscript. The assistance of Rachel Permuth, Robert Bretveld, Christine Cappetta, and Samantha Stalling is also appreciated. This research was supported by a grant from the National Institute of Mental Health (1 RO3 MH 54528-01 ) to the first author. Address correspondence to: Derek Montgomery, Department of Psychology, Bradley Uni- versity, Peoria, IL 61625. Journal of Nonverbal Behavior 20(4), Winter 1996 @ 1996 Human Sciences Press, Inc. 229

Upload: derek-e-montgomery

Post on 10-Jul-2016

227 views

Category:

Documents


2 download

TRANSCRIPT

Page 1: The influence of nonverbal cues associated with looking behavior on young children's mentalistic attributions

THE INFLUENCE OF NONVERBAL CUES ASSOCIATED WITH LOOKING BEHAVIOR ON YOUNG CHILDREN'S MENTALISTIC ATTRIBUTIONS

Derek E. Montgomery, Christy Moran, and Leslie M. Bach

ABSTRACT: The ability of children to use looking behavior as a cue to guide their mentalistic attributions was assessed. In Experiment 1 video displays were pre- sented in which a protagonist faced one of two potential goals, half of the time standing equidistant from both targets and in the remaining trials standing closer to the target not being faced. Preschoolers consistently based their inferences of an- other's attention and goal on the direction in which the protagonist was facing. However, in Experiments 2 and 3 preschoolers experienced difficulty in correctly inferring an actor's desired goal when the protagonist's body was oriented in a direction opposite of where she was looking. Under these conditions, only young elementary school children consistently inferred that the protagonist's goal was the target being looked at (Experiment 2). The results from these three studies suggest that an important development occurring in childhood is the ability to consistently distinguish gaze from body orientation when inferring the goals of another. The implications of this development for children's understanding of the mind are dis- cussed.

One of the most active research areas in recent years in the field of social cognitive development concerns children's developing theory of mind (Flavell & Miller, in press). A theory of mind is the set of implicit assumptions one makes about a hypothesized causal interrelation of mind (e.g., desires, beliefs) and behavior. The syllogism "If an agent desires X (mental state), and sees that it exists, she wil l do things to get X (behavior)" is a core assumption of an adult-like theory of mind (Gopnik & Wellman,

The authors gratefully acknowledge Scott Miller and two anonymous reviewers for very helpful comments on an earlier version of the manuscript. The assistance of Rachel Permuth, Robert Bretveld, Christine Cappetta, and Samantha Stalling is also appreciated. This research was supported by a grant from the National Institute of Mental Health (1 RO3 MH 54528-01 ) to the first author.

Address correspondence to: Derek Montgomery, Department of Psychology, Bradley Uni- versity, Peoria, IL 61625.

Journal of Nonverbal Behavior 20(4), Winter 1996 @ 1996 Human Sciences Press, Inc. 229

Page 2: The influence of nonverbal cues associated with looking behavior on young children's mentalistic attributions

230

JOURNAL OF NONVERBAL BEHAVIOR

1994). This predictive syllogism plays a central role in how the causes of the behavior of oneself and others are conceptualized and this centrality is one reason for the high level of interest in children's developing theories of mind. The absence of such a causal-explanatory system would result in an entirely foreign and largely unrecognizable worldview in which human behavior might instead be explained in terms of physics, previous rein- forcement schedules, teleology, and so forth (Dennett, 1987).

Although determining another's desired goal in order to predict action comprises a core function of a theory of mind, attaining accuracy in this endeavor is neither straightforward nor obvious because mentation, by def- inition, is internal and often private. Typically, the desires and goals under- lying an action are not overtly announced by the actor. However, one po- tential mechanism by which children might overcome the essential privacy of first person mentation rests in the possibility that "internal" mental states are, in fact, overtly perceivable (Butterworth, 1994; Montgomery, 1996). An internal process must have an outward criterion if that process is to be imputed with any consistency and with any standard that defines accuracy (Wittgenstein, 1953). Thus, if mental states are consistently "embodied" in such a way that certain nonverbal behaviors are criteria that reveal an internal state, then the developmental task of attributing mental states to others becomes greatly simplified. For example, a child would not have to relearn the nature of another's desire and goal with every new variation in circumstance and setting if perceptually available criteria revealing those mental states consistently generalize to new situations.

In addressing how the critical ability of reading goals from nonverbal cues might develop, Baron-Cohen (e.g., Baron-Cohen, 1993, 1995; Baron- Cohen & Cross, 1992) has recently drawn attention to the importance of gaze direction as a potentially foundational cue. Gaze direction is closely monitored by human infants, as well as by a number of nonhuman species (e.g., Povinelli & Eddy, 1996), leading Baron-Cohen to argue that monitor- ing the gaze of others is an inborn characteristic in humans. This early perceptual attunement is said to provide an important foundation for the development of an "attention-goal" psychology, which is the implicit belief that "people normally look at (attend to) the object they act on" (Baron- Cohen, 1993, p.74). Put another way, very young children are said to con- sistently infer that gaze is directed toward goals. Concepts such as goal and attention can be thought of as emerging in order to explain why people are more likely to act upon things they are looking at compared to things not being looked at; as a result, one should expect children's early emerging attributions of goal and attention to be constrained by gaze direction (Baron-Cohen, 1995).

Page 3: The influence of nonverbal cues associated with looking behavior on young children's mentalistic attributions

231

DEREK E. MONTGOMERY, CHRISTY MORAN, LESLIE M. BACH

To date, however, direct evidence that children's early mentalistic at- tributions are guided and constrained by another's gaze direction is sparse. This is somewhat surprising given the importance placed by adults on the cue of gaze when making a range of mentalistic inferences (Argyle & Cook, 1976; Kleinke, 1986; Zebrowitz, 1990). For example, increased amounts of gaze toward someone can signify a greater interpersonal liking for that person (e.g., Burgoon, Buller, Hale, & deTurck, 1984; Palmer & Simon, 1995) and can lead to perceptions of such characteristics as friendliness and intimacy (e.g., Kleinke, 1986). All of these inferences presuppose, of course, the belief that gaze direction reflects the thoughts and interests of the person varying his or her looking behavior. In other words, an adult assumes that others purposefully direct their gaze toward a goal or person because of an interest in, and desire to learn more about, the target upon which gaze is focused.

Presently, the only direct test of whether young children also share the assumption that gaze direction is volitional and goal-directed suggests competence among children as young as age 3 (Baron-Cohen, Campbell, Karmiloff-Srnith, Grant, & Walker, 1995). Baron-Cohen et al. (1995) pre- sented to preschoolers and autistic children a display in which a cartoon face was surrounded by different candies depicted in four corners of the display. Each candy was equidistant from the face; however, the eyes of the face were clearly directed toward one of the candies. The 3- and 4-year- olds, but not the autistic children, consistently inferred that the candy the cartoon character wanted was the one his gaze was directed toward, even when an arrow was pointing to a candy different from where the eyes were directed. These results were interpreted as revealing that by age 3 children infer that persons tend to look at things they wish to obtain.

However, although informative, these findings are based upon chil- dren's interpretation of drawings of faces in which the eyes are especially prominent. In addition to this relative lack of ecological validity, the Baron- Cohen et al. (1995) finding does not rule out the possibility that any bodily association between person and target may be thought by children to sig- nify a desire for the target and that gaze direction holds no special signifi- cance to them. To illustrate, since gaze direction is typically aligned with the lower body, both gaze and body orientation are physical associations between actor and goal and each may influence children's mentalistic attri- butions. Another example of a general association between actor and goal is that, typically, the nearer (physically) one is to an object, the more likely one can be assumed to have an interest in it. For example, if two persons are facing an object, preschoolers infer that the person moving closer to the object is the one who desires to obtain it (Lyon, 1993). To examine the

Page 4: The influence of nonverbal cues associated with looking behavior on young children's mentalistic attributions

232

JOURNAL OF NONVERBAL BEHAVIOR

depth of children's understanding of the relation between gaze direction and goal, it is necessary to portray situations in which a protagonist is associated with one potential goal because her gaze is directed toward it and is also physically associated with another potential goal because a salient (yet, to adults, conceptually less meaningful) nonverbal cue is di- rected toward it. The purpose of the following research is to construct such scenes to assess the extent of the influence of gaze direction on children's mentalistic attributions of goal.

The possibility that preschoolers' success on the Baron-Cohen task may not reflect a belief that attention (as indexed by gaze) signifies goal is made credible by recent research revealing relatively little understanding of the relation between mentation and visual perception. For example, pre- schoolers generally indicate that a person's mind is ~not doing anything" when that person is visually examining a picture book (Flavell, Green, & Flavell, 1995a). Young children also often overestimate the attentional ca- pacity of others, saying that someone is "thinking about" or "paying atten- tion to" an object or event peripheral and irrelevant to a problem that person is attempting to resolve (Flavell, Green, & Flavell, 1995b). For ex- ample, preschoolers indicate that someone who is counting the stripes on a scarf being worn by a toy animal is simultaneously attending to peripheral features of the toy such as the sunglasses it is wearing, if, as Flavell and colleagues suggest, preschoolers' attributions of attentional focus are rela- tively indiscriminate, then it is questionable that they interpret gaze as re- vealing one's desired goal because of the understanding that persons gener- ally focus their attention upon objects of interest. Instead gaze direction may only be a learned associate of goal-directed activity. If so, then gaze direction may become less influential in attributions of goal when it con- trasts with other associates of goal-directed action.

In sum, the objective of this research is to establish whether young preschoolers' early mentalistic attributions are indiscriminate or system- atically based upon the cue of where one is looking. Specifically, this issue is addressed by presenting realistic displays of humans and examining young children's inferences of goal when gaze either covaries or contrasts with other potentially influential nonverbal cues such as spatial proximity and body orientation. Experiment 1 contrasts the influences of a general orientation toward something (gaze and body) with the influence of spatial proximity. Experiments 2 and 3 examine the relative influences of gaze direction and body orientation on goat detection. Of particular interest is comparing the accuracy of children's goal detection when these two cues are combined and when they are contrasted.

Page 5: The influence of nonverbal cues associated with looking behavior on young children's mentalistic attributions

233

DEREK E. MONTGOMERY, CHRISTY MORAN, LESLIE M. BACH

Experiment 1

Method

Participants. in this experiment 18 children in each of two age groups were tested. The age groups were 3-year-olds (range 3-1 to 3-1 1 ; M = 3 - 7) and 4-year-olds (range = 4-0 to 5-2; M = 4-7). There were 9 females and 9 males in the younger group and 11 females and 7 males in the older group. Participants were recruited from local preschools, each serving a predominantly middle-class population. One child, a 3-year-old, exhibited attentional difficulties and was subsequently replaced. All participants were tested individually in a quiet room.

Materials. Children watched as videotaped scenes were presented on an 8 in. x 11 in. color screen. Each video display depicted a college-aged protagonist (each gender was equally represented across the six videos) standing erect with a neutral facial expression. The protagonist was stand- ing in a neutral location that was positioned between a green room (lo- cated to the viewing subjects' right) and, to the left, an orange room. The rooms, so named because of their floor and wall colors, were noticeably distinct and served as the potential goals that the protagonist might wish to obtain. The clothing of the protagonist was neither orange nor green, thereby preventing children from associating the protagonist with one of the rooms on this superficial basis. These goals are deliberately neutral because young children might have difficulty understanding that one of two goals is not desired by the protagonist if both goals are perceived as highly desirable. The camera was positioned at the midpoint of the two rooms, providing the viewer a profile view of the protagonist and thereby making the eye direction clearly visible (pilot testing of these displays re- vealed that preschoolers had no difficulty indicating where the protagonist was looking).

In both the Standard and Contrast trials, the gaze and body orientation were fully directed toward the room being faced, meaning that the pro- tagonist was turned 180 degrees away from the non-goal. In the Standard trials the protagonist was equidistant from the green and orange rooms in that she was standing exactly 3 ft from both rooms. In one variant of this trial type the protagonist was facing the green room and in the other vari- ant the protagonist was facing in the-opposite direction (toward the orange room). In the two Contrast trials, the cues of gaze/body orientation and spatial proximity were opposed so that the protagonist was noticeably closer to the room that was not being faced (3 ft) than to the room being faced (9 ft). Otherwise, the Contrast and Standard trials were identical in

Page 6: The influence of nonverbal cues associated with looking behavior on young children's mentalistic attributions

234

JOURNAL OF NONVERBAL BEHAVIOR

the positioning and appearance of the protagonist. There were also two variants of the Contrast trials as the protagonist again faced either the green room or the orange room. In the third trial type (Attention), children were asked which room the protagonist was "paying attention to." There were two variations of this trial, one identical to the Contrast trial and the other identical to the Standard trial. Half of the time in each variant the actor was facing the orange room and the other half of the time was facing the green room.

Procedure. The three trial types were arranged in all possible combi- nations, resulting in six orders of presentation altogether. Both variants per trial type were blocked and presented together and their order of presenta- tion was alternated randomly across subjects. Altogether, six trials were administered to each child.

Testing began by administering a brief pretraining task familiarizing children with the term "paying attention." Subjects watched as the first experimenter (El) instructed the second experimenter (E2) to create mental images and then "pay attention" to various aspects of those images. E2 shared the nature of the imagined features by describing them verbally (see Flavell et ai., 1995b). Although young preschoolers have been found to use the mental verb "want" accurately in a number of natural settings (Bartsch & Wellman, 1995), a brief pretraining was administered for this term also to equate it in salience with the phrase "paying attention." To this end, E1 described the mental verb "want" by saying, "When I want something and I get it, then I'm happy. If I want something and i don't get it, then I'm not happy." Lastly, children were asked to point to both of the different-colored rooms. Only one child had difficulty distinguishing the colors and was subsequently replaced.

The purpose of the Standard and Contrast trials was explained by tell- ing children they would be asked to indicate either what a person "wants to do" or "is paying attention to." The video picture was then shown and stilled as participants were told, "There are two rooms, an orange room and a green room. This person really wants to go into one of these rooms. Just one. He/she wants to go into the special room." Following this descrip- tion the Desire question was posed, asking subjects to, "Point to which room this person wants to go into."

The Attention trials were introduced in a similar fashion, telling chil- dren, "There are two rooms. This person is paying attention to just one of the rooms. She is thinking about one of the rooms and paying attention to it." The Attention Question asked children to, "Point to which room this person is paying attention to." As noted earlier, there was one Attention

Page 7: The influence of nonverbal cues associated with looking behavior on young children's mentalistic attributions

235

DEREK E. MONTGOMERY, CHRISTY MORAN, LESLIE M. BACH

trial in which the protagonist was standing equidistant from both rooms and one in which the protagonist was closer to the room not being faced.

Lastly, to address whether children were capable of predicting conse- quent actions based upon knowledge of a protagonist's desired goal (a key function of a theory of mind), an Action question was posed. One of the four tapes from the Contrast and Standard trials was chosen randomly and children were told the protagonist "wants" a favorite toy located in one of the two rooms. The location of the toy was then revealed by the experi- menter (in whichever room the protagonist was facing) and subjects were asked to, "Point to where she'll go to get the toy." This question was al- ways presented after the final test trial.

Results and Discussion

Participants received one point when inferring that the protagonist de- sired to obtain (or was attending to) the goal toward which her gaze and body were directed. Responses to the Action question were counted as correct if children pointed to the room containing the toy as the location where the protagonist needed to go to satisfy his/her goal. Preliminary an- alyses revealed that there was no difference in performance between the two variants (goal as either orange or green room) for either age group across each of the three trial types, ps > .10, sign test. Also, children's inferences of the desired goal were not affected by whether the Standard or Contrast trials were presented first, ps > .10. Preliminary analyses of vari- ance also revealed that there was not a main effect of gender and that it did not interact with trial type, ps > .10.

An analysis of variance was conducted on a 2 x 2 mixed design with trial type (Standard and Contrast) as the within-subjects variable and age (3- and 4-year-olds) as the between-subjects factor. The Attention trials were analyzed separately since they were made up of both a Standard and Contrast trial and therefore were not entirely independent from these trial types. There was no overall difference between the age groups, F(1, 34) = .18, p > .10, as the 3-year-olds (M = 3.44 out of 4) and 4-year-olds (M = 3.56 out of 4) both performed quite well on the Standard and Contrast trials. Also, the accuracy of responding among the children did not differ between the two trial types, F(1, 34) = .21, p > .10. As Table 1 illustrates, both 3- and 4-year-olds consistently based their attributions of goal on the direction in which the protagonist was facing significantly more often than would be expected by chance alone (50%) in both of the conditions, ts(17) ranging from 4.24 to 7.80, ps < .001 (two-tailed). Clearly, children consis- tently used the combination of gaze direction and body orientation to de-

Page 8: The influence of nonverbal cues associated with looking behavior on young children's mentalistic attributions

236

JOURNAL OF NONVERBAL BEHAVIOR

tect another's goal, doing so even when these cues contrasted with spatial proximity. As would be expected given the consistently high performance of both age groups across all trial types, age and trial type did not interact, F(1, 34) = .21, p > .10.

Given the absence of any differences between the Standard and Con- trast trial types, it is not surprising that children performed similarly across the two different Attention trials, ps > .10 by sign test, in both age groups. Altogether, their attributions of attentional focus for each trial were based upon where the protagonist was facing significantly more often than would be expected by chance for each age group, binomial test, ps < .01.

Lastly, with the exception of one 3-year-old and one 4-year-old, every child correctly answered the Action question (94.4% correct altogether). This high level of success indicates that preschoolers apparently assume that action is a consequence of desire. In other words, even the youngest children in our study appear to have been conceptualizing the desire to obtain a goal as a construct upon which they could make causal predic- tions of subsequent action.

Experiment 2

One limitation of Experiment I was that the influences of body orientation and gaze on goal detection could not be assessed independently. Thus, it is unclear whether it was the gaze of the protagonist that was guiding chil- dren's mentalistic inferences. This is an important issue because, concep-

TABLE 1

Number of Children Correctly Responding on 0, 1, or 2 Trials in Experiment 1

Number of trials answered correctly

Standard Contrast Attention Age group 0 1 2 01 2 0 1 2

3-year-olds 1 4 13* 1 5 12" 0 5 13* 4-year-olds 0 4 14" 2 0 16" 1 2 1.5"

*The number of children answering both trials correctly within a trial type exceeds chance level (25%), binomial test, ps < .05.

Page 9: The influence of nonverbal cues associated with looking behavior on young children's mentalistic attributions

237

DEREK E. MONTGOMERY, CHRISTY MORAN, LESLIE M. BACH

tualty, an attention-goal psychology should be more influenced by the di- rection of one's gaze than by the orientation of one's body. Recall that an attention-goal psychology holds that one normally acts to obtain the target he or she is looking at. Accordingly, Experiment 2 presents scenarios to children in which the front of a protagonist's body (e.g., torso and legs) is facing toward one object while the head is turned so that the gaze is di- rected toward a different object. Of additional interest in Experiment 2 is the influence of body orientation when it is combined with spatial prox- imity since both are arguably associates of goal-directed behavior and yet conceptually less meaningful than gaze direction in revealing goal. The purpose of Experiment 2, therefore, is to assess whether children concep- tually distinguish among these nonverbal cues such that gaze is regarded as the more meaningful cue of goal compared to body orientation and to the combination of body orientation with spatial proximity.

Method

Participants. in this experiment 54 children, 18 in each of three age groups, participated. The age groups were 3-year-olds (range 3-1 to 4-0, M = 3-7), 4-year-olds (range 4-2 to 5-1, M = 4-7), and young grade school children (range 8-7 to 9-5, M = 9-0). There were 9 males and 9 females in each of the three age groups. Participants were recruited from local preschools and elementary schools that serve a predominantly mid- die-class population. Each child was tested individually in a quiet room. The 8- and 9-year-olds were added in this study because children at this age have been shown to clearly exhibit adult-like conceptions of the selec- tive and focused nature of attention (e.g., Flaveil et al., 1995b). This age group is included to validate our assumption that, from an adult stand- point, a person is more likely to want to obtain something she is looking at even when her body is facing something else and, further, even when she is also standing closer to the object not being looked at.

Materials. Children again watched as videotaped scenes were pre- sented on an 8 in. x 11 in. color screen. Each video display depicted a college-aged protagonist (each gender was equally represented across the six videos) standing erect with a neutral facial expression. The protagonist was standing in a room in which there were two tables 30 in. tall (about waist high). A separate item was displayed on ea~,h table.

The six video displays were evenly divided among three trial types. In one trial type the protagonist's body (B) faced one object but he was gazing (G) at the second object (G vs. B trials) (see Figure 1). Thus, the front of the

Page 10: The influence of nonverbal cues associated with looking behavior on young children's mentalistic attributions

238

JOURNAL OF NONVERBAL BEHAVIOR

Figure 1. An illustration of the placement of the protagonist and target objects in the G vs. B trials (Experiments 2 and 3).

protagonist's body (the torso and legs were visible) was facing one table but the head was turned at an approximate 90 degree angle toward the second table. Each table was located exactly 3 ft from the protagonist. In one trial the protagonist looked at the item located to the viewer's left (jacket) and in a second trial he looked at the item located to the viewer's right (pillow). In a second trial type the protagonist was gazing at the object her body was oriented toward (G + B trials). The positioning of the pro- tagonist, tables, and objects was identical to the G vs. B trials. In one trial the protagonist both faced and looked at the item on the table located on the left side of the screen (pants) and in the other trial she looked at and faced the item located to the right (sweater). In the third trial type a pro- tasonist looked at one object but was spatially proximal (SP) to the object that her body faced (G vs. B/SP). The distance from the object being looked at remained 3 ft but the spatially proximal object was only 1 ft from the protagonist, thus preserving the 3:1 ratio from Experiment 1. The objects being displayed in this trial type were a plaid blanket (to the viewer's right) and a shirt (located on the left side of the screen).

Page 11: The influence of nonverbal cues associated with looking behavior on young children's mentalistic attributions

239

DEREK E. MONTGOMERY, CHRISTY MORAN, LESLIE M. BACH

Procedure. The three trial types were arranged in all possible combi- nations, resulting in six orders of presentation altogether. Both variants per trial type were blocked and presented together, their sequence of presenta- tion being alternated randomly.

Although there is much evidence that 3-year-olds have a relatively sophisticated grasp of the meaning of the mental verb "want," we included the brief pretraining used in Experiment 1 to ensure consistency between the two studies. After administering this brief pretraining the purpose of the experimental task was stated by informing children, "Today I'm going to show you some videos. In each video, one person really wants to have something. What I want you to do is tell me which thing the person wants to have."

Then, the experimenter described each scene displayed on the screen in a fashion similar to the narration in Experiment 1. First, children were told, "This is a store and this person is going to buy something today! There are two things in the store, an (X) and a (Y)." The order in which these two items were mentioned was randomized and children were asked to point to each item before the procedure continued. None of the participants had any difficulty identifying the items being displayed. The narration pro- ceeded as in Experiment 1, telling children that the protagonist depicted on the screen "really wants to have one of these things. Just one. He/she wants one of these things a lot." The procedure concluded by stating the Test Question: "Point to which thing this person wants to have."

Results

Participants received one point for each time the desired goal was inferred to be the object that was being looked at by the protagonist. Thus, scores ranged from 0-2 per trial type and 0-6 altogether. Preliminary an- alyses revealed thatperformance was not affected by the location of the goal (right or left) within any of the trial types within each age group, all ps > .05, sign test. Also, analyses were conducted to determine whether per- formance on each trial type was affected by whether it was presented be- fore or after each of the remaining two trial types. There were 6 compari- sons within each of the age groups and none of these comparisons was significant using the Bonferroni correction, all ps > .008. Lastly, there was neither a main effect of gender nor an interaction between gender and any other variables of interest, ps > .10.

An analysis of variance was conducted on a 3 x 3 mixed design with trial type as the within-subjects variable and age as the between-subjects factor. There were main effects of age, F(2, 51) = 9.63, p < .01, and trial

Page 12: The influence of nonverbal cues associated with looking behavior on young children's mentalistic attributions

240

JOURNAL OF NONVERBAL BEHAVIOR

type, F(2, 102) = 6.14, p < .01. There was not an Age X Trial Type interac- tion, F(4, 102) = 1.42, p > .10.

The main effect of trial type was caused by the superior performance among children on the G + B trials (M = 1.65) compared to the G vs. B trials (M = 1.30), [(53) = 3.42, p < .01, and the G vs. B/SP trials (M = 1.31), [(53) = 2.75, p < .01. There was no difference in accurate respond- ing when comparing performance in the G vs. B trials with performance in the G vs. B/SP trials, p > .10. Clearly, children found it much easier to infer the protagonist's desired goal when the cues of gaze and body orientation were combined than when the two cues were opposed. It is also clear that the influence of body orientation on goal detection was not enhanced when it was combined with the cue of spatial proximity. The other main effect, of age, reflects the superior overall performance of the oldest age group (M = 5.5) compared to the 4-year-olds (M = 4.06), t(34) = 2.70, p = .01, and 3-year-olds (M = 3.22), t(34) = 4.47, p < .01. The overall difference in accuracy between 3- and 4-year-olds was not significant, t(34) = 1 .57 , p > .10.

As Table 2 illustrates, the observed age differences reflect the 8- and 9-year-olds' ability to consistently use gaze direction as the definitive cue of goal across all three trial types at a rate significantly greater than would be expected by chance alone (50%), ts(17) ranging from 5.10 to 8.00, all ps < .001. In contrast, the overall performance of 4-year-olds is greater than would be expected by chance only for the G + B trials (M = 1.72), t(17) = 6.67, p < .01 (two-tailed). For the remaining two trial types the 4-year-olds proved to be a heterogeneous group. Their overall performance did not exceed chance responding for the G vs. B trial type (M = 1.22), [(17) = 1.00, p > .10, and the G vs. B/SP trial type (M = 1.11), [(17) = .98, p > .10; however, at an individual level a significant number of 4-year-olds correctly answered both trials within each of these trial types (binomial test, ps < .01) (see Table 2). As a group, 3-year-olds did mar- ginally better than chance (M = 1.33), [(17) = 2.06, p = .055 in the G + B trials; moreover, a significant number of these children answered both trials correctly (binomial test, p < .05) (see Table 2). However, as a group 3-year-olds did not use the cue of gaze to infer goal more often than would be expected by chance in either the G vs. B trials (M = .83) or the G vs. B/SP trials (M = 1.06), ps > .10.

Discussion

Altogether, four important points can be drawn from the results of Ex- periment 2. First, 8- and 9-year-old children unequivocally weight the di-

Page 13: The influence of nonverbal cues associated with looking behavior on young children's mentalistic attributions

241

DEREK E. MONTGOMERY, CHRISTY MORAN, LESLIE M. BACH

TABLE 2

Number of Children Correctly Responding on 0, 1, or 2 Trials in Experiment 2

Number of trials answered correctly

G + B G vs. B G vs. B/SP Age Group 0 1 2 0 1 2 01 2

3-year-olds 2 8 8 * 6 9 3 5 76 4-year-olds 0 5 1 3 " 62 I0" 7 2 9 * 8/9-year-olds I 017* 1116* 2 0 1 6 *

*The number of children answering both trials correctly within a trial type exceeds chance level (25%), binomial test, ps < .05.

Note. G = Gaze direction; B = Body orientation; SP -- Spatial proximity.

rection in which one is looking as significantly more revealing of a person's desired goal than (a) body orientation and (b) body orientation combined with spatial proximity. Their performance illustrates a clear recognition that the direction of one's gaze is a significant cue that serves to limit and guide the possible inferences made about that person's desired goal.

Secondly, preschoolers once again demonstrated their ability to con- sistently infer another's desired goal when that person's gaze direction and body orientation were combined and facing a single target. The 4-year-olds approached ceiling on the G + B trials and a significant number of 3-year- olds also did quite well on this trial type although, on the whole, their performance was more variable than was true for the 4-year-otds. It may have been that 3-year-olds had a bit more difficulty inferring goal when gaze direction and body orientation were combined in Experiment 2 com- pared to Experiment 1 because the object not being faced was located 90 degrees from the front of the head in Experiment 2 rather than 180 degrees as in Experiment 1. Also, the Test question in Experiment 1 asking where the protagonist "wants to go" required an action judgment, whereas the Test question in Experiment 2 asked what the protagonist "wants to have." It may be easier for 3-year-olds to associate facing something with a desire to act rather than with a desire to have. Future research is needed to ad- dress these possibilities more closely. Nevertheless, across two studies 3-year-olds correctly inferred goal when gaze direction and body orienta- tion were combined on 75% of the trials, clearly suggesting that children

Page 14: The influence of nonverbal cues associated with looking behavior on young children's mentalistic attributions

242

JOURNAL OF NONVERBAL BEHAVIOR

even at this young age are learning to consistently use these nonverbal cues to infer another's goal.

Thirdly, the absence of a difference between the G vs. B trials and the G vs. B/SP trials suggests that the impact of body orientation on judgments of goal was not influenced by the relative closeness of the protagonist to the object the body was facing. These results are consistent with the ab- sence of an effect of spatial proximity when it was isolated as an indepen- dent variable in Experiment 1. Coordinating the protagonist's location in space by calculating relative distances may be an overly complicated method of determining goal for young children. They may find it easier to base mentalistic attributions primarily upon features intrinsic to the body (e.g., body orientation or gaze direction). That is not to say that spatial proximity never influences judgments of goal. We did not, for example, test whether it might affect judgments when combined with gaze direction. Moreover, increasing the ratio of distances between potential goals (e.g., 10:1 rather than 3:1) may make spatial proximity a more salient factor than gaze direction and thereby increase its influence on children's judgments. Future research is needed to determine conclusively the possible role of spatial proximity in children's mentalistic attributions. Our findings, how- ever, suggest it is not a major factor in children's detection of goal.

Lastly, 3-year-olds did not consistently use gaze direction to infer goal when that cue contrasted with body orientation; in contrast, 4-year-olds gave some evidence of successfully distinguishing between these two cues. However, there was enough variability in 4-year-olds' overall performance in the G vs. B (and G vs. B/SP) trials such that, as a group, they did not consistently infer the object being looked at as the protagonist's goal at a level exceeding chance. Thus, the results of the 4-year-olds in the G vs. B trials are somewhat ambiguous. This is unsatisfying since a primary goal of Experiment 2 was to determine preschoolers' use of gaze direction when it contrasted with body orientation. Therefore, we conducted a third experi- ment, in which we doubled the number of G vs. B trials, with the hope of getting a clearer picture of the extent to which 4-year-olds should be cred- ited with understanding that goal is revealed by gaze direction even when it contrasts with body orientation.

Experiment 3

Method

Participants. In this experiment 24 4-year-olds participated. They ranged in age from 4 -0 to 5 -0 and their mean age was 4-6. Thirteen

Page 15: The influence of nonverbal cues associated with looking behavior on young children's mentalistic attributions

243

DEREK E. MONTGOMERY, CHRISTY MORAN, LESLIE M. BACH

males and 11 females participated. Participants were recruited from local preschools, each serving a predominantly middle-class population.

Materials. Children again watched as videotaped scenes were pre- sented on an 8 in. x 11 in. color screen. All four videos depicted a G vs. B scenario. Two videos were identical to those presented in the G vs. B trials in Experiment 2. The other two videotaped scenes were the same in the positioning of the protagonist and the items being displayed. Different per- sons were in each of these new videos, however, and two different items (shirt and pants) were displayed. Once again, for half of the trials a pro- tagonist was looking at the item on the left side of the screen and for the other half a protagonist was looking at the item on the right side of the screen.

Procedure. The pretraining and instructions preceding the presentation of each video were identical to those given in Experiment 2. The videos were presented so that the location of the item being looked at, on either the right or left side of the screen, was arranged in one of the 4 possible orders (RRLL, LLRR, LRLR, RLRL). The Test question again asked children to, "Point to which thing this person wants to have."

Results and Discussion

Children received one point each time they correctly indicated that the protagonist's desired goal was the object toward which her gaze was directed. Preliminary analyses revealed that there was not a significant dif- ference in performance between the two genders, p > .10. Further, there was no preferential responding toward the right or left side of the screen when indicating the protagonist's goal, p > .I0.

As Table 3 reveals, the majority of children answered correctly on only two out of four attempts. Not surprisingly, therefore, the mean number of correct responses (M = 2.00) does not significantly differ from chance level responding (50%), t(23) = 0.00, p > .10. These results are consistent with the finding that 4-year-olds, as a group, did not perform significantly above chance-level on the G vs. B trials in Experiment 2. Taken together, 4-year-olds were correct on 61% (22/36) of the G vs. B trials in Experiment 2 and 50% (48/96) of the trials in Experiment 3, hardly suggesting that they should be credited with a firm grasp of the awareness that gaze direction reveals goal even when contrasted with body orientation.

Only a small minority of the children (16.67%) answered four out of four Test questions correctly in Experiment 3. The success of these few

Page 16: The influence of nonverbal cues associated with looking behavior on young children's mentalistic attributions

244

JOURNAL OF NONVERBAL BEHAVIOR

TABLE 3

Number of Children Responding Correctly in Experiment 3

Number of G vs. B trials answered correctly

Age group 0 I 2 3 4

4-year-olds 3 5 9 3 4

Note. G ~ Gaze direction; B =- Body orientation.

children likely reflects an understanding on their part of the importance of gaze relative to body orientation since the probability of answering all four questions correctly is .0625. Despite the success of this small minority, however, the results from the group as a whole clearly indicate that 4-year- olds do not generally regard gaze direction as more revealing of a person's desired goal than body orientation when these two cues contrast.

it is unlikely that children's difficulties in Experiment 3 were due to procedural difficulties such as failing to understand either the experimen- ter's instructions or the meaning of the Test question; otherwise, these fac- tors would also have prevented the success preschoolers had in Experi- ments 1 and 2 when gaze and body orientation were combined. Relatedly, to ensure that children were able to tell where the protagonist was looking in our video displays, we asked a subset of our participants in Experiment 3 (12 altogether) to point out where the protagonist was looking following the final trial. Every child answered this question correctly, ruling out the possibility that children's difficulties in the G vs. B trials occurred because they could not tell which object the protagonist's gaze was directed to- ward.

Thus, we interpret 4-year-olds' performance in Experiment 3 as signi- fying a limit to their mentalistic reading of the cue of gaze direction. Pre- schoolers do not appear to completely understand that the meaning of gaze direction as a cue of goal is distinct, in principle, from body orienta- tion. Children appear to believe that facing something reveals interest in the target being faced (Experiments :1 and 2) without specifically under- standing the unique role visual perception plays in the focus and nature of one's thoughts. As a result, they exaggerate the importance of body orienta- tion relative to gaze direction.

Page 17: The influence of nonverbal cues associated with looking behavior on young children's mentalistic attributions

245

DEREK E. MONTGOMERY, CHRISTY MORAN, LESLIE M. BACH

General Discussion

The purpose of this research was to examine the influence of nonverbal cues associated with looking behavior on young children's mentalistic attri- butions. Of particular interest was presenting realistic displays of humans in which gaze direction contrasted with other overt associates of goal-di- rected behavior (i.e., spatial proximity and body orientation). The results of Experiments 1 and 2 revealed that the attributions of goal by children as young as 4, and to a large extent even as young as age 3, were consistently guided by a concern for where another was facing. That is, young children inferred another's desired goal to be the target toward which both gaze and body orientation were aligned. Spatial proximity had relatively little influ- ence when it contrasted with the direction in which one was facing (Exper- iment 1) and when it was combined with body orientation (Experiment 2). The results of Experiments 2 and 3 additionally revealed that when body orientation contrasted with gaze, then preschoolers were generally unclear as to the protagonist's goal. In comparison, there was little ambivalence among young grade school children as to the greater importance of gaze direction, relative to body orientation, in revealing another's goal (Experi- ment 2).

Given the developmental sequence occurring across these three studies we tentatively suggest the following levels of competence. First, preschoolers can infer that persons are more likely to be mentally con- nected (to want, to think about) something being faced rather than some- thing not being faced. That is, they appear to associate cognitive activity (attention, goal-based desire) with the target toward which the head and body are directed. However, as discussed below, preschoolers may not yet fully grasp the causal interrelations among gaze direction, attention, and goal-based desires. Thus, a second level of development is appreciating that gaze direction is unique in revealing goal, in part because visual per- ception is intimately related to information-gathering and the focus of one's thoughts. This uniqueness of gaze direction separates it from body orienta- tion and presumably is the reason why, by grade school, children come to overwhelmingly regard gaze direction as more revealing of goal than body orientation.

The first part of this developmental sequence, judging another's goal based upon where that person is facing, is an impressive early develop- ment. This is especially true considering that the attributions made in Ex- periments 1 and 2 were toward novel characters in a novel setting and, further, were made in the absence of a situational context (e.g., past behav- ioral regularities such as "Johnny always goes into that room") that might

Page 18: The influence of nonverbal cues associated with looking behavior on young children's mentalistic attributions

246

JOURNAL OF NONVERBAL BEHAVIOR

typically accompany everyday situations. This means the principle, "what one physically faces usually reveals one's desired goals (i.e. what one wants to act upon)," apparently generalizes to novel situations and likely brings coherence and meaning to otherwise disparate situations. The sensi- tivity of young children to the nonverbal cue of facing something almost certainly serves as an important step into the communicative milieu that characterizes social interactions. Learning to anticipate what someone wants by simply observing the physical orientation of that person likely facilitates children's development across a broad array of important social functions such as perspective-taking, prosocial behavior, and interpersonal communication. The importance of this development can be highlighted by considering that children with autism may be impaired in their ability to perceptually learn the significance of nonverbal cues (Baron-Cohen, 1995; Loveland, 1991) and this failure may be associated with the severe diffi- culty they experience in the areas of social behavior and social inter- relatedness.

However, despite preschoolers' impressive ability to read goal from the combination of gaze and body orientation, the results of Experiments 2 and 3 call into question the claim that preschoolers have an attention-goal psychology (Baron-Cohen, 1995). This is so because preschoolers' accu- racy in inferring another's desired goal dropped noticeably when gaze di- rection contrasted with the direction of body orientation. Their difficulty under such circumstances would not be expected if they were holding an attention-goal psychology and, therefore, assuming that persons are more likely to direct their gaze toward goals than toward non-goals. As noted above, preschoolers were perhaps limited in the G vs. B trials by their incomplete understanding of the conceptual interdependence among var- ious mental constructs. The conceptual underpinning of an attention-goal psychology would seem to entail at least some realization of the following: perception informs thinking, which in turn influences the formation of one's desires and goals, both of which, when formed, subsequently lead to goal-directed action (Wellman, 1990). Accordingly, if preschoolers do not fully grasp the interrelations among visual perception, attention, and goal (e.g., Flavell et al., 1995a, 1995b), then they may not be entirely sure why gaze direction reveals interest and goal. The conceptual justification for singling out gaze direction as more important than body orientation is missing.

Their difficulty may be a reflection of the relative complexity and indi- rectness of the relation between goal and gaze direction in their own expe- riences. For example, to the extent there is a relation between gaze direc- tion and goal, the influences can be construed as bidirectional. It may be

Page 19: The influence of nonverbal cues associated with looking behavior on young children's mentalistic attributions

247

DEREK E. MONTGOMERY, CHRISTY MORAN, LESLIE M. BACH

that (a) the more something is looked at the more it increases one's desire for that object or (b) because of a desire and intent that had previously been formed one examines an object for an extended period. The correla- tional nature between seeing and wanting also means that, at times, the two factors may be inversely related; for example, while looking at some- thing for a period of time one might be determining that the object is of little interest or worth. The indirectness of the relation between vision and goal may be confusing to young children, with the variations and subtleties difficult to appreciate. Future research might reveal that preschoolers readily understand the significance of gaze direction when the relation be- tween visual perception and the resultant mental state is more direct. For example, preschoolers may find it relatively easy to say that a person will remember or learn more about what is being gazed at than about what is only being faced with the lower body.

In a different vein, future research might also investigate children's attributions of goal in "gaze only," "body orientation only," and "spatial proximity only" conditions. Such research would reveal the extent to which each of these cues is influential independent of the others. For ex- ample, in situations in which gaze direction and spatial proximity are neu- tral, both children and adults may default to the cue of body orientation to make mentalistic inferences. Our research was primarily directed toward addressing whether young children use gaze direction to reveal goals even when it is contrasted with other associates of goal-directed activity. Thus, we did not explore the effects of all of the possible combinations of gaze direction, spatial proximity, and body orientation on mentalistic attribu- tions. An extension of our findings in that direction would be worthwhile and provide a more comprehensive picture of the relative contributions of the situational features that children use to make mentalistic attributions.

In sum, the development of the ability to infer another's goals based upon that person's gaze direction is a critical foundation underlying social interactions and the ability to decode the nonverbal communication of a wide range of mentalistic concepts such as intents, thoughts, and desires. The research reported here suggests that the meaning derived from the important cue of gaze direction develops gradually and is understood as superseding body orientation only after the preschool years. Preschoolers may therefore be better described as having a "facing-goal" psychology rather than the more specific "attention-goal" psychology. The preschool years are thus an intriguing mixture of strengths and shortcomings regard- ing the detection of goal. The meaning of gaze direction is not entirely understood early in childhood and is still subject to construction. A strength however, demonstrated by the presence of a "facing-goal" psy-

Page 20: The influence of nonverbal cues associated with looking behavior on young children's mentalistic attributions

248

JOURNAL OF NONVERBAL BEHAVIOR

chology, is that very young children can systematically use nonverbal cues to make mentalistic inferences. This implies that children's developing con- ception of the mind is interrelated with their perception of the body. The chal lenge for future research is to investigate why this interrelation occurs and the extent to which the social construction of the meaning of mentalis- t ic concepts is constrained and guided by features of the body.

References

Argyle, M., & Cook, M. (1976). Gaze and mutual gaze. Cambridge, UK: Cambridge Univer- sity Press.

Baron-Cohen, S. (1993). From attention-goal psychology to belief-desire psychology: The de- velopment of a theory of mind and its dysfunction. In S. Baron-Cohen, H. Tager-Flusberg, & D. J. Cohen (Eds.), Understanding other minds: Perspectives from autism (pp. 59-82). Oxford: Oxford University Press.

Baron-Cohen, S. (1995). Mindblindness. Cambridge, MA: MIT Press. Baron-Cohen, S., Campbell; R., Karmiloff-Smith, A., Grant, J., & Walker, J. (1995). Are chil-

dren with autism blind to the mentalistic significance of the eyes? British Journal of De- velopmental Psychology, 13, 379-398.

Baron-Cohen, S., & Cross, P. (1992). Reading the eyes: Evidence for the role of perception in the development of a theory of mind. Mind and Language, 6, 173-186.

Bartsch, K., & Wellman, H. M. (1995). Children talk about the mind. Oxford: Oxford Univer- sity Press.

Burgoon, J. K., Buller, D. g., Hale, J. L., & deTurck, M. A. (1984). Relational messages associ- ated with nonverbal behaviors. Human Communication Research, 10, 351-.378.

Butterworth, G. (1994)~ Theory of mind and the facts of embodiment. In C. Lewis & R Mitch- ell (Eds.), Children's early understanding of mind (pp. 115-132). Hillsdale, NJ: Erlbaum.

Dennett, D. C. (1987). The intentional stance. Cambridge, k,~: MIT Press. Flavell, J. H., Green, F. L., & Flavell, E. R. (1995a). Young children's knowledge about think-

ing. Monographs of the 5ociety for Research in Child Development, 60 (1, Serial No. 24_3).

Flavell, J. H., Green, F. L., & Flavell, E. R. (1995b). The development of children's knowledge about attentional focus. Developmental Psychology, 3 I, 706-712.

Flavell, J. H., & Miller, R H. (in press). Social cognition. In D. Kuhn & R. S. Siegler (Eds.), W. Damon (Series Ed.), Handbook of child psychology: VoL2. Cognition, perception, & lan- guage. New York: Wiley.

Gopnik, A., & Wellman, H. M. (1994). The 'theory' theory. In L. A. Hirschfeld & S. A. Gelman (Eds.), Mapping the mind: Domain specificity in cognition and culture (pp. 257-293). Cambridge, UK: Cambridge University Press.

Kleinke, C. L. (1986). Gaze and eye contact: A research review. Psychological Bulletin, 100, 78-100.

Loveland, K. (1991). Social affordances and interaction lh Autism and affordances of the human environment. Ecological Psychology, 3, 99-119.

Lyon, T. D. (1993). Young children's understanding of desire and knowledge. Unpublished doctoral dissertation, Stanford University.

Montgomery, D. E. (1996). Wittgenstein's private language argument and children's under- standing of the mind. Unpublished manuscript.

Palmer, M. T., & Simmons, K. B. (1995). Communicating intentions through nonverbal behav- iors: Conscious and nonconscious encoding of liking. Human Communication Research, 22, 128-160.

Page 21: The influence of nonverbal cues associated with looking behavior on young children's mentalistic attributions

249

DEREK E. MONTGOMERY, CHRISTY MORAN, LESLIE M. BACH

Povinelli, D. j., & Eddy, 1". J. (1996). Chimpanzees: Joint visual attention. Psychological Sci- ence, 7, 129-135.

Wellman, H. M. (1990). The child's theory of mind. Cambridge, MA: MIT Press. Wittgenstein, L. (1953). Philosophical investigations. Oxford: Blackwell. Zebrowitz, L. A. (1990). 5ocial perception. Pacific Grove, CA: Brooks/Cole.