Child Development, March/April 2010, Volume 81, Number 2, Pages 597–615
Preschoolers’ Implicit and Explicit False-Belief Understanding: Relations With Complex Syntactical Mastery Jason Low Victoria University of Wellington
Three studies were carried out to investigate sentential complements being the critical device that allows for false-belief understanding in 3- and 4-year-olds (N = 102). Participants across studies accurately gazed in anticipation of a character’s mistaken belief in a predictive looking task despite erring on verbal responses for direct false-belief questions. Gaze was independent of complement mastery. These patterns held when other low-verbal false-belief tasks were considered and the predictive looking task was presented as a time-controlled film. While implicit (gaze) knowledge predicted explicit (verbal) false-belief understanding, complement mastery and cognitive flexibility also supported explicit reasoning. Overall, explicit false-belief understanding is complexly underpinned by implicit knowledge and input from higher-order systems of language and executive control.
According to de Villiers (2000), children must understand and reproduce complement syntax before achieving an understanding of false beliefs. Complements are statements embedded under propositional attitude verbs within a sentence. In a recent meta-analysis of studies on false-belief understanding, Milligan, Astington, and Dack (2007) uncovered a stronger direction of effect from measures of individual language ability to falsebelief comprehension than the reverse. The largest effect size was observed for children’s mastery of embedded complement structures, but this finding remains unclear for two reasons. First, the effect size may be overstated as only four studies in the meta-analysis assessed the unique role of sentential complements. Second, the correlation between false-belief reasoning with mastery of sentential complements did not differ significantly from correlations with other measures of language ability (e.g., synonym production; Perner, Stummer, Sprung, & Doherty, 2002). The relation between
This research was funded internally at Victoria University of Wellington by a School of Psychology Research Grant (SoP22817). The author is grateful to Janet Astington, Helena Gao, Sophie Jacques, Josef Perner, Ted Ruffman, and Karen Salmon for advice, and three anonymous reviewers for insightful and constructive comments on the original article. The author also thanks the children who generously participated in the research, and Steve Cochran and the many research assistants for their valuable help in collecting the data. Correspondence concerning this article should be addressed to Jason Low, School of Psychology, Victoria University of Wellington, P.O. Box 600, Wellington 6140, New Zealand. Electronic mail may be sent to
[email protected].
standard false-belief tasks with a variety of individual language skills may partly reflect implicit–explicit changes in representational format underpinning children’s understanding of mental states. As such, we need to clarify the link between complement mastery and false-belief reasoning to better identify the constraints on individual language ability relating to mental state understanding. Complementation Enables Cognition de Villiers singled out children’s mastery of the grammatical argument structure of mental state and communication verbs taking realis complements (clauses with overt present or past tense) with their associated marking of points of view as prerequisite for representing false beliefs (e.g., de villiers 2004, 2005; de Villiers & de Villiers, 2000, 2003). Consider the sentence: ‘‘Homer said the cat yodeled.’’ While the complement embedded under the verb is false (the cat yodeled), it is still true that Homer said something. The complement structure also licenses the protagonist’s terms of reference even when they are not ours: ‘‘Homer said the cat yodeled, but in fact it meowed.’’ de Villiers contends that overt evidence of truth or falsity from the verb say is easier to conceptualize and may be a
2010, Copyright the Author(s) Journal Compilation 2010, Society for Research in Child Development, Inc. All rights reserved. 0009-3920/2010/8102-0015
598
Low
precursor to understanding sentential complements that include analogic mental state verbs (e.g., think). de Villiers and Pyers (2002) found that memory for the content of complement sentences embedded under communication and mental state verbs at an earlier test point predicted later false-belief comprehension and not the reverse. Training on communication verb sentential complements alone can also improve children’s false-belief performance (and not vice versa; Hale & Tager-Flusberg, 2003). Schick, de Villiers, de Villiers, and Hoffmeister (2007) also found predictive correlations linking grasp of communication verb (told) complement structure in American Sign Language with falsebelief understanding among deaf children. A criticism of the memory for complements task is that even though children are simply asked to restate the false utterance of the protagonist, those utterances may cry out for an explanation in terms of the deceptive motive of the actor. However, Lohmann and Tomasello (2003) found that even training children to talk about objects using complement frames that do not highlight deceptive experiences can facilitate false-belief understanding. These findings fit with the strong view that ‘‘sufficient reasoning ability (for false-belief computation, prediction and explanation) is defined by having language of the appropriate degree of complexity’’ (de Villiers & de Villiers, 2003, p. 338). Implicit False-Belief Knowledge One theoretic complication to de Villiers’ account of false-belief development remains relatively unexplored: children’s implicit knowledge. Clements and Perner (1994) ran a change-in-location falsebelief task whereby they videoed children’s eye gaze in anticipation of the protagonist reappearing to look for the target object. Children between 2 years 11 months and 3 years 7 months looked more at the correct location despite providing an incorrect verbal false-belief judgment. Older children (by 4 years) matched their verbal response to an overt false-belief question with their visual orienting gesture. It is possible that young children possess an implicit understanding of false beliefs, and implicit knowledge of belief might precede and contribute to explicit understanding. The possibility of a distinction between implicit and explicit falsebelief knowledge fits with nondevelopmental research. For example, blindsight patients show no awareness of objects that fall into the affected portion of their visual field and yet are able to accurately intuit features of the objects (Perner &
Clements, 2000). On a cautionary note, however, Clements and Perner’s (1994) anticipatory eye gaze results might instead index a weaker form of explicit knowledge. One way to disestablish the ‘‘implicit’’ nature of early false-belief knowledge is to argue that young children’s visual orienting could reflect less confident (but explicit) expectations. However, Ruffman, Garnham, Import, and Connolly (2001) found that the young children who gave incorrect verbal falsebelief predictions did not even want to bet a modest number of counters on the location as indicated by their correct eye gaze. These results suggest that children are not conscious of the knowledge conveyed in their eye gaze. A second objection is that anticipatory gaze may reflect a seeing = knowing rule (e.g., O’Neill, 1996). Garnham and Ruffman (2001) redesigned the predictive looking false-belief task to comprise three hiding locations wherein the protagonist first explored the middle hiding location before placing the object in the left-hand location. A second character then moved the object to the right-hand location. If predictive gaze is based on a sensitivity to the protagonist not knowing, children should look in anticipation equally often to the left and irrelevant middle locations. However, children looked in anticipation clearly to the left-hand location; anticipatory gaze appears to be based on implicit falsebelief understanding. A third objection stems from studies, involving deaf children, uncovering how performances on low-verbal and traditional verbal measures of false belief are correlated with complement mastery (e.g.,de Villiers & de Villiers, 2003; Schick et al., 2007). However, the critical issue may not be whether tasks are verbal or low verbal per se, but whether a task taps into implicit versus explicit knowledge. In the low-verbal object-hiding theoryof-mind task used by Schick et al., (2007), participants were instructed to point to which of two identical boxes contained a target object upon determining a helper’s visual access to the hiding event. When children point, they are explicitly aware of having gestured, and this may not be equivalent to eye glances that are factuality implicit (Goldin-Meadow & Alibali, 1999). The low-verbal and verbal false belief tasks used by Schick et al. may have correlated with complement mastery because both methods tapped explicit knowledge. Finally, Onishi and Baillargeon (2005) found that 15-month-olds showed increased looking times to an event when the actor’s search behavior violated a false belief compared with when it fitted a false
Implicit and Explicit False-Belief Understanding
belief. These findings have been replicated with 13month-olds even when similarities to last personobject-location encoding are controlled for (e.g., Surian, Caldi, & Sperber, 2007). In Clements and Perner’s (1994) study, however, very young 2-year-olds did not show implicit understanding via anticipatory gaze. In the predictive looking task, the target object is present; 2-year-olds’ knowledge of the actual location of the object may interfere with their implicit knowledge of where a protagonist with a false belief might look. This interference could also partly reflect young children’s fledgling ability to inhibit or select attention away from a true-belief suggestion (e.g., Leslie, German, & Pollizi, 2005). Of course, this does not mean that the predictive looking task is not appropriate for tapping into implicit false-belief knowledge per se; it may simply mean that the task involves other demands that may swamp the rudimentary nature of 2-year-olds’ false-belief sensitivity. Using a simpler version of the predictive looking task where the target object was removed altogether, Southgate, Senju, and Csibra (2007) found that 2-year-olds looked first (and longer) to the location where an actor was expected to search given a false belief. The developmental timeline suggests continuity in implicit false-belief understanding from 2 to 4 years of age and, further, that early knowledge may be formatted ‘‘in a rudimentary and implicit form’’ (Onishi & Baillargeon, 2005, p. 257). A step forward for theory building is to understand how implicit knowledge underpins explicit understanding, and how factors such as language ability make explicit knowledge more accessible. Rise of Explicit Understanding One view is that through the endogenous cyclical process of redescription, implicit representations are recoded into explicit ones (with the highest format providing access to declarative verbal reporting; Karmiloff-Smith, 1992). By hypothesis, variation in implicit social-perceptual understanding may reasonably correlate with the extent of explicit verbal false-belief reasoning. Continuity (i.e., a correlation in performance) between implicit and explicit false-belief knowledge is bolstered by certain lines of evidence. First, implicit knowledge is a precondition for receptivity to explicit false-belief training (e.g., Clements, Rustin, & McCallum, 2000). Second, other incipient theory of mind insights such as the duration of infants’ correct looking during referential acts are correlated with higher level
599
cognitive outcomes (e.g., Wellman, Phillips, Dunphy-Lelii, & LaLonde, 2004). To provide a concrete mechanism for how changes in explicit reflection might occur, Zelazo (1994) suggested that redescription could also take place over the course of problem solving when children demonstrate cognitive flexibility or the ability to construct a higher order rule to simultaneously integrate multiple conflicting representations of a single event. Performance on the Dimensional Change Card Sort (DCCS) task is closely correlated to explicit false-belief task performance, independent of age and general intelligence (e.g., Frye, Zelazo, & Palfai, 1995; Mu¨ller, Zelazo, & Imrisek, 2005). Zelazo, Gao, and Todd (2007) explain that the two kinds of tasks may be related because both require cognitive flexibility in relating two incompatible perspectives (e.g., in the DCCS—how it is possible to sort the same cards by shape and then by color; in the representational change false-belief task—how it is possible to directly respond with a past or present self-perspective). Preschoolers may become more proficient at reasoning through both types of problems when they flexibly integrate opposing knowledge into an embedded hierarchical ‘‘if-if-then’’ rule. The link between cognitive flexibility and explicit false-belief understanding must, however, be qualified in several ways. First, studies by Carlson and Moses testing other aspects of executive control reveal a strong association between conflict inhibition tasks with standard verbal falsebelief performance (e.g., Carlson & Moses, 2001; Carlson, Moses, & Claxton, 2004). Second, Perner and Lang (2000) pointed out that explicit verbal false-belief tasks can be passed without necessarily reformulating two separate rules in a hierarchical three-term fashion (e.g., if I am looking for the marble, then here; if Sally is looking for the marble then there). Third, the evidence on whether explicit mental state understanding or cognitive flexibility over rule switching emerges first is equivocal (e.g., Kloo & Perner, 2003). Nonetheless, even views concentrating on the cognitive flexibility facet of executive control in relation to advancing mental state awareness maps onto the broader point made here that the grammar of complement constructions cannot be the critical device that allows for false-belief representation. Jacques and Zelazo (2005) have suggested that the relation between individual language ability and explicit false-belief answering can instead be meaningfully recast in terms of language working to service cognitive flexibility in promoting explicit perspective taking. To the extent that cognitive
600
Low
flexibility in rule use refers to linguistically mediated reasoning for promoting explicit control of thought and action (Carlson, 2003), then such a view even fits with research indicating that verbally labeling cards before sorting on the DCCS task can improve inhibitory control to support explicit attentional focus on multiple and conflicting perspectives (e.g., Kirkham, Cruess, & Diamond, 2003). One integrated view may be that, alongside an implicit understanding of false belief, skill sets that include individual language ability (e.g., complement mastery) and cognitive flexibility can help make emergent explicit understandings of false belief. Only Tantalizing? Unfortunately, studies emphasizing early implicit sensitivity to false-belief understanding remain tantalizing because they do not include any measure of children’s individual language abilities (e.g., Clements & Perner, 1994; Clements et al., 2000; Ruffman et al., 2001). If it turns out that visual orienting does not uniquely tap implicit knowledge but is correlated with complement mastery, it is possible to maintain that certain syntactic structures provide the breakthrough for any false-belief understanding (e.g., de Villiers, 2000, 2004; de Villiers & de Villiers, 2003). If eye gaze is not correlated with mastery of complement structure, it would mean that although certain grammatical platforms may play a role in harnessing explicit levels of false-belief understanding that are more complex and abstract, complementation is unlikely to provide the conceptual foundation for children’s metarepresentational ability (Surian et al., 2007). Implicit knowledge is also likely to contribute to explicit false-belief understanding. For the present investigation, two distinct but connected premises were submitted for analysis: (a) explicit false-belief understanding has distinctive roots in implicit knowledge (via anticipatory eye gaze) and (b) cognitive flexibility and complex language ability uniquely contribute to underpinning children’s explicit false-belief understanding.
Study 1 The first study examined whether complement mastery was correlated with anticipatory eye gaze and verbal judgments on standard false-belief tasks. Based on studies maintaining that visual orienting and verbal answers can potentially tap into implicit
and explicit false-belief understanding, respectively (e.g., Clements & Perner, 1994; Ruffman et al., 2001), two main hypotheses were submitted. First, early false-belief sensitivity was expected to show up on indirect anticipatory eye gaze but not on direct verbal answering. Second, eye gaze was hypothesized to be language independent, but anticipatory gaze and complement mastery were each expected to uniquely contribute to predicting explicit false-belief understanding. Method Participants.. Data came from 24 participants: twelve 3-year-olds (M = 3.8, SD = 2.09 months, range = 3.5–3.10; 4 girls and 8 boys) and twelve 4-year-olds (M = 4.7, SD = 2.62 months, range = 4.1–4.10; 6 girls and 6 boys). A total of 30 children were interviewed but 6 children (four 3-year-olds and two 4-year-olds) were not included in the final analysis as they failed to answer control questions in the various false-belief tasks correctly and did not sit still during the eye gaze task to allow examination of predictive looking. All participants in this study (and also in Studies 2 and 3) were Pakeha (New Zealanders of European descent), spoke English as their first language, and were from middleclass families. Procedure.. All children were individually interviewed in a quiet room at their kindergarten, and each of the 24 data participants took part in four 10- to 15-min fixed-order sessions with an interval of about 1 week between each. In Session 1, the eye-gaze false-belief task was presented. In Session 2, the Peabody Picture Vocabulary Test (PPVT)–III was administered. In Session 3, the Matrices and complement mastery tasks were presented. In Session 4, the unexpected contents, deceptive appearance, and change-in-location false-belief tasks were presented. Verbal and nonverbal ability.. Children’s verbal ability was assessed using the PPVT–III (Dunn & Dunn, 1997). Nonverbal ability was measured using Raven’s Colored Matrices (Raven, 1993). Raw scores of both measures were used. Complementation task.. The task measuring memory for complements embedded under a communication verb (say) was derived from Hale and Tager-Flusberg (2003). Each story involved different characters saying they did one activity when they were doing something else. Stories were worded to avoid any deceptive context or invitations to consider the motives of the false utterance. The order in which the information about what the central
Implicit and Explicit False-Belief Understanding
character said and what he or she did was counterbalanced across trials. Each trial was accompanied by two drawings, and at the end of each story participants were asked what the central character said. In one trial the experimenter presented a picture of a boy by himself baking cookies and said, ‘‘James was baking some cookies for his friend Lucy.’’ The experimenter replaced the first picture with a second one showing the boy and his friend without any action and said, ‘‘Lucy came along and asked him what he was doing. James said, ‘‘I’m baking some bread.’’ Participants were asked the test question, ‘‘What did James say he was baking?’’ The remaining three trials were about a girl saying she was cutting paper when she was really cutting her hair, a boy saying he kissed his mother when he really kissed Big Bird, and a girl saying she played with a cat when she really played with a dog. Each correct answer received 1 point for a total score of 4. Eye gaze.. Ruffman et al.’s (2001) predictive gaze task was used to measure implicit false-belief understanding. A 3D model (80 cm length · 35 cm width · 50 cm height) of a house was displayed on a table. At the front there was a ladder going to the top and a slide exit on either side (red at left, green at right). Slide exits at the front were separated by a distance of 70 cm and were adjacent to the slide entrances hidden at the back of the model. A red and green box (7 cm · 7 cm · 7 cm) was positioned 2 cm from either side of the respective slide exits. Toy objects were used to enact the event. The child sat on the floor in front of the model, with a
Ladder
70 cm Red Slide Red Box
Green Slide Green Box
Figure 1. Eye gaze apparatus setup with illustrative frames of gaze direction.
601
camera above it recording eye gaze (see Figure 1). An assistant hidden behind the model house helped the experimenter with the staging of the event. The experimenter first demonstrated to participants that when the central protagonist wanted to look in the red box he would come down the red slide, and when he wanted to look in the green box he would come down the green slide. All children remembered where the protagonist would emerge from to look in the respective boxes. For the falsebelief story, the experimenter enacted a narrative whereby the protagonist hid his toy car in the red box and then went to sleep behind the house (the assistant’s hand appeared from behind to remove the protagonist upon climbing to the top of the house). Then the experimenter enacted the next episode where another character came along and moved the toy from the red to the green box and left the scene. The experimenter then sat behind the child. The assistant hidden behind the model house then announced that the central character was waking up and wanted to go play with his toy. The experimenter then wondered aloud to elicit anticipatory gaze, ‘‘I wonder which slide [protagonist] will come down to look for his car.’’ Four seconds later, children were asked an explicit false-belief question: ‘‘Which slide will [protagonist] come down to look for his car?’’ A true-belief control story with the protagonist (a different character) witnessing the target (a different object) changing locations from the red to green container was also enacted. False- and true-belief trials were counterbalanced. Participants passed the true-belief explicit question and looked longer to the green slide location, providing confidence that gaze on the falsebelief trial was not because of participants retracing story events and looking to where the protagonist first put the object. All participants passed control questions asked after the standard explicit falseand true-belief questions: reality (where is the [object] now?), memory (where did [protagonist] put his [object] in the beginning?), and perceptual access (did [protagonist] see [displacer] move the [object]?). Videotapes of participants’ eye gaze for the false-belief trial over a 4-s period following the ‘‘I wonder’’ prompt were analyzed to determine the number of frames during which children looked to a given location when anticipating the character’s descent. Miyuki and Chisato (2006) pointed out that the whites of human eyes produce strong contrast with the pupils when looking left or right, and in that sense, detection of gaze direction can be
602
Low
relatively straightforward. Using VideoReDo Plus, one rater coded all tapes in slow motion and scored gaze direction on a frame-by-frame basis (looking times were accurate to 0.04 s; 25 frames ⁄ s). Another rater coded 20% of the videotapes, and raters agreed 100% of the time as to whether or not the child was looking more to the correct location. The gaze task captures rich variance in children’s implicit false-belief knowledge via latency of visual orienting over 4 s; explicit understanding, however, was only tapped by verbal answers to a single direct false-belief question. To ensure sufficient variance in the measurement of explicit knowledge, three other direct false-belief tasks were included. Unexpected contents and deceptive appearance tasks.. Single trials of the unexpected contents and deceptive appearance tasks were given (Bialystok & Senman, 2004; Gopnik & Astington, 1988). For unexpected contents, participants were shown an M&Ms tube and asked what it contained (all said ‘‘M&Ms’’). They were then shown that it contained a spoon instead of M&Ms. Participants were asked about their own false belief (When you first saw this tube, what did you think was inside?) as well as to predict what an out of sight puppet (Ernie) would think was in the tube (When Ernie sees this tube, what will he think is inside?). Participants were then asked the reality control question (What is inside the tube right now?). For deceptive appearance (involving a sponge rock), participants were shown the object placed on a table and asked, ‘‘What is this?’’ All children answered correctly (rock or stone). Next the object’s hidden property was revealed and two direct false-belief questions of the object’s appearance were asked: ‘‘What did you think this was when you first saw it?’’ and ‘‘Ernie did not hear or see what we were doing; what will Ernie think this is?’’ Participants were then asked the reality control question (What is this really?). Change-in-location task.. The unexpected contents and deceptive appearance tasks contributed a maximum of 2 points each to explicit understanding (two direct false-belief test questions in each context). Verbal answers to the single direct question in Ruffman et al.’s (2001) gaze task contributed 1 point to explicit false-belief understanding as presented in a change-in-location context; thus, a single trial of Baron-Cohen, Leslie, and Frith’s (1985) Sally-Ann change-in-location task was also administered. Toy characters were used and the hiding locations employed a basket and a jug with a small plastic fruit serving as the target object. The explicit test question involved children answering where
the central protagonist (Sally) would first look for the (displaced) object. Then participants were asked two control questions: ‘‘Where is the fruit now?’’ and ‘‘Where did Sally put the fruit?’’ Overall, six explicit false-belief questions were asked and their contexts (unexpected contents, deceptive appearance, and change in location) contributed equal subscores to explicit false-belief answering. All participants answered the associated control questions correctly. A composite explicit false-belief score was calculated for each participant by adding the points across all direct test questions yielding a maximum score of 6 (a = .82). Results and Discussion Preliminary analyses revealed that there were no gender effects in Study 1 (or in Studies 2 and 3), and including gender in the regression models did not affect the results. Consequently, gender was not included in all analyses reported here. Eye gaze.. The location of children’s initial look following the anticipatory ‘‘I wonder’’ prompt was analyzed first. Significantly more children showed correct first looks (N = 21 ⁄ 24; 87%) than expected by chance based on the binomial distribution (p < .001). As a second approach to the data, eye gaze was assessed in terms of the total amount of time spent looking at the correct and incorrect locations. There was no significant age group difference for the amount of time spent looking at the correct location or the incorrect location (see Table 1 for mean scores). Nonetheless, collapsed across age groups, children spent a longer total amount of time looking at the correct location (M = 1.45 s, SD = 0.78) than the incorrect location (M = 0.83, SD = 0.69), F(1, 23) = 8.02, p < .01, gp2 = .26. Finally, an analysis of participants’ visual orienting preferences relative to their ability to give a correct verbal answer on the single explicit false-belief question for the gaze task was carried out. A conservative metric was used to capture the degree of correct preferential looking whereby participants were classified as passing eye gaze only if the looking time difference for the false-belief trial ([looking at correct location] ) [looking at incorrect location]) was at least +0.15 s. There was a significant link between anticipatory gaze and giving a correct verbal answer for the explicit question, v2(1, N = 24) = 4.51, p < .05, Cramer’s / = .43. Ten children (42%) passed eye gaze and correctly answered the single explicit false-belief question while 5 children failed both (21%). Critically, 37% (n = 9) of
Implicit and Explicit False-Belief Understanding
603
Table 1 Mean Performance on Various Measures of Interest (Standard Deviations Shown in Parentheses) Measure Study 1 (N = 24) Eye gaze (s) At correct location At incorrect location Matrices Composite explicit false-belief PPVT Complement mastery Study 2 (N = 36) Eye gaze (s) At correct location At incorrect location Matrices Composite explicit false-belief PPVT Complement mastery DCCS Study 3 (N = 42) Eye gaze (s) At correct location At incorrect location Matrices Composite explicit false-belief PPVT Complement mastery DCCS
3-year-olds
4-year-olds
Total
Range
1.22 0.77 10.42 2.25 57.75 1.83
1.67 0.88 9.42 3.92 66.17 2.83
(0.68) (0.80) (4.74) (1.66) (9.64) (0.83)
1.45 (0.78) 0.83 (0.69) 9.92 (4.30) 3.08 (2.04) 61.96 (10.23) 2.33 (1.09)
0–2.52 0–2.68 2–18 0–6 50–85 0–4
1.31 (1.05) 0.75 (0.72) 10.00 (3.64) 0.61 (0.57) 55.39 (16.96) 1.89 (1.45) 1.11 (0.76)
1.61 (0.76) 0.53 (0.53) 11.78 (4.14) 1.43 (0.66) 68.94 (15.05) 2.83 (1.15) 1.94 (0.64)
1.46 (0.92) 0.64 (0.63) 10.89 (3.95) 1.02 (0.73) 62.17 (17.23) 2.36 (1.38) 1.53 (0.81)
0–3.68 0–2.88 3–20 0–2 18–93 0–4 0–3
ns ns ns F(1, 34) F(1, 34) F(1, 34) F(1, 34)
= = = =
15.95***, gp2 = .32 6.43*, gp2 = .16 4.68*, gp2 = .12 12.71**, gp2 = .27
1.77 (0.88) 0.59 (0.50) 13.05 (2.51) 2.10 (1.67) 40.38 (13.69) 1.61 (1.20) 1.33 (0.86)
1.98 (0.79) 0.66 (0.68) 14.05 (2.94) 3.38 (1.96) 57.57 (17.23) 2.57 (1.29) 2.38 (0.60)
1.87 (0.83) 0.63 (0.59) 13.55 (2.75) 2.74 (1.91) 49.00 (17.90) 2.10 (1.32) 1.86 (0.90)
.20–3.60 0–2.28 10–19 0–6 22–83 0–4 0–3
ns ns ns F(1, 40) F(1, 40) F(1, 40) F(1, 40)
= = = =
5.23*, gp2 = .12 12.37**, gp2 = .24 6.14*, gp2 = .13 21.32***, gp2 = .35
(0.83) (0.58) (3.96) (2.09) (9.32) (1.11)
Age difference
ns ns ns F(1, 22) = 4.63*, gp2 = .17 F(1, 22) = 4.72*, gp2 = .18 F(1, 22) = 6.19*, gp2 = .22
Note. PPVT = Peabody Picture Vocabulary Test; DCCS = Dimensional Change Card Sort. *p < .05. **p < .01. ***p < .001.
children passed eye gaze while erring on their verbal answers to the explicit false-belief question but no participant showed the reverse relation. Group differences on explicit false-belief and other measures.. Mean scores on each of the other measures are also reported in Table 1. A one-way ANOVA indicated that 4-year-olds had a higher aggregate explicit false-belief score than 3-yearolds. Similar analyses also revealed significant age group differences in PPVT performance and complement mastery. There was no significant age group difference in general nonverbal ability on the Matrices task. Correlations and hierarchical regressions.. For the remaining analyses in Study 1, to examine empirical connections between implicit and explicit falsebelief understanding, the difference in looking time variable was used as the candidate metric for reflecting implicit knowledge. The reasoning is as follows. First, a looking time difference metric converges with approaches used in other relevant developmental literature, for instance, gaze follow-
ing and social attention, where correlations between visual orienting durations and higher cognitive task performance are computed and interpreted (e.g., Brooks & Meltzoff, 2005; Wellman et al., 2004). Other metrics such as the correct first look, while important for determining that children’s predictive looking behaviors do not reflect chance responding, do not yield sufficient variation to avoid spuriously low correlations with explicit false-belief understanding (or complement mastery; cf. Bornstein & Sigman, 1986). Finally, looking-time difference provides a clearer index of a state of mind as reflected in the degree of preference in children’s visual orienting responses (Ruffman, 2000) and, further, duration of difference in looking time by yielding variation fits with arguments that there is real, though subtle, growth in implicit knowledge (Munakata, McClelland, Johnson, & Siegler, 1997). Full and partial correlations are reported in Table 2 (top row). Full correlations indicated that explicit falsebelief understanding (composite score), aside from being related to chronological age, was related to
604
Low
Table 2 Bivariate and Full Partial Correlations (in Parentheses), Among Variables of Interest in Studies 1 (N = 24), 2 (N = 36), and 3 (N = 42) Variable Study 1 Age Matrices PPVT Complements Explicit false belief Eye gaze Variable
1
2
—
).10 —
1
2
3
.45* ).22 —
3
Study 2 (above diagonal) and study 3 (below diagonal) Age — .26 .47** Matrices .20 — .30 PPVT .58*** .20 — Complements .31* ).04 .55*** Cognitive flexibility .50** ).15 .34* Explicit false belief .42** ).07 .49** Eye gaze .17 ).11 .21
4
.67*** .21 .38 —
5
6
.61** .12 .51* .71*** (.44*) —
4
5
.41* .26 .35* — .48** (.38*) .60*** (.40*) .13 ().01)
.43** ).02 .08 .47** (.42*) — .53*** (.37*) .23 (.12)
.24 ).06 .23 .13 ().06) .54** (.53**) — 6
.56*** .12 .41* .58*** (.46**) .49** (.38*) — .55*** (.50**)
7
.27 .37* .23 .31 (.19) .04 ().04) .48** (.45**) —
Note. Full partial correlations control for age alongside scores on the PPVT and Matrices. PPVT = Peabody Picture Vocabulary Test. *p < .05. **p < .01. ***p < .001.
eye gaze and language ability (both general vocabulary and complementation). There was no significant bivariate correlation between eye gaze and complement mastery. When the effects of chronological age and general developmental abilities were controlled, explicit false-belief reasoning was related to eye gaze and complement mastery. Partial correlation results supported the hypothesis that spontaneous anticipatory eye gaze marked implicit knowledge that was not significantly related to individual differences in complement mastery. There appear to be dissociations in preschoolers’ false-belief knowledge between the behavioral and verbal levels. A hierarchical regression analysis was conducted to determine whether eye gaze and complement mastery each contributed unique variance in predicting explicit understanding (see Table 3). Age, and PPVT and Raven’s scores were entered as the first block of independent variables, and overall, they accounted for 49% of explicit falsebelief variance. Analysis of regression coefficients revealed age, b = .48, t(20) = 2.67, p = .015, to be a focal predictor of explicit false-belief understanding. Eye gaze performance was entered in the second step, and variation in implicit understanding uniquely accounted for a further 14% of variance in explicit reasoning. Complement mastery was entered at the third step and accounted for further unique variance (11%) in explicit reasoning. Analy-
Table 3 Results of Hierarchical Regressions for Variables Predicting Explicit False-Belief Performance in Study 1 (N = 24) Variable Step 1 Age Matrices PPVT Step 2 Age Matrices PPVT Eye gaze Step 3 Age Matrices PPVT Eye gaze Complements
B
SE B
b
T
DR2
DF
Sign. DF
.16 .11 .07
.06 .08 .04
.48 .24 .35
2.67* )0.45 1.93
.49
6.51
.00
.14 .12 .06 .75
.05 .07 .03 .27
.41 .24 .29 .39
2.59* 1.72 1.82 2.73*
.14
7.43
.01
.03 .04 .04 .79 .93
.06 .06 .03 .24 .33
.09 .09 .20 .41 .50
0.54 0.69 1.45 3.34** 2.81*
.11
7.89
.01
Note. PPVT = Peabody Picture Vocabulary Test. *p < .05. **p < .01.
sis of regression coefficients confirmed that, even at the third step, eye gaze performance continued to exert unique predictive value to explicit false-belief reasoning, b = .41, t(18) = 3.34, p = .004. Implicit knowledge and complement mastery appeared to uniquely scaffold preschoolers’ explicit false-belief understanding.
Implicit and Explicit False-Belief Understanding
Study 2 While the findings of Study 1 were consistent with theoretic predictions, three issues needed to be addressed further. First, the sample size was small, so it is important to replicate the findings. Second, de Villiers (e.g., de Villiers & de Villiers, 2003) argued that extant findings on anticipatory looking are difficult to align with evidence gained from deaf studies involving less verbal measures of false belief. If less or nonverbal tasks do not pattern after verbal ones, then deaf children should demonstrate increased skills on less verbal measures because such participants are more visually oriented, but this does not appear to be the case (e.g., Schick et al., 2007, using the object-hiding task; Woolfe, Want, & Siegal, 2002, using the thought-bubbles task). However, even with such less verbal falsebelief tasks, children are still asked to make a direct prediction based on a protagonist’s epistemic state, suggesting that those tasks essentially draw on explicit knowledge. In contrast, eye glances are not recorded in response to a direct question; rather, they are implicitly elicited in anticipation of a character’s actions (Ruffman, 2000, 2004). To disentangle these ideas, Study 2 provided children a larger battery of false-belief measures that included the object-hiding and thought-bubbles tasks. Following appeals to implicit and explicit false-belief understanding, it was predicted that complement mastery would be correlated with explicit false-belief understanding (including performance on the object-hiding and thought-bubbles tasks) but not with implicit (anticipatory looking) knowledge. While the results of Study 1 supported the first theoretic submission that explicit false-belief decisions have roots in unique implicit knowledge and that grasp of complex language contributes to conscious understanding, I also submitted a second premise that explicit knowledge may be reconfigured at higher levels of disembodiment through cognitive flexibility and natural language. While implicit knowledge via eye gaze was predicted to correlate with explicit verbal false-belief reasoning (as in Study 1), DCCS performance and complement mastery were predicted to each correlate with explicit understanding. Method Participants.. New data came from 36 participants: eighteen 3-year-olds (M = 3.6, SD = 2.38 months, range = 3.4–3.11; 5 girls and 13 boys) and eighteen 4-year-olds (M = 4.4, SD = 2.94 months,
605
range = 4.1–4.10; 6 girls and 12 boys). A total of 43 children were tested but five 3-year-olds failed to answer control questions in the false-belief tasks correctly and two 3-year-olds did not attend the final testing session, and so those individuals were not included in the final analysis. Procedure.. Participants were individually tested in a room at their kindergarten across six 10- to 15min sessions with an interval of 1 week between each. Dovetailing with Study 1, tasks were presented in the following order. In Session 1, the eyegaze task was presented. In Session 2, the PPVT–III was administered. In Session 3, the Matrices and complement mastery tasks were given. In Session 4, the unexpected contents, deceptive appearance, and change-in-location false-belief tasks were presented. In Session 5, the DCCS and thought-bubbles tasks were given. In Session 6, the object-hiding task was presented. DCCS (full version).. Participants were shown two target cards (blue rabbit and red boat) attached to one of two target trays and told of two preswitch rules (e.g., by color) for sorting a series of bivalent cards (red rabbits and blue boats). After successfully completing a practice trial sorting two cards, participants started the preswitch phase (e.g., ‘‘If it’s blue it goes here, and if it’s red it goes there. This is a blue one, where does it go?’’). No feedback was given. Participants needed to sort all six preswitch trials correctly in order to receive a score of 1 (Zelazo, 2006). For the postswitch phase participants were told: ‘‘Now we are not going to play the color game anymore. We are going to play a shape game now. In the shape game, all rabbits go here and all boats go there. If it’s a rabbit it goes here, and if it’s a boat it goes there.’’ The experimenter picked up a card and labeled it by its relevant dimension, and the child had to sort cards independently (e.g., ‘‘This is a rabbit, where does it go?’’). Participants had to sort at least five of six trials to pass the postswitch phase (Zelazo, 2006). The full DCCS task comes with a third phase (the border trials) that requires a higher level of cognitive flexibility. Some of the test cards were presented with black borders and the others were normal: ‘‘Now we are playing a different game. If there is a black line around the card like this one (red rabbit with a border), we play the color game. In the color game, all red ones go here and all blue ones go there. This card is red, so we put it over here. If there is no black line around the card, like this one (red rabbit without a border), we play the shape game. In the shape game, all rabbits go here
606
Low
and all boats go there. This one is a rabbit, so we put it here.’’ This phase comprised 12 trials (three red rabbits and three blue boats without borders, three red rabbits and three blue boats with borders). On each trial, the rule was repeated. Children must sort at least 9 of 12 cards correctly to pass this phase (Zelazo, 2006). The full DCCS was administered to capture continuous changes on cognitive flexibility over rule use (Hongwanishkul, Happaney, Lee, & Zelazo, 2005). Test cards were presented such that no more than two cards of the same type were presented consecutively. Each child’s performance was scored as 0 (failed preswitch phase), 1 (passed pre- but failed postswitch phase), 2 (passed pre- and postswitch, but failed the border phase), or 3 (passed all phases; Zelazo, 2006). False-belief tasks.. With respect to Ruffman et al.’s (2001) eye-gaze task, all participants passed the true-belief control trial. One rater coded all tapes for the false-belief trial in slow motion and scored gaze direction on a frame-by-frame basis (looking times were accurate to 0.04 s). Another rater coded 20% of the videotapes, and raters agreed 100% of the time as to whether or not the child was looking more to the correct location. As in Study 1, single trials of the standard M&Ms, sponge-rock and Sally-Ann false-belief tasks were also administered. Overall, correct answers on the direct false-belief test questions for the task contexts of unexpected contents, deceptive appearance, and change in location each provided a subtotal score of 2, respectively. The new false-belief tasks were thought bubbles (Woolfe et al., 2002) and object hiding (Call & Tomasello, 1999). False belief: Thought bubbles.. A pretest was carried out to check that participants understood what thought bubbles meant. Two pictures were shown: one of boy taking a dog for a walk and the other of a boy thinking about a dog (with a thought bubble over his head containing a dog). The experimenter asked: ‘‘Which boy is thinking about a dog?’’ All children correctly pointed to the picture depicting a thought bubble. The task proceeded with two falsebelief and two true-belief (control) trials (counterbalanced). An example false-belief trial showed a boy whose hand was reaching inside a partially shut kitchen cupboard. The experimenter said, ‘‘Billy is reaching into the cupboard for a glass of water.’’ Then the experimenter placed her palm over the protagonist’s face and said, ‘‘Let’s cover Billy up so he does not see what is inside the cupboard.’’ The cupboard door-flap was opened: ‘‘Look, there is a mouse in the cupboard.’’ (The pic-
ture shows Billy’s hand reaching toward a mouse.) The flap was covered back, and the experimenter removed her palm covering the protagonist. She then presented another picture of Billy with an empty thought bubble over his head. A series of forced-choice picture items were shown underneath: a glass of water (false belief), a mouse (reality), an apple, and a bag (distracters). Children were instructed: ‘‘Point to what Billy thinks is inside the cupboard.’’ The other false-belief trial involved a boy who was fishing but whose line snagged a boot hidden behind reeds. True-belief trials involved a girl who sees a tall boy looking at her from behind a fence (the fence was removed to reveal a tall boy), and a boy walking along the beach who sees a fish tail from behind a bush (the bush was removed to reveal a fish). All participants passed the true-belief trials. The maximum correct false-belief score for the thought-bubbles task was 2. False belief: Object hiding.. Participants had to work out and point to which of two identical boxes contained an object. Children were told that the communicator (seated behind the hider) would help them find where an object (a balloon) was hidden by pointing to the correct box. Following Call and Tomasello (1999), three control trials were first administered. In the visible displacement trial, the hider concealed the balloon (carried out behind a raised barrier) and presented the two boxes to the child. The communicator then pointed to the box the balloon was in. Then the communicator turned around to have a drink of water and searched noisily in her bag. This distraction was natural while being obvious enough to ensure that the child knew the communicator was distracted and not paying attention to what would be happening next. In full view of the child, the hider opened up both boxes, transferred the balloon from one to the other, and closed the lids to the boxes. Then the communicator (timing was well practiced) turned around, and the hider asked the child to point to where the balloon was. This control trial confirmed that all participants could successfully keep track of the hider taking the object out from one box and moving it into the other. The next control trial was nearly identical to the first except that the hider simply switched the location of the boxes in full view of the child without taking the balloon out. The hider then asked the child to point to where the sticker was. This second control trial confirmed that all participants were able to successfully keep track of the object when the hider changed the location of the box in which the object was hidden. In the final control trial, after the hider hid the balloon and
Implicit and Explicit False-Belief Understanding
presented the boxes to the child, the communicator before pointing to the location of the object turned around to drink water and search in her bag. At this point, the hider opened both boxes and switched the balloon from one container to another and closed the lids to the containers, all in full view of the child. Then the communicator turned around and pointed to the (incorrect) container she had seen the balloon hidden. Then the hider asked the child to point to where the balloon was. This last control trial confirmed that all participants were able to successfully ignore the communicator’s signal when it was wrong. In the verbal false-belief trials, children watched the communicator observing the hider conceal the balloon. Then the hider presented the two boxes toward the child. Before the communicator pointed to which box the balloon was in, she turned around to drink and search in her bag. Then, in full view of the child, the hider opened the lid of the loaded box and moved the balloon into the other container and put the lids back on. After the hider switched the balloon, she asked the critical verbal false-belief question, ‘‘When [communicator] turns back around, where will she think the balloon is?’’ After the child pointed, the communicator turned back around and pointed to the box where she thought the balloon was. Two of such trials were administered and the scores were averaged. In the two nonverbal false-belief trials, the hider hid the balloon in one of the boxes behind the barrier and presented them to the child. The communicator turned around to drink and search in her bag. The hider then swapped the location of the boxes in full view of the child. Then the communicator faced forward to signal which location she had seen the balloon hidden; thus, the box that she pointed to was incorrect, as she did not see the hider swapping the boxes. The child, who had no knowledge of the balloon’s whereabouts until that moment was then asked to point to where the balloon was. Call and Tomasello (1999) argued that this trial represented a nonverbal measure of false belief because participants need to interpret the nonverbal information provided by the communicator in an appropriate way: (a) she initially saw the balloon being hidden in one of the boxes, but (b) did not see the boxes swapping locations, and (c) must have marked the particular box upon turning back around based on her outdated false belief, and so (d) the balloon must be in the other box. Two nonverbal trials were given and scores averaged. Presentation of the verbal and nonverbal false-belief blocks was counterbalanced. Average scores on the
607
verbal and nonverbal false-belief blocks were recombined to produce a full object-hiding task score (maximum of 2). Results and Discussion Eye gaze and explicit false-belief performance.. Replicating Study 1, a majority of the children displayed correct first looks (N = 32 ⁄ 36) than expected by chance (p < .001, binomial test). There was no significant age difference for the amount of time spent looking at the correct or the incorrect location (see Table 1 for mean values). However, children still spent a longer total amount of time looking at the correct location (M = 1.46 s) than the incorrect location (M = 0.64), F(1, 35) = 16.62, p < .001, gp2 = .32. As in Study 1, contingencies between passing ⁄ failing visual orienting versus giving a correct ⁄ incorrect verbal answer to the single explicit false-belief question found in Ruffman et al.’s (2001) predictive looking task were examined. There was a significant relation between eye gaze and verbal responding on the direct falsebelief question, v2(1, N = 36) = 4.98, p < .05, Cramer’s / = .37. Sixteen children (44%) passed eye gaze and correctly answered the single explicit false-belief question while 7 children failed both (20%). However, many more children passed eye gaze while erring on the verbal answer to the explicit false-belief question (33%, n = 12) but only 1 participant (3%) showed the reverse relation. Ruffman (2004) has argued that the object-hiding task is equivalent to standard false-belief measures tapping into explicit knowledge in that children are directly asked a question about the central protagonist’s belief. Fitting with Ruffman’s argument, many children passed eye gaze while erring on all of the false-belief test questions on the full objecthiding task (53%; n = 19) but only 1 participant (3%) showed the reverse relation of passing the object-hiding task but failing eye gaze. Similarly, more children passed eye gaze while erring on all of the false-belief questions on the thought-bubbles task (31%; n = 11) but only 2 participants (6%) showed the reverse relation of passing thought bubbles but failing eye gaze. In short, the dissociation between accurately gazing in anticipation of a character’s mistaken belief while erring on verbal responses to the direct false-belief question in the predictive looking task was also found when contingencies between anticipatory gaze and overt pointing responses given for the object-hiding and thought-bubbles tasks were considered. Furthermore, not only were the verbal and nonverbal
608
Low
subparts of object-hiding task correlated (r = .71, p < .001), but children’s scores on the overt falsebelief questions asked for unexpected contents, deceptive appearance, change in location, thought bubbles, and verbal and nonverbal parts of the object-hiding task were intercorrelated (a = .87). In relation to bivariate associations with complex language ability, children’s scores to overt falsebelief questions for the change-in-location, deceptive appearance, and unexpected-contents task contexts were each correlated with complement mastery (r = .48, p < .01; r = .54, p < .01; and r = .36, p < .05, respectively). Children’s tallies on the thought-bubbles task alongside scores on the verbal and nonverbal parts of the object-hiding task were also correlated with complement mastery (r = .49, p < .01; r = .40, p < .05; r = .38. p < .05, respectively). These results replicate work with deaf children (e.g., Schick et al., 2007) and could be claimed to fit with the view that complement mastery is important for any false belief reasoning. However, these particular false-belief tasks tap explicit knowledge by posing direct questions to children; as such, it is not surprising that they should all correlate positively with complement mastery (Ruffman, 2000, 2004). Critically, and anticipating the rest of the results, eye gaze remained separate by being not correlated with complement mastery. Consequently, a composite explicit falsebelief understanding score was calculated by averaging subscores on answers to direct questions for unexpected contents, change in location, deceptive appearance, thought bubbles, and the full objecthiding task (hence, average maximum composite total = 2). Four-year-olds obtained significantly higher composite explicit false-belief scores compared to 3-year-olds (see Table 1). Correlations and hierarchical regressions.. Mean scores on the remaining measures of interest are also found in Table 1, and the direction of age group differences replicate those uncovered in Study 1. The analyses then turned to investigate correlations between anticipatory eye gaze (measured by the difference metric, as per Study 1) and composite explicit false-belief understanding with other variables of interest. Bivariate and partial correlations are reported in Table 2 (second row, above diagonal). At the bivariate level, PPVT performance, complement mastery, cognitive flexibility, and explicit false-belief understanding were related to chronological age. PPVT and complementation scores were also interrelated, and explicit false-belief reasoning was correlated with both PPVT and complementation performance. Explicit
false-belief reasoning was further correlated with anticipatory eye gaze and cognitive flexibility. General nonverbal ability on the Matrices was also correlated with anticipatory gaze. Complement mastery was also correlated with cognitive flexibility at the bivariate level. Upon partial correlation analysis, only the following results remained: (a) explicit false-belief reasoning was related to anticipatory eye gaze performance, (b) explicit falsebelief reasoning was related to cognitive flexibility and complement mastery, and (c) complement mastery was correlated with cognitive flexibility. Two hierarchical regression analyses were conducted to determine whether implicit knowledge reflected in anticipatory eye gaze, together with cognitive flexibility and complementation, contributed unique variance to the prediction of explicit falsebelief understanding. Age, and PPVT and Raven’s scores were entered as the first block of independent variables. In the first regression analysis, eye gaze performance and cognitive flexibility were entered in the second and third steps, respectively. The second regression analysis was similar to the first except that complement mastery was entered at the final step. I did not pit cognitive flexibility against complement mastery when considering their unique effects because they were correlated, and it is possible that complex language might service the work of cognitive flexibility in promoting explicit false-belief reasoning (e.g., Mu¨ller et al., 2005; a point taken up in the General Discussion). Regressions are summarized in Table 4. In the first regression analysis, age was the only significant predictor at Step 1, and accounted for 34% of explicit false-belief variance. Implicit (eye gaze) knowledge and cognitive flexibility contributed uniquely to explicit false-belief variance after effects associated with age, PPVT performance, and nonverbal reasoning on the Matrices were jointly removed. They accounted for a further 13% and 10% of the variance, respectively. For the second regression analysis, complement mastery significantly accounted for a further 9% of the variance after the effects of implicit knowledge was removed. Overall, implicit knowledge, cognitive flexibility and complement mastery contributed unique variance to predicting explicit false-belief understanding.
Study 3 The predictive looking task set up in Studies 1 and 2 involved live experimenters fiddling about
Implicit and Explicit False-Belief Understanding Table 4 Results of Hierarchical Regressions for Variables Predicting Explicit False-Belief Performance in Study 2 (N = 36) Variable Regression 1 Step 1 Age Matrices PPVT Step 2 Age Matrices PPVT Eye gaze Step 3 Age Matrices PPVT Eye gaze Cog. flexibility Regression 2 Step 3 Age Matrices PPVT Eye gaze Complements
B
SE B
b
T
DR2 DF Sign. DF
.06 ).01 .01
.02 .03 .01
.48 2.91** .34 5.51 ).07 )0.45 .21 1.24
.00
.05 ).04 .01 .24
.02 .03 .01 .09
.41 2.74* .13 7.86 ).19 )1.32 .18 1.21 .40 2.80**
.00
.03 ).03 .01 .25 .33
.02 .03 .01 .08 .12
.22 1.43 .10 7.24 ).16 )1.16 .23 1.66 .42 3.18** .36 2.69*
.01
.04 ).04 .01 .21 .19
.02 .03 .01 .08 .07
.32 2.22* ).22 )1.61 .13 0.90 .34 2.52* .35 2.56*
.01
.09 6.56
Note. PPVT = Peabody Picture Vocabulary Test. *p < .05. **p < .01.
with the cardboard house and dolls in front of the child. It is possible that the timing and accentuation of events was not as rigorously controlled as it would be with a fixed video. It is important to re-examine anticipatory gaze while ensuring that all participants witnessed the unfolding of the episodes in the predictive looking task in a time-controlled fashion. Moreover, Study 2 was the only one that included the DCCS measure; it is important to replicate the findings that cognitive flexibility, alongside complement mastery, contributed unique variance to explicit false-belief understanding. A third study was carried out where Ruffman et al.’s (2001) predictive looking task was staged and filmed first and only then played back on a monitor to all participants. This change ensured that the timing and accentuation of events to elicit eye gaze was rigorously controlled for all participants. This study again included standard measures of explicit false-belief understanding together with the complement mastery and DCCS tasks. It was predicted that eye gaze would not be correlated with complex language and that cognitive flexibility and comple-
609
ment mastery would contribute unique variance to the prediction of explicit false-belief reasoning. Method Participants.. New data came from 42 participants: twenty-one 3-year-olds (M = 3.4, SD = 3.78 months, range = 3.0–3.10; 13 girls and 8 boys) and twenty-one 4-year-olds (M = 4.5, SD = 2.79 months, range = 4.1–4.10; 9 girls and 12 boys). A total of 46 children were tested but three 3-year-olds and one 4-year old were not included in the final analysis as they failed to answer control questions in the false belief tasks correctly. Procedure.. Participants were individually tested in a quiet room at their kindergarten, and all took part in four 10- to 15-min sessions spaced about 1 week apart from each other. Tasks were in the same order as in Studies 1 and 2. Except for two methodological changes, Study 3 replicated the tasks used in Study 2. First, with respect to the predictive looking task, the only difference was that the episodes in the gaze task (for false- and true-belief trials) were enacted and filmed, and the digitized film clips with accompanying audio track were then played back to children on a 15.4-in. (diagonal) monitor. The film clips were digitized into mpeg formats such that most of the screen was filled upon playback—this ensured that the correct and incorrect locations in the false- and true-belief stories were mapped toward either ends of the monitor. At least to a certain extent, then, the linear conditions of observing the screen was maximized to ensure that the whites of participants’ eyes would produce strong contrast with the pupils upon looking left or right. In this way, steps were taken to ensure that anticipatory eye gaze patterns were, as much as possible, straightforward to interpret and analyze. Each participant sat in front of the computer monitor (on a table) and a camera was placed above it to record eye gaze. One rater coded all recordings of the false-belief trial in slow motion and scored gaze direction on a frame-by-frame basis (all participants passed the true-belief control predictive looking trial). Another rater coded 20% of the videotapes, and raters agreed 100% of the time as to whether or not the child was looking more to the correct location. Second, for the explicit false-belief battery, the thought-bubbles and object-hiding tasks were not used; Study 2 showed that those tasks mapped onto standard measures of direct false-belief reasoning. Following Study 1, then, only single trials of the
610
Low
standard M&Ms, sponge-rock, and Sally-Ann falsebelief tasks were administered. Correct verbal answers on the direct test questions based on the contexts of unexpected contents, deceptive appearance, and change in location each contributed a subtotal of 2 points to explicit false-belief knowledge; this yielded a combined maximum score of 6 for each participant (a = .70). Results and Discussion Eye gaze.. Replicating Study 2, significantly more children showed correct first looks (N = 37 ⁄ 42) than expected by chance based on the binomial distribution (p < .001). There were no significant age group differences for the amount of time spent looking at the correct or the incorrect location (see Table 1). As in Studies 1 and 2, children still spent a longer total amount of time looking at the correct location (M = 1.87 s) than the incorrect location (M = 0.63), F(1, 41) = 60.14, p < .001, gp2 = .60. There was also a significant link between accurately gazing in anticipation of a character’s mistaken belief and giving a correct verbal answer for the explicit falsebelief question, v2(1, N = 42) = 4.68, p < .05, Cramer’s / = .33. Nineteen children (45%) passed eye gaze and correctly answered the single explicit false-belief question while 5 children failed both (12%). However, 43% (n = 18) of children passed eye gaze while erring on their verbal responses to the explicit false-belief question, but no participant showed the reverse relation. Beyond eye gaze, group differences on the remaining measures of interest also replicated the findings of Study 2 (see Table 1). Correlations and hierarchical regressions.. Full bivariate and partial correlations are reported in Table 2 (second row, below diagonal). As in Study 2, when the effects of chronological age and general developmental abilities were controlled, explicit falsebelief reasoning was related to eye gaze, cognitive flexibility, and complement mastery. Anticipatory eye gaze continued to show no significant correlation with complement mastery. Cognitive flexibility and complement mastery remained correlated during full partial correlation analysis. Following analyses in Study 2, two hierarchical regressions were conducted to clarify whether implicit (gaze) knowledge, alongside cognitive flexibility and complex syntactical understanding contributed unique variance to the prediction of explicit false-belief reasoning (see Table 5). In the first regression analysis, PPVT performance was a significant predictor at Step 1, and
Table 5 Results of Hierarchical Regressions for Variables Predicting Explicit False-Belief Performance in Study 3 (N = 42) Variable Regression 1 Step 1 Age Matrices PPVT Step 2 Age Matrices PPVT Eye gaze Step 3 Age Matrices PPVT Eye gaze Cog. flexibility Regression 2 Step 3 Age Matrices PPVT Eye gaze Complements
DR2
DF
Sign. DF
.30
5.38
.00
.05 ).08 .03 .81
.04 .19 1.28 .18 12.62 .09 ).12 )0.95 .02 .31 2.07* .23 .45 3.55**
.00
.01 ).03 .03 .75 .68
.04 .04 0.25 .07 .09 ).04 )0.32 .02 .28 1.97 .22 .41 3.46** .29 .32 2.31*
5.36
.02
.05 ).04 .01 .81 .60
.03 .19 1.41 .12 10.18 .08 ).05 )0.46 .02 .07 0.46 .20 .44 3.99** .19 .41 3.19**
.00
B
SE B
b
T
.06 .04 .23 1.39 ).13 .10 ).19 )1.35 .04 .02 .39 2.31*
Note. PPVT = Peabody Picture Vocabulary Test. *p < .05. **p < .01.
accounted for 30% of explicit false-belief variance. Implicit knowledge and cognitive flexibility contributed uniquely to explicit false-belief variance after effects associated with age, PPVT performance, and nonverbal reasoning on the Matrices were jointly removed. They accounted for a further 18% and 7% of the variance, respectively. Interestingly, in Step 3 when all variables were considered in the first regression, PPVT performance no longer contributed unique variance to predicting explicit falsebelief scores. General lexical knowledge via the PPVT may have predicted explicit false-belief performance at Step 1 because, on some superficial level, those verbal theory-of-mind tasks require some level of language ability (e.g., Chandler, Fritz, & Hala, 1989). However, the PPVT taps not only into word knowledge but also discourse semantics (e.g., words have a labeling function). Jacques and Zelazo (2005) explained that labeling can also help theory-of-mind performance by facilitating cognitive flexibility, which in turn promotes higher level consciousness in thought and control over explicit false-belief attribution. They also posited that
Implicit and Explicit False-Belief Understanding
understanding the arbitrary and symbolic nature of labels themselves could allow for psychologic distancing, which in turn promotes cognitive flexibility. Either way, the effects of semantic ability (lexical and discourse semantics) via the PPVT for explicit false-belief performance may be partly because of effects on cognitive flexibility. This may explain why PPVT scores were no longer significant once cognitive flexibility performance was entered at the final step of the regression. At the final step of the second regression analysis, complement mastery significantly accounted for 12% of the variance after the effects of implicit knowledge was removed. Even when all variables were considered, implicit knowledge continued to uniquely predict explicit false-belief understanding.
General Discussion According to de Villiers (2000, 2004, 2005), the structural properties of tensed complement syntax provide the critical breakthrough format that allows false-belief understanding. Embedded within this view are the related claims that children’s nonverbal and verbal false-belief task performance pattern after each other in their dependency upon complement mastery and that children are not likely to demonstrate any false-belief reasoning that is fugitive from linguistic propositional representation. Fitting with previous research (e.g., Milligan et al., 2007), Studies 1–3 revealed a significant relation between complement mastery and false-belief understanding. However, the current findings go beyond extant research by revealing that the relation is only evident for tasks that directly ask children to make false-belief decisions and not for tasks that indirectly elicit false-belief understanding. These findings elevate theorizing on language and theory of mind by weaving in data on children’s implicit and explicit understanding. Implicit understanding of false belief was confirmed across various analyses of eye gaze: correct first look, longer total looking to the correct location, and in terms of the richer metric of difference in looking time when noise because of fixations at the incorrect location was removed. Despite the relatively course-grain analyses of eye gaze, the findings follow suit with an accumulating number of studies showing that looking behavior can provide important information about distinctions between implicit and explicit knowledge (e.g., Clements & Perner, 1994; Onishi & Baillargeon, 2005; Ruffman et al., 2001; Southgate et al., 2007; Surian
611
et al., 2007). The use of formal automatic eye-tracking technology to measure gaze fixation on welldefined points could expand the scope of analysis on the grade of implicit knowledge being tapped as a function of exact ocular spatiotemporal trajectories. Nonetheless, the present investigation provides important evidence that anticipatory looking is implicit in a new sense: Complement mastery is not correlated with gaze performance, and complex syntactical understanding is only correlated with direct false-belief performance. These results were replicated across three studies with larger sample sizes and when the predictive looking task was presented as a film to remove timing confounds. The dissociation signature whereby grasp of complementation structure relationally impacts on one form of knowledge but not another is evidence for separate implicit and explicit representational systems. In Study 2 where other less verbal false-belief tasks such as object hiding and thought bubbles were included, the pattern of findings replicated Schick et al.’s (2007) work showing a significant relation between responding on these kinds of tasks with complement mastery. While anticipatory gaze performance was correlated with object-hiding and thought-bubbles false-belief scores, eye gaze on the predictive looking task still did not correlate with complement mastery. Performance on those lowverbal false belief tasks was, however, strongly correlated with verbal answers provided for traditional false-belief tasks (i.e., change in location, unexpected contents, and deceptive appearance). It seems that nonverbal false-belief attribution patterns after direct or verbal answers only insofar as they tap into explicit theory of mind understanding (Ruffman, 2000, 2004). In the present investigation, there was no significant difference between 3- and 4-year-olds’ implicit false-belief knowledge as tapped by the predictive looking task. The issue of whether implicit knowledge, despite its earlier manifestation, should show ontogenetic maturation is a gnawing one. If the present investigation had studied a wider age range (e.g., included 2-year-olds), there may be significant group differences in the extent of implicit knowledge as tapped by the usual predictive looking task, and the difference in looking time variable may even then correlate with chronological age (e.g., Clements & Perner, 1994; Ruffman, 2000). Gradual change in implicit knowledge may also be suggested when we consider children’s performance on a variety of indirect false-belief tasks. For example, 15-month-olds show implicit false-belief
612
Low
sensitivity via the violation-of-expectancy methodology (e.g., Onishi & Baillargeon, 2005), and 2-yearolds show implicit false-belief sensitivity when the predictive looking task is simplified by completely removing the target object from the scene (e.g., Southgate et al., 2007). It is possible that children may be building up implicit representations of false belief with a greater number of processing units and stronger connection weights, as embodied in a neural network, and the different types of indirect false-belief tasks may necessitate different grades of implicit network representation. Mechanismwise, implicit knowledge could develop in part through statistical pattern learning. Children may detect regularities across events where people often successfully locate objects where they last put them. Across most events, these regularities work well, but occasionally children may be confronted by actions where people look for objects in the wrong place. Other inferential systems such as goal or intentional tracking could potentially help parse these actions and contexts so that an implicit false-belief interpretation is discerned over event irregularities where people end up going to look for an object not where it necessarily is. Also, because such events are less common compared to people successfully looking for objects where they currently are, it takes time for implicit false-belief knowledge to be built up with sufficient network complexity (Clements et al., 2000). Consequently, there will be phases in knowledge construction whereby young children show fragile anticipatory gaze, and also phases where young children may show dissociation in false-belief understanding at the levels of spontaneous visual orienting and verbal judgment. Further detections of exceptions to behavioral regularity may result in a state of cognitive disequilibrium that critically pushes systemwide integration of all extant knowledge representations into a higher order format that supports conscious and verbally correct false-belief judgments. Such a theoretic possibility could potentially explain why, as found in Studies 2 and 3, explicit false-belief understanding also correlated significantly with flexibility in handling higher order rules that is partly tapped by the DCCS. Of course, the current data on early implicit false-belief understanding in 3- and 4-year-olds cannot decisively arbitrate on whether the basis of such rudimentary knowledge should be viewed as constructivist and hard-won or nativist in origin. One could instead argue that the high accuracy in first look reflects a modular theory-of-mind mechanism (e.g., Leslie et al.,
2005). Leslie et al. (2005) also view the theory-ofmind mechanism to be assisted by its own onboard selection processor that inhibits true-belief suggestions. As young children have limited inhibitory control, they may automatically self-correct in looking between the correct (false belief) and incorrect (true belief) locations, leading to variation over the difference in looking time metric, as was found here. To the extent that an early rudimentary understanding of false belief might be supplied by a modular mechanism, and basic recognition is progressively made explicit through intersections with higher order systems of language and aspects of executive control, there is overlap between constructivist and nativist discussions of the current findings. Reiterated rounds of internal redescription of implicit knowledge yielding stronger representations of explicit knowledge may come about when children engage in cognitive flexibility over the course of problem solving (Karmiloff-Smith, 1992; Zelazo, 1994). Replicating previous work (e.g., Frye et al., 1995; Mu¨ller et al., 2005), Studies 2 and 3 revealed that DCCS performance was positively correlated with composite explicit false-belief performance. This association could have partly come about through cognitive flexibility in constructing an explicit higher order conditional if-if-then rule helping to embed different perspective settings so that conflicting knowledge can be accurately applied. For instance, in the object-hiding direct false-belief task, embedded conditionals might be: if friend turned around (Setting 1), and if she indicated to the object being here, then predict that object is there; but if friend kept observing (Setting 2), and if she indicated to the object being here, then predict that object is here. The result that complement mastery also predicted explicit reasoning could be interpreted to dovetail with Zelazo and colleagues’ broader view that complex language may potentially off-load cognitive flexibility for explicit false-belief reasoning (e.g., Jacques & Zelazo, 2005; Mu¨ller et al., 2005). It is possible that cognitive flexibility over explicit false-belief reasoning may be more effective if rules are carried out in linguistic formulations; discursive thought may create richer psychologic distance, thereby helping children explicitly understand that perspectives (and rules) are arbitrary or necessarily representational. Rule talking one’s way through false-belief problems could also bring out explicit thinking indirectly via the effect of linguistic formulations making psychologic perspectives an object of consideration.
Implicit and Explicit False-Belief Understanding
However, we need to carefully consider the implication that mastery of complement syntax per se also predicted children’s explicit false-belief reasoning scores. Indeed, in interpreting the correlation between DCCS performance and complement mastery found in Studies 2 and 3, it is difficult to think of how preschoolers could have solved the memory for complements task by necessarily using if-if-then propositional language. Proficiency in hierarchical rule use, as partly tapped by the DCCS task, could provide for accurate conscious falsebelief problem solving, but only within limits. Preschoolers could also explicitly compare two representations of a situation through syntactic constructions involving a single verb of belief and an embedded tensed complement (e.g., ‘‘Sally thinks the ball is in the basket’’). In the same way, for the object-hiding task in Study 2, preschoolers could explicitly recognize perspectives and make an accurate judgment when they verbally formulate the communicator’s signal as ‘‘the friend says (indicates) ⁄ thinks the object is there.’’ Understanding richness of minds may also be partly explicated using communication and mental state verb-complement constructions. Nonetheless, there is still the question of how precisely cognitive flexibility and complement mastery might come to developmentally interface or intersect to underpin explicit falsebelief understanding. One modest interpretation of the current results may simply be that linguistically mediated thought involving diverse syntactic constructions at different levels of analysis (e.g., from pruned complements to if-if-then rules) is partly responsible for conscious flexibility in shifting from one response perspective to another (e.g., Carlson, 2003; Smith, Apperly, & White, 2003). The empirical finding that complement mastery was uniquely related to explicit false-belief understanding raises a further thorny issue: Even though participants were not invited to consider the motives of the false utterance, the misstatements in the complements tasks potentially involve misrepresentation of an actual state of affairs. Complement mastery and explicit false-belief reasoning may be correlated because they both involve misrepresentation (as opposed to a more specific role for complex syntax per se). The complements tasks could even be viewed as a simplified (but explicit) measure of false belief. It will be crucial to develop other viable analogs of the complement mastery task to delineate the extent to which it is the implications of certain syntactic constructions that partly lead children to explicitly understand differences in minds. The development of implicit measures
613
(e.g., capitalizing on looking time) could open up investigations as to whether young infants and preschoolers may show nascent sensitivity to how a false proposition can be legally attached to a verb of communication. In doing so, we stand to discover whether an implicit understanding of complement syntax may be related to an implicit understanding of false belief. Finally, we need to test children on multiple measures of indirect theory-of-mind tasks that also include controls for confounds because of associative encoding (e.g., Surian et al., 2007)—higher reliability from composite scores of implicit false belief may uncover whether there are weak, but significant, correlations between rudimentary false-belief knowledge and complement mastery. Overall, it will be important for future research to discover how early implicit understandings that intertwine the beginnings of theory of mind and language acquisition metamorphosize into more explicit knowledge (Astington, 2005). In conclusion, a richer understanding of children’s theory of mind will require collaborative efforts that use multiple methods, theories, and levels of analysis to illuminate how children’s conceptual representations may be differentiated and integrated along the implicit–explicit continuum. Explicit false-belief understanding was directly underpinned by unique implicit false-belief knowledge as tapped by anticipatory eye gaze. This finding is important because it means that the causal role of complex language in the development of false-belief reasoning does have to be constrained, to refer not to any reasoning but only to explicit answering. Indeed, aside from preschoolers being able to operate on the basis of automatic implicit knowledge, explicit verbal false-belief responding was also multiply supported by complement mastery and cognitive flexibility. Generally, to the extent that explicit false-belief understanding is open to input from language and higher order systems of executive control, the current findings on distinctions between implicit and explicit falsebelief reasoning have significant empirical and theoretic fit with different levels of analysis interested in how growth in reflection makes possible significant cognitive advances. While the answers uncovered in this investigation turned out to be fairly complex, integrative work of the kind attempted here serve as an important start to answering questions on how representational changes in mental state understanding develop as a function of intersections between language, cognitive control, and early core knowledge.
614
Low
References Astington, J. W. (2005). The developmental interdependence of theory of mind and language. In N. J. Enfield & S. C. Levinson (Eds.), Roots of human sociality: Culture, cognition and interaction (pp. 179–206). Oxford, UK: Berg. Baron-Cohen, S., Leslie, A. M., & Frith, U. (1985). Does the autistic child have a ‘‘theory of mind’’? Cognition, 21, 37–46. Bialystok, E., & Senman, L. (2004). Executive processes in appearance-reality tasks: The role of inhibition of attention and symbolic representation. Child Development, 75, 562–579. Bornstein, M. H., & Sigman, M. D. (1986). Continuity in mental development from infancy. Child Development, 57, 251–274. Brooks, R., & Meltzoff, A. N. (2005). The development of gaze following and its relation to language. Developmental Science, 8, 535–543. Call, J., & Tomasello, M. (1999). A non-verbal false-belief task: The performance of children and great apes. Child Development, 70, 381–395. Carlson, S. M. (2003). Executive function in context: Development, measurement, theory and experience. Monographs of the Society for Research in Child Development, 68(Serial No. 274). Carlson, S. M., & Moses, L. J. (2001). Individual differences in inhibitory control and children’s theory of mind. Child Development, 72, 1032–1053. Carlson, S. M., Moses, L. J., & Claxton, L. J. (2004). Individual difference in executive functioning and theory of mind: An investigation of inhibitory control and planning ability. Journal of Experimental Child Psychology, 87, 299–319. Chandler, M., Fritz, A. S., & Hala, S. (1989). Small-scale deceit: Deception as a marker of two, three-, and fouryear-olds’ early theories of mind. Child Development, 60, 1263–1277. Clements, W. A., & Perner, J. (1994). Implicit understanding of belief. Cognitive Development, 9, 377–395. Clements, W. A., Rustin, C. L., & McCallum, S. (2000). Promoting the transition from implicit to explicit understanding: A training study of false-belief. Developmental Science, 3, 81–92. de Villiers, J. (2000). Language and theory of mind: What are the developmental relationships? In S. BaronCohen, H. Tager-Flusberg, & D. J. Cohen (Eds.), Understanding other minds: Perspectives from developmental cognitive neuroscience (2nd ed., pp. 83–123). New York: Oxford University Press. de Villiers, J. G. (2004). Why language first? Invited paper, Web conference on co-evolution of language and theory of mind. Retrieved August 19, 2008, from http:// www.interdisciplines.org/coevolution/papers/5 de Villiers, J. G. (2005). Can language acquisition give children a point of view? In J. W. Astington & J. A. Baird (Eds.), Why language matters for theory of mind (pp. 186–219). New York: Oxford University Press.
de Villiers, J. G., & de Villiers, P. A. (2000). Linguistic determinism and the understanding of false-beliefs. In P. Mitchell & K. J. Riggs (Eds.), Children’s reasoning and the mind (pp. 189–226). Hove, UK: Psychology Press. de Villiers, J. G., & de Villiers, P. A. (2003). Language for thought: Coming to understand false-beliefs. In D. Gentner & S. Goldin-Meadow (Eds.), Language in mind: Advances in the study of language and thought (pp. 335– 384). Cambridge: MIT Press. de Villiers, J. G., & Pyers, J. E. (2002). Complements to cognition: A longitudinal study of the relationship between complex syntax and false-belief understanding. Cognitive Development, 17, 1037–1060. Dunn, L. M., & Dunn, L. M. (1997). Peabody Picture Vocabulary Test (PPVT–III) (3rd ed.) Circle Pines, MN: American Guidance Service. Frye, D., Zelazo, P. D., & Palfai, T. (1995). Theory of mind and rule based reasoning. Cognitive Development, 19, 483–528. Garnham, W. A., & Ruffman, T. (2001). Doesn’t see, doesn’t know: Is anticipatory looking really related to understanding of belief? Developmental Science, 4, 94–100. Goldin-Meadow, S., & Alibali, M. W. (1999). Does the hand reflect implicit knowledge? Yes and no. Behavioral and Brain Sciences, 22, 766–767. Gopnik, A., & Astington, J. W. (1988). Children’s understanding of representational change and its relation to the understanding of false-belief and the appearancereality distinction. Child Development, 59, 26–37. Hale, C. M., & Tager-Flusberg, H. T. (2003). Influence of language on theory of mind. Developmental Science, 6, 346–359. Hongwanishkul, D., Happaney, K. R., Lee, W. S. C., & Zelazo, P. D. (2005). Assessment of hot and cool executive function in young children: Age-related changes and individual differences. Developmental Neuropsychology, 28, 617–644. Jacques, S., & Zelazo, P. D. (2005). Language and the development of cognitive flexibility: Implications for theory of mind. In J. W. Astington & J. A. Baird (Eds.), Why language matters for theory of mind (pp. 144–162). New York: Oxford University Press. Karmiloff-Smith, A. (1992). Beyond modularity: A developmental perspective on cognitive science. Cambridge, MA: MIT Press. Kirkham, N. Z., Cruess, L., & Diamond, A. (2003). Helping children apply their knowledge to their behavior on a dimension-switching task. Developmental Science, 6, 449–476. Kloo, D., & Perner, J. (2003). Training transfer between card sorting and false belief understanding: Helping children apply conflicting descriptions. Child Development, 74, 1823–1839. Leslie, A. M., German, T. P., & Pollizi, P. (2005). Beliefdesire reasoning as a process of selection. Cognitive Psychology, 50, 45–85.
Implicit and Explicit False-Belief Understanding Lohmann, H., & Tomasello, M. (2003). The role of language on the development of false-belief understanding: A training study. Child Development, 74, 1130– 1144. Milligan, K., Astington, J. W., & Dack, L. A. (2007). Language and theory of mind: Meta-analysis of the relation between language ability and false-belief understanding. Child Development, 78, 622–646. Miyuki, K., & Chisato, Y. (2006). Understanding gaze perception and control. Japanese Journal of Psychonomic Science, 25, 77–84. Mu¨ller, U., Zelazo, P. D., & Imrisek, S. (2005). Executive function and children’s understanding of false belief: How specific is the relation? Cognitive Development, 20, 173–189. Munakata, Y., McClelland, J. L., Johnson, M. H., & Siegler, R. S. (1997). Rethinking infant knowledge: Toward an adaptive process account of successes and failures in object permanence tasks. Psychological Review, 104, 686–713. O’Neill, D. K. (1996). Two-year-old children’s sensitivity to parent’s knowledge state when making requests. Child Development, 67, 659–677. Onishi, K. H., & Baillargeon, R. (2005). Do 15-month-old infants understand false beliefs? Science, 308, 255–258. Perner, J., & Clements, W. A. (2000). From an implicit to an explicit theory of mind. In Y. Rossetti & A. Revonsuo (Eds.), Beyond dissociation: Interaction between dissociated implicit and explicit processing (pp. 273–294). Amsterdam: John Benjamins. Perner, J., & Lang, B. (2000). Theory of mind and executive function: Is there a developmental relationship? In S. Baron-Cohen, H. Tager-Flusberg, & D. J. Cohen (Eds.), Understanding other minds: Perspectives from autism and developmental cognitive neuroscience (pp. 150– 181). New York: Oxford University Press. Perner, J., Stummer, S., Sprung, M., & Doherty, M. (2002). Theory of mind finds its Piagetian perspective: Why alternative naming comes with understanding belief. Cognitive Development, 17, 1451–1472. Raven, J. C. (1993). Colored Progressive Matrices. Oxford, UK: Oxford Psychologists Press.
615
Ruffman, T. (2000). Nonverbal theory of mind: Is it important, is it implicit, is it simulation, is it relevant to autism? In J. W. Astington (Ed.), Minds in the making: Essays in honor of David R. Olson (pp. 456–479). Oxford, UK: Blackwell. Ruffman, T. (2004). Theory of mind and language ability: Understanding the bigger picture. Invited paper, Web conference on co-evolution of language and theory of mind. Retrieved August 19, 2008, from http://www.interdis ciplines.org/coevolution/papers/8 Ruffman, T., Garnham, W., Import, A., & Connolly, D. (2001). Does eye gaze indicate implicit knowledge of false-belief? Charting transitions in knowledge. Journal of Experimental Child Psychology, 80, 201–224. Schick, B., de Villiers, P., de Villiers, J., & Hoffmeister, R. (2007). Language and theory of mind: A study of deaf children. Child Development, 78, 376–396. Smith, M., Apperly, I. A., & White, V. (2003). False belief reasoning and the acquisition of relative clause sentences. Child Development, 74, 1709–1719. Southgate, V., Senju, A., & Csibra, G. (2007). Action anticipation through attribution of false beliefs by two-yearolds. Psychological Science, 18, 587–592. Surian, L., Caldi, S., & Sperber, D. (2007). Attribution of beliefs by 13-month-olds. Psychological Science, 18, 580–586. Wellman, H. M., Phillips, A. T., Dunphy-Lelii, S., & LaLonde, N. (2004). Infant social attention predicts preschool social cognition. Developmental Science, 7, 283–288. Woolfe, T., Want, S. C., & Siegal, M. (2002). Signposts to development: Theory of mind in deaf children. Child Development, 73, 768–778. Zelazo, P. D. (1994). From the decline of development to the ascent of consciousness. Behavioral and Brain Sciences, 17, 731–732. Zelazo, P. D. (2006). The Dimensional Change Card Sort (DCCS): A method of assessing executive function in children. Nature Protocols, 1, 297–301. Zelazo, P. D., Gao, H. H., & Todd, R. (2007). The development of consciousness. In P. D. Zelazo, M. Moscovitch, & E. Thompson (Eds.), Cambridge handbook of consciousness (pp. 405–432). New York: Cambridge University Press.
This document is a scanned copy of a printed document. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material.
