The link between preschoolers’ executive function and theory of mind and the role of epistemic states

Journal of Experimental Child Psychology 108 (2011) 513–531

Contents lists available at ScienceDirect

Journal of Experimental Child Psychology journal homepage: www.elsevier.com/locate/jecp

The link between preschoolers’ executive function and theory of mind and the role of epistemic states Anne Henning a,⇑, Frank M. Spinath b, Gisa Aschersleben a a b

Developmental Psychology Unit, Saarland University, 66041 Saarbrücken, Germany Differential Psychology and Psychological Diagnostics Unit, Saarland University, 66041 Saarbrücken, Germany

a r t i c l e

i n f o

Article history: Available online 30 November 2010 Keywords: Executive function Theory of mind Child temperament Dimensional change card sort Theory-of-mind scale EAS temperament inventory

a b s t r a c t The aim of this study was to assess the specific relation between 3- to 6-year-olds’ performance on a task measuring executive function (EF), the Dimensional Change Card Sort task (DCCS), and different developmental attainments in their theory of mind (ToM) by employing a battery of scaled ToM tasks that were comparable in task format and task demands. In addition, individual differences on the temperamental dimensions emotionality, activity, sociability, and shyness were assessed by parental rating. The main findings show that children’s (N = 195) performance on the DCCS related to their overall performance on the ToM scale but that this relation was specific to those ToM tasks that tap children’s understanding of epistemic states such as knowledge access, diverse beliefs, and false beliefs regarding content and location. The relation between children’s EF and overall ToM performance remained significant after controlling for age, sentence comprehension, child temperament, and parental education. Individual differences in child activity showed consistent negative relation to EF and ToM abilities. The findings point to a differential involvement of the various EF components in reasoning about different mental concepts. Ó 2010 Elsevier Inc. All rights reserved.

Introduction The view of people as mental agents whose actions are causally mediated by their internal states such as intentions, emotions, and beliefs is commonly construed as a theory of mind (ToM). A milestone in children’s ToM development takes place at around 4 years of age, when children understand ⇑ Corresponding author. Fax: +49 0 681 302 3871. E-mail address: [email protected] (A. Henning). 0022-0965/$ - see front matter Ó 2010 Elsevier Inc. All rights reserved. doi:10.1016/j.jecp.2010.10.006

514

A. Henning et al. / Journal of Experimental Child Psychology 108 (2011) 513–531

that beliefs about the world may be true or false and, therefore, may lead to successful or erroneous actions (Wellman, Cross, & Watson, 2001). A key ability in false belief reasoning is the ability to hold in mind two conflicting representations regarding the same situation (e.g., true belief vs. false belief) and to inhibit the prepotent response tendency exerted by one of the two (e.g., one’s own true belief vs. another person’s false belief). Around the same time during the preschool years, children also show a marked improvement in higher level cognitive processes, termed executive function (EF), that are involved in planning and controlling goal-directed behavior (Zelazo, Carter, Reznick, & Frye, 1997). This rather heterogeneous set of skills seems to group into three core components of EF: working memory, inhibitory control, and mental set shifting (see Garon, Bryson, & Smith, 2008, for a review). In addition to sharing a similar developmental timetable with a marked improvement during the preschool period, empirical evidence of the past two decades points to an ontogenetic link between children’s EF and ToM (e.g., Carlson, Mandell, & Williams, 2004; Flynn, 2007; Frye, Zelazo, & Palfai, 1995; Hala, Hug, & Henderson, 2003; Hughes & Ensor, 2005; Hughes & Ensor, 2007; Müller, Zelazo, & Imrisek, 2005; Perner, Lang, & Kloo, 2002). Typically, the correlations found between individual differences in EF performance and ToM tasks are robust and remain significant after controlling for covarying effects of child age and verbal ability. Moreover, the findings of a study comparing Chinese and U.S. preschoolers suggest that this link is culturally universal and not influenced by differences in socialization goals regarding children’s ability for self-control (Sabbagh, Xu, Carlson, Moses, & Lee, 2006). Although Chinese children outperformed U.S. preschoolers on the EF tasks but not on the ToM tasks, the correlation between EF and ToM performance was robust in both cultural groups. Various explanations have been put forward to account for this association between children’s performance on EF and ToM tasks (see, e.g., Perner & Lang, 1999, for an overview). On the one hand, expression accounts (Hughes & Russell, 1993; Russell, Mauthner, Sharpe, & Tidswell, 1991) posit that successful performance on classical ToM tasks requires some level of executive skills to disengage from salient reality cues as well as from one’s own representations of events to reflect on the mental states of self and others. On the other hand, emergence accounts (Carlson & Moses, 2001; Perner, 1998; Russell, 1996) argue for a functional dependency between EF and ToM in development but offer different proposals regarding developmental direction. For example, Perner (1998) hypothesized that some metarepresentational insight into the causal relation between actions and underlying mental states is fundamental for the development of self-monitoring and inhibitory control. By contrast, Russell (1996) assumed that a certain level of executive control on goal-directed actions is a prerequisite for children’s growing ability to reflect on the mental states underlying these actions. A third account, the cognitive complexity and control (CCC) theory (Frye, 2000; Frye et al., 1995), explains the relation between EF and ToM by reference to developmental changes in children’s domain-general ability to use hierarchically embedded rule systems for reasoning about complex problems. According to this view, EF tasks such as the Dimensional Change Card Sort task (DCCS) (Frye et al., 1995) and classic ToM tasks both are complex problem-solving tasks that require the use of a higher order if–if–then rule to integrate two lower order conflicting representations or rules. Finally, there is some evidence suggesting that the maturation of common brain structures underlying EF and ToM may account for the observed correlations (Ozonoff, Pennington, & Rogers, 1991). Perner’s (1998) position that insight into mental states fosters EF development is supported by evidence showing that children’s performance on the DCCS related to their performance on false belief prediction tasks as well as false belief explanation tasks, in which the pull of the real is less salient (Perner et al., 2002). However, the longitudinal evidence available to date suggests that earlier EF skills predict later ToM abilities rather than the reverse (Carlson, Moses, & Breton, 2002; Carlson et al., 2004; Flynn, 2007; Flynn, O’Malley, & Wood, 2004; Hughes, 1998; Müller et al., 2005). Still, there is no conclusive evidence supporting either an expression or emergence executive account of ToM development, especially given that the test batteries commonly employed comprise a variety of tasks within each area (EF and ToM) that are not necessarily comparable in task formats and task demands. The work of Carlson and Moses (2001) and Carlson et al. (2002) suggests that different components of EF skills show differential relations with children’s ToM development. Specifically, EF tasks that recruit not only inhibitory control but also working memory required for flexible shifting between conflicting propositions regarding one and the same situation (termed conflict tasks) show stronger relations with children’s ToM performance than do EF tasks that mainly recruit inhibitory control

A. Henning et al. / Journal of Experimental Child Psychology 108 (2011) 513–531

515

(termed delay tasks) (see also Hala et al. (2003), for similar findings). In addition, most conclusions regarding the developmental link between EF and ToM are based on correlations between averaged scores from test batteries employed for each area. These batteries usually comprise a variety of tasks assessing epistemic states such as beliefs, especially false beliefs, and only rarely tasks assessing other mental constructs such as desires and emotions. On a closer look, high correlations are generally found between EF development and FB understanding, whereas evidence regarding understanding of mental constructs thought to develop before 4 years of age, such as desires, is mixed (see, e.g., Carlson et al., 2004). Finally, the findings of Sabbagh, Moses, and Shiverick (2006) suggest that especially executive skills are required in reasoning about representations that are normatively thought to be an upto-date reflection of a true state of affairs (beliefs and signs) as compared with reasoning about representations of a past state of affairs (photographs). With reference to this normative aspect, and by drawing on a distinction made in speech act theory between different directions of fit between the world and a representation thereof, Searle (1983) distinguished between epistemic states, such as beliefs that have a mind-to-world fit (i.e., the belief may be at fault, but not the world), and desires and volitional states that have a world-to-mind fit (i.e., the world may be at fault in that it does not comply with or fulfill the desire or volitional state). In addition to these theoretical considerations, there is some evidence pointing toward distinct components of mental reasoning, with one component involved in reasoning about epistemic concepts, such as knowledge and beliefs, and another component involved in reasoning about desires and emotions. Dunn and colleagues, for example, found that emotion understanding and false belief understanding were differentially related to family measures, such as parental education and social class (Cutting & Dunn, 1999; Dunn, Brown, Slomkowski, Tesla, & Youngblade, 1991), as well as to developmental outcome (Dunn, 1995). Furthermore, neither type of mental understanding contributed to the explanation of each other’s variance over and above children’s age, language ability, and family background (Cutting & Dunn, 1999). Based on Searle’s theoretical considerations as well as on the above reviewed findings suggesting possible differential associations between different types of EF tasks and different types of ToM tasks, the main goal of the current work was to assess the specific relation between preschoolers’ EF development as assessed with a conflict task, the DCCS, and different developmental attainments in their theory of mind by employing a battery of scaled ToM tasks that were comparable in task format and task demands. Based on previous findings (e.g., Carlson et al., 2004; Sabbagh et al., 2006), we expected to find stronger relations between the EF task and performance on ToM tasks tapping beliefs, especially false beliefs, than between the same EF task and performance on ToM tasks tapping desire and emotion understanding. It is also still an open question whether individual differences in child temperament may account for the link between EF and ToM development. Temperament is commonly defined as personality traits that emerge early in ontogeny, show stability throughout childhood into adulthood, and have a substantial genetic component (Buss & Plomin, 1984, see Henderson and Wachs (2007), for a review). Buss and Plomin (1984) proposed three temperamental dimensions underlying individual trait differences: emotionality, activity, and sociability. Emotionality refers to a tendency for easy and intensive autonomous arousal that may result in expressions of fear or anger. Activity is defined in terms of energy consumed by motor movement and refers to a tendency for incessant and restless body movement. Sociability denotes a tendency to prefer companionship over solitude and is conceptually differentiated from shyness. According to Buss and Plomin, shyness refers to avoidance of social contact specifically with unfamiliar persons, whereas a shy person may very well prefer the company of familiar persons over solitude. Individual differences in temperament are already observable in early infancy and show an increasing stability after 2 years of age (Henderson & Wachs, 2007). So far, this is also the youngest age for which a correlation between EF and ToM abilities has been reported (Hughes & Ensor, 2005; Hughes & Ensor, 2007). More important, evidence for stable individual differences in EF and ToM across 2, 3, and 4 years of age (Hughes & Ensor, 2007) raises the question of whether child temperament may account in part for the correlation between EF and ToM tasks. Few studies have assessed the relation between child temperament and EF performance (Carlson & Moses, 2001; Carlson et al., 2004) or ToM development (Banerjee & Henderson, 2001; Wellman, Lane, LaBounty, & Olson, in press). To our knowledge, none of the many studies showing a link between EF

516

A. Henning et al. / Journal of Experimental Child Psychology 108 (2011) 513–531

and ToM controlled for the influence of child temperament. However, there is some evidence suggesting that temperament characteristics (which support a contemplative and approaching stance), as opposed to strong reactivity accompanied by negative affective valence and externalizing behavior, may be related to both EF and ToM development. Banerjee and Henderson (2001), for example, reported evidence showing that socially anxious school children have specific difficulties in mental reasoning concerning the links between different mental concepts such as emotion, intention, and belief. Also Wellman et al. (in press) hypothesized that specific social–emotional temperament characteristics may interfere with ToM development. The authors derived their hypothesis from recent comparative research on social cognition by Hare and Tomasello (2004, 2005; see also Hare, 2007) showing that although chimpanzees read social communicative intentions (e.g., gaze direction) in competitive situations, humans clearly outperform chimpanzees, especially in cooperative contexts. With reference to evidence suggesting that domesticated species (e.g., domesticated foxes and dogs) outperform their wild counterparts in reading human attentional cues (Hare et al., 2005), Hare and Tomasello argued that the ancestors of today’s product of domestication, our dogs, were likely those wolves that reacted less aggressively toward humans. Taking up this proposal of Hare and Tomasello (2005) that the initial difference in phylogeny might have been in temperament, Wellman et al. (in press) hypothesized that during child development, an initial difference in temperament may lead to differences in interactive behavior and social experiences, which in turn may foster or interfere with the development of mental understanding. The authors reported supporting evidence showing that specific social–emotional temperament characteristics, such as nonaggressiveness, a shy–withdrawn stance, and perceptual sensitivity, at 3 years of age predicted a more advanced false belief understanding 2 years later. Similarly, there is evidence suggesting that temperament dimensions underlying a contemplative and less reactive stance are positively related to children’s EF development. Carlson and Moses (2001), for example, reported a high congruence between parental ratings of inhibitory control and several EF tasks, including the DCCS, whereas only one item of the inhibitory subscale (doing well on self-control games such as Simon Says) related to ToM performance. In another study, Carlson and colleagues (2004) reported a positive relation between children’s EF performance and parental ratings on attentional focus at 24 months of age and parental ratings on attentional focus, inhibitory control, and perceptual sensitivity at 39 months of age. Interestingly, parental ratings on inhibitory control predicted EF performance but not the reverse. This is consistent with research on adults showing that individual differences in impulsiveness could be predicted from performance on EF tasks (Whitney, Jameson, & Hinson, 2004). In the current work, we employed the EAS Temperament Inventory (Buss & Plomin, 1984) to explore the influence of children’s individual differences in the dimensions emotionality, activity, sociability, and shyness (EAS) on their performance in EF and ToM tasks. A validation study of the German adaptation of this inventory (Angleitner, Harrow, Hempel, & Spinath, 1991) in a sample of 354 pairs of 2- to 14-year-old twins showed good internal consistency as well as significant agreement between maternal and paternal ratings (Spinath, 2000). Furthermore, in addition to a rating form for parents, the EAS inventory also offers a rating form for educators that allowed us to collect ratings on child temperament separately from parents and kindergarten teachers to test for stability in ratings across life contexts. Although findings from previous research converge on a beneficial influence of temperament characteristics that are related to a contemplative and less reactive stance, these findings predict differential relations between the different EAS subscales and children’s EF and ToM. Whereas Carlson and colleagues’ work (Carlson & Moses, 2001; Carlson et al., 2004) suggests that individual differences in emotionality and activity, but not sociability or shyness, might be related to children’s EF skills, Banerjee and Henderson’s (2001) findings point to a specific negative relation between children’s ToM and the dimensions sociability and shyness. In addition, Wellman et al. (in press) study suggests such a specific, albeit positive, relation. Finally, Hughes, Dunn, and White (1998) suggested that children with better EF skills might have good social and communicative skills and, thus, may in turn have more opportunity to learn about other people’s mental states. According to this explanation of the link between EF and ToM, the dimensions sociability and shyness should be related to performance on both EF and ToM tasks. In addition to controlling for the influence of child temperament on the link between EF skills and their ToM development, we also controlled for two family factors, siblings and parental education, that

A. Henning et al. / Journal of Experimental Child Psychology 108 (2011) 513–531

517

are thought to influence ToM and EF development. A number of studies present evidence for a positive influence of siblings on ToM (Jenkins & Astington, 1996; McAlister & Peterson, 2006; McAlister & Peterson, 2007; Perner, Ruffman, & Leekam, 1994); however, the evidence is far from consistent. Whereas some researchers reported such a positive influence on ToM development for older siblings but not for younger ones (Lewis, Freeman, Kyriakidou, Maridaki Kassotaki, & Berridge, 1996; Ruffman, Perner, Naito, Parkin, & Clements, 1998), others showed that this effect varied as a function of the age difference between siblings and variety of siblings (Cassidy, Fineberg, Brown, & Perkins, 2005; Peterson, 2000), and still others did not find any effect (Cole & Mitchell, 2000; Cutting & Dunn, 1999; Peterson & Slaughter, 2003). McAlister and Peterson (2006) showed that the number of siblings had a positive influence on performance in both ToM and EF tasks. Interestingly, although the number of siblings contributed independently to individual differences in ToM, there was no convincing evidence for the number of siblings to mediate the effect of EF on ToM. Similarly, whereas some studies reported a strong relation between maternal educational level and children’s ToM development (Pears & Moses, 2003; see also Cutting & Dunn, 1999), others did not replicate this finding (Ruffman, Perner, & Parkin, 1999). In addition, the evidence regarding the influence of paternal educational level is inconclusive (Cutting & Dunn, 1999; Dunn & Brown, 1994). In summary, the first aim of the current study was to explore whether the developmental link reported in the literature between preschoolers’ EF function and ToM is specific to epistemic states or whether it also applies to other developmental attainments such as understanding desires and real versus apparent emotions. Therefore, we tested preschoolers’ performance on the DCCS and the ToM scale (Hofer & Aschersleben, 2007; Wellman & Liu, 2004) and related their performance on the DCCS to their overall performance as well as to their performance on each of the individual tasks in the battery that maintained comparable task formats and demands. In addition to exploring more closely the relation between preschoolers’ EF and ToM development by taking different mental constructs into account, the second and third aims of the current study were to assess the possible influences of internal child characteristics and of external contextual factors on this developmental relation. Therefore, information on child temperament and regarding siblings and parental education was collected by parental and educators’ reports. Method Participants The final sample comprised the data of 195 3- to 6-year-olds (M = 4;11 [years;months], SD = 1;1, range = 3;0–6;9, 100 girls and 95 boys). An additional 19 children were tested but excluded from analysis because of reported developmental disorders (n = 12), child fussiness (n = 3), experimenter error (n = 3), or insufficient German language ability (n = 1). Of the 195 children, 86 (40 girls and 46 boys) came from a medium-sized city in the southwest of Germany and were tested either in a separate room in their kindergartens (n = 67, 4 kindergartens) or in a university lab (n = 19). The remaining 109 children (60 girls and 49 boys) came from a medium-sized city and neighboring medium-sized town in the west of Germany and were tested in a separate room in their kindergartens (5 kindergartens). With regard to siblings, 23% of the children were single children, 50% had one sibling, 21% had two siblings, and 6% had between three and five siblings. According to parental report, 140 children were monolingual and 55 were bilingual (including 3 who were trilingual). All children were reported to speak German. Children were given a small gift for participating. In addition, kindergartens received a gift and parents received an expense allowance when bringing their children to the university lab. Parents were contacted either through their children’s kindergarten or by telephone from a list of families who had expressed interest in volunteering for research on child development. Tasks and materials Dimensional Change Card Sort To assess children’s EF development, the DCCS (Frye et al., 1995; Zelazo, 2006) was employed. In the standard version of this card-sorting task, children were required to sort bivalent test cards first

518

A. Henning et al. / Journal of Experimental Child Psychology 108 (2011) 513–531

according to one dimension (e.g., shape) and then according to the other dimension (e.g., color). In the more challenging border version of this task, children were required to apply one of the two rules (sorting according to shape or color) according to the presence (e.g., shape) or absence (e.g., color) of a black border framing the test card. The construction of the 23 cards and 2 sorting trays, as well as the setup and procedure, closely followed the protocol of Zelazo (2006). The set of cards comprised 2 target cards, 14 test cards used for the standard version, and 7 test cards for the border version in which the images were framed by a black border. Of the 14 test cards for the standard version, 7 were also used in the more challenging border version. The target cards were affixed to the two sorting trays that were placed side by side in reaching distance in front of the children. During the demonstration phase, the experimenter introduced the preswitch rule (e.g., sort the cards according to shape) and sorted 2 test cards, one of each type, accordingly by putting them face down in the respective trays. During the immediately following preswitch phase, children were asked to sort 6 test cards one by one according to the preswitch rule. The experimenter then introduced the postswitch rule (e.g., sort the cards according to color), and children were asked to sort another 6 cards according to the postswitch rule. Children who passed the postswitch phase proceeded immediately to the border version of the task and were required to sort another 12 test cards. The order of sorting dimensions was counterbalanced across children. During the preswitch phase, 99 children sorted according to color and 96 sorted according to shape. In the border version, test cards with the black border always needed to be sorted according to the respective preswitch rule. Test cards were administered in a fixed semirandom order, making sure that the same type of test card was not selected on more than two consecutive trials. Theory of mind scale A battery of six ToM tasks was employed to assess children’s ToM development. Five of these tasks were closely modeled to the original scale of Wellman and Liu (2004), assessing a range of different developmental attainments (diverse desires, diverse beliefs, knowledge access, contents false belief, and real–apparent emotion). These tasks have been shown to produce a coherent Guttman scale for typical preschoolers in the United States and China (Wellman, Fang, Liu, Zhu, & Liu, 2006; Wellman & Liu, 2004) as well as in Germany when the German version of this scale (Hofer & Aschersleben, 2007) was administered to a sample of 107 German 3- to 5-year-olds (Kristen, Thoermer, Hofer, Aschersleben, & Sodian, 2006). For each task, laminated colored cards and toy figurines were employed as indicated in the manual (Hofer & Aschersleben, 2007). False belief understanding has been widely used as a marker of preschoolers’ ToM development (e.g., Astington & Jenkins, 1999; Dunn et al., 1991; Lalonde & Chandler, 1995). Following Aschersleben, Hofer, and Jovanovicˇ (2008), a second false belief task (explicit false belief) regarding the location of an object was included and was comparable to the other ToM tasks in task format and demands. This allowed us to calculate a combined false belief score and, thus, to specifically investigate the relation between children’s executive control and their false belief understanding in addition to investigating the relation between children’s executive control and their overall ToM development. In the Diverse Desires task, the child needs to differentiate his or her own desire about an object (e.g., a cookie) from another person’s differing desire about the same object to correctly predict the other person’s snack choice (e.g., the carrot and not the cookie). In the Diverse Beliefs task, the child needs to differentiate his or her own belief about the location of an object (e.g., the cat is hiding in the bush) from another person’s differing belief about the location of same object (e.g., the cat is hiding in the garage) to correctly predict the other person’s action (e.g., look for the cat in the garage). The Knowledge Access task requires an understanding of the causal relation between seeing and knowing. In this task, the child is first shown the content of a box and then asked to judge the knowledge of another person who did not have visual access to the content of the box. In the Explicit False Belief task, the child is told the correct location of an object (e.g., the gloves are in the backpack) and that another person thinks that this object is in a different location (e.g., the closet). The child needs to understand that false beliefs lead to erroneous actions to correctly predict that the other person will search in the wrong location. In the False Belief task, the child is shown an unexpected content (toy pig) of a candy box (Smarties), and he or she is told that another person did not have visual access to the content of the box. The child needs to judge the other person’s false belief about the content

A. Henning et al. / Journal of Experimental Child Psychology 108 (2011) 513–531

519

of the box (Smarties), differentiating it from his or her own true belief (toy pig). In the Real–Apparent Emotion task, the child needs to understand that a person may feel one emotion (e.g., sadness) but display a different emotion (e.g., happiness) due to interpersonal constraints. In this task, a pretest assessed the child’s ability to differentiate among three emotional facial expressions (sad, neutral, and happy) given as possible answers to the test questions. If the child failed the pretest, his or her data were excluded for this task. All tasks included a focal test question as well as at least one control question to ensure that children comprehended and remembered all of the relevant story information necessary for producing a meaningful response to the test question. For each task, children needed to pass the test questions and all associated control questions to count as passing the task. Following Wellman and Liu (2004), tasks were administered in a fixed order with increasing degrees of difficulty: Diverse Desires, Diverse Beliefs, Knowledge Access, Explicit False Belief, False Belief, and Real–Apparent Emotion. In addition, a previous study showed no differences in the scaling of the tasks comprising the ToM scale when comparing two different orders of presentation (Kristen et al., 2006; see also Wellman & Liu, 2004, for varying order of tasks). Sentence comprehension: SETK 3–5 To control for language abilities, children’s receptive language ability was assessed with the subscale on sentence comprehension of the Sprachentwicklungstest für Kinder (SETK) [Language Development Test for Children]. This German language development test for 3- to 5-year-olds (Grimm, 2001) comprises six subscales. Each subscale assesses a specific aspect of children’s syntactic and morphological competences and yields subscale-specific standardized scores (T values) as well as percentile ranks. Due to time constraints, in the current study, only one of the subscales (sentence comprehension) was employed to assess children’s comprehension of sentences that differ in complexity. In the first part of the subscale of this test, children were asked to identify one image out of four possible images that corresponded to the content of a sentence read out loud by the experimenter. In the second part of this test, children were asked to manipulate a set of objects according to the instruction given by the experimenter. EAS Temperament Inventory To assess children’s temperament, the EAS Temperament Inventory (Buss & Plomin, 1984) was employed in its German adaptation (Angleitner et al., 1991; Spinath, 2000). In addition to the 20-item parental rating form given to parents, a 19-item rating form for educators was filled in by each child’s kindergarten teacher (except for the 19 children tested in the university lab). This questionnaire measures the dimensions emotionality (e.g., ‘‘cries easily’’), activity (e.g., ‘‘is always on the go’’), sociability (e.g., ‘‘likes to be with people’’), and shyness (e.g., ‘‘tends to be shy’’). For each item, parents and teachers were asked to rate, on a 5-point Likert scale, the degree to which a statement applied to the child (1 = not characteristic of the child, 5 = very characteristic of the child). Sociodemographic questionnaire In addition to the EAS Temperament Inventory, parents were also asked to complete a questionnaire on sociodemographic information. Maternal and paternal education was assessed on an ordinal scale ranging from 0 to 4 according to whether parents did not complete schooling (0); completed secondary schooling on the lowest level (1 = Hauptschulabschluss [secondary school certificate, lower track]), middle level (2 = Realschulabschluss [secondary school certificate, middle track]), or highest level (3 = Abitur [secondary school certificate, top track] or Fachabitur [vocational technical diploma]); or received an academic degree (4 = Hochschulabschluss [university degree] or Fachhochschulabschluss [university of applied sciences degree]). In the German school system, after elementary school, students are put on one of three tracks according to their level of achievement: lower (Hauptschule), middle (Realschule), and top track (Gymnasium). Procedure Children were tested in a quiet room by two experimenters. After a brief warm-up phase with both experimenters, Experimenter 1 administered the ToM scale while Experimenter 2 quietly

520

A. Henning et al. / Journal of Experimental Child Psychology 108 (2011) 513–531

recorded children’s performance on the protocol sheet. Following a short break, Experimenter 2 administered the remaining tasks while Experimenter 1 filled in the protocol. A total of seven pairs of trained experimenters participated in the study. The pairing of the individual experimenters and the assignment of the tasks to each individual experimenter were kept constant across the testing sessions and testing sites. Preliminary analyses showed that children’s performance on the ToM tasks, the DCCS, and the SETK 3–5 did not vary as a function of experimenter pair (all p values >.05). Scoring Dimensional Change Card Sort Following Zelazo (2006), to pass the standard version of the DCCS, children needed to correctly sort at least 5 of the 6 test cards during the postswitch phase. In addition, to pass the standard version, children also needed to correctly sort at least 5 of the 6 test cards during the preswitch phase. To pass the border version of the DCCS, children needed to correctly sort at least 9 of the 12 test cards. Only children who passed the postswitch phase proceeded to the border version. For each child, two scores were calculated. A child was assigned either a 0 (fail) or a 1 (pass) for performance on the postswitch phase (postswitch score). In addition, a combined DCCS score for performance on the preswitch phase, postswitch phase, and border version ranging from 1 to 3 was calculated based on whether children passed only the preswitch phase (1), passed the postswitch phase but not the border version (2), or passed the border version (3). For the postswitch phase, the data of 6 children were excluded because of fussiness (n = 2) or because they did not pass the preswitch phase (n = 4). For the border version, the data of 3 children were excluded due to experimenter error (n = 2) or fussiness (n = 1). Theory of mind scale Children needed to answer the focal test question, as well as the control questions, to count as passing a task. Thus, for each task, children were assigned either a 0 (fail) or a 1 (pass). Consequently, total ToM scale scores ranged from 0 (no task solved) to 5 (all tasks solved). In addition, a combined False Belief score (FB score) ranging from 0 to 2 was calculated by summing the scores of the two False Belief tasks. The ToM scale score could not be calculated for 32 children because of missing data for one or more tasks due to experimenter error (n = 10), fussiness (n = 4), or no or unclear answers (n = 5) or because children did not pass the pretest on recognition of facial expressions in the Real–Apparent Emotion task (n = 13). The FB score could not be calculated for 6 children because of missing data for one task due to experimenter error (n = 2), fussiness (n = 1), or no or unclear answers (n = 3). Sentence comprehension: SETK 3–5 As a measure of language ability, the children’s T values of the SETK 3–5 subscale on sentence comprehension were used for the analyses. The data on sentence comprehension of 4 children was excluded due to experimenter error (n = 3) or fussiness (n = 1). An additional 5 children gave unclear answers on one item (n = 4) or two items (n = 1) of the SETK 3–5. These missing data points were replaced with the modal value of the performance of same-aged children (± 1 month) on the respective items. EAS Temperament Inventory The average score for each of the four subscales (Emotionality, Activity, Sociability, and Shyness) was calculated for parents’ and kindergarten teachers’ ratings. Of the 195 EAS questionnaires handed out to parents, 16 were not returned. In addition, 9 of 20 items were missing in 1 questionnaire, and so that questionnaire was not included in the analyses. Finally, for 3 children, the data of one subscale (Emotionality [n = 2] or Sociability [n = 1]) were excluded because answers to more than one of five items were missing. Of the 176 EAS questionnaires handed out to kindergarten teachers, 16 were not returned.

A. Henning et al. / Journal of Experimental Child Psychology 108 (2011) 513–531

521

Results In this section, first descriptive results are reported for children’s performance on the DCCS, the ToM scale, and the SETK 3–5 as well as for parental and kindergarten teachers’ ratings on the EAS Temperament Inventory. After that, results regarding the relation between children’s executive functioning and their ToM abilities are presented, as are the relations of these two measures with parental ratings of infant temperament, presence of (older) siblings, and parental educational level. Finally, specific relations between the DCCS postswitch scores and the single ToM tasks are investigated. Executive function As shown in Table 1, a total of 133 children passed the postswitch phase of the DCCS, and only 16 children also passed the more challenging border version. Most of the 4- to 6-year-olds, but less than half of the 3-year-olds, passed the postswitch phase, and the majority of 3- and 4-year-olds failed the more challenging border version of the DCCS. In addition, considerably less than half of the 5- and 6year-olds passed the border version. The combined DCCS score ranged from 1 to 3 (M = 1.78, SD = 0.59, n = 186). Performance did not differ as a function of order of sorting rules in the pre- and postswitch phases, t(187) = 1.23, p > .22. Theory of mind The ToM scale score ranged from 0 to 5 (M = 2.83, SD = 1.34, n = 163). As can be seen in Table 1, the general pattern of results replicated the findings obtained in a previous study with German preschoolers (Kristen et al., 2006). In addition, the average score obtained in the current study was nearly identical to the average score reported previously for a German sample of 4-year-olds (M = 2.8) by Aschersleben and colleagues (2008). The combined FB score ranged from 0 to 2 (M = 0.73, SD = 0.79, n = 189), with 91 children failing both False Belief tasks, 40 children passing both tasks, and 58 children passing one of the two tasks. Sentence comprehension The children in the current sample scored in the normal range of language development on the subscale on sentence comprehension of the SETK 3–5 (T values: M = 47.84, SD = 9.39). In addition, when children were grouped into four age groups (see Table 1 for respective age ranges), mean T values ranged from 47.84 to 50.69 and standard deviations ranged from 9.12 to 11.73, indicating that children in each age group scored in a similar normal range of language development. Temperament Table 2 displays parents’ and kindergarten teachers’ average scores for each of the four subscales of the EAS Temperament Inventory. Pearson correlations showed that there were significant positive relations between parents’ and kindergarten teachers’ ratings for the subscales Emotionality, Sociability, and Shyness and a trend for the Activity subscale. In addition, comparisons between group averages showed that parental ratings differed significantly from those of kindergarten teachers on each subscale. On average, parents rated their children as more social, more emotional, more active, and less shy than did their children’s kindergarten teachers. EF and ToM development In a first step, the relation between children’s EF development and their ToM development and each EF and ToM measure’s relation with child temperament and sociodemographic variables were assessed. A series of bivariate correlations (Spearman’s rank or point–biserial) was performed for each ToM measure with each EF measure, children’s age (in months), children’s score for sentence

522

ToM scale

DCCS

Age group

n

Diverse desires

Diverse beliefs

Knowledge access

Explicit false belief (location)

False belief (content)

Real–apparent emotion

Postswitch phase

Border version

3;0–3;11

52

4;0–4;11

46

5;0–5;11

58

6;0–6;11

39

Total

195

42 (80.8) n = 52 39 (84.8) n = 46 55 (94.8) n = 58 38 (97.4) n = 39 174 (89.2) N = 195

32 (61.5) n = 52 27 (58.7) n = 46 46 (79.3) n = 58 32 (82.1) n = 39 137 (70.3) N = 195

16 (33.3) n = 48 23 (50.0) n = 46 46 (82.1) n = 56 28 (73.7) n = 38 113 (60.1) N = 188

11 (21.2) n = 52 12 (26.1) n = 46 27 (46.6) n = 58 18 (46.2) n = 39 68 (34.9) N = 195

6 (12.0) n = 50 7 (16.3) n = 43 38 (66.7) n = 57 20 (51.3) n = 39 71 (37.6) N = 189

4 (10.3) n = 39 6 (15.0) n = 40 11 (20.0) n = 55 14 (36.8) n = 38 35 (20.3) N = 172

22 (46.8) n = 47 29 (63.0) n = 46 46 (80.7) n = 57 36 (92.3) n = 39 133 (70.4) N = 189

0 (0) n = 44 1 (2.2) n = 46 6 (10.5) n = 57 9 (23.1) n = 39 16 (8.6) N = 186

Note. Percentages are in parentheses.

A. Henning et al. / Journal of Experimental Child Psychology 108 (2011) 513–531

Table 1 Frequencies (and percentages) of children in each age group who passed the single ToM tasks, the DCCS postswitch phase, and the border version of the DCCS.

523

A. Henning et al. / Journal of Experimental Child Psychology 108 (2011) 513–531

Table 2 Comparison between parents’ and kindergarten teachers’ average rating scores for each of the four subscales of the EAS Temperament Inventory and intercorrelations between parents’ and kindergarten teachers’ ratings. EAS subscale

Emotionality Activity Sociability Shyness

Parents

Kindergarten teachers

M

SD

n

M

SD

n

t Test

Pearson’s correlation coefficient

n

3.01 3.90 3.79 2.33

0.69 0.74 0.60 0.85

176 178 177 178

2.55 3.41 3.58 2.63

0.89 0.89 0.56 0.79

160 160 160 160

t = 5.73, df = 142, p < .001 t = 6.70, df = 144, p < .001 t = 3.16, df = 143, p = .002 t = 4.27, df = 144, p < .001

r = .27, r = .15, r = .24, r = .40,

143 145 144 145

p = .001 p = .072 p = .005 p < .001

Note. The parental EAS form comprises 20 items. In the EAS form for educators, item 9 of the Activity subscale in the parental form is missing because it refers to children’s activity right after getting up in the morning. Therefore, parental averages calculated across only these four activity items of the educator’s form were used for the t test and the correlation coefficient regarding the Activity subscale. The average scoring of parental ratings on the Activity subscale score comprising only the four items included in the educators’ form is M = 4.01, SD = 0.74.

comprehension, parental score on each EAS subscale, sibling variables, and parental education. Group differences as a function of sibling variables between children passing or failing the postswitch phase were assessed using chi-square tests. In a second step, a series of four regression analyses was conducted. Two regressions with children’s DCCS postswitch score as a dependent variable were performed: once with their total ToM scale score and once with their FB score as a predictor variable. After that, two regressions with children’s DCCS postswitch score as a predictor variable were performed: once with their total ToM scale score as a dependent variable and once with their FB score as a dependent variable. In all regression analyses, the following variables were included as additional predictors: children’s age, children’s score for sentence comprehension, and those sociodemographic variables that were significantly related to the dependent variable in the previous correlational analyses. Finally, to assess the specific relations between children’s EF development and their performance on each of the individual ToM tasks, phi coefficients were computed for each ToM task and the postswitch score. For each analysis in this section, only children with complete data for the relevant measures were included. Bivariate correlations: EF and ToM Children’s ToM scale score correlated significantly with their combined DCCS score, rsp(160) = .40, p < .001, their postswitch score, rpb(162) = .36, p < .001, their age, rsp(163) = .49, p < .001, and their score for sentence comprehension, rsp(159) = .26, p = .001. Similarly, children’s combined FB score correlated significantly with their combined DCCS score, rsp(180) = .38, p < .001, their postswitch score, rpb(183) = .36, p < .001, their age, rsp(189) = .41, p < .001, and their score for sentence comprehension, rsp(185) = .19, p = .010). Bivariate correlations: Temperament Correlations with parental scores on each EAS subscale showed that parental ratings on the Activity subscale were consistently related to children’s performance on the DCCS as well as on both ToM tasks, whereas there were no significant relations and correlation coefficients were rather small for the subscales Emotionality, Sociability, and Shyness (see Table 3). Overall, children who were rated as more active by their parents had lower scores on the combined DCCS score, the ToM scale, and the combined False Belief tasks. In addition, children who did not pass the postswitch phase had higher parental ratings on the Activity subscale (M = 4.06, SD = 0.67) than children who passed the postswitch phase (M = 3.80, SD = 0.76). Bivariate correlations: Siblings The presence or absence of siblings was not significantly related with any of the measures assessing the development of EF and ToM (all p values > .18). Also, there was no significant difference in the postswitch score between children with siblings (106 of 147 children passed) and children without

524

A. Henning et al. / Journal of Experimental Child Psychology 108 (2011) 513–531

Table 3 Correlations between parental ratings of child temperament and children’s ToM and EF development. EAS subscale

Emotionality Activity Sociability Shyness + *

ToM development

EF development

ToM scale score (Spearman’s rank)

n

Combined FB score (Spearman’s rank)

n

Postswitch score (point– biserial)

n

Combined DCCS score (Spearman’s rank)

n

.10 .22* .08 .09

145 146 145 146

.04 .18* .06 .10

170 172 171 172

.07 .16* .07 .02

170 172 171 172

< .01 .15+ .02 .02

167 169 168 169

p < .10. p < .05.

siblings (27 of 42 children passed) (p > .05). Similarly, there was no significant relation between the number of siblings and children’s performance on the DCCS or on the ToM tasks (all p values > .36). Finally, children with siblings (n = 151) were grouped according to whether they had one or more older siblings (n = 96) or not (n = 55). There was no difference in performance on the postswitch phase between children with an older sibling (71 of 94 children passed) and children without an older sibling (34 of 53 children passed) (p > .05). In addition, there was no significant relation between the presence or absence of an older sibling and children’s performance on the ToM tasks (all p values >.14). However, children with an older sibling had higher combined DCCS scores (M = 0.81, SD = 0.76) than children without an older sibling (M = 0.71, SD = 1.13), rpb(144) = .17, p = .038. This result is likely not due to children’s age at testing given that children’s age was not related to the presence or absence of an older sibling, rpb(151) = .07, p = .005. Bivariate correlations: Parental education A trend indicated a positive relation between maternal education and the ToM scale, rsp(159) = .13, p = .098, but not between maternal education and the combined FB score, rsp(185) = .06, p = .409. Maternal education was positively related to children’s combined DCCS score, rsp(183) = .19, p = .012. In addition, mothers of children who passed the postswitch phase had a higher educational level (M = 2.52, SD = 1.09) than mothers of children who did not pass the postswitch phase (M = 2.18, SD = 1.13). This relation was marginally significant, rpb(185) = .14, p = .055. Paternal education was positively related to children’s ToM scale score, rsp(146) = .19, p = .021, as well as to the combined FB score, rsp(168) = .17, p = .027. The positive relation between paternal education and children’s combined DCCS score was marginally significant, rsp(166) = .15, p = .050. In addition, fathers of children who passed the postswitch phase had a higher educational level (M = 2.58, SD = 1.26) than fathers of children who did not pass the postswitch phase (M = 2.20, SD = 1.18). This relation was marginally significant, rpb(168) = .14, p = .073. Maternal educational level was significantly correlated with paternal educational level, rsp(171) = .59, p < .001. Regression analyses A binomial logistic regression was conducted with children’s DCCS postswitch score as a dependent variable and with their total ToM scale score, their age, their score for sentence comprehension, and parental ratings on the Activity subscale as predictors. Nagelkerke’s R2 = .33 indicated that 33% of variance was explained by the model. The hit rate of correctly classified children was 80.1%. Children’s total ToM scale score (B = .59, Wald statistic = 8.42, p = .004) and their score for sentence comprehension (B = .06, Wald statistic = 6.25, p = .012) were significantly related to the criterion variable. Children’s age (p > .06) and Activity subscale ratings (p > .69) were not significant predictors (see Model 1 in Table 4). Another binomial logistic regression was conducted with the DCCS postswitch score as a dependent variable and with children’s combined FB score, their age, their score for sentence comprehension, and parental ratings on the Activity subscale as predictors. Nagelkerke’s R2 = .38 indicated that 38% of variance was explained by the model. The hit rate of correctly classified children was 77.8%. Children’s combined FB score (B = 1.06, Wald statistic = 9.40, p = .002), their age (B = .06, Wald statistic = 9.26, p = .002), and their score for sentence comprehension (B = .07, Wald statistic = 9.60,

525

A. Henning et al. / Journal of Experimental Child Psychology 108 (2011) 513–531

Table 4 Relations of children’s ToM measures and their DCCS postswitch scores controlling for age, sentence comprehension, child temperament, and parental education. Independent variable

B

B SE

b

sr2

Exp(B)

Model 1: DCCS postswitch score (DV), N = 141; model fit: ToM scale score 0.59* Age (months) 0.04 Sentence comprehension 0.06* Activity subscale (parental rating) 0.13

2 log likelihood = 125.802, Nagelkerke’s R2 = .33 0.20 0.02 0.03 0.33

1.80 1.04 1.06 0.88

Model 2: DCCS postswitch score (DV), N = 162; model fit: Combined FB score 1.06* Age (months) 0.06* Sentence comprehension 0.07* Activity subscale (parental rating) 0.15

2 log likelihood = 142.527, Nagelkerke’s R2 = .38 0.34 0.02 0.02 0.32

2.87 1.06 1.07 0.86

Model 3: ToM scale score (DV), N = 141, R2 = .35 DCCS postswitch score 0.73* Age (months) 0.04** Sentence comprehension 0.02* Activity subscale (parental rating) 0.17 Paternal educational level 0.003

0.23 0.01 0.01 0.13 <0.01

.24 .38 .15 .09 .08

.08 .15 .03 .01 .01

Model 4: combined FB score (DV, dichotomized), N = 109; model fit: 2 log likelihood = 97.950, Nagelkerke’s R2 = .43 DCCS postswitch score 2.54* 1.08 12.66 Age (months) 0.08* 0.02 1.08 Sentence comprehension 0.04 0.03 1.05 Activity subscale (parental rating) 0.23 0.35 0.79 Paternal educational level 0.002 0.01 1.00 Note. Only those variables of child temperament and sociodemographic background that were significantly related to the dependent variable in the previous correlational analyses were included. Exp(B) indicates that for every 1-unit increase in the predictor, the odds of passing the postswitch phase, or of passing both False Belief tasks, increased by a factor equal to Exp(B). DV, dependent variable; FB, False Belief. * p < .05. ** p < .001.

p = .002) were significantly related to the criterion variable. Activity subscale ratings were not a significant predictor (p > .63) (see Model 2 in Table 4). For the multiple regression with children’s total ToM scale score as the criterion variable and with children’s DCCS postswitch score, their age, their score for sentence comprehension, parental ratings on the Activity subscale, and paternal educational level as predictors, the R of the overall model was significantly different from zero, F(5, 135) = 14.39, p < .001, R2 = .35. Children’s DCCS postswitch score (B = .73, t = 3.13, p = .002), their age (B = .04, t = 4.96, p < .001), and their score for sentence comprehension (B = .02, t = 2.12, p = .036) were significantly related to the criterion variable. Activity subscale ratings (p > .20) and paternal education (p > .29) were not significant predictors (see Model 3 in Table 4). The unique contributions to the total variance of the ToM scale score were 8, 15, and 3% for children’s DCCS postswitch score, their age, and their score for sentence comprehension, respectively (indicated by squared part correlation [sr2]). Finally, a multinomial logistic regression was performed with children’s dichotomized FB score1 as a dependent variable and with children’s DCCS postswitch score, their age, their score for sentence comprehension, parental ratings on the Activity subscale, and paternal educational level as predictors. Nagelkerke’s R2 = .43 indicated that 43% of variance was explained by the model. The hit rate of correctly classified children was 76.1%. Children’s DCCS postswitch score (B = 2.54, Wald statistic = 5.49, p = .019) and their age (B = .08, Wald statistic = 10.58, p = .001) were significantly related to the criterion variable. Children’s sentence comprehension score (p > .14), parental Activity subscale ratings (p > .49),

1 In the case of a categorical variable with more than two categories, the statistical software used (SPSS 17) automatically dichotomizes the variable by taking one category of the variable as a reference. Therefore, we created a theoretically plausible dichotomized FB belief score by including only children who either failed both FB tasks (receiving a score of 0) or passed both tasks (receiving a score of 1).

526

A. Henning et al. / Journal of Experimental Child Psychology 108 (2011) 513–531

and paternal educational level (p > .85) were not significant predictors (see Model 4 in Table 4). In sum, regression analyses showed that the relations between the ToM measures and the DCCS postswitch score remained significant after controlling for children’s age, their score on sentence comprehension, child temperament, and parental education. Specific relations between the DCCS postswitch score and the single ToM tasks A central aim of the current study was to explore the specific relations between children’s performance on a conflict EF task, the DCCS, and different developmental attainments in their understanding of mental states as measured by the different tasks of the ToM scale. Children’s understanding of epistemic states, but not of diverse desires or the relation between felt and displayed emotions, was strongly related to their EF development (see Table 5). Children’s performance on both False Belief tasks, as well as on the Diverse Beliefs and Knowledge Access tasks, showed a significant positive relation with their performance on the postswitch phase of the DCCS (all p values 6 .001). Unlike ToM tasks assessing epistemic states, children’s performance on the Diverse Desires and Real–Apparent Emotion tasks was not significantly related to their postswitch score (p > .40 and p = .097, respectively). Because the two tasks that showed the least variance, Diverse Desire and Real–Apparent Emotion, were also the two tasks that did not show a significant correlation with children’s performance on the DCCS postswitch score, we conducted the same set of correlational analyses on two subsamples of children who demonstrated greater variability. In Subsample 1, 19 children who failed the Diverse Desire task (mean age = 4;2, SD = 0;10, range = 38–75 months, 10 girls and 9 boys, 12 children residing in the west of Germany and 7 children residing in the southwest of Germany; order in DCCS: 4 children in color first and 15 children in shape first) were matched individually to one of the children who passed this task (mean age = 4;2, SD = 0;10, range = 38–75 months, 11 girls and 8 boys, 12 children residing in the west of Germany and 7 children residing in the southwest of Germany; order in DCCS: 11 children in color first and 8 children in shape first). An additional 2 children failed the Diverse Desire task but were not included in this subsample due to missing data for the DCCS postswitch phase. In Subsample 2, 35 children who passed the Real–Apparent Emotion task (mean age = 5;6, SD = 1;0, range = 38–82 months, 14 girls and 21 boys, 17 children residing in the west of Germany and 18 children residing in the southwest of Germany; order in DCCS: 16 children in color first and 19 children in shape first) were individually matched to one of the children who did not pass this task (mean age = 5;6, SD = 0;11, range = 39–80 months, 14 girls and 21 boys, 13 children residing in the west of Germany and 22 children residing in the southwest of Germany; order in DCCS: 18 children in color first and 17 children in shape first). Children were matched individually regarding age, gender, and residence area (west or southwest of Germany). Given that perfect matches on all characteristics were not possible for all cases due to the post hoc nature of the procedure, age (in years and months) was given priority over gender and residence area.

Table 5 Relation between passing or failing the postswitch phase and passing or failing each individual ToM task. ToM task

Diverse Desires Diverse Beliefs Knowledge Access Explicit False Belief (location) False Belief (content) Real–Apparent Emotion

DCCS postswitch

Fail Pass Fail Pass Fail Pass Fail Pass Fail Pass Fail Pass

Fail

Pass

n

Phi coefficient

7 49 26 30 34 21 48 8 46 9 42 6

12 121 30 103 38 91 75 58 66 62 92 29

189

phi = .053, p = .468

189

phi = .239, p = .001

184

phi = .304, p < .001

189

phi = .281, p < .001

183

phi = .302, p < .001

169

phi = .128, p = .097

A. Henning et al. / Journal of Experimental Child Psychology 108 (2011) 513–531

527

For each subsample, phi coefficients were computed to explore the specific relations between children’s performance on the DCCS and their performance on the different ToM tasks. Again, only the data of children with complete data for the relevant measures were included. In both subsamples, children’s understanding of diverse desires as well as their understanding of the distinction between felt and displayed emotions did not relate to their performance on the postswitch phase of the DCCS (Subsample 1: Diverse Desire, phi = .054, p > .73, n = 38, and Real–Apparent Emotion, phi = .115, p > .57, n = 24; Subsample 2: Diverse Desire, phi = .000, p = 1.0, n = 70, and Real–Apparent Emotion, phi = – .082, p > .49, n = 70). With regard to the tasks assessing children’s understanding of epistemic states, in Subsample 1, children’s performance on both False Belief tasks showed a significant positive relation with their performance on the DCCS (Explicit False Belief: phi = .384, p = .018, n = 38; False Belief: phi = .387, p = .024, n = 34). The relation for the Knowledge Access task showed a trend (phi = .293, p = .07, n = 38), and the relation for the Diverse Beliefs task was not significant (phi = .119, p > .46, n = 38). In Subsample 2, analyses yielded significant positive relations for the Diverse Beliefs (phi = .361, p = .003, n = 70), Knowledge Access (phi = .269, p = .027, n = 68), and False Belief (phi = .321, p = .009, n = 66) tasks, but not for the Explicit False Belief task (phi = .129, p > .28, n = 70). In sum, these additional analyses with subsamples that demonstrated greater variance, once in the Diverse Desires task and once in the Real–Apparent Emotion task, yielded patterns of results that were very similar to the original pattern of results. Discussion The aim of the current study was to assess the specific relation between 3- to 6-year-olds’ EF development, as assessed with a card-sorting task, and different developmental attainments in their ToM by employing a battery of scaled ToM tasks that were comparable in task format and task demands. The main findings showed a link between EF and ToM development even after controlling for age and sentence comprehension as well as for covarying effects of child temperament and parental education. Importantly, however, this relation was specific to those ToM tasks that tap children’s understanding of epistemic states such as knowledge access, diverse beliefs, and false beliefs regarding content and location. Children’s performance on the postswitch phase as well as on the border version of the DCCS was overall consistent with previous findings (Carlson, 2005; Garon et al., 2008; see Zelazo, 2006, for specific predictions regarding the DCCS). Similarly, children’s performance on the ToM scale replicated previous patterns of results suggesting a developmental progression from understanding discrepant desires, to understanding beliefs, to understanding false beliefs, and finally to understanding the distinction between real and apparent emotion (Kristen et al., 2006; Wellman & Liu, 2004; Wellman et al., 2006). Whereas the majority of children in each age group passed the Diverse Desires and Diverse Beliefs tasks, the classical drop in 3-year-olds’ performance can be observed for the two False Belief tasks, for which 4- to 6-year-olds’ performance was better but still far from ceiling. The majority of children, even at 6 years of age, had difficulties in passing the Real–Apparent Emotion task. This failure is due in part to the relatively high task demands on children’s working memory given that many children failed the control questions that ascertain children’s memory of the relevant story information. However, of those children who correctly remembered the story line (n = 84), less than half of them (n = 35) correctly answered the test questions. Overall, the average scores on the ToM scale as well as on the FB tasks were comparable to those reported for the validation study of the German ToM scale (Kristen et al., 2006; see also Aschersleben et al., 2008). Correlational analyses showed that children’s score on the postswitch phase and their combined DCCS score were positively related to their performance on the ToM scale as well as to their combined FB score. Regression analyses showed that these relations remained significant after controlling for effects of children’s age, their scores on sentence comprehension, child temperament, and parental education. These findings are in line with the wealth of cross-sectional and longitudinal evidence suggesting a robust link between EF and ToM development (e.g., Carlson et al., 2004; Frye et al., 1995; Hughes, 1998; Lang & Perner, 2002; Müller et al., 2005). In addition, consistent with predictions, when testing the specific relations between children’s EF development and their performance on each of the individual ToM tasks, an intriguing pattern

528

A. Henning et al. / Journal of Experimental Child Psychology 108 (2011) 513–531

emerged. Children’s performance on the postswitch phase of the DCCS showed significant positive relations with those ToM tasks that tap their understanding of diverse beliefs, knowledge access, and false belief regarding content and location. Correlation coefficients for the Diverse Desires and Real–Apparent Emotion tasks were also positive but far from statistically significant for Diverse Desires and only indicating a trend for Real–Apparent Emotion. This pattern of findings is consistent with Sabbagh and colleagues’ (2006) proposal that executive skills, especially inhibitory control and working memory, are required for reasoning about representations that are expected to reflect a true state of affairs. However, the current evidence is inconclusive regarding the hypothesis that the link between EF and ToM development is based on acquiring the ability for complex rule-based reasoning. The tasks comprising the ToM scale are similar in task format, which in all tasks except the Real– Apparent Emotion task may be conceived in terms of hierarchically embedded levels. In other words, in the tasks tapping epistemic states as well as the Diverse Desires task, discrepant propositions regarding one and the same situation may be integrated when taking into account the difference in mental perspectives between the self and the other. Our pattern of findings is in line with previous theoretical considerations (Searle, 1983) as well as empirical evidence (e.g., Cutting & Dunn, 1999) suggesting a component view of mental understanding. Although both belief and desire understanding require the understanding that different persons may have different mental states regarding the same object or state of affairs in the world, epistemic states also require an understanding that mental representations may differ from reality. Abraham, Rakoczy, Werning, von Cramon, and Schubotz (2010) reported important evidence supporting a dissociation between belief and desire understanding at a functional level. Their functional magnetic resonance imaging (fMRI) study showed that processing scenarios involving false beliefs, but not scenarios involving unfulfilled desires, was selectively accompanied by activation in a brain region thought to be involved in the critical process of decoupling the representation of a belief from the representation of the reality that this belief is about (Leslie, 1987). In a similar vein, Pineda and Hecht (2009) recently proposed a dual-component model of ToM that posits two distinct components underlying mental state reasoning. Whereas one earlier developing component is involved in reasoning about person–object-directed states such as attention and desire, a second later developing component is thought to be presentation based and involved in language and theory building. The authors argued that the first component may rely in part on a simulation mechanism (i.e., one may simulate how one would feel about a certain situation), whereas this simulation mechanism cannot be employed to correctly judge whether a mental state is a faithful representation of a state of affairs. In sum, it may well be that reasoning about those mental states that involve judging whether a mental state faithfully represents a state of affairs draws more heavily on executive skills than those mental states that cannot be false by definition. Parental ratings of child temperament were positively related to kindergarten teachers’ ratings with high statistical significance for the dimensions emotionality, sociability, and shyness and a trend for the dimension activity. These correlations indicate relative stability in children’s temperament in different social contexts. The size of the correlation coefficients are in line with temperament research showing that agreements between parents and kindergarten teachers tend to be moderate due to obvious differences in interactional contexts and size of reference population (see, e.g., Funder & West, 1993). Comparisons between group averages showed that parents rated their children as more social, more emotional, more active, and less shy than did their children’s kindergarten teachers. Given the overall positive relations between parents’ and kindergarten teachers’ ratings, a possible explanation for these differences in group averages might be that parents were more prone to socially desirable response tendencies than kindergarten teachers. A few previous studies (Banerjee & Henderson, 2001; Carlson & Moses, 2001; Carlson et al., 2004; Wellman et al., in press), which investigated individual differences in child temperament in relation to either children’s EF skills or their ToM development, suggested a differential relation between temperamental factors and performance on the DCCS and ToM scale. In the current study, individual differences in the temperamental dimension activity were consistently related to both EF and ToM development. Children rated by their parents as more active had more difficulties in switching between conflicting rules and reasoning about mental states. Contrary to predictions, the temperamental dimensions sociability and shyness were not related to either EF or ToM development. Therefore,

A. Henning et al. / Journal of Experimental Child Psychology 108 (2011) 513–531

529

these findings are incongruent with Hughes and colleagues’ (1998) suggestion that influence of EF skills on ToM development is mediated by children’s social skills. In line with previous work showing a predictive influence of inhibitory control on children’s EF development (e.g., Carlson et al., 2004), the current pattern of findings rather suggests a mediating influence of the dimension activity on this developmental link. Alternatively, this pattern of results may simply stem from a general difficulty of children who exhibit greater motor activity to participate successfully on all sorts of tasks employed in developmental research. Because directionality of effects cannot be established from the correlational data in this study, future research is needed to further investigate the relations between child temperament and EF and ToM development. In addition to controlling for the influence of child temperament on the link between EF skills and children’s ToM development, we also controlled for two family factors, siblings and parental education, that are thought to especially influence ToM development. Previous research pointed to a beneficial effect of siblings on ToM (e.g., Cassidy et al., 2005; Ruffman et al., 1998) as well as on EF development (McAlister & Peterson, 2006). In the current study, the presence and number of siblings as well as the presence of one or more older siblings did not have a positive effect on either children’s ToM development or their EF development, consistent with previous studies showing no effect for siblings on ToM development (e.g., Peterson & Slaughter, 2003). Following the argument of Cutting and Dunn (1999), it is likely that the quality of interaction, rather than the number or age of siblings per se, is relevant for children’s ToM development. Regarding parental education, in the current study, children whose parents had a higher educational level tended to have better EF and ToM abilities. Thus, the current findings are in line with previous work showing a positive influence on ToM development (e.g., Pears & Moses, 2003). However, somewhat unexpectedly, the size of relations between educational level and child performance showed reverse patterns for maternal education as compared with paternal education. Relations with ToM development were overall significant for paternal education compared with a trend for the relation between maternal education and overall ToM development and virtually no relation between maternal education and children’s false belief understanding. Conversely, relations with EF were overall stronger for maternal educational level compared with paternal educational level. A possible explanation for such a relation is that parents with a higher educational level tend to have children with higher general cognitive abilities (see also Pears & Moses, 2003). Although the current findings do not suggest a strong influence of parental education on children’s EF and ToM development, it is important to note that the link between children’s EF and ToM development was also robust after controlling for parental educational level. In summary, the findings of the current study replicate previous research indicating a robust link between children’s EF and ToM development during the preschool years. Moreover, the investigation of the specific relations between children’s EF and their understanding of different mental constructs extends previous research by suggesting a special role of executive skills for understanding epistemic states in comparison with understanding desires and emotions. Importantly, comparability of results across different ToM tasks is given by maintaining parallel formats and task demands. However, a limitation of the current study involves the use of only a single EF task. Future research needs to assess whether this pattern of findings generalizes across different EF tasks as well as across different components of executive functioning (i.e., inhibitory control, working memory, and mental set shifting). In addition, future research is needed to address the question of whether there is an underlying cognitive ability common to conflict EF tasks and ToM tasks tapping epistemic states and, if that is the case, whether this ability is domain general or specific to social cognition. Finally, further research also needs to more closely explore a possible moderating influence of specific social–emotional temperament characteristics on children’s developing EF skills and mental understanding.

Acknowledgments The authors thank the psychology undergraduate students of the Empiriepraktikum 2007–2008 as well as Alicia Daniels, Julia Daniels, and Saskia Schwadtke for help with data collection and coding. We are especially grateful to the kindergarten staff members for their support and to the parents and

530

A. Henning et al. / Journal of Experimental Child Psychology 108 (2011) 513–531

children who participated in the study. We further thank Stephanie Carlson and two anonymous reviewers for their constructive and helpful comments on earlier versions of the manuscript. References Abraham, A., Rakoczy, H., Werning, M., von Cramon, D. Y., & Schubotz, R. I. (2010). Matching mind to world and vice versa: Functional dissociations between belief and desire mental state processing. Social Neuroscience, 5, 1–18. Angleitner, A., Harrow, J., Hempel, S., & Spinath, F. M. (1991). Deutsche Adaptation Des Emotionalitäts-Aktivitäts-SoziabilitätsTemperamentinventars (Elternbeurteilungsform) [Emotionality-activity-sociability temperament inventory (parental rating form)]. Unpublished questionnaire, Bielefeld University. Aschersleben, G., Hofer, T., & Jovanovicˇ, B. (2008). The link between infant attention to goal-directed action and later theory of mind abilities. Developmental Science, 11, 862–868. Astington, J. W., & Jenkins, J. M. (1999). A longitudinal study of the relation between language and theory-of-mind development. Developmental Psychology, 35, 1311–1320. Banerjee, R., & Henderson, L. (2001). Social–cognitive factors in childhood social anxiety: A preliminary investigation. Social Development, 10, 558–572. Buss, A. H., & Plomin, R. (1984). Temperament: Early developing personality traits. Hillsdale, NJ: Lawrence Erlbaum. Carlson, S. M. (2005). Developmentally sensitive measures of executive function in preschool children. Developmental Neuropsychology, 28, 595–616. Carlson, S. M., Mandell, D., & Williams, L. (2004). Executive function and theory of mind: Stability and prediction from ages 2 to 3. Developmental Psychology, 40, 1105–1122. 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., & Breton, C. (2002). How specific is the relation between executive function and theory of mind? Contributions of inhibitory control and working memory. Infant and Child Development, 11, 73–92. Cassidy, K. W., Fineberg, D. S., Brown, K., & Perkins, A. (2005). Theory of mind may be contagious, but you don’t catch it from your twin. Child Development, 76, 97–106. Cole, K., & Mitchell, P. (2000). Siblings in the development of executive control and a theory of mind. British Journal of Developmental Psychology, 18, 279–295. Cutting, A. L., & Dunn, J. (1999). Theory of mind, emotion understanding, language, and family background: Individual differences and interrelations. Child Development, 70, 853–865. Dunn, J. (1995). Children as psychologists: The later correlates of individual differences in understanding of emotions and other minds. Cognition and Emotion, 9, 187–201. Dunn, J., & Brown, J. R. (1994). Affect expression in the family, children’s understanding of emotions, and their interactions with others. Merrill–Palmer Quarterly, 40, 120–137. Dunn, J., Brown, J., Slomkowski, C., Tesla, C., & Youngblade, L. (1991). Young children’s understanding of other people’s feelings and beliefs: Individual differences and their antecedents. Child Development, 62, 1352–1366. Flynn, E. (2007). The role of inhibitory control in false belief understanding. Infant and Child Development, 16, 53–69. Flynn, E., O’Malley, C., & Wood, D. (2004). A longitudinal, microgenetic study of the emergence of false belief understanding and inhibition skills. Developmental Science, 7, 103–115. Frye, D. (2000). Theory of mind, domain specificity, and reasoning. In P. Mitchell & K. J. Riggs (Eds.), Children’s reasoning and the mind (pp. 149–167). Hove, UK: Psychology Press. Frye, D., Zelazo, P., & Palfai, T. (1995). Theory of mind and rule-based reasoning. Cognitive Development, 10, 483–527. Funder, D. C., & West, S. G. (1993). Consensus, self–other agreement, and accuracy in personality judgment: An introduction. Journal of Personality, 61, 457–476. Garon, N., Bryson, S. E., & Smith, I. M. (2008). Executive function in preschoolers: A review using an integrative framework. Psychological Bulletin, 134, 31–60. Grimm, H. (2001). Sprachentwicklungstest für Kinder (SETK 3–5) [Language development test for 3- to 5-year-old children]. Göttingen, Germany: Hogrefe. Hala, S., Hug, S., & Henderson, A. (2003). Executive function and false-belief understanding in preschool children: Two tasks are harder than one. Journal of Cognition and Development, 4, 275–298. Hare, B. (2007). From nonhuman to human mind: What changed and why? Current Directions in Psychological Science, 16, 60–64. Hare, B., Plyusnina, I., Ignacio, N., Schepina, O., Stepika, A., Wrangham, R., et al (2005). Social cognitive evolution in captive foxes is a correlated by-product of experimental domestication. Current Biology, 15, 226–230. Hare, B., & Tomasello, M. (2004). Chimpanzees are more skillful in competitive than in cooperative cognitive tasks. Animal Behaviour, 68, 571–581. Hare, B., & Tomasello, M. (2005). Human-like social skills in dogs? Trends in Cognitive Sciences, 9, 439–444. Henderson, H. A., & Wachs, T. D. (2007). Temperament theory and the study of cognition–emotion interactions across development. Developmental Review, 27, 396–427. Hofer, T., & Aschersleben, G. (2007). ‘‘Theory of Mind’’ Skala für 3- bis 5-jährige Kinder [German theory of mind scale for 3- to 5-yearold children] (3rd ed.). Munich, Germany: Max Planck Institute for Human Cognitive and Brain Sciences. Hughes, C. (1998). Finding your marbles: Does preschoolers’ strategic behavior predict later understanding of mind? Developmental Psychology, 34, 1326–1339. Hughes, C., Dunn, J., & White, A. (1998). Trick or treat? Uneven understanding of mind and emotion and executive function among ‘‘hard to manage’’ preschoolers. Journal of Child Psychology and Psychiatry, 39, 981–994. Hughes, C., & Ensor, R. (2005). Theory of mind and executive function in 2-year-olds: A family affair? Developmental Neuropsychology, 28, 645–668. Hughes, C., & Ensor, R. (2007). Executive function and theory of mind: Predictive relations from ages 2 to 4. Developmental Psychology, 43, 1447–1459.

A. Henning et al. / Journal of Experimental Child Psychology 108 (2011) 513–531

531

Hughes, C., & Russell, J. (1993). Autistic children’s difficulty with mental disengagement from an object: Its implication for theories of autism. Developmental Psychology, 29, 498–510. Jenkins, J. M., & Astington, J. W. (1996). Cognitive factors and family structure associated with theory of mind development in young children. Developmental Psychology, 32, 70–78. Kristen, S., Thoermer, C., Hofer, T., Aschersleben, G., & Sodian, B. (2006). Skalierung von Theory of Mind Aufgaben [Scaling of theory of mind tasks]. Zeitschrift für Entwicklungspsychologie und Pädagogische Psychologie [Journal of Developmental Psychology and Educational Psychology], 38, 186–195. Lalonde, C. E., & Chandler, M. J. (1995). False belief understanding goes to school: On the social–emotional consequences of coming early or late to a first theory of mind. Cognition and Emotion, 9, 167–185. Lang, B., & Perner, J. (2002). Understanding of intention and false belief and the development of self-control. British Journal of Developmental Psychology, 20, 67–76. Leslie, A. M. (1987). Pretense and representation in infancy: The origins of ‘‘theory of mind’’. Psychological Review, 94, 412–426. Lewis, C., Freeman, N. H., Kyriakidou, C., Maridaki Kassotaki, K., & Berridge, D. M. (1996). Social influences on false belief access: Specific sibling influences or general apprenticeship? Child Development, 67, 2930–2947. McAlister, A., & Peterson, C. (2006). Mental playmates: Siblings, executive functioning, and theory of mind. British Journal of Developmental Psychology, 24, 733–751. McAlister, A., & Peterson, C. (2007). A longitudinal study of child siblings and theory of mind development. Cognitive Development, 22, 258–270. Müller, 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. Ozonoff, S., Pennington, B. F., & Rogers, S. J. (1991). Executive function deficits in high-functioning autistic children: Relationship to theory of mind. Journal of Child Psychology & Psychiatry, 32, 1081–1105. Pears, K. C., & Moses, L. J. (2003). Demographics, parenting, and theory of mind in preschool children. Social Development, 12, 1–19. Perner, J. (1998). The meta-intentional nature of executive functions and theory of mind. In P. Carruthers & J. Boucher (Eds.), Language and thought (pp. 270–283). Cambridge, UK: Cambridge University Press. Perner, J., & Lang, B. (1999). Development of theory of mind and executive control. Trends in Cognitive Sciences, 3, 337–344. Perner, J., Lang, B., & Kloo, D. (2002). Theory of mind and self-control: More than a common problem of inhibition. Child Development, 73, 752–767. Perner, J., Ruffman, T., & Leekam, S. R. (1994). Theory of mind is contagious: You catch it from your sibs. Child Development, 65, 1228–1238. Peterson, C. C. (2000). Kindred spirits: Influence of siblings’ perspectives on theory of mind. Cognitive Development, 15, 435–455. Peterson, C. C., & Slaughter, V. (2003). Opening windows into the mind: Mothers’ preferences for mental state explanations and children’s theory of mind. Cognitive Development, 18, 399–429. Pineda, J. A., & Hecht, E. (2009). Mirroring and mu rhythm involvement in social cognition: Are there dissociable subcomponents of theory of mind? Biological Psychology, 80, 306–314. Ruffman, T., Perner, J., Naito, M., Parkin, L., & Clements, W. A. (1998). Older (but not younger) siblings facilitate false belief understanding. Developmental Psychology, 34, 161–174. Ruffman, T., Perner, J., & Parkin, L. (1999). How parenting style affects false belief understanding. Social Development, 8, 395–411. Russell, J. (1996). Agency: Its role in mental development. Hove, UK: Lawrence Erlbaum. Russell, J., Mauthner, N., Sharpe, S., & Tidswell, T. (1991). The ‘‘windows task’’ as a measure of strategic deception in preschoolers and autistic subjects. British Journal of Developmental Psychology, 9, 331–349. Sabbagh, M. A., Moses, L. J., & Shiverick, S. M. (2006). Executive functioning and preschoolers’ understanding of false belief, false photographs, and false signs. Child Development, 77, 1034–1049. Sabbagh, M. A., Xu, F., Carlson, S. M., Moses, L. J., & Lee, K. (2006). The development of executive functioning and theory of mind. Psychological Science, 17, 74–81. Searle, J. R. (1983). Intentionality: An essay in the philosophy of mind. Cambridge, UK: Cambridge University Press. Spinath, F. M. (2000). Temperamentsmerkmale bei Kindern: Psychometrische Güte und verhaltensgenetische Befunde zum deutschen Emotionalitäts-Aktivitäts-Soziabilitäts Temperamentinventar (EAS) nach Buss und Plomin (1984) [Temperament traits in children: Psychometric properties and behavioral-genetic results with the German Emotionality-activitysociability temperament inventory (EAS, Buss & Plomin, 1984)]. Zeitschrift für Differentielle und Diagnostische Psychologie [Journal of Differential and Diagnostic Psychology], 21, 65–75. Wellman, H. M., Cross, D., & Watson, J. (2001). Meta-analysis of theory of mind development: The truth about false belief. Child Development, 72, 655–684. Wellman, H. M., Fang, F., Liu, D., Zhu, L., & Liu, G. (2006). Scaling of theory-of-mind understandings in Chinese children. Psychological Science, 17, 1075–1081. Wellman, H. M., Lane, J. D., LaBounty, J., & Olson, S. L. (in press). Observant, nonaggressive temperament predicts theory-of-mind development. Developmental Science. doi:10.1111/j.1467-7687.2010.00977.x. Wellman, H. M., & Liu, D. (2004). Scaling of theory-of-mind tasks. Child Development, 75, 523–541. Whitney, P., Jameson, T., & Hinson, J. M. (2004). Impulsiveness and executive control in working memory. Personality and Individual Differences, 37, 417–428. Zelazo, P. D. (2006). The dimensional change card sort (DCCS): A method of asserting executive function in children. Nature Protocols, 1, 297–301. Zelazo, P. D., Carter, A., Reznick, J. S., & Frye, D. (1997). Early development of executive function: A problem-solving framework. Review of General Psychology, 1, 1–29.