Word counts:
Short Abstract 85
Long Abstract 209
Main text 16,639
References 5,485
Entire text 22,582
Jeremy I. M. Carpendale
Department of Psychology
Simon Fraser University
8888 University Drive
Burnaby, BC Canada V5A 1S6
email: jcarpend@sfu.ca
Charlie Lewis
Department of Psychology
Lancaster University,
Lancaster, LA1 4YF, UK
email: c.lewis@lancaster.ac.uk
Abstract (short abstract)
Theories of children’s developing understanding of mind tend to emphasize either individualistic processes of theory formation, maturation or introspection, or the process of enculturation. We propose an alternative account, according to which the development of children’s social understanding occurs within triadic interaction involving the child's experience of the world as well as communicative interaction with others about their experience and beliefs. We contend that it is through such triadic interaction that children gradually construct knowledge of the world as well as knowledge of other people.
Abstract (long abstract)
Theories of children’s developing understanding of mind tend to emphasize either individualistic processes of theory formation, maturation, or introspection, or the process of enculturation. However, such theories must be able to account for the accumulating evidence of the role of social interaction in the development of social understanding. We propose an alternative account, according to which the development of children’s social understanding occurs within triadic interaction involving the child's experience of the world as well as communicative interaction with others about their experience and beliefs (Chapman, 1991, 1999). It is through such triadic interaction that children gradually construct knowledge of the world as well as knowledge of other people. We contend that the extent and nature of the social interaction children experience will influence the development of children’s social understanding. Increased opportunity to engage in cooperative social interaction and exposure to talk about mental states should facilitate the development of social understanding. We review evidence suggesting that children’s understanding of mind develops gradually in the context of social interaction. Thus, we need a theory of development in this area that accords a fundamental role to social interaction, yet does not assume that children simply adopt socially available knowledge, but rather children construct an understanding of mind within social interaction.
Key Words: Language; Piaget; Social interaction; Theories of mind; Vygotsky; Wittgenstein.
Constructing an Understanding of Mind:
The Development of Children’s Social Understanding within Social Interaction
1. Introduction
Debate concerning how children come to understand the social and psychological world, now often known as children’s "theories of mind," has become increasingly concerned with the influence of social interaction. Recent studies have found that individual differences in preschoolers' competence on measures of false belief understanding are correlated with aspects of children's socialization history. By suggesting that social interaction may influence the development of children’s mentalistic understanding, this research has reopened issues which have long been discussed in the debate over how children come to understand the mind.
As an example of this research, Perner, Ruffman and Leekam (1994) reported that preschoolers with siblings demonstrated false belief understanding at an earlier age than children without siblings. This "sibling effect" was replicated by Jenkins and Astington (1996), although they found that it was less pronounced for children with more advanced linguistic abilities. Lewis, Freeman, Kyriakidou, Maridaki-Kassotaki and Berridge (1996) also found an association between number of siblings and performance on false belief tests in one study, but overall they found a more consistent effect of older siblings and kin on the development of false belief understanding. In a series of experiments with a large number of participants beneficial effects were found for older but not younger siblings (Ruffman, Perner, Naito, & Parkin, 1998). The "sibling effect", however, was not replicated in two more recent studies involving working class families (Cole & Mitchell, 2000; Cutting & Dunn, 1999). This research suggests the possibility that it is the nature of the relationships that children experience that influences development rather then just the number of people in the household (Hughes, Deater-Deckard & Cutting, 1999). There is further evidence of correlations between social cognitive development and parenting style (Hughes et al., 1999; Ruffman, Perner & Parkin, 1999; Vinden, 2001), aspects of parent-child conversation (Sabbagh & Callanan, 1998), attachment (Fonagy, Redfern & Charman, 1997; Meins, 1997; Meins, Fernyhough, Russell & Clark-Carter, 1998; Symons & Clark, 2000), mothers’ education (Cutting & Dunn, 1999), and socio-economic circumstances (Holmes, Black & Miller, 1996).
Furthermore, a number of studies have found correlations between language and social understanding (e.g., de Villers, 2000; Cutting & Dunn, 1999; Happé, 1995; Jenkins & Astington, 1996). In longitudinal studies, forms of family talk about mental states have been found to be related to later success on false belief tests (e.g., Brown, Donelan-McCall, & Dunn, 1996; Dunn, Brown, Slomkowski, Tesla, & Youngblade, 1991; Moore, Furrow, Chiasson, & Patriquin, 1994; Ruffman, Slade & Crowe, 2002). In addition, mothers who think of their children in mentalistic terms ("mindmindedness"), and thus, presumably talk to their children about the psychological world, have children who are more advanced in understanding beliefs than other children (Meins & Fernyhough, 1999; Meins et al., 1998). Similar correlations between family interaction and the development of children’s understanding of emotions have also been reported (e.g., Dunn, Brown, & Beardsall, 1991; Hooven, Gottman, & Katz, 1995; Kuebli, Butler, & Fivush, 1995; Steele, Steele, Croft & Fonagy, 1999). In a longitudinal study, Astington and Jenkins (1999) found that earlier language abilities predicted later false belief performance, but earlier false belief competence did not predict later language abilities, supporting the conclusion that language is important in social cognitive development.
Another source of evidence that social interaction is important for the development of a mentalistic understanding comes from research with deaf children. A number of studies have now shown that deaf children with hearing parents are delayed in the development of false belief understanding, whereas deaf children with deaf parents are not delayed (Peterson & Siegal, 2000; Woolfe, Want & Siegal, 2002). This seems to be because deaf parents are native users of sign language and thus their children are exposed to normal conversation, but hearing parents are less fluent in sign language and therefore their children are not exposed to complex conversation about everyday events involving people’s actions, beliefs and emotions. Conversation about the mental world may well be essential for the development of social understanding.
This accumulating evidence that aspects of social interaction are correlated with social understanding must be explained by a complete account of social cognitive development. The contemporary body of research in this area derives from one of the dominant theories, according to which children come to understand their own and others’ minds by formulating an implicit "theory" of the mind (Astington, Harris & Olson, 1988). In this article we will, in general, use the broader phrase, social understanding (Dunn, 1988), as we present an alternative position. All theories of the development of social understanding have had to recognize the research of social interaction on social understanding, but they do so in very different ways, and most accounts still begin from an individualistic perspective. It has seemed that the only possible alternative is to contrast individualistic accounts with unspecified statements about the ‘enculturation’ of the child. In this article we argue for an alternative account of the development of mental state understanding that integrates the social and individual dimensions of development. We contend that the child does not merely adopt socially available knowledge. Rather, within social interactions an understanding of mental states develops and is constituted. Our approach is based on Chapman’s (1991, 1999) reformulation of Piagetian theory, drawing on Vygotsky and Wittgenstein, and it is also consistent in some ways with other contemporary approaches (Hobson, 1993, 2002; Montgomery, 1997).
To introduce our alternative theory we set it in the context of continuing debates and competing theories within the "theory of mind" tradition (section 2). We first discuss the more general issue concerning the relative contributions of social versus individual processes in development (section 2.1), and then turn to attempts by competing theories to explain the role of social interaction in the development of children’s social understanding (section 2.2). In the third section we introduce our alternative constructivist approach in which we argue that social interaction is essential in the development of cognitive, social and moral knowledge. We briefly discuss the development of infants’ ability to engage in triadic interaction involving the self, others, and the physical world (section 3.1). This basic yet essential level of social understanding allows children to engage actively in social interaction, to acquire language and, thus, gradually to develop a more sophisticated mentalistic understanding by learning the criteria for the use of words referring to the mental world (section 3.2). In the fourth section we address two important issues that distinguish our account from those we criticize. We explore what has been taken as the crucial evidence for the dominant theoretical approach and suggest that the explanatory framework presented here explains the data more completely (section 4.1). Not only does the constructivist perspective better account for the range of experimental evidence, it can also explain the role of relationships in social development, like the sibling effect and the influence of attachment patterns (section 4.2). We conclude (section 5) by making some suggestions for future directions in research that follows from our theoretical perspective.
2 The problem with the "Theory of mind" Tradition
The problem with the "theory of mind" literature is that the majority of approaches to the issue are individualistic. This is usually contrasted with a straw person, referred to as ‘enculturation’. In the two parts of this section we first describe a solution to the individual-social dichotomy before articulating how the problem of individualism manifests itself in contemporary theories.
2.1 Theory construction vs. social construction: Individualism, collectivism, and relationalism.
At the broadest level, debate regarding the development of children’s social understanding concerns the relative contributions of social and individual processes. This is the issue of whether theories start with the individual, or focus on the influence of the social group on development. We first illustrate how this issue has emerged in research on the development of children’s social understanding, and then set this debate in the context of the more general issue in order to derive an approach to resolve this problem.
In the context of social cognitive development, the developmental question, according to Raver and Leadbeater (1993), is "whether the true starting point is to be located in the single, isolated, free mind of the individual or in a social, communal world of shared experience or language" (p. 355). Raver and Leadbeater (1993) contrasted the "theory of mind" and social development research programs and suggested that,
Theory-of-mind research focuses on the single mind of the individual child as a rational empiricist, processing incoming perceptual data and reporting the results of those observations. In contrast, social developmentalists focus on the interaction of at least dyads, and the development of social minds of children communicating in a peopled world (Raver & Leadbeater, 1993, p. 355).
This tension has also been articulated by Astington and Olson (1995) as occurring between theory construction and social construction. They suggest that an alternative to theory construction in which "children construct a theory about human talk and action" (p. 185) is a process of enculturation in which "children internalize the folk psychology of their particular culture" (p. 184), and, therefore, an understanding of mind acquired in such a fashion is "a set of cultural norms" (p. 185). In this view, social construction is equivalent to enculturation and the ability to "participate in a kind of interpretive discourse…. In the one case the child is seen as constructing concepts, in the other as internalizing social understanding" (p. 185). Internalization, in this socialization approach, involves making external social norms internal. This enculturation approach, which has been attributed to Wittgenstein, is clearly stated by Astington and Gopnik (1991, p. 19-20): "On this view folk psychology is … what Wittgenstein would call a ‘form of life’, a set of social and cultural practices and conventions. The mechanism for development, on this view, would be socialization or enculturation—children would learn how to psychologize appropriately in the way that they learn to dress properly or eat politely," or learn that "forks go on the left" (Gopnik, 1993, p. 3). Astington and Gopnik (1991, p. 20) further specify the difference between the view they attribute to Wittgenstein and the "theory formation" view by stating that "theories always develop with reference to the outside world; put very simply, a theory former wants to get closer to the truth. Cultural practices, on the other hand, are, at least largely, self-constitutive: they make themselves the case. Theories are true or false, cultural practices are right or wrong."1
The implication of this interpretation of Wittgenstein is that social cognition is relativistic and thus forms of social understanding are culture-specific and free to vary across cultures, just as ways to dress properly or eat politely vary. We, however, argue in section 3.2 below for a different interpretation of Wittgenstein. Briefly, although we acknowledge cultural variation (Lillard, 1998; Vinden, 1996, 1999), the basic forms of understanding of self and others that we are concerned with in this article are built onto forms of shared practice, or everyday activities, that are necessarily part of human interaction, such as adults caring for infants and, thus, are likely to be common across cultures (Canfield, 1993).
Astington and Olson (1995), as well as Raver and Leadbeater (1993), suggest the need for an integration of the two perspectives—i.e., the isolated individual formulating a theory of social behavior versus the child passively adopting concepts available from the culture. The conflict between these two contrasting positions, often referred to as individualism and collectivism, is a general issue concerning the role of social interaction in the development of knowledge that arises in many areas of development (Carpendale & Müller, in press). Given that he is often mislabeled as the archetypal ‘individualist’ developmental theoretician, it might seem surprising that it was this "perennial problem" that Piaget (1977/1995, p. 184) grappled with in a series of essays published as the "Sociological Studies" (Chapman, 1986; Kitchener, 1986; Smith, 1995). The question for Piaget (1977/1995, p. 184) was "Do the operations by means of which we attain what rational consciousness calls truth depend on society and, if so, in what sense?" Piaget (1977/1995) criticized individualism for neglecting the role of social life in transforming the individual’s cognitive structures. In Piaget’s words, "human knowledge is essentially collective, and social life constitutes an essential factor in the creation and growth of knowledge, both pre-scientific and scientific" (Piaget, 1977/1995, p. 30).
On the other hand, collectivism is also problematic for a number of reasons. One difficulty is that if knowledge is simply passed on from the collective to the individual, how is that individuals can question and criticize collective beliefs? "If truth is something social, how can one distinguish legitimate common representations from collective beliefs not based on reason?" (Piaget, 1977/1995, p. 197). Piaget also criticized collectivism for not considering the influence of different types of social relationships on development (Carpendale & Müller, in press; Duveen, 1997).
As an alternative to individualism or collectivism Piaget argued for a third possibility, according to which it is relations between individuals that is primary (Kitchener, 1986, ch. 4; Smith, 1995; see also, Bunge, 2000; Elias, 1978): "The primary fact, from this point of view, is neither the individual not the set of individuals but the relationships among individuals, a relationship constantly modifying individual consciousnesses themselves" (Piaget, 1977/1995, p. 136). This position leads to a consideration of different types of relationships and their influence on development: "One will no longer be content to say that ‘society’ is the basis of logic but will ask exactly what social relationships are involved" (Piaget, 1977/1995, p. 136). The two types of relationships that Piaget (1932/1965) described, forming the extremes of a continuum, were constraint and cooperation. Constraint involves inequality and views being imposed by authority, whereas cooperation involves interaction among equals.
Our approach to the development of children’s social understanding focuses on the relations between people. From a relational, action-based perspective the developing child is embedded in social interaction, and an involvement in social activity itself is an integral part of constructing knowledge of this activity. This activity simultaneously involves operative interaction with the world as well as communicative interaction with other people (Chapman, 1991, 1999). It is the mutuality of operative and communicative interaction thatforms the basis of the constructivist position proposed here.
Although we have focused on Piaget, a relational, embodied, constructivist approach to development could also be based on other theorists (Bunge, 2000; Elias, 1978; Overton, 1998). There are a number of approaches that recognize the importance of social interaction in development (e.g., Cole, 1992; Rogoff, 1997, 1998), but this is not the place to exhaustively review and critique them.2 The perspective we take is on the psychological development of infants and children with the assumption that such development occurs within the infant or child’s activity—an activity matrix made up of biological, social/cultural and psychological dimensions (Overton, 1994). From this perspective, development begins from a point of relative lack of differentiation, and the child’s distinctions between self, other and the world of objects emerge through activity in such a matrix.3
An issue that arises in regard to constructivism is that such approaches may appear to imply relativism. That is, if knowledge is constructed within social interaction then, since social interaction may vary across cultures, it would appear that such knowledge would be specific to particular cultural groups. Moreover, there would be no way to evaluate different forms of knowledge; all forms would be equally valid. That is, there are just different forms of knowledge and there is no way to be able to say one is better than another. This "constructionist" position (Gergen, 1994), therefore, collapses the distinction between development and mere change, and amounts to relativism (for further discussion see Chandler, 1997). However, although the "constructivist" position we endorse accepts that knowledge is constructed within interaction, there is still development in the sense that one form of knowledge is better, more complete, or more adequate than other forms of knowledge (Chapman, 1988b).
As a way to resolve this difficulty with relativism, Chapman (1988b) distinguished progressivity in development from directionality toward a predetermined end point. According to Chapman’s (1988b) reading of Piaget’s equilibration theory, this theory was meant to address the question of "how one form of thinking or knowing might be judged more or less ‘advanced’ than another" (p. 97). After the necessary first step of describing forms or stages of thinking, the next step for Piaget was to explain the process of development from one form of knowledge to another. Development from this approach is progressive in the sense that partial and incomplete perspectives are coordinated resulting in more coherent and complete forms of knowledge. The process of equilibration is assumed to be universal and progressive, but this does not imply a particular predetermined end-state. Rather, development involves movement away from an initial starting point of lack of knowledge (Chapman, 1988b).
In the next sub-section we discuss attempts to account for the influence of social interaction on the development of children’s social understanding at the level of particular theories of children’s social cognitive development.
2.2 Accounting for the Influence of Interaction on Children's Social Understanding
The issue of how to address the social dimension in development is of importance within debates over which is the best theoretical account of children's understanding of mind because each of the competing accounts must explain the recent evidence suggesting that social interaction influences such development. Until recently, the discussion concerning theories of children's "theories of mind" has primarily consisted of debate among three dominant theories: theory-theory, modularity theory, and simulation theory. The pace and excitement of debate between each of these factions over the past decade has been frantic (for reviews see e.g., Hala & Carpendale, 1997; Flavell & Miller, 1998; Lewis & Carpendale, 2002). However, since recent analyses have witnessed much cross-fertilization between perspectives (see Carruthers & Smith, 1996) the aim of this section will be to highlight common stress points within these positions. This section briefly describes how the dominant theoretical perspectives are still highly controversial. There are sufficient critiques of each perspective for s not to dwell on individual theories. Instead the focus will be on a fundamental flaw shared by all, thereby justifying the need for the alternative put forward in section 3.
The phrase "theories of mind" signals the fact that the dominant theoretical position in research on mental state understanding has been that children construct theories of mental states which are similar to theories or paradigms in science. Change occurs either because children acquire a more sophisticated capability to represent mental state phenomena (Perner, 1991), or because they modify their theories in the light of evidence that is not compatible within their current framework of understanding (Gopnik, 1993; Gopnik & Wellman, 1992, 1994; Wellman, 1990). Theory-theorists claim that the child's understanding of mental states is theory-like because this knowledge is coherent and domain-specific, it shows a complex relation with the available evidence, postulates unobservable entities, and, most importantly, undergoes 'paradigm shifts' when the child abandons one set of principles for another in the face of sufficient disconfirming evidence.
Stich and Nichols (1992) characterize a theory as a "body of rules or principles or propositions ... which serve to guide the execution of the capacity to be explained" (p. 35). Similarly, Gopnik and Wellman (1992, 1994) state that a theory is a system of interrelated laws or rules that can be used to derive or infer predictions or explanations. In other words, understanding a particular action requires applying the theory’s laws and deriving a solution. The editors of the special issue of Mind and Language (1992) in which these articles appeared suggested that such an approach is the most common strategy in cognitive science for explaining the ability to negotiate a particular psychological domain. For example, this is the dominant approach in the area of language, but even here it has been severely criticized (e.g., Baker & Hacker, 1984; Tomasello, 1995b).
Critics have long argued about whether the child's understanding can be equated with theoretical postulates within scientific movements (e.g., Feldman, 1992; German & Leslie, 2000; Harris, 1994; Hobson, 1991; Scholl & Leslie, 2001; Nelson, Plesa, & Henseler, 1998). Russell (1992) argues that describing children’s understanding of mind in terms of a series of theories does not provide an account of development. Gellatly (1997) claims that drawing an analogy between cognitive development and historical change in theories conflates different levels of discourse, and neglects the social dimension to child development and theory change in science. Campbell and Bickhard (1993, p. 33) suggest that if the term ‘theory’ is used loosely it is general enough to include any form of human knowledge.
The view that children's understanding of mind is best explained in terms of one or more innate modules resulting in a "hard-wired" theory draws on arguments that the development of such social understanding is simply too important to be left to chance and this understanding is acquired by children very early and without apparent effort or explicit teaching (e.g., Baron-Cohen, 1995; German & Leslie, 2000). The position that an innately given module or set of modules is triggered at some point in development would seem to be the least consistent with developmental changes in children’s understanding of mind (Gopnik & Wellman, 1994). It lives uneasily with the accumulating evidence, such as the "sibling effect" described above, showing that the nature of social interaction children experience is closely related to the development of their social understanding. A further example is that secure attachment is associated with early false belief understanding (Fonagy, Redfern & Charman, 1997; Meins, 1997; Symons & Clark, 2000). This might be accounted for from the innate perspective by assuming that attachment depends on innate temperament, but this explanation is ruled out because the same infant can form different attachments with different adults (Meins et al., 1998). In addition, Tomasello (1999b) has argued that there has not been enough evolutionary time for the series of four mechanisms proposed by Baron-Cohen to have evolved. Further, Moore (1996) has pointed out difficulties with each of the sources of evidence (i.e., animal lesion studies, a study of neuroimaging and studies of patients with brain damage) advanced by Baron-Cohen (1995) for the existence of modules or mechanisms. As in the debate over language development, we assume that there is agreement that children must be biologically prepared to develop an understanding of mind. The debate concerns whether the solution to this problem of understanding the mind is innately given, or whether it is the capacity to develop such a solution that has evolved.
According to simulation theory, children develop an understanding of mental states through introspection, and use their imagination to reason about psychological matters (Harris, 1991, 2000). We agree that imagination must play an important role in children’s reasoning about the mind, but the introspective aspect of this approach is problematic for a number of reasons. Simulation theory faces the problem of how we can ever be confident that we can generalize our introspection to others (see Russell, 1996, for a discussion). A more important difficulty for simulation theory is that the assumption that children learn the meanings of mental state terms through introspection is vulnerable to Wittgenstein's (1968) private language argument (Chapman, 1987a; Montgomery, 1997; Russell, 1996). Wittgenstein’s argument—actually an anti private language argument or arguments (Russell, 1996)—seems to establish that children cannot learn the meanings of mental state terms just by introspecting on our own inner experience.
In reviewing Wittgenstein’s arguments we now move from the problems of the individual theories to a critique of a fundamental assumption shared by all theories discussed so far. That is the assumption in these theories that the problem the child must overcome, through inference or introspection, is to figure out what is going on in the private and hidden realms of other minds. German and Leslie (2000, p. 230) set up the "‘fundamental problem of theory of mind’ as follows: given that beliefs, desires and pretends can be neither seen heard nor felt, how does the young brain succeed in learning about them?" It is just this causal psychological view of the mind that Wittgenstein rejected. Wittgenstein (1968) argued that our language "bewitches" people into thinking of psychological matters as inner objects and his aim was to dispel misconceptions that our language leads one into. People assume that language refers to things, and in the case of beliefs, desires and intentions we assume inner objects, or representations, that are causally related to behavior. We contend that this assumption is common to most theories of children’s "theories of mind" and, thus, Wittgenstein’s philosophy can be applied as a general critique of many current theories of social cognitive development (Racine, in press). Wittgenstein’s goal, especially his private language argument, was therapeutic because his intent was to clear the ground of misconceptions and we aim to use his remedy to propose a constructivist cure for the currently ailing approaches to mental state understanding.
Although there is general agreement about the importance of Wittgenstein’s private language argument, there is considerable controversy regarding the meaning and implications, and even the location of the argument, or arguments, in Wittgenstein’s (1968) Philosophical Investigations (ter Hark, 1990).4 By a private language Wittgenstein did not mean a new language that has not yet been taught to anyone else, or a dying language only spoken by one last speaker. Rather, a private language is radically private in the sense that it is not possible to teach to others because it is based on private ostensive definition. Wittgenstein’s rejection of the idea that one could define words privately is part of his critique of what he referred to as Augustine’s view of language, according to which ostensive definition is the foundation for language because "it correlates words with things" (Hacker, 1990, p. 99). Among the problems with this view of language, Wittgenstein (1968, para. 28, emphasis in original) showed that "an ostensive definition can be variously interpreted in every case." An Augustinian might argue that the word-thing connection in language can be saved by the fact that one can ‘mentally point’ – i.e. when it comes to inner sensations one simply directs our attention. Thus, privately or mentally pointing would be unambiguous. However, this too would not work because it would require a person to "individuate a particular mental activity—sensation or impression—concentrate his attention on it, and label it" (Williams, 1999, p. 19). Hacker (1996, p. 132) argues that there is nothing like mental pointing that could be similar to public pointing. One cannot concentrate one’s attention inwardly and label a sensation. Thus, regarding internal private states, nothing "could logically count as remembering correctly or incorrectly" (Hacker, 1990, p. 108). This is just one aspect of Wittgenstein’s general rejection of the whole view of beliefs, desires and intentions as inner objects that one could introspect upon and label.
Perhaps a more fundamental reason that words cannot be defined privately is that meaning is not attached to words, utterances, or representations in a mechanistic or dyadic manner (Goldberg, 1991). Instead, meaning is based on use, on shared practices with others. Wittgenstein (1968, para. 580) argued that "an ‘inner process’ stands in need of outward criteria."
According to Wittgenstein, the way to find out the meaning of psychological expressions is not to look inside the self, but to look at the function that the respective words and concepts play in our language. This is because the meaning of such expressions is determined by the ways in which they are typically used, not by the particular subjective impressions that we happen to have in these situations (Chapman, 1987a, p. 107).
It is important to emphasize that the Wittgensteinian view of beliefs is very different from that assumed by most work in the "theory of mind" tradition. In Racine’s words (personal communication, November 2002), "when people in the field speak of beliefs as mental contents, they reify belief into a new form of mental content that is independent of activity but yet causes activity. There are no such contents. Belief exists and is created in action, not in the head. We should not take the development of an ability to re-present activities off line to be the development of an ability to experience inner states of belief" (see also, Racine, in press). It is this Wittgensteinian view of beliefs and intentions that we assume in the rest of this article.
We should be clear that Wittgenstein was arguing against the idea that introspection could be the way one learns the meaning of psychological terms. But he did not rule out introspection in some cases once someone has learned to talk about the psychological world, such as in the following example: "Does it make sense to ask: ‘Do I really love her, or am I only pretending to myself?’ and the process of introspection is the calling up of memories; of imagined possible situations, and of the feelings that one would have if …." (Wittgenstein, 1968, para. 587; see also Hacker, 1996, p. 133). We should also emphasize that Wittgenstein was not denying that one experiences sensations privately. This was not the target of his private language argument. Rather he rejected the idea that it would be possible to learn the meaning of psychological words through introspection.
We have pointed out some difficulties with the dominant theoretical accounts of children's mentalistic understanding (for further criticism see e.g., Gellatly, 1997; Hobson, 1991; Nelson et al., 1998; Russell, 1996). There are, however, different versions of these positions and we lack the space to do justice to the complexity of these theoretical approaches. We also recognize that criticisms may be countered with auxiliary hypotheses. These theoretical positions converge on some points. That is, most theorists would agree that infants must be innately prepared to learn about the social world, that their own inner experience must play some role in the development of this understanding, that such knowledge is interconnected and changes with development, and that imagination is important in the process of social reasoning. Within such general agreement, however, there is still much controversy and there is an urgent need for an alternative theoretical account that is more consistent with the accumulating evidence of the important role of social interaction in the development of children's social understanding.
To summarize our argument thus far, a common problem with the dominant perspectives in the field is that each focuses upon the cognitive architecture of mental state reasoning, without reflecting upon the social landscape in which such reasoning is constructed (Astington, 1996; Lewis et al., 1996). It is not by coincidence that recent discussion has begun to consider this social dimension as more central to development (Astington & Baird, in preparation). Astington’s (1996) Vygotskian approach more explicitly includes such a social dimension. However, as discussed in the previous section, we must carefully distinguish social constructionist approaches that are equivalent to passive enculturation from social constructivist positions that also recognize the social dimension as essential but claim that passive adoption of cultural norms cannot account for all development. So far, Astington (1996) has not been explicit on this issue, and leaving it ambiguous tends to lead to the enculturation interpretation. Although Vygotsky could be read as an enculturation theory, more complex readings are possible and more plausible (e.g., Fernyhough, 1996). In the next section we construct an account that is in keeping with such a reading.
3. A Solution: Constructing an Understanding of Mind
In the preceding sections we contrasted approaches based on the assumption that the development of an understanding of mind is an individual process of introspection, maturation, or the formation of a theory, with the contrasting position that children acquire culture-specific concepts regarding the mind that are passed on from the social group. Instead of choosing between these two contrasting positions we begin from a different starting point and endorse an alternative perspective emphasizing the relations between people (Piaget, 1977/1995). Concepts about the mind are not just passed on from the social group, nor are they completely formed by individual child-theorists. Instead, children gradually construct social understanding through the regularities they experience in interacting with others.
Our approach to the development of children's social understanding is based on Chapman's (1991, 1999) reformulation of Piagetian theory emphasizing the social dimension in development. Piaget is now known for his emphasis on subject-object interaction, although in his early work (e.g., Piaget, 1924/1928), and continuing throughout his career (Piaget, 1977/1995), he was also concerned with the social dimension to development (Chapman, 1988a; Lourenço & Machado, 1996). Vygotsky (1978, 1934/1986), on the other hand, is known for his concern with the role of social interaction in the development of higher mental functions. An integration of subject-object interaction and social interaction results in a triangle "consisting of an active subject, the object of knowledge, and a (real or implicit) interlocutor, together with their mutual relations", which Chapman (1991, p. 211) termed the "epistemic triangle". Chapman (1991, p. 212) argued that "although Piaget recognized the importance of both operative and communicative forms of interaction in various phases of his work, he never integrated those components in a single model." This epistemic, or "knowing" triangle preserves the insights of Piaget and Vygotsky, and facilitates thinking about how social interaction and subject-object experience can be reconciled in development. The epistemic triangle has much in common with Hobson’s (1994, 2002) notion of a "relatedness triangle", except that the latter focuses almost exclusively upon affective engagement, particularly in infancy.
Chapman employed the notion of the epistemic triangle to describe the role of social interaction in the development of knowledge in the area of children's reasoning on concrete operational tasks (Carpendale, 1999a; Carpendale, McBride & Chapman, 1996; Chapman & Lindenberger, 1992). We take it to be the lowest common denominator for understanding development in general and it is equally applicable to children's reasoning regarding social matters. The fundamental point is that social interaction is essential in the development of cognitive, social and moral knowledge. Children "construct a new conception of reality by coordinating their object-oriented operative interactions with those of other people, by means of their communicative interactions with the individuals in question" (Chapman, 1999, p. 34, emphasis in original). At the same time that children are constructing knowledge of the physical world they are also constructing knowledge of other people. This is because it is through communicative interaction that children discover that others sometimes have different beliefs about the world. We take it that children maintain the assumption of a stable external world that remains the same for themselves and other people. In order to maintain this assumption when they encounter others who may have different beliefs about aspects of the world, such as the location or identity of objects, children change their expectations, or their understanding of mind and how beliefs are formed. At some point children realize that access to information, for example through seeing, is essential in the formation of beliefs. That is, children construct an understanding of how they and other people acquire knowledge of the world and they may achieve comparable levels of development at similar ages because of commonalities in their experience.
From this perspective, knowledge is gradually constructed within social interaction. Children’s social knowledge is based on action; it is not theoretical in the sense of a set of laws formed on the basis of observation to explain the doings of other people. Understanding others is at first practical. It is gradually constructed through regularities in interaction with others. At first the interaction between infant and other (e.g., the caregiver or sibling) is dyadic and not yet referential, but by the latter part of the first year this interaction becomes triadic, between the infant, the caregiver, and objects.5 That is, aspects of the world become increasingly included in the interaction between infant and caregiver. The infant is embedded in this triadic interaction but only gradually differentiates the self, other people, and the world of objects. In keeping with this gradual view of development, we expect to see children exhibit early forms of knowledge about the mind first in interaction and only later will children become able to reflect on such knowledge.
In the following sub-sections we flesh out this view of development in more detail, first briefly in development during infancy and next when children begin to learn words.
3.1 The Development of the Epistemic Triangle: Social Understanding in Infancy
Triadic interaction between the child, another person and the world is essential in the development of knowledge in general, and here we focus on social knowledge. This leads to the question of how and when the capacity for triadic interaction develops from an infant’s dyadic interaction with either a parent or the world of objects. There is much debate and discussion concerning how and when infants achieve the capacity for triadic interaction—or move from "primary intersubjectivity" to "secondary intersubjectivity" (Trevarthen, 1979; Trevarthen & Hubley, 1978).6 Examples of behavior involving triadic interaction begin toward the end of the first year of life and include social referencing, gaze following and different forms of pointing. One of the early manifestations of triadic interaction between the infant, an adult, and objects during infancy occurs at about 12 months of age when infants start to point to objects (Schaffer, 1977). Bates, Camaioni and Volterra (1976) distinguished between infants’ use of pointing gestures in order to make requests ("proto-imperatives") and pointing to direct adults' attention to objects ("proto-declaratives"). At this age, infants appear to be actively directing an adult’s attention. These are two forms of behavior that at least appear to involve joint visual attention. This capacity for joint attention plays an essential role in word learning during the second year (e.g., Baldwin, 1995; Lock, 1980). The triangle between the infant, parent and the world in action can be observed even earlier, at about 9 months of age, in the phenomenon of social referencing, in which an infant apparently refers to a parent's facial expression to determine their evaluation of an ambiguous situation (e.g., Walden & Ogan, 1988).
Any claim about the age at which joint attention behaviors such as pointing, gaze following and social referencing develop will be controversial. This is because there are different forms of these behaviors that develop at different ages and because there is much debate about what these behaviors may reveal concerning infants’ social understanding; both rich and lean interpretations of this apparent evidence of early social understanding in infancy are possible (Baldwin & Moses, 1996; Moore & Corkum, 1994). Richly interpreted, these behaviors involving joint visual attention seem to be evidence of infants’ understanding that other people have intentions and attention that can be directed to various aspects of the world (Carpenter, Nagell, & Tomasello, 1998; Tomasello, 1995a). The behaviors mentioned above have been interpreted as evidence that infants at this age already have an "implicit" "theory of mind" (Bretherton, 1991; Bretherton, McNew, & Beeghly-Smith, 1981). From the perspective of this rich interpretation, infants require a mentalistic understanding of attention in order to engage in joint attention with adults and, therefore, this behavior reveals such an insight.
Theorists taking a rich view of joint attention can be seen as falling within an individualist framework. According to that framework, social development begins with the first person perspective and the individual’s own self-knowledge which then must be extended to others; i.e., "knowledge of other minds is parasitic on our knowledge of our own mind" (Jopling, 1993,p. 291). With this starting point for self-knowledge, the route to an explanation of understanding other minds is usually through the analogical argument. This argument has an extensive history. A current use of it by Meltzoff, Gopnik and Repacholi (1999, p. 35) is the following: "(a) When I perform that bodily act I have such and such a phenomenal experience, (b) I recognize that others perform the same type of bodily act as me, (c) the other is sharing my behavioral state; ergo, perhaps the other is having the same phenomenal experience." Tomasello (1999b, p. 70-76, 1999a, p. 68) also relies on the analogical argument in his theorizing about infancy (although his theorizing concerning language is consistent with ours).
The analogical argument relied on by the individualist framework is problematic for a number of reasons (eg., ter Hark, 1990). For example, it seems to attribute too much to the infant in counterfactual reasoning and the ability to take an external perspective on their own experience (Soffer, 1999). Analogical reasoning is possible once children have the concept of self and other, but the argument already presupposes this distinction and thus cannot logically be its source (Scheler, 1954, pp. 240-241). (For further criticism of the analogical argument see Müller & Carpendale, in press.)
However, more recent accounts support a leaner interpretation. For example, Baldwin and Moses (1996) explored development in triadic interaction in the context of social referencing and found that infants look at their mothers under conditions of stress but not to gather information. This suggests that triadic interaction may be important in further development of infants’ social understanding, but at 9-12 months it is only very rudimentary and is more likely to be a manifestation of the attachment system than social understanding.
The most well known lean account is that of Moore and Corkum's (1994; Corkum & Moore, 1995, 1998). They show that infants engage in joint visual attention, or end up in situations in which their attention is coordinated with adults, through processes such as conditioning without having yet developed an understanding of attention (for a similar argument see Kaye, 1982). Rather than joint attention necessarily revealing infants’ understanding of other people's attention, Moore and Corkum (1994) suggest that there are developmental changes in infants' understanding of joint visual attention. They provide evidence that around the end of the first year infants rely primarily on observing the adult’s head turn without apparently understanding the importance of eye direction. This appears to call into question the commonsense view that infants of this age understand that the other person is paying attention to the object. Later, during the first half of the second year, infants start to use eye direction information as well as the head turn to achieve joint visual attention. This suggests that experience in which infants’ attention is coordinated with adults’ attention occurs before infants understand much about attention (Moore, 1999). The infant becomes embedded in interaction with some success and only through the experience of that interaction later develops a more complete understanding: "Once the infant is reliably engaging in joint attention, the experiences offered by this new triadic interactive form of behavior will afford progress in the infant’s conceptualization of attention and set the child on the road to the discovery of all forms of social life that rest on sharing attention with others" (Corkum & Moore, 1995, pp. 81-82).
This position is also supported by research on infants' pointing (following Murphy & Messer, 1977). Desroches, Morissette and Ricard (1995) report that in a longitudinal study the majority of infants began to use noncommunicative pointing (i.e., pointing without looking at their mother) at 12 months of age. However, only beginning at 15 months did these same infants begin to use pointing that seemed communicative in intent, for example, by looking at their mothers to ensure they were successfully communicating. Desroches et al. (1995) document the development of increasingly sophisticated forms of pointing between the ages of 12 to 18 months. Here again the evidence suggests that infants engage in action such as pointing, which may result in achieving joint visual attention or the coordination of attention with an adult because the parent may attend to the object to which the infant is pointing. However, these young infants do not, as yet, appear to understand what they are doing because they do not look at their parent’s face. In further exploring the development of pointing, Moore and D’Entremont (2001) found that infants at one year were more likely to point when their parent was looking at them, whereas at two years of age infants tended to point when their parent had not seen an interesting sight. This suggests that young infants point to enhance interaction, but older infants understand pointing as directing others’ attention.
Another reason for caution concerning the acceptance of overly rich interpretations of the meaning of early joint attention behaviors comes from research with nonhuman primates. Although chimpanzees can follow gaze, and thus on the surface their behavior looks surprisingly similar to infants’, Povinelli (1999) has shown through a series of experiments that chimpanzees do not appear to understand the psychological significance of gaze direction. For example, chimpanzees trained to beg with a hand out gesture and given the opportunity to beg for food from one of two experimenters sitting side-by-side are initially just as likely to beg from someone with her eyes closed, turned away, or even with her head covered by a bucket, as they are to beg from someone with her eyes open. It takes many trials for the chimpanzee to learn to beg from the person who can actually see their begging gesture (Povinelli, 1999).
Proponents of the lean position have much in common with our position on infant social understanding. In particular, Moore (1999, p. 48) sets the origins of infant social understanding within triadic interaction (equivalent to subject, interlocutor and object): "One might say that the 12-month-old recognizes intentional relations but only to the extent that she can share such a relation with an interactive partner. For the 1-year-old, intentional relations exist in the interaction and are not a property of, or descriptive of, individuals". However, Barresi and Moore’s (1996) more detailed account starts from an individualistic perspective in which infants have different information about self and other and the integration of these two sources of information leads to the development of an understanding of self-other equivalence: "An individual organism’s information about intentional relations is available from two different sources. An organism has direct first person information about its own intentional relations and direct third person information about the intentional relations of other agents" (Barresi & Moore, 1996, p. 108). This implies that a self-other distinction is already in place, whereas, from our perspective, it is this distinction that requires an explanation.
An alternative approach that avoids these problems within an individualist framework is the relational framework (Jopling, 1993), according to which the development of social understanding begins from relations between people rather than individual self-knowledge. The starting point in development is a position of relative non-differentiation between self, other, and the world (Baldwin, 1906; Hobson, 1993; Merleau-Ponty, 1964; Piaget, 1936/1963, 1937/1971; Vygotsky, 1998; Werner & Kaplan, 1963). From an observer’s point of view one sees the infant interacting with others and the world, but at first the infant has not constructed these distinctions. The differentiations between inner and outer, subject and object, and self and other occur gradually within interaction (Müller & Runions, in press). The process by which the infant shifts from participating in triadic interactions to an understanding of the distinction between their own and others’ intentions remains a holy grail within this area of theorization. However, only the constructivist perspective starts from a point where this problem is not already ‘solved’ – as stated above accounts within the "theory of mind" tradition assume self-other differentiation from the outset.
A central assumption from a constructivist perspective is that knowledge originates in activity. Infants interact dyadically with the world of objects as well as with people and through this activity they develop sensorimotor action schemes. Such schemes embody knowledge because they are modified as a function of the differences between what is anticipated and what the infant actually experiences (Chapman, 1999). With further development, two or more action schemes may be combined, resulting in more objectivity or separateness from the infant’s own action. The infant develops expectations about what can be done with objects as well as expectations about their interactions and routines with people. And infants begin to regard people as independent "centers of causality" (Piaget, 1937/1971); people do not always do what the infant expects or wants (Baldwin, 1906). Different patterns of activity are possible with people compared to objects. There has been a great deal of research attention devoted to the study of face to face dyadic interaction between infant and parent. In such interaction infant and parent respond to each other and this could be described as a form of communication but it does not yet refer to anything outside of this dyadic interaction (Chapman, 1991). Objects may begin to be included in this interaction in routines such as giving and taking. From this perspective, infants first experience triadic interaction that is supported by the parent or through processes such as infants learning to associate head turns with interesting sights (Corkum & Moore, 1998). At this point infants still do not have a clear understanding of others’ attention and infants construct such an understanding through interaction (Müller & Runions, in press). Development involves a process of gradually constructing action schemes and combining them. Infants slowly differentiate and coordinate their attentional directedness toward objects with others’ directedness toward objects (Müller & Carpendale, in press).
Hobson’s (2002) "relatedness triangle" is a similar approach that emphasizes the importance of the infant’s emotional engagement. To this important insight, we add that the infant’s understanding of causality and spatial relations are also critical in the infant’s ability to engage in joint attention. This point is reflected in the various forms of gaze following that are observable from about 3 months to 18 months of age. More complex forms of this social behavior, such as following an adult’s gaze to an object behind the infant, require the infant to have an understanding of herself as an object within a world of other objects (Müller & Carpendale, in press).
When infants have developed the capacity to coordinate attention with others they can refer to aspects of the world by directing others’ attention, as well as understand other people’s referential intent. Reference involves more than just forming an association between a sign, usually verbal, and an action, as is the case in the few triadic chimpanzee gestures (Tomasello & Camaioni, 1997). Instead, signs can be used flexibly to convey different meaning in different contexts, and this requires an ability to infer others’ referential intent.7 Recent "theory of mind" research on infancy has worked with inventive experimental procedures, but neglects an older tradition (e.g., Lock, 1978; Schaffer, 1984) that attempts to chart the development of such signs within the richness of the infant’s daily interaction rituals and experiences. Our position is in keeping with this tradition and the more recent view of Reddy(e.g., 1991), that the rituals that preverbal children engage in during "teasing" and "mucking about" show the beginnings of the process by which children come to share attention and to construct a practical or lived form of social understanding or knowledge of other people.
The ability to achieve joint attention and to understand other people’s referential intent is the foundational insight on which language is based (e.g., Baldwin, 1995; Sinha, 1999). Tomasello (1999b; Tomasello, Kruger & Ratner, 1993) argues that it is this uniquely human ability that allows people to make use of culture to a far greater extent than other animals. This ability is based on an initial level of social understanding that gives infants the capacity to engage in triadic interaction, and such interaction then provides a context for further social cognitive development. Infants now start to use actions and words to refer to other things. Their communication at this point becomes referential: that is, infants indicate what they mean the adult to attend to and are also able to determine adults’ referential intent. At this point, "words begin to mediate children’s sensorimotor acquaintance with reality. Children no longer know reality solely in terms of what they could do with it, but also in terms of what they could potentially say (or hear) about it" (Chapman, 1999, p. 34 emphasis in original). All of the components of the epistemic triangle are now in place. In the next section we discuss how this new level of knowing provides a context for further development in children’s understanding of the social world.
3.2 Outer Criteria for Inner Processes: Talking about the Psychological World
As infants develop the capacity to coordinate attention with others they begin to learn language. Children’s word use is built onto shared practices that they have already developed, such as following or directing another’s attention, making requests and conveying their intentions or goals. Learning to talk is grafted onto prior ways of acting and interacting with others. This is a view of language as activity, not just exchanging information. In this section we are concerned with the development of children’s ability to talk about the psychological world. This raises the issue of the relations between language and thinking. A common assumption is that concepts are acquired first and then the words for such concepts. In contrast, from a Wittgensteinian perspective, language and thinking are much more tightly intertwined. Wittgenstein (1981, para. 324) asks, "Does a child learn only to talk, or also to think? Does it learn the sense of multiplication before or after it learns multiplication?" We assume that the child learns about the concepts and the words at the same time. That is, learning the meaning of words to do with multiplication involves learning the operations that are criterial for those words.
The research on children’s language and their social understanding initially used language as a window on their understanding and tended to focus on words that seem most obviously to be mental state terms such as ‘think’ and ‘know’. Other words, however, that are not obviously mental state terms at all, such as ‘hide’, involve some understanding of how people know things and, indeed, how people can be prevented from knowing things (Turnbull & Carpendale, 1999a: for similar discussion see Russell, 1992). In fact, it would be difficult to draw a clear line between words that do or do not turn on some understanding of the psychological world. Hacker (1991) points out that saying that a robot or an automatic door opener ‘sees’ something is parasitic on the primary use of such a verb. With reference to human agents, simple words such as ‘look’ and ‘see’ seem to refer to perception, but their correct use is tied up with an understanding of the perceiver’s psychological attributes when they attend to something. Learning the use of such words is rooted in children’s everyday experience with coordinating attention with others. Children can be asked the false belief test question by referring only to where Maxi would look (the "look question") without use of the word ‘think’. In fact, in some research young 3-year-olds seem to pass this form of the test question before the ‘think’ version of the question (Chandler & Hala, 1994; Hala & Chandler, 1996). Similarly, 3-year-olds pass a false belief question phrased with ‘say’ before a similar question phrased with "think" (Nelson, Skwerer, Goldman, Henseler, Presler, & Walkenfeld, in press). The general point is that a great deal of talk that is based on an understanding of knowledge acquisition and the mental world may not, in fact, involve what we would usually think of as mental state terms (Turnbull & Carpendale, 1999a).
In describing how children learn to talk about the psychological world it is necessary to draw on Wittgenstein's notion of criteria. According to Chapman’s operational semantic theory (1999; Carpendale, McBride, & Chapman, 1996), children learn about the meaning of relational words such as ‘longer’ through learning the operations that are criterial for the use of such words. Children may come to understand that the meaning of ‘longer’, in the case of length, involves the operation of checking to see which of two sticks protrudes further. Only through elaboration do children learn the additional criterion of first lining up the ends of the sticks. Terms like ‘longer’ are also involved in other language games referring to time. In the present case of learning how to talk about the psychological world the criteria for the use of such words are embedded within social interaction. As discussed above, Wittgenstein's (1968) private language argument stresses that it is not possible to learn the meanings of words referring to inner processes through introspection (Chapman, 1987a; Montgomery, 1997; Russell, 1996). It follows from this that children learn about inner processes (their own and other people's) through public criteria, not only through their subjective experience; that is, they learn the appropriate contexts for the use of various words referring to the psychological world. Children learn the pattern of interaction for which it is appropriate to use a particular term, either mental, emotional, or dealing with pain, etc. Wittgenstein's view of how children learn the meaning of sensation terms is that
words are connected with the primitive, the natural, expressions of the sensation and used in their place. A child has hurt himself and he cries; and then adults talk to him and teach him exclamations and, later, sentences. They teach the child new pain-behavior (Wittgenstein, 1968, para, 244).
Wittgenstein's argument regarding sensation-words also applies to psychological words (Chapman, 1987a). Children learn to talk about the psychological world in a variety of ways including discussions about themselves and other people. That children appear to develop the ability to understand another’s belief at the same time as their own (Gopnik, 1993) would suggest that they come to learn the criteria for each at about the same time. Criteria for talk about the mental world are behavioral evidence for inner experiences such as sensations, emotions, thinking, remembering, imagining, etc.: "Criteria are those publicly observable circumstances which might be used in teaching the correct use of the expression to a child or someone else learning our language" (Chapman, 1987a, p. 105).
From a Wittgensteinian perspective, the use of psychological terms becomes part of earlier, or more "primitive", prelinguistic behavior (Hacker, 1997). Words such as ‘see’ and ‘look’ may be grafted onto to earlier joint attention behaviors. These words can then be used as new ways to direct and follow attention. Similarly, emotion words may be grafted onto the child’s prior reactions or behaviors. The circumstances surrounding the use of psychological words become the criteria for their use. Talk about the mental world is built onto prior activities as a refinement (Canfield, 1993, 1999; Malcolm, 1991; Turnbull & Carpendale, 2001).
It might seem that if we endorse the Wittgensteinian idea that children learn about the mental world through learning how to express their feelings, plans, and goals, etc. in language, and learning the criteria for the third person use of various psychological terms, then we must endorse an enculturation position in which mentalistic concepts are imported from the social world to the individual. As mentioned above, this is one interpretation of Wittgenstein and it implies a cultural relativism, by which children would just learn the mental concepts used in their particular culture. However, we do not endorse this interpretation for two reasons. First, at the basic level of social understanding (e.g., seeing, looking, intentions, desires and beliefs) children’s understanding is built onto shared practices that we expect would be common across cultures because these are common aspects of human experience (for a parallel argument within the theory view see, Wellman, 1998). This does not rule out that there may be cross-cultural variability, such as in complex emotions (Lillard, 1998).
Secondly, and central to our argument, children’s development is constructive. This involves combining the epistemic triangle approach to development with Wittgenstein’s concept of criteria. That is, children do not acquire an understanding of talk about the psychological world in an all or none fashion. Concepts are not passed on, ready-made, through language. Criteria are multiple and children may initially acquire a subset, which enables them to use words apparently correctly in supportive contexts. It is only through communicative interaction with others about beliefs (sometimes differing, sometimes concordant) about the world that children gradually construct an understanding of belief. In other words, although young children often use and seem to understand words based on an understanding of the process of knowledge acquisition (e.g., ‘see’, ‘remember’ and ‘guess’), their understanding is as yet incomplete and is based on a partial set of the criteria for the correct use of such words (Montgomery, 1997). For example, Lillard (1993b) showed that when young children first begin to use the word ‘pretend’ they focus on the more obvious criterion of the action being performed rather than on the individual’s knowledge or intentions. This view is consistent with the generally protracted nature of language development (e.g., Nelson, 1997).
We now turn to a consideration of other accounts of the relation between language and social understanding. The finding that various aspects of children’s linguistic ability are associated with false belief performance has already been mentioned (e.g., de Villiers, 2000; de Villiers & de Villiers, 2000; Cutting & Dunn, 1999; Happé, 1995; Jenkins & Astington, 1996). One approach is that language can be used as a "window" to reveal the child’s social understanding. Bartsch and Wellman’s (1995, see also Sabbagh & Callanan, 1998) work exemplifies the usefulness of such research, in that it shows that children produce terms referring to desires before they refer to beliefs. However, one should be careful of the assumption that mental terms refer to mental states. Even a simple word like ’want‘, which would be coded as a desire term, can be used in many ways, such as to make requests or to make offers. Budwig (2002) found that mothers and their children used the word ’want‘ in different ways. Regarding syntax, the de Villiers (e.g., de Villiers & de Villiers, 2000) show that that the child’s grasp of the syntax of complementation (the fact that mental state verbs necessarily take complementary clauses when they refer to metal states, like other non-mental state verbs: e.g., He thinks that he had eggs for breakfast; He said that he had eggs for breakfast) is correlated with false belief test performance. Such analyses suggest important connections between mental state understanding and the ability to parse statements about mental states. Longitudinal evidence indicates that language competence, as measured in standardized tests, appears to predate the development of false belief understanding (Astington & Jenkins, 1999).
Each piece of research shows how the child’s understanding of mental states is likely to be related to their general proficiency in understanding and producing words and sentences. However, such correlations raise the issue of causal direction, third factors and the nature of any influence. Indeed they are somewhat antithetical to our approach because language and social understanding are so intertwined that it is somewhat artificial to separate human activity into parts and call one part language and the other social understanding. This position follows, in part, from viewing language as activity (Turnbull, in press; Turnbull & Carpendale, 1999a). However, if we do talk in terms of language and social understanding, then, from the perspective of the epistemic triangle, language is important for two reasons. Astington (2001, p. 686) has described these two roles, and the need for a resolution, as "a means for representing false belief in contradistinction to the evidence given in reality and it is also the means by which children become aware of beliefs, both content and attitude…. What is needed is a new conceptualization that reconciles and combines insights from both views." Although we may differ from Astington in what we mean by representation - from our perspective language mediates children’s knowledge of reality—children learn the criteria for words to talk about human activity and then can reflect on the psychological world. Language, or communicative interaction, is the means through which children learn about other people’s experience and so develop a more complete set of criteria.
Parallel arguments have been made within more traditional accounts. Harris (1996), in particular, has the insight that conversation is important. Rather then mere exposure to mental state terms, Harris argues conversation is important because it is a constant reminder that other people have different perspectives. We agree that this is one role that conversation plays but this is not sufficient. Communicative interaction helps children realize when they only have a partial set of the criteria for using words such as ‘look’, ’ ‘see’ , ‘think’ and ’know’ correctly, and their understanding of the patterns of interaction in which these words are used is incomplete. Conversation has a pivotal role within the epistemic triangle, but as part of an integrated system not as an independent "factor". . In order to maintain the assumption that one lives in a common, stable, external world that is the same for the self and others—one of the basic presuppositions required in any attempt at communication--children must develop an increasingly sophisticated understanding of the nature of the mind. Maintaining the assumption of an independent external world, in the face of evidence of other people's differing experience of the world will require that children modify their understanding of the relations between their own and others' beliefs about the world to include the notion that one's beliefs about the world depend on the information to which one is exposed.
Once children can talk about the social, emotional, and psychological world they can begin to reflect on or think about people’s activity in psychological terms (Chapman, 1991; Piaget, 1945/1962; Vygotsky, 1934/1986). This raises the issue of how we view the process of reasoning. In the theories we reviewed above the process of reasoning involves simulation or the application of rules that are either innate or formulated by the child on the basis of observation. In contrast, we suggest that children understand talk about the psychological world in terms of the patterns of activity that are criterial for the use of such mental state terms, i.e., the pattern of interaction for which we use these words. Reasoning, then, involves the coordination of these activities. Thus, as well as the child’s understanding of questions in terms of human interaction, they also require the ability to imagine and coordinate activities that may not be immediately present. From this perspective, reasoning is not based on the application of rules, but particular instances of rules would be manifest in the process of reasoning (Carpendale, McBride, & Chapman, 1996).
4 Using the Epistemic Triangle to Explain the ‘Core Findings’ in ‘Theory of Mind’ and the Influence of Relationships
In this section we address the ‘so what?’ questions that necessarily follow in response to the proposed position in section 3. In 4.1 we take the paradigm that appears to show most clearly that a clear conceptual shift takes place, the false belief test, and compare a traditional theory-theory account with the one offered here. In 4.2 we explore whether the theory presented here can go further than previous ones in accounting for the correlations between a range of social relationships, like the sibling effect, and mental state understanding. We suggest that it can do so more convincingly than any other account.
4.1 The Gradual Acquisition of Criteria Within Conversation.
We are not the first to criticize the dominant accounts of ‘theory of mind’ development by suggesting that such skills are acquired gradually. Such gradualism has been explained within a framework of other arguments, like the protracted development of language (Nelson, 1997), and from within a philosophical critique of the theory-theory approach (Woodfield, 1996; see also Russell, 1996). It is thus important to outline how gradualism follows from our theoretical perspective. Theory-theorists might depict a series of theories, "rudimentary prototheories" (Ruffman, 2000, p. 263) like "preliefs" (Perner, Baker & Hutton, 1994) or processing demands in different procedures. This account can only go so far. However, an argument based upon gradualism cannot be an excuse for fuzziness of thinking. It needs both a theoretical foundation and to provide an explanation of the data.
The notion of gradualism underlies the type of developmental approach taken by Piaget, and also Vygotsky. For example, as Chandler (2001) has pointed out, in Piaget and Inhelder's (1948/1967) research on the development of children's understanding of visual perspective taking they described a sequence of different forms of understanding. This developmental approach has often been lost in the subsequent role taking literature and the current "theory of mind" research . The strong tendency has been to collapse this protracted development into a single transition point (Chandler, 2001; Chandler & Carpendale, 1998).
The view that an understanding of mind unfolds gradually may appear to be a non-starter when the collected data on the false belief test are examined. The recent meta-analysis conducted by Wellman, Cross & Watson (2001) attempted to identify the factors that reliably facilitate performance in the false belief test. In an analysis of 178 experiments with 591 conditions, Wellman and colleagues found that factors like the type of task used, or whether the target question focused on the protagonist’s thoughts or his or her actions, were stable across studies. There seemed to be a clear conceptual shift in children’s performance at around age 4, and this is consistent with the theory-theory account of developmental change. However, consistent patterns of performance on one procedure do not necessarily inform us about the nature of change – from our theoretical perspective they may simply show that four-year-old children across the many studies engage in sufficient interaction with others about the veracity of beliefs to pass this test. As Scholl and Leslie (2001) note the meta-analysis might only show the sorts of improvement with age seen on most developmental tasks.
Indeed proponents of the theory-theory perspective have long warned us about the problem of "neurotic task fixation" (Gopnik, Slaughter & Melzoff, 1994, p. 157) and there are dangers in relying upon only one task, particularly those versions of that task which do not test the variations in performance. In Wellman et al.’s meta-analysis some types of studies were not considered. Moses (2001) points out that among the types of studies not included are studies involving explanations of false beliefs, deceptive behavior, and eye movements rather than verbal responses. Even within the homogeneity of tests that were likely to lead to homogenous results, there was still considerable variability among samples, and the authors found that "several task manipulations do increase young children’s performance: framing the task in terms of explicit deception or trickery, involving the child in actively making the key transformations, and high-lighting the salience of the protagonist’s mental state or reducing the salience of the contrasting real-world state or affairs, all help young children to perform better" (Wellman et al., 2001, p. 672). Two issues emerge from Wellman et al.’s interpretation of the collected false belief data, which are typical of the ‘theory-theory’ approach to the literature: the extent of task variation and its implications.
In contrast to the theory-theory, the constructivist view actually predicts that there will be variations in the child’s acquisition of knowledge within a domain. For example, when exploring the development of children's social understanding, Selman, Schorin, Stone, & Phelps (1983) found different levels of perspective taking when the child was more involved in an activity than when reflecting upon it. The same analysis can be applied to the types of data reported by Wellman et al. and the findings, listed above, which they chose to omit from their analysis. For example, Wellman et al. found that if the protagonist’s motive is made explicit or if the child actively participates in the procedure (e.g., Chandler & Hala, 1994; Hala & Chandler, 1996) then preschoolers pass the test in greater numbers. Similar results are found when child participants act out the answer by demonstrating with a doll (Freeman, Lewis, & Doherty, 1991). We propose that children do better on false belief tasks when they are actively involved because their understanding of the events is supported by the social interaction. Not only is their attention directed by the experimenter (within the epistemic triangle, the subject-interlocutor interaction), but also their involvement in important aspects of the events commits them to acting upon the object of knowledge (the subject-object of knowledge part of the triangle). Thus children’s thinking is supported by their practical activity within a dynamic triadic process. With further development in their ability to hold different aspects of events in mind and coordinate them, children will be able to deal with false belief situations at a more abstract level. This type of research provides a snapshot of mental state understanding in the making. It reveals a microcosm of relationships, which current studies are showing to be vital and which explain patterns of findings to be revisited in section 4.2.
A constructivist perspective would predict that the performance of children who are within the process of developing these insights would be influenced by many aspects of social interaction. Recent research shows that there are other influences of the context of the assessment procedure, which highlight such gradualism. For example, there is inconsistency between laboratory based assessment procedures—standard false belief tests—and naturalistic observation of young children. Children typically pass standard false belief tests only by the age of approximately four, whereas in the home parents report observing their 2 1/2 or young 3-year-olds apparently demonstrating an understanding of false belief (Astington, 2000; Newton, Reddy & Bull, 2000). In a diary study Newton et al., (2000) found that parents reported many incidents in which their young children were involved in deception even though these children were known to have failed false belief tasks. In agreement with our position, Newton et al. suggest that rather than early deception reflecting insight, young children engage in deceptive acts with only partial understanding and such experience is the context for learning about deception.
Wellman et al. claimed that although factors such as the child’s active involvement improved performance on false belief tasks this evidence "failed to fit an early competence model" (p. 674) because they "do not raise the youngest children’s performance to systematically above-chance performance" (p. 674 ) and there were no interactions between these task manipulations and age. Wellman et al. derive "the essential claim that such task factors mask early competence" (p. 672). The difficulty with this interpretation is that Wellman et al. appear to have interpreted arguments for gradualism within their own way of thinking about competence, resulting in the straw person they term the "early competence model". Wellman et al’s inference is based upon a prediction that the child either ‘has’ or does not have false belief understanding, which may be hidden by performance factors. This is based on the assumption that competence is a hidden underlying factor that causes performance (Chandler, 1991). However, there are sufficient data showing how children are able to understand false beliefs at different times to support the claim presented here that we need to explore the process of change in greater detail. The general problem is that if an individual has developed a particular competence why do they demonstrate it in some situations but not others? Is this variability measurement error or evidence of gradual development?8
Variability in performance across different tests is usually explained in terms of performance factors masking the child's competence. But, as Chapman (1987a) pointed out, it is not always possible to separate performance factors clearly from the underlying competence. The assumption that there is one competence which is presumed to arrive all in one piece, like a theory or a set of rules, is what Chapman (1987a) referred to as the "measurement model". As an alternative approach he proposed the "membership model", according to which there are varying degrees of understanding that are assessed with different procedures. Here we could say that degrees of membership, or competence with mental state concepts, are related to children’s increasing grasp of the multiple criteria for words referring to the mental world and their ability to understand events outside of supported social contexts in their immediate practical activity.
The perspective we apply draws on Vygotskian (1978, 1934/1986) ideas according to which children’s initial, fragile social understanding, or "understanding-in-action" (Dunn, 1996), is at first evident when supported by social interaction. Instead of naturalistic observation of early competence (e.g., Newton et al., 2000) being the application of a theory, this type of experience is necessary to gradually construct social understanding. This claim that children begin talking about the mental world with only a partial understanding is consistent with work on language learning in general (Nelson, 1997). This is not, however, a simple adoption of culturally available concepts concerning the mental world; rather, children must come to understand these concepts through becoming competent in using such words in their practical activity. Children’s correct use of psychological words is tied up with their understanding of social situations and human activity. This understanding is facilitated by relationships that help the child understand other points of view. With experience in such social interaction concerning the mental world, children’s understanding becomes progressively consolidated. That is, children are able to think about situations abstracted from their practical context of interaction and are able to anticipate and reason about situations such as the false belief task. The ability to talk about the mental world gives children a resource with which to reflect on and understand others as well as themselves.
The fixation on false belief understanding has prevented us from examining the longer view of development. As Chandler (1988) pointed out long ago, this initial understanding of the nature of beliefs and mind is not yet equivalent to a mature adult understanding of the mind. An understanding of the possibility of false beliefs will not allow children to make sense of situations in which people with access to exactly the same information still arrive at different, but equally legitimate, interpretations of the same information. A child with only an understanding of false beliefs would assume that only one interpretation could possibly be right, and others must, therefore, be wrong. To accommodate to experience of interpretive diversity given the same information, while still maintaining the assumption of a stable, independently existing external world, children slowly modify their understanding of the process of knowledge acquisition to achieve an understanding of the interpretive nature of knowledge. So, when children encounter situations in which different beliefs are apparently based on the very same information they must revise their earlier understanding that beliefs depend on the information one is exposed to (i.e., false belief understanding) and develop an interpretive understanding of the mind (Carpendale & Chandler, 1996; Chandler & Lalonde, 1996; Lalonde & Chandler, in press). This early insight into the interpretive nature of knowledge is a step toward a more mature understanding of mind, but there is still further development in understanding how knowledge is acquired during adolescence and adulthood (Carpendale & Chandler, 1996; Chandler, Hallett, & Sokol, 2001). Proponents of the dominant theories have been notably quiet about what happens in development after the child’s fifth birthday. However, research which explores whether five-year-olds can use simple false belief knowledge to make inferences about their own and others’ perspectives finds that they singularly fail to do so (Varouxaki, Freeman, Peters & Lewis, 1999).
In addition to the content of talk about the mental world, the approach we are proposing also implies that the nature of the parent-child relationship should be important in the development of children’s social understanding through facilitating children’s understanding of the events and human activities being talked about. We turn to this implication of our approach in the next section.
4.2 The Role of Relationships in the Construction of Social Understanding
In this section we return to the issues illustrated by the ‘sibling effect’ discussed at the start of the paper. Any account of how the child comes to understand the psychological world must be able to explain a range of other sources of evidence showing that mental state understanding is significantly correlated with factors in the child’s social environment, like attachments, parenting styles and parent-child communication. In this section we argue that the importance of communication in either facilitating or hindering the understanding of other perspectives directs our attention to the nature of communication within different relationships. The ‘sibling effect’ is important because it is a marker of the types of process hypothesized here, but it is only one such marker. Dunn (1996) has repeatedly reminded us of the important role of relationships in social cognitive development. What is needed is a theoretical explanation of how relationships operate as a vehicle for the child’s construction of the mind. The idea of the epistemic triangle is useful in explaining this development because it is through becoming aware of other people's beliefs and coordinating these often differing perspectives with their own beliefs that children develop an understanding of mind. Through such interaction and confronting others' often differing beliefs about the world children gradually construct a more complete understanding that increasingly coordinates their own experience with that of other people. It is partially the resistance or refractoriness of social interaction that stimulates the development of knowledge. Thus, we would expect that differences in the amount and nature of the social interaction experienced would be related to individual differences in infants' early social understanding and to young children’s further developing mentalistic understanding. We will discuss two ways of describing parent-child relationships: attachment and cooperation.
4.2.1 Affective Engagement and Mental State Understanding
A body of research in which social relationships are shown to correlate with social understanding has been influenced by attachment and psychodynamic theories (Fonagy & Target, 1997; Hobson, 1993, 2002; Meins, 1997, 1999). Fonagy, Redfern and Charman (1997) found that securely attached children, as measured with the Separation Anxiety Test, a projective measure of attachment security, were more competent on "theory of mind" tasks than insecurely attached children. This positive relation between attachment security and false belief understanding has also been found with a Q-sort measure of attachment (Symons & Clark, 2000). In a longitudinal study, Meins (1997) found that children who were classified as securely attached at 11 to 13 months were more likely than insecurely attached children to pass a false belief task at 4 years, and more complex mentalising tasks at 5 years (Meins et al., 1998). A relation between attachment and early social understanding in infancy was also reported by Bretherton, Bates, Benigni, Camaioni, and Volterra (1979). These researchers found that children who were securely attached at 12 months had used more proto-declarative pointing at 11 months than other infants.
There have been a number of potentially compatible explanations proposed for this positive relation between secure attachment and social understanding. Meins (1999; Meins, Fernyhough, Fradley & Tuckey, 2001) suggests that security of attachment and social cognitive development are positively associated because parents of children who develop secure attachments respond sensitively and appropriately. To do this they need to think of their children as persons with thoughts and feelings and treat them in this way. That is, parents whose children develop secure attachments are "mindminded," they tend to "treat their infants as individuals with minds, rather than merely entities with needs that must be met" (p. 332). This way of interacting develops secure attachment, and it also exposes young children to talk about the psychological world. Meins’ (1999) explanation for the connection between attachment security and social cognitive development is that the same parental characteristic that results in secure attachment also results in the "exposure of the infant to mental state language" (p. 337). She found that maternal child-centered ("mindminded") language (but not mindminded language about other people) to 5-month-olds predicted children’s false belief performance at age four (Meins, Fernyhough, Wainwright, Das Gupta, Fradley, & Tuckey, 2002). This evidence requires an explanation of how exposure to mental state language influences social development, like the one we have provided here.
At a level at which theories of affect and cognition are integrated, it is not too difficult to show how analyses of infant-caregiver affect are wholly compatible with data on parent-infant shared attention. Indeed Hobson (2002, p. 147) argues that security of attachment, which is based on the mother’s way of relating to the infant, may influence the infant’s ability to engage with others, and it is this engagement that is essential in interaction in the "relatedness triangle" and in the development of thinking, especially thinking about the social world. From the vantage point of observations of interactions, Baldwin and Moses (1996) point out that internal working models consist of knowledge of other people and relationships. That is, infants develop expectations about other people based on primary relationships. Such views echo one another and fit into the framework of our general developmental theory, which extends beyond infancy. Perhaps we should not be surprised that such a way of viewing internal working models converges with Piagetian ideas about development being rooted in interaction because one likely source of Bowlby’s ideas was Piaget’s notion of schemes. Piaget (1945/1962, pp. 188-189, 206-207) wrote about the "affective schemes" or "personal schemes" that infants develop as "modes of feeling and reacting" to people in ways that sound like internal working models, i.e., the set of expectations infants build up about other people and how they will act. This is early sensorimotor, lived, practical knowledge about people based on expectations acquired through experience of how parents respond.
Research on the development of affective exchange in infancy has greatly illuminated key issues in infant development. The correlations between such exchanges and tests of later mental state understanding suggest a need for a broader explanatory framework. One possibility is that a secure attachment is an indicator of a relatively cooperative parent-child relationship, and, as we will argue in the next section, this is the type of relationship that facilitates the development of knowledge, according to constructivist theory.
4.2.2 Cooperation versus Constraint
Exploration of the nature of parent-child relationships has led to interest in the influence of parenting style on social cognitive development (Astington, 1996; Vinden, 2001). Ruffman, Perner and Parkin’s (1999) data show that parents who reported that they would deal with disciplinary situations by asking their child to "reflect on the victim’s feelings" (p. 406) had children who were more advanced in false belief understanding. In addition, cooperative interaction is related to and may facilitate the development of children’s social understanding. Dunn and colleagues (1991) found that cooperative sibling interaction was related to successful false belief explanations. Also, Brown, Donelan-McCall, and Dunn (1996) reported that cooperative interactions with friends and siblings were related to children’s frequent use of mental state terms. Such evidence needs a theoretical explanation.
If, as Chapman (1991, 1999) argued, knowledge of the world develops through coordinating other people's perspectives with one’s own within the epistemic triangle, then aspects of relationships and communication that facilitate our understanding of other people's perspectives should have a positive effect on development. An aspect of relationships that was important in Piaget's (1932/1965, 1977/1995) work on moral judgment is the degree of constraint versus cooperation present in relationships. This analysis is relevant for the present discussion because development, according to Piaget (1932/1965), is facilitated by relationships of cooperation and mutual respect and hindered by relationships of constraint and unilateral respect because cooperation facilitates understanding. Cooperative relationships have the potential of approaching Habermas' (1983/1990) conception of the ideal conditions of unrestrained communication, allowing all participants to understand each other's positions fully and arrive at solutions to conflicts that everyone can agree with. In our terms if understanding is gradually constructed by the child within triadic interaction, then it follows that the extent and nature of cooperation between the child and others are important.
It follows from a constructivist perspective, that cooperative relationships, allowing free communication between parents and children, should facilitate the development of an understanding of mind. Parents differ in how much they feel obliged to justify their positions and listen to their children's perspective and, thus, in how much cooperation they allow or encourage in their relationships with their children. Relationships among parents and children that are more cooperative than constraining should facilitate children's understanding of other people's points of view. This interpretation is consistent with research showing that parental styles (Baumrind, 1991) differing in the extent to which parents reason with their children differentially affect child development. Cooperative relationships are the ideal context for the development of knowledge, but this alone is not sufficient. Talk about a substantive connection to the world is also required (Döbert, in press)—in this case talk about people’s activity in terms of their goals, beliefs and desires. Various experiences such as cooperative peer interaction in the context of role enactment and shared pretense may contribute to the development of mental state understanding largely due to the stimulation of interaction and conversation.9
Of course, these correlations do not allow us to draw causal conclusions. The parent-child relationship is a system that is influenced by characteristics of both the child and the parents. It is important to acknowledge that relationships involve the child as well as the parent, and, although the influence of parents tends to be emphasized, characteristics of the child will also influence the nature of the parent-child relationship (Bell, 1968; Symons & Clark, 2000). Parental characteristics that may influence parent-child relationships include parents’ beliefs about parenting (e.g., Baumrind, 1991; Ruffman et al., 1999; Vinden, 2001) and emotions (Hooven, Gottman & Katz, 1995), as well as other factors that could affect parents’ ability to engage with their child such as level of stress, depression and psychiatric disorders, patience, social support, socio-economic circumstances, and education.
On the other hand, the extent to which parents can be cooperative and can support the child’s reasoning will partly depend on, and interact with, the child’s abilities and characteristics. The child’s developmental level may influence the nature of the relationship because children who are more advanced in language and social cognitive development may be easier to interact with in a cooperative manner.10 Individual differences on the child's part may vary from extremes such as autism,11 which severely restricts the child's ability to engage in triadic interaction as evidenced by the lack of protodeclaritives at 18 months of age (Hobson, 1993), to differences in ability to focus and maintain attention (e.g., hyperactivity, impulsiveness, distractability). Blindness also makes joint attention more difficult to achieve and thus seems to hinder the development of social understanding (Minter, Hobson, & Bishop, 1998; Peterson, Peterson, & Webb, 2000). We agree with Hobson (1993, 2002) that such factors on the child’s part influence their ability to engage in interpersonal relatedness, as well as the nature of that interaction, and it is such interaction that is essential in social cognitive development. We have restricted our analysis to the study of human development in this article but the same principles could be extended to non-human primates.12
5. Conclusion and Future Directions
What benefits can be gained by adopting a constructivist approach to the development of children’s social understanding? For a start we have summarized examples of a voluminous literature which relates an understanding of mental states to a range of social experiences. Traditional accounts of "theory of mind" can only go so far in adding to the list of social influences on the child simulator/theoretician before coming to realize that a social dimension has to be incorporated into such theories. The theory presented here sets out to do two things. First, it attempts to establish from the start that social processes are a necessary part of any account of how children come to construct an understanding of their social world: from infant dyadic interaction through to complex social skills beyond simple false belief understanding. Secondly, it has revisited and reworked the constructivist account of development and used this to reassert its central focus on action and the potential for examining knowledge acquisition as a social process. For us a constructivist approach to understanding the child’s grasp of the social world is the only possible solution to the impasse between individual and social perspectives on social understanding and the problem this reveals in the assumption which the ‘theory of mind’ metaphor makes about the development of knowledge and meaning, i.e., that the child has to learn about the workings of the mind through inference or introspection. For us, following Wittgenstein, "as a human activity ‘meaning’ is best thought of as a verb, not a noun" (Shotter, 1978, p 46).
By providing a framework that integrates social and cognitive processes, the analysis presented here calls for new directions in research based upon the relations between particular forms of interaction and the development of social knowledge. Here we suggest some required shifts in the research agenda, as a means of illustrating how the implications of our position are very different from those offered by theorists within the "theory of mind" tradition. We refer to five areas of urgently needed research.
This research agenda begins in infancy with the study the development of "shared meaning" (Chapman, 1999, p.34). This is one of the most difficult and important problems in developmental psychology. We have pointed out flaws in other approaches and provided a firm foundation and the tools with which to build an account of the development of infants’ ability to grasp others’ referential intent (see also Müller & Carpendale, in press). But we acknowledge that the sketch we have provided is, as yet, partial. Researchers must begin by addressing the process by which the infant engages in interaction and joint attention. In doing so, we suggest that longitudinal naturalistic observation of infant-parent interaction and infants’ joint attention would complement the current experimental research (e.g., Moore & D’Entremont, 2001). Ironically such naturalistic research was more common 25 years ago (e.g., Bates et al., 1976; Lock, 1978; Schaffer, 1977). What is needed is a detailed analysis of infants’ interaction and the emergence of social referencing, gaze following and pointing, as well as a study of the context of the ongoing sequence of interaction in which the act is embedded. Some contemporary research of this sort does already exist (Reddy, 1991) and we suggest that such work may help solve the riddle of how self-other differentiation and coordination emerges during infancy. According to the view adopted here, and in contrast to the position held in the "theory of mind" accounts of infancy, "the child’s understanding is not just a matter of recognizing a correspondence between the mother’s words and reality, but of grasping her referential intent in that situation – knowing what one is meant to attend to in response to her words and gestures" (Chapman, 1999, p 34, emphasis original).
Secondly, we need to move from the intriguing findings relating individual differences in the social experiences of infants and their social understanding to research which explores the processes involved. For example, the findings of Meins et al. (2002) are intriguing, but how is it that mothers’ tendency to talk about their infants in psychological terms is positively associated with their child’s understanding of false beliefs almost four years later? What is it about the nature of these parents’ interactions with their infants that correlates with the development of social understanding? Such research turns our attention to the nature of parent-child interaction and developing talk about the psychological world, and our theory provides a framework for this research. For example, such studies should examine how the criteria for the use of mental state terms are displayed (Turnbull & Carpendale, 1999a, 2001). Differences between families in the ways in which criteria are made evident could also be studied. In studying talk about the psychological world it is important to remember that researchers should not just be concerned with mental state terms but more broadly with talk about human activity (see section 4.1).
Thirdly, and related to the points above, the gradual view of development implied in an activity based theory such as ours means that researchers need to study forms of understanding even before children are able to explicitly explain false beliefs. One way to approach this issue is by manipulating the tasks to discover what factors facilitate young children’s reasoning within the context of the assessment procedure. This type of research, and a considerable amount of such research has been conducted (e.g., Lewis & Mitchell, 1994), should be used to think about nature of development, not merely to search for "early competence". This would involve an extension of research using novel approaches to assessing children’s grasp of an actor’s beliefs. For example, Call and Tomasello’s (1999) non verbal false belief procedure could well be used to tap early forms of understanding, just as it has been used to reveal a competence in deaf children (Figueras-Costa & Harris, 2001; see also Carpenter, Call, & Tomasello, 2002). There is still scope to make a closer inspection of how children respond during traditional tests of false belief. For example, an interesting source of evidence of an early form of understanding comes from within the theory-theory and is referred to as "implicit knowledge" of false belief (Clements & Perner, 1994; Clements, Rustin & McCallum, 2000; Garnham & Ruffman, 2001). This work seems to demonstrate that young children, even slightly before their third birthday, respond to a prompt in a false belief test by looking in the direction of the correct container, even though these same children then go on to fail a more standard verbal question. Clements and Perner (1994) describe this knowledge as implicit because it cannot be verbalized. It has more recently been referred to as "nonverbal theory of mind" (Ruffman, 2000), but we point out that although the child’s response is not verbalized it is in response to a verbal prompt. Not only might research on this effect illuminate potentially interesting processes by which children come to construct an understanding of the false belief task (e.g., the nature of the criteria they use), it would also free theory-theory from the contortions it has to make in order to incorporate such evidence into its developmental account.
Fourthly, we need to explore more closely just what it means to a child "have" an understanding of belief, in order to move beyond over-simple accounts of theoretical transitions at age four. As well as expecting gradual development in different forms of understanding, we also predict and would investigate unevenness in development between different areas of children’s experience. Some children may have more experience and therefore more understanding of particular aspects of social understanding. For example, in a recent study Peskin and Ardino (in press) found that children’s performance on a false belief task in which an object is moved to a new location without the knowledge of the protagonist ("unexpected transfer task") was positively associated with success in a hide-and-seek game, whereas children’s performance on a false belief task in which a container has unexpected contents ("unexpected contents task") was positively associated with success in keeping a secret from one of the experimenters about a birthday cake wrapped in a bag. Such data are clearly compatible with a constructivist account.
Fifthly and finally, and as stated above, researchers need to move away from age/stage fixation to explore what happens to children after age four. For a start children cannot necessarily use the knowledge they demonstrate in the standard false belief test. For example, a recent study by Maridaki-Kassotaki, Lewis & Freeman (in press) found that in Greek, in which two terms are used synonymously to mean ‘to look for’, a version of the false belief test with one verb appeared to facilitate three-year-olds’ performance, whereas its synonym hampered performance in five-year-olds. Such data reveal that children’s understanding of questions about beliefs are mediated by their understanding of verbs referring to mental states. Further work along these lines would reveal more about the process by which children come to acquire the subtleties of a mature understanding of mind – particularly knowledge and skills that make up the complexity of a culture’s psychological make-up (Lillard, 1998). The approach put forward here stresses the need to explore the matrix of social relationships involving transactions between the child and others and in which the child constructs social understanding.
References
Astington, J. W. (2001). The future of theory-of-mind research: Understanding motivational states, the role of language, and real-world consequences. Child Development, 72, 685-687.
Astington, J. W. (2000). Language and metalanguage in children’s understanding of mind. In J. W. Astington (Ed.), Minds in the making: Essays in honor of David R. Olson (pp. 267-284). Oxford: Blackwell.
Astington, J. W. (1996). What is theoretical about the child’s understanding of mind?: A Vygotskian view of its development. In P. Carruthers & P. K. Smith, (Eds.) Theories of theories of mind, (pp. 184-199). Cambridge: Cambridge University Press.
Astington, J. W. & Baird, J. A. (Eds.). (In preparation). Why language matters for theory of mind. New York: Oxford University Press.
Astington, J. W., & Gopnik, A. (1991). Theoretical explanations of children’s understanding of mind. British Journal of Developmental Psychology, 9, 7-29.
Astington, J. W., Harris, P. L., & Olson, D. R. (Eds.) (1988). Developing theories of mind. New York: Cambridge University Press.
Astington, J. W., & Jenkins, J. M. (1995). Theory of mind development and social understanding. Cognition and Emotion, 9, 151-165.
Astington, J. W., & Jenkins, J. M. (1999). A longitudinal study of the relations between language and theory-of-mind development. Developmental Psychology, 35, 1311-1320.
Astington, J. W., & Olson, D. R. (1995). The cognitive revolution in children’s understanding of mind. Human Development, 38, 179-189.
Baker, G. P., & Hacker, P. M. S. (1984). Language, sense and nonsense. Oxford: Blackwell.
Baldwin, D. A. (1995). Understanding the link between joint attention and language. In C. Moore & P. J. Dunham (Eds.), Joint attention: Its origins and role in development (pp. 131-158). Hillsdale, NJ: Erlbaum.
Baldwin, D. A., & Moses, L. J. (1996). The ontogeny of social information gathering. Child Development, 67, 1915-1933.
Baldwin, J. M. (1906). Thoughts and things, Vol. 1: Functional logic. New York: the MacMillan Company.
Baron-Cohen, S. (1995). Mindblindness: An essay on autism and theory of mind. Cambridge, MA: MIT Press.
Barresi, J., & Moore, C. (1996). Intentional relations and social understanding. Behavioral and Brain Sciences, 19, 107-154.
Bartsch, K., & Wellman, H. M. (1995). Children talk about the mind. Oxford: Oxford University Press.
Bates, E., Camaioni, L., & Volterra, V. (1976). Sensorimotor performatives. In Bates, E. Language and context (pp. 49-71). New York: Academic Press.
Baumrind, D. (1991). Parenting styles and adolescent development. In R. M. Lerner, A. C. Petersen, & J. Brooks-Gunn (Eds.), Encyclopedia of adolescence (Vol. 2). New York: Garland.
Bell, R. Q. (1968). A reinterpretation of the direction of effects in studies of socialization. Psychological Review, 75, 81-95.
Boom, J. (in press). Individualism and collectivism: A dynamic systems interpretation of Piaget’s interactionism. In J. I. M. Carpendale & U. Müller (Eds.), Social interaction and the development of knowledge. Mahwah, NJ: Erlbaum.
Bretherton, I. (1991). Intentional communication and the development of an understanding of mind. In D. Frye & C. Moore (Eds.), Children’s theories of mind: Mental states and social understanding (pp. 49-75). Hillsdale, NJ: Erlbaum.
Bretherton, I., Bates, E., Benigni, L., Camaioni, L., & Volterra, V. (1979). Relations between cognition, communication, and quality of attachment. In E. Bates (Ed.), The emergence of symbols: Cognition and communication in infancy (pp. 223-269). New York: Academic Press.
Bretherton, I., McNew, S., & Beeghly-Smith, M. (1981). Early person knowledge as expressed in gestural and verbal communication: When do infants acquire a "theory of mind"? In M. E. Lamb & L. R. Sherrod (Eds.), Infant social cognition: Empirical and theoretical considerations (pp. 333-373). Hillsdale, NJ: Erlbaum.
Brown, J. R., Donelan-McCall, N., & Dunn, J. (1996). Why talk about mental states? The significance of children’s conversations with friends, siblings, and mothers. Child Development, 67, 836-849.
Bruner, J. (1990). Acts of meaning. Cambridge, MA: Harvard University Press.
Budwig, N. (2002). A developmental-functionalist approach to mental state talk. In E. Amsel & J. P. Byrnes (Eds.), Language, literacy, and cognitive development: The development and consequences of symbolic communication (pp. 59- 86)Mahwah, NJ: Erlbaum.
Bunge, M. (2000). Ten modes of individualism – none of which works – and their alternatives. Philosophy of the Social Sciences, 30, 384-406.
Call, J., & Tomasello, M. (1999). A nonverbal false belief task: The performance of children and great apes. Child Development, 70, 381-395.
Campbell, R. L., & Bickhard, M. H. (1993). Knowing levels and the child’s understanding of mind. Behavioral and Brain Sciences, 16, 33-34.
Canfield, J. V. (1993). The living language: Wittgenstein and the empirical study of communication. Language Sciences, 15, 165-193.
Canfield, J. V. (1999). Folk Psychology versus philosophical anthropology. Idealistic Studies, 3, 153-171.
Carpendale, J. I. M. (1999a). Constructivism, communication, and cooperation: Implications of Michael Chapman's "Epistemic Triangle." In I. Sigel (Ed.), The development of representational thought: Theoretical perspectives (pp. 129-16). Mahwah, NJ: Erlbaum.
Carpendale, J. I. M. (1999b). Symbols and side effects: Commentary on ‘cognitive development as an executive process. Developmental Science , 2, 279-280.
Carpendale, J. I. M. (2000). Kohlberg and Piaget on stages and moral reasoning. Developmental Review, 20, 181-205.
Carpendale, J. I. & Chandler, M. J. (1996). On the distinction between false belief understanding and subscribing to an interpretive theory of mind. Child Development, 67, 1686-1706.
Carpendale, J. I, M., McBride, M., & Chapman, M. (1996). Language and operations in children's class reasoning: The operational semantic theory of reasoning. Developmental Review, 16, 391-415.
Carpendale, J. I. M., & Müller, U (in press). Social interaction and the development of rationality and morality. In J. I. M. Carpendale & U. Müller (Eds.), Social interaction and the development of knowledge. Mahwah, NJ: Erlbaum
Carpenter, M., Nagell, K., & Tomasello, M. (1998). Social cognition, joint attention, and communicative competence from 9 to 15 months of age. Monographs of the Society for Research in Child Development, 63 (Serial No. 255).
Carpenter, M., Call, J., & Tomasello, M. (2002) A new false belief test for 36-month-olds. British journal of Developmental Psychology, 20, 393-420.
Carruthers, P., & Smith, P. K. (Eds.) (1996). Theories of theories of mind. Cambridge: Cambridge University Press.
Chandler, M. J. (1988). Doubt and developing theories of mind. In J. W. Astington, P. L. Harris, & D. R. Olson (Eds.), Developing theories of mind (pp. 387-413). New York: Cambridge University Press.
Chandler, M. J. (1997). Stumping for progress in a post-modern world. In E. Amsel & K. A. Renninger (Eds.), Change and development: Issues of theory, method, and application (pp. 1-26). Mahwah, NJ: Erlbaum.
Chandler, M. J. (1991). Alternative readings of the competence-performance relation. In M. J. Chandler & M. Chapman (Eds.), Criteria for competence (pp. 5-18). Hillsdale, NJ: Erlbaum.
Chandler, M. J. (2001). Perspective taking in the aftermath of theory-theory and the collapse of the social role-taking literature. In A. Tryphon & J. Voneche (Eds.), Working with Piaget: In memoriam—Barbel Inhelder (pp. 39-63). Hove, UK: Psychology Press.
Chandler, M. J., & Carpendale, J. I. M. (1998). Inching toward a mature theory of mind. In M. Ferrari & R. J. Sternberg (Eds.), Self-awareness: Its nature and development (pp. 148-190). New York: The Guilford Press.
Chandler, M. J., & Hala, S. (1994). The role of personal involvement in the assessment of early false belief skills. In C. Lewis & P. Mitchell (Eds.), Children’s early understanding of mind: Origins and development (pp. 403-425). Hove, UK: Erlbaum.
Chandler, M. J., Hallett, D., & Sokol, B. (2001). Competing claims about competing knowledge claims. In B. K. Hofer & P. R. Pintrich (Eds.), Personal epistemologies (pp.145-168). Mahwah, NJ: Lawrence Erlbaum Associates.
Chandler, M. J., & Lalonde, C. (1996). Shifting to an interpretive theory of mind: 5- to 7-year-olds’ changing conceptions of mental life. In A. Sameroff & M. Haith (Eds.), Reason and responsibility: The passage through childhood (pp. 111-139). Chicago: University of Chicago Press.
Chapman, M. (1986). The structure of exchange: Piaget’s sociological theory. Human Development, 29, 181-194.
Chapman, M. (1987a). Inner processes and outward criteria: Wittgenstein's importance for psychology. In M. Chapman & R. A. Dixon (Eds.), Meaning and the growth of understanding: Wittgenstein's significance for Developmental Psychology (pp. 103-127). Berlin: Springer-Verlag.
Chapman, M. (1987b). Piaget, attentional capacity, and the functional implications of formal structure. Advances in Child Development and Behavior, 20, 289-334.
Chapman, M. (1988a). Constructive evolution: Origins and development of Piaget's thought. New York: Cambridge University Press.
Chapman, M. (1988b). Contextuality and directionality of cognitive development. Human Development, 31, 92-106.
Chapman, M. (1991). The epistemic triangle: Operative and communicative components of cognitive development. In M. Chandler & M. Chapman (Eds.), Criteria for competence: Controversies in the conceptualization and assessment of children's abilities (pp. 209-228). Hillsdale, NJ: Erlbaum.
Chapman, M. (1992). Equilibration and the dialectics of organization. In H. Beilin & P. B. Pufall (Eds.). Piaget’s theory: Prospects and possibilities (pp. 39-59). Hillsdale, NJ: Erlbaum.
Chapman, M. (1999). Constructivism and the problem of reality. Journal of Applied Development Psychology, 20, 31-43.
Chapman, M., & Lindenberger, U. (1992). Transitivity judgments, memory for premises, and models of children’s reasoning. Developmental Review, 12, 124-163.
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.
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.
Cole, M. (1992). Context, modularity, and the cultural constitution of development. In L. T. Winegar & J. Valsiner (Eds.), Children’s development within social context: Vol. 2, Research and methodology (pp. 5-31). Hillsdale, NJ: Erlbaum.
Corkum, V., & Moore, V. (1995). Development of joint visual attention in infants. In C. Moore & P. Dunham (Eds.), Joint attention: Its origins and role in development (pp. 61-83). Hillsdale, NJ: Erlbaum.
Corkum, V., & Moore, V. (1998). The origins of joint visual attention in infants. Developmental Psychology, 34, 28-38.
Cutting, A. L., & Dunn, J. (1999). Theory of mind, emotion understanding, language, and family background: Individual differences and interrelations. Child Development, 70, 853-865.
Desrochers, S., Morissette, P., & Ricard, M. (1995). Two perspectives on pointing in infancy. In C. Moore & P. J. Dunham (Eds.), Joint attention: Its origins and role in development (pp. 85-101). Hillsdale, NJ: Erlbaum.
de Villiers, J. (2000). Language and theory of mind: What are the developmental relationships? In S. Baron-Cohen, H. Tager-Flusberg, & D. J. Cohen (Eds.), Understanding of minds: Perspectives from developmental cognitive neuroscience (pp. 83-123). 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. 191-228). Hove, UK: Psychology Press.
Döbert, R. (in press). The development and overcoming of "universal pragmatics" in Piaget’s thinking. In J. I. M. Carpendale & U. Müller (Eds.), Social interaction and the development of knowledge. Mahwah, NJ: Erlbaum.
Dunn, J. (1988). The beginning of social understanding. Cambridge, MA: Harvard University Press.
Dunn, J. (1996). Children’s relationships: Bridging the divide between cognitive and social development. Child Psychology and Psychiatry, 37, 507-518.
Dunn, J., Brown, J., & Beardsall, L. (1991). Family talk about feeling states and children’s later understanding of others’ emotions