Theoretical Framework: God Representations and Attachment Theory

Attachment Relationships and Religion

Attachment can be defined as a close emotional bond between two individuals, initially between an infant and the significant primary caregiver or attachment figure, usually the parents (Bowlby, 1969/1982, 1973, 1980). Secure attachment relationships are characterized by love, intimacy, and a strong predisposition on the part of the child to seek or maintain proximity to the attachment figures. Attachment figures serve as safe havens in times of distress and as secure bases from which the child can explore the world and develop new mental and physical skills (Ainsworth, 1985; Bowlby, 1969/1982, 1973, 1980; Granqvist, 2010). Within the psychology of religion, GodFootnote 1 (as perceived by the individual) is regarded as a symbolic attachment figure with the same dual function (Granqvist & Kirkpatrick, 2016; Proctor et al., 2009).

During early development, the infant learns to internalize the dyadic experiences with the attachment figures and develops corresponding mental representations, which John Bowlby, the founding father of attachment theory, termed internal working models (IWMs). IWMs involve both a model of the self (“I am okay”—“I am not okay”) and a model of the other (“She/he is okay”—“she/he is not okay”). These IWMs guide perceptions, expectations and behaviours in future relationships (Bowlby, 1969/1982, 1973, 1980; Hall & Gorman, 2003; Hall & Fujikawa, 2013). To a large extent, they function automatically on an implicit level of awareness, although they also have a more explicit layer that can be communicated verbally (Hall & Fujikawa, 2013).

Attachment styles can be defined on the basis of different configurations of positively and negatively valenced IWMs of self and other and the accompanying dimensions of anxiety and avoidance. Four attachment styles are discerned: secure attachment (low anxiety, low avoidance, positively valenced models of self and other); insecure, preoccupied attachment style (negative model of self, “I am bad/worthless…”, positive model of other, high anxiety combined with low avoidance, which is often reflected in clinging or claiming behaviour); dismissive or avoidant attachment (positive model of self, negative model of the other, who is experienced as unreliable, not responsive, and unsafe, low anxiety, high avoidance); and fearful attachment (high levels of anxiety and avoidance, negative models of self and others, often as a result of traumatic experiences, abuse and neglect) (Bartholomew & Horowitz, 1991; Granqvist & Kirkpatrick, 2013; Priel & Besser, 2001). While 60–70% of individuals in normal samples have a secure attachment style, 30–40% are typified by one of the three insecure attachment styles (Granqvist, 2010). For a more comprehensive introduction to attachment theory, see Cassidy & Shaver, 2016.

The attachment system is active across the life span (Bowlby, 1969/1982) and IWMs are fairly stable over time and across situations. As IWMs generalize across relationships, both religion and spirituality in general, and the relationship with God (or the divine) as perceived by the child in particular, capitalize on the operation of the attachment system. In this context, representation of God can be defined as specific types of IWMs—i.e. as representational models of God as (divine) other and models of self in relation to God. These God representations, which can be classified according to various styles of attachment to God, reflect the God that children experience and believe in personally (in contrast to the “official” God of religious traditions) (Granqvist & Kirkpatrick, 2013; Proctor et al., 2009; Schaap-Jonker, 2018). As models of self and other have both affective-relational and cognitive components, and they function on both implicit and explicit levels of awareness, God representations comprise both emotional and cognitive understandings of God, as well as implicit and explicit forms of memory (see Schaap-Jonker et al., 2016, 2017 for a more detailed discussion of the multidimensional and multifaceted nature of God representations). Most regular methods of examining God representations, such as self-report questionnaires and interviews, have difficulties tapping into implicit aspects of God representations (Davis et al., 2016). Hence, the notion that drawings of God could make a valuable contribution in this regard.

Within the psychology of religion, two opposing hypotheses have been formulated about the functions of God representations in the context of children’s psychological and spiritual well-being. According to the correspondence hypothesis, differences in God representations correspond to and reflect differences in attachment styles, attachment history, and IWMs (Kirkpatrick, 2005; cf. Granqvist & Kirkpatrick, 2016). Children with secure IWMs of self and other perceive God as supportive and experience emotions of love, closeness, trust, and security in relation to God, while those with insecure attachment styles view God as punitive, rigid, distant, passive, unreliable, or fear-provoking. In addition to this IWM correspondence, there is evidence for socialized correspondence, as secure attachment facilitates religious socialization: Those who have warm and sensitive caregivers are more likely to adopt their parents’ religion and internalize similar religious beliefs, values, and behaviours than those who are insecurely attached (Granqvist & Hagekull, 1999; Granqvist, 2002a, 2002b). The acquisition of epistemic trust, which is the capacity to trust others as a source of knowledge about the world, may play a part here: In the context of religious socialization, a secure child wants to consider their parents’ communication of religious knowledge as trustworthy and relevant to the self (cf. Fonagy et al., 2015; Fonagy et al., 2016, pp. 780, 793).

The compensation hypothesis states that for insecurely attached children an attachment relationship with God functions as a substitute for inadequate and disappointing human attachment figures (Kirkpatrick, 2005). Compensation may take the form of explicit religious compensation, which means that insecurely attached children tend to show higher levels of religiousness than those who are securely attached (e.g. Granqvist, 2002b). In addition, God as a substitute attachment figure may fulfil an affect-regulating function and provide emotional compensation, including longitudinally (Granqvist, 2002a). The corresponding and compensatory functions of God representations may occur on different levels of awareness; it is possible that compensation is observed on an explicit level, while the God representation corresponds with insecure IMWs on an implicit relational level (cf. Hall & Fujikawa, 2013).

Before we turn to children’s drawings of God, we should discuss another attachment-related construct that is relevant to the study of children’s drawings of and narratives about God, namely the mentalizing function. The infant develops the capacity to mentalize within the context of the attachment system when the caregiver treats her/him as a mental agent, unconsciously ascribing a mental state to the infant (“You have to cry? You feel so frightened?”). Provided with mind-minded interactions through mirroring and affect-regulating behaviour, the infant learns to distinguish its own mental world from both the physical world and the mental world of others and learns to understand that mental world (Allen et al., 2008, pp. 80–102; Steele & Steele, 2008). As a result, the infant is able to mentalize—i.e., to imaginatively perceive or interpret one’s own or others’ behaviour—as conjoined with intentional mental states, such as personal reasons, beliefs, feelings, needs and desires. Several developmental modes of mentalizing are distinguished, namely the teleological mode, psychic equivalence mode, pretend mode and the mentalizing or reflective mode (Allen, 2006; Muthert & Schaap-Jonker, 2015; cf. Schaap-Jonker & Corveleyn, 2014). In the mentalizing mode, mental states can be experienced as representations. A mentalizing child is involved in an imaginary activity, constructing meaning on the basis of feelings, thoughts, wishes, and expectations that are not directly observable. In this form of grounded imagination, the mental, subjective world is both separated from objective reality and simultaneously anchored in it (Allen, 2006, pp. 3–30). Thus, inner and outer realities are seen as linked, though distinctive (Fonagy, 2008, pp. 27, 33). The child (or adult) includes the representation of mind states in their understanding of self and others, being aware that their own and another’s mental states reflect a certain viewpoint on reality and that those representations of reality do not fully equate with reality. This distance results in a sense of representingness (Bogdan, 2005, p. 190), the awareness that knowledge is representational. One can use mental states without losing the “as if” condition (see below). Therefore, in a mentalizing mode, an individual is able to reflect on their emotions, understanding them while feeling them (Allen et al., 2008, p. 59; Bateman & Fonagy, 2006, p. 4; Fonagy et al., 2002, pp. 7, 96; Jurist, 2005, pp. 426–444).

Mentalizing is an essential psychological mechanism, which is a prerequisite for relating and representing, including in the religious domain. In addition, mentalizing is involved in the capacity for imagination, transcendence, symbolizing, empathy, taking perspective, and meaning-making (see also Schaap-Jonker & Corveleyn, 2014). Hence, mentalizing seems to be one of the mechanisms through which the attachment system affects the God representations that children draw and talk about. However, insecure attachment may hinder the development and functioning of the mentalizing capacity.

Children’s Drawings of Gods and Attachment

Different attachment styles imply different types of relationship between self and other, and thus different ways of being receptive to and connecting with the external social and cultural world, including religion. We therefore assume that children with a secure attachment style will be more open and responsive to (religious) culture regarded as “not-me”, as they have acquired epistemic trust (cf. Fonagy et al., 2015; Fonagy et al., 2016, pp. 780, 793). Furthermore, a securely attached child is able to mentalize and thus integrate external ideas, thoughts, stories, images, rituals, and symbols regarding God into their own thinking, speaking and experiencing of God, while at the same time being connected to personal experiences, needs, and longings (cf. Boyatzis, 2013). It is by integrating these internal and external worlds that one can use symbols to refer to otherness in a way that is personally meaningful (Bateman & Fonagy, 2012; Fonagy, 2008; Schaap-Jonker & Corveleyn, 2014; Muthert, 2020; cf. Granqvist & Kirkpatrick, 2013, p. 140). With regard to children’s drawings of Gods, we expect that those who are securely attached will use more religious symbols and that these symbols will have a referring and self-transcending character (It is as if …; cf. Bogdan’s (2005) sense of representingness and Peirce’s (1977) category of thirdness in semiotics; see also Ogden, 1986, p. 224ff; 2005). In contrast, insecure attachment and the accompanying limited capacity to mentalize (Allen, 2013) will be associated with symbolic equation, in which symbols are used as signs (It is what it is; cf. Peirce’s (1977) category of secondness in semiotics and the psychic equivalence mode; see also Ogden, 1986 p. 224ff; 2005), without a referring quality, or without using signs and symbols. In addition to the psychological literature, art therapy also distinguishes between cognitive meaning-making strategies and emotion- and/or relation-focused strategies in artistic expressions (Einarsdottir et al., 2009; Stanczak, 2007; Ring, 2006). Drawings are therefore considered to be expressions of meaning and understanding, in which the communicative power of both the drawing and the drawing process is stressed. To investigate drawings properly, narratives about the drawings are seen as crucial. In the narratives of insecurely attached children, we would expect fewer references to the symbols present and less explanation of their use of religious symbols than in the narratives of children with a secure attachment style. Regarding artistic development, age is not related in a linear way to the development of symbolization; rather, a U-shaped development of graphic symbolization was found in different studies that included young children (Davis, 1997, Harvard Project Zero; Haanstra et al., 2013).

Although affective aspects of religious and spiritual development in childhood have largely been neglected in research (Granqvist & Dickie, 2006), and research on attachment to God has been dominated by adult studies (Cassibba et al., 2013; Granqvist & Kirkpatrick, 2016), there are several studies of children’s God representations in relation to attachment (Cassibba et al., 2013; De Roos et al., 2001a, 2001b, 2004; Granqvist et al., 2007). However, none of these studies used drawings of Gods as a method to investigate aspects of children’s representations of God (the divine). At the same time, existing studies that focused on children’s drawing of God (see Chap. 2, this volume) did not include attachment factors in the analyses. Our study therefore aims to contribute both to the study of attachment and religion and of children’s religiousness and spirituality, and to the study of children’s drawings of God.

Empirical Research: Process, Findings, and New Questions

If secure and insecure attachment styles influence mentalizing activity, one would expect to recognize differences in meaningful and expressive communication forms such as the Dessins de dieux protocol-led drawings and narratives. Therefore, we aimed to study drawings by Dutch children from the perspective of attachment theory. To assess children’s attachment style, the brief Experiences in Close Relationships Scale-Revised Child version (ERC-RC) was added to the standard procedure and questionnaire (see Chap. 2, this volume). This questionnaire is a reliable and valid instrument to measure anxious and avoidant attachment to parents (Brenning et al., 2011). In our study, the 12-item version was administered twice to measure the attachment to both father and mother. In this section, we report our data collection process and the decisions we made during this process.

Five graduate students of the University of Groningen collected drawings and/or wrote their Master’s theses on this topic. In addition, 79 Bachelor’s students interviewed younger children in the context of a small research assignment. We explicitly included drawings from god-talk contexts as well as non-god-talk contexts. Our sample included children across a range of ages (in the scope of the overall design, 5–13 years old) in both religious (mainly Christian) and general contexts, in towns and cities and other parts of the northern Netherlands.

We predicted that the different ages of the children involved would lead to drawings from distinctive artistic and/or technical stages (Malchiodi, 1998, 2003; Van de Vijfeijken, 2001). Based on the overall prevalence of secure and insecurely attached children (60–70% and 40–30% respectively; Altenhofen et al., 2013; Moullin et al., 2014; Wolff et al., 2017), we assumed that both groups would automatically be present in our sample. An initial in-depth study of ten drawings by 7/8-year-old children (Van der Vegt, 2014) concluded that there were some limitations in directly linking the social-emotional attachment qualities of fear and trust to children’s drawings of God. Clear differences were only observed in cases of quite strong indicators of anxiety and avoidance, which were based on the ERC-scale mean items scores (1 or 2: (very) low anxiety/avoidance; 4 or 5: (very) high anxiety/avoidance).

However, the drawing process, including the child’s behaviour during the process and the use of materials in addition to the narrative, seemed to be much more decisive than the drawing itself. This finding underlines the importance of the mixed method approach to the general design and of contextual information. We concluded that in cases of more diffuse ERC scores, we would need more drawings in order to be able to comment reliably on secure and insecure attachment. We would also prefer process descriptions.

To gain greater insight into the drawings of the children with insecure attachment, as reflected in their ERC scores, our next question (Van Neijenhof, 2016) was whether the drawings showed particular aspects of insecure attachment (and if so, which ones), and to what extent aspects of compensation and/or correspondence could be observed (Granqvist & Kirkpatrick, 2013; Hall & Fujikawa, 2013). Only 12 children (based on the 140 drawings that were collected at that time!) fitted the criteria (one aged almost 5, five 6/7-year-olds and six children aged 12/13) of being insecurely attached to both father and mother (other children reported no anxiety or avoidance at all, or only in relation to either the father or mother). An age-related difference was revealed. Primary school children showed clear aspects of insecure attachment in their drawings, such as the absence of expressive flowering and little in-fill, almost no padding of the paper (cf. Meykens & Cluckers, 2006), while high school pupils did not. Perhaps the drawings by the older children needed to match reality as seen in the outside world, whereas for the younger children, being engaged in their drawing activity was far more important than the result, which means that intuitive and/or unconscious elements would occur more naturally and frequently. This relates both to the implicit relational level and the developmental artistic/technical drawing stages (Malchiodi, 2003; Meykens & Cluckers, 2006; Werner, 1957; Lusebrink, 1990). Based on an analysis of (1) the drawing, (2) the narrative and (3) a personalized description of the drawing procedure, Van Neijenhof tentatively concluded that compensation was seen in god-talk contexts with only one exception: coming from a non-god-talk context, the youngest participant showed her trust in a good and loving God. One could hypothesize that, in line with the children’s religious affiliation, compensation is only observed on the explicit level in some cases, while correspondence predominates on the implicit relational level (cf. Hall & Fujikawa, 2013). However, the differentiation in insecure attachment styles proved difficult and in the case of religious affiliation the polar distinction between secure or insecure is not particularly solid based on a single drawing.

On the basis of the above findings, we concluded that we needed more indicators of insecure attachment. We therefore added a family drawing to the general protocol as different studies have shown that children’s drawings of family are indicators of early attachment (cf. Procaccia et al., 2014; Attili et al., 2011; Fury et al., 1997; Madigan et al., 2003; Kaplan & Main, 1986). For example, open arms, a drawn floor and a full drawing area are indicators of secure attachment, while incomplete, floating and/or small figures are expressions of insecure attachment. We expected that a comparison of two drawings of objects that children relate to would also improve our analyses in instances of more diffuse attachment styles in terms of ERC scores.

With this new format, drawings were collected from younger children in “multi-problem” families in the free Baptist community in Groningen, from a primary school for asylum seekers (without administering the ERC-RC because of a lack of valid translations in Arabic and Tigrinya; both by De Kraker-Zijlstra (2017) and from a Protestant community (Pruis, 2018). Ambivalent aspects confirmed the value of a multi-perspective approach. Adding the family drawing also seemed fruitful in that findings attached to the first drawing did not necessarily resemble the analysis of the second. From the perspective of cultural diversity (Aronsson & Andersson, 1996; Gernhardt et al., 2013; Malchiodi, 1998; Tharp, 1991), new questions arose, as the literature regards intracultural differences as far more decisive than intercultural differences. However, various styles of upbringing stress and value other goals. If obedience and respect are highly valued because of community interests (Gernhardt et al., 2016), resulting in avoiding attachment patterns, what exactly are we doing if we promote secure attachment based on assigning a high priority to the development of the self? Which elements relating to attachment are recognized as essential in particular cultural contexts? What does it mean when we view secure attachment and related aspects as being healthier than aspects in the other categories, bearing in mind that many children are regarded as being insecurely attached? How do these questions relate to God representations?

If ambivalence emerges as an important aspect in drawing God (especially in contexts where one would expect more insecure attachment styles), might this suggest that a God representation that includes ambivalent aspects could be valued as healthier (from a relational point of view) because this God is still able to embody secure and insecure aspects? Here, we are adhering to the idea that tolerating ambivalence is one of religion’s major cultural contributions (cf. Westerink, 2017; Jongsma-Tieleman, 1996; Winnicott, 1971). If so, how do we change our attachment models and figures concerning God representations in such a way that this ambivalence becomes visible and applicable instead of emphasizing the polarities of secure and insecure? In other words, how could particular culturally diverse meaning issues become incorporated more seriously in attachment theory?

Comparing Drawings by Securely and Insecurely Attached Children: A Qualitative Analysis

Research Questions, Hypotheses and Attachment Questionnaire

Starting from the question of how attachment qualities are reflected in children’s drawings of and narratives about God representations in different subcultures in which god talk is either present or absent, we compared 12 drawings by insecurely attached children (Ins1–12) with 12 drawings by securely attached children (S1–12). The drawings were selected from our total Dutch collection of 233 drawings and narratives (although not all participants filled in the attachment questionnaire).

Our analysis focused on qualitative aspects of children’s drawings of God and their use of religious symbols. Attachment style was assessed with the ERC-RC (see above). The groups were equivalent in terms of age and gender. We were not able to make an overall match of god-talk/non-god-talk contexts. Each group consisted of nine girls (2 aged 7; 2 aged 10; 2 aged 12; 3 aged 13) and three boys (1 aged 10; 1 aged 12; 1 aged 13). Our research question was: How do different attachment styles relate to the use of symbols when drawing God? Our hypothesis was that securely attached children would use more God representation-related symbols in their drawings than children who were insecurely attached, and that these symbols would have a referring and self-transcending character.


In order to gain an understanding of the use of symbols in drawing God representations in relation to attachment style, all drawings and narratives by the securely (N = 12) and insecurely attached (N = 12) children were independently double coded by both authors of this chapter. The narrative was coded first. We calculated the number of independent symbols that were written down by the children themselves. The main criterion for coding an aspect as a symbol was that the child used referring and/or self-transcending language. The child shows awareness of the fact that what was drawn is not a symbolic equation of God but it contains imaginative “as if” elements (in the way Peirce (1977) described thirdness, cf. Raguse (1994)): it is a representation. An example of a referring aspect is found in drawing 4 of the secure sample (S4): “I drew my thoughts about God with the hand of God who holds the world… because God holds and protects you”. In this example, God is not identical to the drawn hand. But the hand refers to God’s assumed protection. Other narrative examples show some difficulty in using referring aspects in drawing God: InS11: “I wrote God down in letters. I could not imagine something else.” InS5: “If you look at my ‘drawing’ [blank] you do not see that much. Because I do not think anything… God does not exist for me.” Subsequently, the drawings were coded by calculating independent symbols as well. The most important criterion for coding was again the potential referring quality of the drawn aspects as seen by the researchers, but still informed by the child’s narrative.

In addition, we rated (1) the padding of the paper on a five point-scale: very small, small, average, more than average, and completely padded; (2) we noted the specific position on the paper of the main drawn god aspect (Eshleman et al., 1999); (3) we described the place where God was situated figuratively; (4) we marked the anthropomorphic/non-anthropomorphic design of the drawings; and (5) we coded our interpretation of the nature of the attachment relationship to God as reflected in each child’s drawing (human beings-God and/or me-God) on the basis of positive, negative (punishment), and ambivalent aspects, as well as concrete relational features. After independent coding of all these aspects, differences in coding were discussed and negotiated until consensus was reached. In cases of doubt, we chose the lowest number. Calculations were based on the shared consensus. Although the significance of process descriptions and family drawings had become obvious to us during the data collection process, these sources of information are not included in this qualitative exploration because they were not available for all the participants in our sample.


The results of the coding of symbols are summarized in Table 11.1. In general, securely attached children showed more independent symbols in their drawings than insecurely attached children (drawings Ins26/S34; narratives Ins16/S26).

Table 11.1 Number of coded symbols in the drawings and narratives of the participants in the secure and insecure subgroup and measured padding of the paper on a five-point scale

Table 11.2 shows the results on the basis of age. In the 10–13 age group, securely attached children drew roughly twice as many symbols as the insecurely attached group, with one exception: a 13-year-old girl (Ins9) showed Islam (instead of God) in a full padded drawing with 12 independent coded symbols. In this insecure group, three children did not draw any symbols (InS 5,6,8). All drawings by the 7-year olds were difficult to analyse in terms of our criteria.

Table 11.2 Comparison of coded symbols as seen by the researcher (based on the drawing, informed by the narrative) and by the children themselves (based on the narrative)

Because of the separate coding for symbols in the drawings and narratives, we were also able to study the differences. The narratives (written by the children themselves) contained fewer symbols (Stotal.narratives = 28/Instotal.narratives = 16) than the coded symbols in the drawings (Stotal.drawings = 34/InStotal.drawings = 26 including InS9; InStotal.drawings = 14 without InS9). The difference could be explained by the fact that children can forget to write about aspects in the case of rich drawings. The protocol did not explicitly ask them to write about symbols. Another possibility is that researchers and children make different interpretations. If we focus on the 10–12-year-olds, the secure group showed 19 and 16 independent coded symbols in the drawings and narratives respectively; the insecure group showed 9 symbols in both the drawings and narratives.

Table 11.1 also includes a rough rating of the padding of the paper, scored on a five-point scale: very small, small, average, more than average, fully padded. The secure group seems to have used more padding (total score = 47; mean score = 3.9) than the insecure group (total score = 31; mean score = 3.1; blank sheets are excluded). Thus, the securely attached children in our small sample showed more symbols in both their drawings and narratives and their drawings had more infill than those of the insecurely attached children. It is possible that the symbols they used required more space. Six securely attached and three insecurely attached children used a complete infill.

In addition, we found that four children in the insecure group drew an anthropomorphic image (Ins2–4,12), and nine children in the secure group (S1–5;7,8,10,12). In this subsample of nine, three drawings contained partly anthropomorphic images such as arms, hands or a huge head reaching from the clouds (S4,5,7).

Table 11.3 shows the localization of (the main symbol connected to) God. We found five categories: extraterrestrial, in heaven, between heaven and earth, on earth, and drawings and narratives without explicit references to the localization of God. The secure group used the different categories more or less equally. The insecure group showed more references to a localization on earth or did not use references at all. When God was localized in heaven, the insecure group drew God on the lower half of the paper, whereas the secure group used the upper half (see Chaps. 6 and 7, this volume).

Table 11.3 Imagined localization of (the main symbol connected to) god

Table 11.4 presents our interpretations of the attachment relationship to God, as reflected in the drawings and narratives. The majority of the drawings in both subgroups did not provide adequate information in terms of the attachment relationship aspects we were looking for based on the literature (positive, ambivalent, negative/punishing; explicit relational features). However, when there was explicit input on this subject, the secure group used more positive connotations, whereas the insecure group tended to use more ambivalent or negative aspects.

Table 11.4 Attachment to God


See Figs. 11.1, 11.2, 11.3, 11.4, 11.5, 11.6, 11.7, 11.8, 11.9, 11.10, 11.11, 11.12, 11.13, 11.14, 11.15, 11.16, 11.17, 11.18, 11.19, 11.20, 11.21, 11.22, 11.23, and 11.24.

Fig. 11.1
figure 1

Fig. 11.2
figure 2

Fig. 11.3
figure 3

Fig. 11.4
figure 4

Fig. 11.5
figure 5

Fig. 11.6
figure 6

Fig. 11.7
figure 7

Fig. 11.8
figure 8

Fig. 11.9
figure 9

Fig. 11.10
figure 10

Fig. 11.11
figure 11

Fig. 11.12
figure 12

Fig. 11.13
figure 13

Fig. 11.14
figure 14

Fig. 11.15
figure 15

Fig. 11.16
figure 16

Fig. 11.17
figure 17

Fig. 11.18
figure 18

Fig. 11.19
figure 19

Fig. 11.20
figure 20

Fig. 11.21
figure 21

Fig. 11.22
figure 22

Fig. 11.23
figure 23

Fig. 11.24
figure 24

Discussion and Conclusion

For the relationship between attachment style and children’s drawings of Gods, we hypothesized that securely attached children would use more God representation-related symbols in their drawings than children who are insecurely attached and that these symbols would have a referring and self-transcending character. The results of our qualitative analyses seem to support this hypothesis. Both the drawings and the narrative of those who were securely attached according to the attachment questionnaire (ECR-RC) contained more God representation-related symbols than those of the insecurely attached children. This finding could be explained in terms of an open and responsive attitude towards the (religious) environment, epistemic trust being a possibly mediating mechanism (Fonagy et al., 2015; Fonagy et al., 2016, pp. 780, 793). (The drawings and narratives of the four 7-year-olds were difficult to rate. A possible explanation is that the writing and drawing skills of that particular age group do not meet our criteria. However, this subgroup is too small to make generalizations.) In addition, the drawings by the group with secure attachment showed more infill of the page, in line with findings regarding drawings of families (Attili et al., 2011). Their drawing of God is positioned more in the upper part of the picture than in the drawings by the insecure group. This suggests that children with a secure attachment style feel freer to take up space, facing others and the world with their heads held high, which are expressions of basic trust.

However, our sample was very small—just 24 children. This is an recognized limitation of our study, which was designed to be of an exploratory nature. The results should therefore be interpreted with considerable caution. Follow-up studies involving bigger samples are needed. Taking this limitation into account, in instances where attachment relationship aspects were explicitly present, the secure attachment sample displayed different themes and articulations than the insecure sample. The secure group showed more positive connotations and used different figurative places in connection with God.

An important point of discussion among the researchers concerned the scoring of symbols. How do we know, for example, whether a sun is part of the God representation that should be interpreted in relation to the divine, or is simply a standard part of the drawing, because the child is always drawing the sun, whatever the subject of the drawing may be? In this context, interpretation has a subjective component, as it is largely culturally determined. Therefore, follow-up studies also need intra- and interculturally diverse exchanges when it comes to counting and rating.

We rated the symbols to which children referred in their narrative. However, those who collected the drawings and narratives were not instructed to ask questions about (potential) symbols in the drawings, as this was not part of the standard protocol in the international project. If new data is collected in the future, the protocol should include explicit attention to the meaning of the different aspects of the drawing as seen by the child in view of these referring and or self-transcending qualities. The process description should also be included in the protocol.

The material was collected in different contexts. Most drawings were collected in schools, and some in churches, but the degree to which children felt safe and happy at these places may have affected the results, possibly also in relation to the ECR-RC-scores. Based on research data for the general population, we predicted higher numbers of insecurely attached children. This discrepancy can be partly explained by the fact that not every institute where we collected our drawings accepted our questionnaire. In some cases, research assistants showed resistance in using the questionnaire because they did not feel sufficiently well-equipped. Another explanation could be that specific social contexts influence the narrating and drawing activities. We suggest that the individual processes should also be taken into account in the general protocol.

In our study, we made a clear distinction between securely and insecurely attached children. We only included children who were either securely or insecurely attached to both father and mother and excluded those who reported secure attachment scores for one parent and insecure scores for the other. However, from a theoretical perspective, the capacity to tolerate ambivalence is an indication of mature development. This may also apply to toleration of ambivalence regarding attachment to different parents. Cultural norms and values could play a part in this context. Whereas relatedness, confidence and closeness between parents and child are stressed in the Western world of the twenty-first-century, obedience and independence (cf. the distinctions between relatedness and self-definition (Blatt, 2008), agency and communion (Bakan, 1966), autonomy and surrender (Angyal, 1951) and the accompanying developmental tasks are seen as more important in some other cultures. This suggests that secure and insecure attachment could (or even should) be operationalized across more dimensions than the degree of anxiety and avoidance.

In sum, our qualitative exploration suggests that children’s drawings of God and their use of God-representational symbols reflect their attachment styles. Hence, drawings of God can be considered a valuable method for assessing children’s religious experience in relation to their relational-affective functioning, which is a promising research line for the psychology of religion.