Attachment and autism spectrum conditions: Exploring Mary Main’s coding notes

Assessment: ASC

Currently, ASC is considered a lifelong neurodevelopmental condition which, at its core, is characterized by atypical social communication and restricted and repetitive behavior across different contexts (American Psychiatric Association, 2013). Early clinical descriptions regarded autism as a rare discrete condition (e.g., Kanner, 1943). However, the latest edition of the The Diagnostic and Statistical Manual of Mental Disorders (5th ed.; DSM-5; American Psychiatric Association, 2013) marked a significant formal change in how diagnosticians conceptualize these features. That is, the nature of the variable, that is, autism, has changed from a categorical condition to a continuum, thus making it a spectrum. A key to this change has been the acknowledgment of the heterogeneity of symptom manifestations and diagnostic issues with comorbidity.

OPEN IN VIEWER

Table 1.

Discriptive Statistics and Attachment Classifications.

Case ID	Sex	D autism	Primary classification	Alternative classification
31	M	D autism	D	C1
23	F	Not assigned	D	B2
20	M	Not assigned	B1	CC
26	M	D autism	D	CC
19	M	Not assigned	D	C
7	F	Unclear	D	B4/CC
34	F	D autism	B	D
41	F	D autism	B1	CC/C
5	M	D autism	B2	A2
29	M	Not assigned	B1	B3
28	F	D autism	D	B/A1
35	M	D autism	B3	Uncodable for D due to parental intervention*,**
37	M	D autism	A1	Not assigned
42	M	D autism	A2	A1

*Signifies that there was difficulty coding D in the child.

**Case 90 where the overall classification was unavailable.

In terms of social communication, the DSM offers several examples of features that are considered expressions of ASC. Difficulties with social reciprocal conversation, atypical eye contact, and difficulties with understanding social relationships are among some of the features described in the diagnostic manual. However, conditions characterized by atypical social communication are not uncommon in the DSM (e.g., social communication disorder and social anxiety disorder). Therefore, it is perhaps unsurprising that the other domain, ritualized and repetitive behaviors, is not only hallmark characteristics of ASC but also often differential when considering alternative socio-communication difficulties. Examples of restricted and repetitive behaviors include stereotypical movements, echolalia, atypical sensory interests, and emphasis on routine (American Psychiatric Association, 2013).

From an assessment perspective, clinicians rely on a range of standardized behavioral observations, interviews, and survey measures. While a range of these tools exists, it is widely acknowledged that a combination of the Autism Diagnostic Observation Schedule (ADOS-2; Lord et al., 2012) and Autism Diagnostic Interview Schedule (ADI-R; Rutter, Le Couteur, & Lord, 2003) constitutes “the gold standard” for assessing ASC (de Bildt et al., 2004; Le Couteur, Haden, Hammal, & McConachie, 2008; Rutter et al., 2003). While there is little doubt that these tools have played a vital role in helping to refine understanding of ASC, they are often unable to differentiate between ASC and symptomatically similar presentations such as issues around attachment (Woolgar & Scott, 2014).

Assessment: Child attachment

The attachment system is a motivational system with the principle function of gaining access or proximity to the attachment figure (Bowlby, 1969). This biologically based behavioral system is activated when the child recognizes real or perceived threat or separation. Attachment behaviors are varied (e.g., crying, clinging, running) and change with development, but all serve the function of gaining access to the attachment figure. The majority of studies on childhood attachment use the Strange Situation Procedure (SSP) to classify attachment in infancy (Ainsworth, Blehar, Waters, & Wall, 1978). This laboratory-based procedure provides a snapshot of infant attachment related and exploratory behavior following two separations and reunions with their caregiver (Ainsworth et al., 1978). The following are the sequential episodes of the assessment:

The child and caregiver enter the room.

The function of these different episodes is to offer a baseline and critical moments from which to consider behavior within the attachment-relevant contexts of separation from and reunion with a familiar caregiver.

Ainsworth, Blehar, Waters, & Wall (1978) observed that infants typically displayed one of three patterns: Secure (B), Avoidant (A), and Ambivalent or resistant (C). According to Ainsworth and colleagues, infants were considered Securely (B) attached if they used their caregiver as a secure base for exploration, showed signs of missing their parent during separation, and greeted their parent with an overt gesture and actively sought parent on reunion (Ainsworth et al., 1978; Solomon & George, 2008). Children who were classified as Avoidant (A) showed little sign of affect toward their caregiver upon entering the room, showed little distress during the separation, and actively avoided their caregiver during reunion. Despite inhibiting their reactions, Ainsworth et al. (1978) pointed to physiological evidence to suggest that the attachment system is still being activated in these children. Based on her observations of these infant–caregiver dyads at home, Ainsworth et al. proposed that their avoidant behavior was a response to common experiences of rebuff from their caregiver when distressed. Ambivalent–resistant children (C) on the other hand showed weak exploration of the room, showed overt signs of distress during separation, and combined displays of distress and anger and attempts to cling to the caregiver on reunion (Ainsworth et al., 1978). Based on her observations of these infant–caregiver dyads at home, Ainsworth et al. proposed that their ambivalent and resistant behavior was a means to maintain the attentiveness of a caregiver who, at home, tended to delay responding or whose availability might be inconsistent. Meta-analytic reviews have found that, except for samples of families under particular stress, cross-culturally, around 70% of children are classified as (B) Secure, 20% as Avoidant (A), and 10% as Ambivalent/resistant (Mesman, van IJzendoorn, & Sagi-Schwartz, 2016).

Yet through the 1980s, researchers began to note that some of the patterns of behavior the children were displaying did not align with those described in the original Ainsworth system. Following a systematic exploration of 200 SSP videotapes, Main and Solomon (1990) proposed a new “disorganized” (D) attachment classification. In their protocol for coding the new classification, they distinguished seven distinct indices (Main & Solomon, 1990):

Sequential display of contradictory behavioral patterns;

Disorganized attachment behaviors are not necessarily pervasive and may only become apparent for brief moments during the SSP. As a result, children who receive a “disorganized” primary classification also receive, where possible, a secondary alternate “organized” classification. As guidance for coders in handling the issue of pervasiveness, Main and Solomon (1990) constructed a 9-point scale, with scores of more than 5 leading to a disorganized classification. In low-risk community samples, rates of disorganized attachment are around 15%; but this can increase to the majority of a sample among families known to social services for maltreatment, among families with traumatized parents, and among families facing multiple compounding socioeconomic risks. Main and colleagues theorized that the display of these behaviors, especially on reunion with the caregiver, understood to reflect a child’s experience of conflict or fear regarding that caregiver. Yet there can be many reasons for conflict or fear, as shown by the diversity of factors that increase its prevalence across different samples (Granqvist et al., 2017). It is also suspected that the behaviors may well also have a different meaning if they occur when a child is more alarmed or less alarmed than the degree evoked by the Strange Situation; as such, Granqvist and colleagues (2017) caution against coding disorganization using the behaviors specified by Main and Solomon outside the context of the SSP, as their validity for such application is presently unclear.

Phenotypic similarities: ASC and attachment-related difficulties

One of the key features of ASC is “deficits in social reciprocity from abnormal social approach to reduced emotions or affect to failure to initiate or respond to social interactions” (DSM-5; American Psychiatric Association, 2013). This avoidance of social interactions appears similar to that characterized by the classification of thoroughgoing avoidance (A1) in the Ainsworth system, which describes failure to initiate or respond to social interactions with the caregiver on reunion. On this basis, it would be reasonable to expect that children with ASC would be overrepresented in the avoidant category. However, published studies indicate a lower prevalence of avoidant attachment in this group, with between 7% (Capps, Sigman, & Mundy, 1994) and 15% (Naber et al., 2007) of children with ASC classified as insecure-avoidant.

However, there are also a number of symptomatic similarities between the behaviors listed under indices IV of Main and Solomon’s classification scheme for disorganized attachment and the diagnostic criteria for ASC. For example, “rhythmical, repeated movements without visible function” is an indicator of disorganization (Main & Solomon, 1990), while “stereotyped or repetitive motor movements, use of objects, or speech (e.g., simple motor stereotypies, lining up toys or flipping objects, echolalia, idiosyncratic phrases)” are features of ASC behavior (American Psychiatric Association, 2013, p. 50; Lord et al., 2012; Rutter et al., 2003). Main and Solomon (1990) describe a case in which the infant would put her hands to her ears in response to being picked up by a caregiver (p. 143). In this case, the caregiver was found to be abusive. Interestingly, the ADOS-2, the direct behavioral observation used to diagnose children with ASC, gives a similar example in the “Stereotyped Behaviours and Restricted Interests’ section part D4. The instructions state “putting his or her hands over and/or fingers in his or her ears should be coded here” (ADOS-2, Module 3, Coding system, Lord et al., 2012). This specific example highlights an important difference between the coding of the SSP and the ADOS-2 and the need for informed clinical judgment. In the SSP, Main and Solomon (1990) are clear that stereotypy is coded as indicating disorganization when it occurs in contexts of relational stress such as reunion more than at pre-separation baseline and that it cannot be coded in the absence of the parent. In the ADOS, however, it remains unclear whether the observer would make this distinction when coding this behavior. On similar note, Main and Solomon (1990) propose that “asymmetries of facial expression directly upon the appearance of the parent” are indicative of disorganized attachment. Facial expressions are also deemed important in the diagnosis of ASC. For instance, Items 51, 57, and 58 of the ADI-R, the parental interviews used to assess and diagnose ASC, are concerned with the range and ability of the child to use facial expressions in the section on social development and play (Rutter et al., 2003). In this diagnostic interview, parents report on their experiences of the child’s social responses. It is likely that the asymmetries in social expression associated with disorganized attachment could receive a mark on Item 58 of the ADI-R, which describes “facial expressions slightly or occasionally inappropriate or odd” (ADI-R interview protocol: 52; (Rutter et al., 2003). Similarly, Main and Solomon (1990) observed asymmetrical facial expressions, but these were among children from maltreated samples on reunion with their caregiver in the SSP. Such overlap is potentially troubling from the point of view of differential diagnosis, even if both the ADI-R and the Main and Solomon protocols state that any such behavior must be considered in the context of other behaviors and features to warrant a classification. Good diagnostic practice always takes context seriously, but where behavioral markers overlap and the context for interpreting them may be ambiguous or difficult to discern, difficulties for clinicians will arise.

There have been calls to catalog the overlapping and distinguishing features of ASC versus attachment-related difficulties. Based on clinical observations, Moran (2010) developed the Coventry Grid to initiate a conversation around how to go about making this differentiation. The Grid outlines some of the overlapping domains of difficulties for children with both conditions (e.g., focus on routine, repetitive language, food sensitivity, attachment to preferred objects, atypical play, or atypical social interactions). It is not uncommon to hear references to the Grid in clinical practice, and there have been attempts to repurpose the Grid into a formal clinical interview (Coventry Grid Interview; Flackhill, James, Soppitt, & Milton, 2017). In a recent critical review of the literature, McKenzie and Dallos (2017) state that although the Grid does neatly describe the difficulties clinicians face when differentiating ASC from attachment-related difficulties, one problem with the Grid is that it combines the insecure attachment classifications: avoidance, resistance and disorganization, which represent quite a different phenomenon. There is also a paucity of research into the psychometric properties of the Grid, though the unpublished doctoral thesis of Kendall-Jones (2014) offers favorable indications regarding the internal consistency of the subscales as well as adequate face and content validity. Nevertheless, it is worth restating the original mission described by Moran (2010) which was to ignite a discussion regarding the difficulties in this complicated area of diagnostic practice.

Assessment: Attachment in children with ASC

Adding to complexity for clinicians is the fact that not only may it be difficult to distinguish certain symptoms of ASC from disorganized attachment, but it is quite possible for the two phenomena to co-occur. Early commenters mistakenly presumed that ASC was incompatible with attachment (e.g., Bettelheim, 1967; Cohen, Paul & Volkmar, 1987; Kanner, 1943, 1960; Mahler & Furer, 1968; Rutter, 1978). However, this early hypothesis has been discredited by a host of studies (see Rutgers, Van Ijzendoorn, Bakermans-Kranenburg, & Swinkels, 2007). Researchers have demonstrated that children with a diagnosis of ASC show a range of attachment behaviors toward caregivers. Gestures, vocalizations, and preferential behaviors including proximity seeking for caregivers have been seen in children with ASC (Rogers, Ozonoff, & Maslin-Cole, 1993; Shapiro, Sherman, Calamari, & Koch, 1987), and many children with ASC are classified as securely attached by blind coders of the SSP. In the first study to apply the disorganized classification for a sample of children with ASC, Capps, Sigman, and Mundy (1994) looked at the attachment profiles of 19 children with autism using a modified SSP where the child in which the second separation was curtailed as the stranger did not leave the child alone in the room. Fifteen of the sample were deemed “classifiable” on the basis of the Ainsworth coding protocols, while the entire sample displayed evidence of attachment to caregiver. All 15 of the “classifiable” children had sufficient markers of disorganized attachment to be given this as their primary classification. Secondary classifications were secure (40%), insecure-avoidant (7%), insecure-ambivalent (13%), and insecure-undetermined (20%). Yet, in fact, only three of the children (20%) were considered to be truly disorganized when potential ASC traits were excluded from the analysis (Capps et al., 1994). Such findings agree with concerns among clinicians that ASC behaviors may be misidentified as behaviors related to disorganized attachment, and vice versa. Attempting to respond to this issue, Willemsen-Swinkels, Bakermans-Kranenburg, Buitelaar, Van Ijzendoorn, and Van Engeland (2000) distinguished among the Main and Solomon indices in a study of children with Pervasive Developmental Disorder (PDD). When Index IV behaviors (stereotypies) were removed from the analysis of disorganization/disorientation, only 16.5% of the children were classified as disorganized compared to 10.5% of normally developing children. Interestingly, even with Index IV removed, the prevalence of disorganization increased to 53.8% among children with both PDD and an intellectual disability (Willemsen-Swinkels, Bakermans-Kranenburg, Buitelaar, Van Ijzendoorn, & Van Engeland, 2000)

In a meta-analytic review, Rutgers, Bakermans-Kranenburg, van Ijzendoorn, and van Berckelaer-Onnes (2004) found that 53% of children with ASC develop secure attachments. In a more recent review, Teague, Gray, Tonge, and Newman (2017) found rates of secure attachment ranged from 40% to 63% with a mean rate of 47% (Teague et al., 2017) for children with ASC. Although this figure is lower than rates of secure attachment among children without ASC, it nevertheless makes clear that infants with ASC can and often do form secure attachment relationships. It is also worth noting that the same review found that the discrepancy between children with ASC and neurotypical children disappeared when controlling for cognitive ability or level of severity of ASC symptoms (Rutgers et al., 2004). Level of intellectual functioning and severity of ASC symptoms is recurrently linked to lower attachment security (e.g., Naber et al., 2007; Rogers & DiLalla, 1990). In addition to the lower rates of attachment security, the review by Teague and colleagues (2017) also found that children with ASC had higher rates of disorganized attachment: approximately 22% compared with 15% in low-risk neurotypical samples. Most recently, Rozga et al. (2018) note that the theoretically expectable link between parental sensitivity and infant security is present in samples of children with ASC with relatively fewer cases coded as disorganized (e.g., Oppenheim, Koren-Karie, Dolev, & Yirmiya, 2012), whereas in samples with ASC where most children are also coded as disorganized (e.g., Van IJzendoorn et al., 2007), there is no relationship between parental sensitivity and the child’s attachment classification.

McKenzie and Dallos (2017) have challenged the prevalence literature, arguing that “autism and attachment difficulties result in similar symptoms and even very experienced clinicians find identifying which symptoms are attributable to autism extremely challenging” (p. 636). Merely cutting out Index IV behaviors, the strategy pursued by Willemsen-Swinkels and colleagues is regarded by McKenzie and Dallos as insufficient to adequately address the issue, though in our view this remains an open, empirical question. Nonetheless, we agree with McKenzie and Dallos that further efforts would be valuable to untangle the behavioral properties of ASC from disorganized attachment, even if absolute distinctions at an individual level will likely always retain some uncertainty. Although by no means able to offer a model for making such distinctions, this article will nonetheless attempt to speak to these concerns, making use of archival materials to explore strategies employed by Mary Main, the leading authority on coding disorganized attachment. In this, our approach is aligned with Chang’s (2017) argument for the potential of “history as complementary science” on the basis of three kinds of benefits that may be gained from history for science:

increasing scientists’ critical awareness of the contingency of the present and expanding their understanding of the concepts they depend upon (what Chang terms “extension”);

Stereotypies and repetitive behaviors: Context and frequency

In line with the written Main and Solomon (1990) protocols, Main consistently treated the context and frequency of stereotypies and repetitive behaviors as important and differential. Specifically, behaviors that appeared throughout the SSP and not exclusively during key moments—such as reunion—were typically considered features of ASC by Main rather than related to attachment relationship. This is evident in Case 41 where, during Episode 2, the child shows atypical vocalizations and has her “arms extended” and moves her wrists in a way that is described by Main as “autistic.” Then during Episode 4, when the mother leaves, the child expresses some atypical vocalizations and complex hand mannerisms. Main likewise does not consider these stereotypies a signifier of “true disorganization.” During Episode 8, however, Main observes that the child engages in hair twisting upon the entrance of her mother. Since this is the first instance of this behavior, and it occurs directly on reunion with the caregiver, Main considers this to be indicative of disorganization. Overall, this child was classified as autistic and securely attached (B) with their caregiver, despite displaying sufficient behavior from the indices of disorganized attachment that she would otherwise have likely received this classification.

An example of frequency playing a differential role can be seen in Case 35. In the early, pre-separation episodes, which are treated as a “baseline” for understanding a child’s ordinary behavior, the child engaged in “hand flapping” and atypical movements. Main reflects on this continuity between behavior pre- and post-separation in her remarks on Episode 5 where she observes the child “goes to the door, grinning, jumping, and hand flapping.” Main comments “we have seen this too often to call it regular D, it seems often to simply be his style of movement.” The behavior does not seem differentially related to key attachment stimuli such as separation and reunion, giving Main confidence that the behavior is solely expressive of ASC and not indicative of disorganized attachment. Again, the child was classified as autistic and as having a secure attachment with their caregiver.

Case 34 is much less clear-cut for Main. Here the child displays atypical movements and “stereotypy” in Episodes 2 and 3. During Episode 5, Main notes that as the caregiver enters, the child displays “many gestures of autism/excitement, legs spread, arms shaking as M enters, but these seem to be looks of pleasure.” Main muses to herself: “This is almost impossible to code.” She continues,

it would definitely be called D in the system for low risk samples, especially with throwing her head back, but on the other hand she just seems overwhelmed by excitement/ pleasure and unable to remain contained in her body. So, we code D = 5.

Prone postures

The coding notes suggest that falling prone or lying to the side was a feature which was highly prevalent in this sample in the Strange Situation. In all cases, Main gave cases where a child fell prone a “True D” score—though the score it received depended on the specific posture. It appeared that if the child was fully prone, this led to the assignment of a greater “D score” on the 1–9 scale than if the child was observed lying on their side or leaning.

Movement while assuming the posture appeared to be of importance to Main; her published discussions suggest that she regards falling prone as a kind of tonic immobility in the context of dissociation (Main & Morgan, 1996), and so a child showing movement whist prone would reduce confidence that it was a dissociated state. For instance, in Case 19, during Episode 3, the child “lies on the sofa and bangs the orange toy and then lies still in an odd posture, but some movement, on stomach, slowly moving the toy.” For this episode of abnormal posture, he was given a “D score” of 3. This is lower than his “D score” of 6 in Episode 4 when “he moves around prone the other way, depressed posture, stilling, some slight hand movement.” Stilling, an Index V indicator of disorganization, appeared to be closely related to falling prone in this sample. However, since stilling by definition entails a lack of movement, in line with the Main and Solomon (1990) protocols that specify duration as an important criterion, a major additional factor used by Main in determining the “D score” for stilling was the length of time the posture was adopted. A good example of this is Case 28, Episode 7 where Main observes: “He then moves to M’s empty chair and pulls his whole body onto it, face turned in, anomalous and yet deliberate, depressed and waiting.” This is given a “D score” of 9. Main goes on to reflect that “we have never seen a child manage to do this before.” As this is scored during Episode 7, this did not contribute to the child’s overall score.

In Case 7, Main describes another child, whose attachment is difficult to classify. In this case, the child demonstrates some atypical gestures and posture in Episode 2. Episode 4 is characterized by “crying” and aggressive play with the toys. On the first reunion, the child “wipes hand across her eyes” and repeats this behavior when the caregiver moves closer. The child repeats this behavior in Episodes 6 and 7. During the final reunion in Episode 8, the child is observed to “move away” from the caregiver, which Main states “seems like a momentary sequential display of contradictory behaviors.” This is followed by the child wiping her hand across her eyes again. In the comments section, Main remarks “we are not familiar enough with autism to know whether the child’s tantrum and movements are characteristic” of ASC. However, she states that the behaviors in Episode 5 make an “organized” classification “difficult to assign.” This appears to be because the stereotypies intensify on reunion, suggesting relatively decisively that the behavior indicates tension or conflict about the relationship, especially since in Episode 8 the behavior is followed by direct conflict behavior regarding proximity with the caregiver.

In coding prone behavior, Main also observes that the response by the caregiver can make this uncodable. Potential disorganization of the attachment system can be camouflaged by parental intervention. This was noted in Case 35 during Episode 8 where the child briefly collapses to the floor in a depressed position. Six seconds later, the mother intervenes by correcting the child, sitting her upright. Main comments that the mother might be “breaking up the real D behaviour.” This intervention was consequential as Main concludes that, as a result, the procedure was “uncodable for real D.” The SSP requires that the parent’s behavior is sufficiently standardized that attachment behavior shown by the child toward the parent can be displayed without interruption.

Aggressive behavior

Aggressive behavior toward the caregiver was mentioned by Main in several of cases. In the written coding system, aggressive behavior toward the caregiver while the child is apparently in a “good mood” is mentioned (Main & Solomon, 1990, p. 136). The coding notes give the impression that the criterion requiring “good mood” may have been relaxed, as aggressive behavior toward the caregiver without apparent good mood was nonetheless used as an indicator of disorganized attachment when the behavior was coupled with other indices of disorganization such as atypical postures and fear smiling.

In Case 23, there is “quasi-aggressive behaviour” with shrieking, kicking of legs and throwing of pillows, as well as some “clasping of her left wrist with her right hand.” The antecedents for this behavior are unclear as the caregiver and child are playing a game. In Case 31, the aggression is clearer. Main states that during reunion

he stares down at her aggressively and seems to hit at her with the cup. She says ow and he flails the cup down at her again but doesn’t hit her face. She’s restraining him whilst he flails the cup at her aggressively.

Previous research has established that atypical sensory interests or difficulties are a prevalent feature of ASC (e.g., Ben-Sasson et al., 2009). Sensory sensitivity, rather than sensory seeking, has been associated with aggressive behavior in this population (e.g., Mazurek, Kanne, & Wodka, 2013). In Case 31, as the mother picks the child up for a hug, “he seems to raise up to flail away, multiple jerks away from her” and then shows aggressive behavior. This is given a “True D” score of 5. Main states that this is indicative of true disorganization since the aggression seems to arise out of nowhere. Without having seen the tapes, it is impossible to appraise Main’s assessment of the particular case. It could well be that there appeared little relationship between the caregiver’s behavior and the child’s. However, we are struck that items regarding negative/aggressive responses to touch feature in a number of ASC assessment tools (e.g., The Diagnostic Interview for Social and Communication Disorders; DISCO; Wing, Leekam, Libby, Gould, & Larcombe, 2002). Therefore, it is possible that the same behavior, seen by a DISCO coder, would have been assigned as a marker of ASC rather than related to child attachment. In her coding notes, Main herself acknowledges that, without specialist training in assessment of ASC, she runs up against the limits of her expertise in such attempts to judge whether aggressive behavior toward the caregiver should be regarded as regarded as signally disorganization, ASC, or—potentially—both.

Atypical facial expressions

From the coding notes, atypical facial expressions appear to have been prevalent in this sample. Commonly, these unusual facial expressions were described by Main as either “aggressive smiles” or “fear smiles.” These were not typically thought by Main to be autistic in nature and were instead either uncoded or coded as “True D.” A contributing factor was evidently that there was an abundance of these abnormal facial expressions directly on reunion, which suggested to Main that they offered a window into the quality of the attachment history with that caregiver. For instance, Case 26 displayed repeated and sustained atypical facial expressions. In Episode 5, it is noted that C “gets a very odd facial expression, makes anomalous almost inhuman noises with fear-aggressive grin at the wall. This goes on and on.” This extract leads to a very high score of D = 9. In Episode 5, the child’s odd facial expressions continue but seem to coincide with an element of contradictory behavior, which adds to the coder’s belief that the behaviors reflect disruption of the attachment system. “He moves away from, then approaches M with the same strange and stretched fear/aggression tooth-filled grin he had when distressed, putting his arms up to be hugged, she bends to hug him, his face is extremely odd.” This extract again scores highly with a score of D = 6.

The cause of the atypical facial expressions in this sample is explored by Main in her reflections on Case 41. This case is unusual in that the father happens to enter at the end of the normal SSP, which up to that point had involved the mother. It is also an anomalous case in that the child’s facial expressions have not been classified as disorganization, the implication being that they are instead due to ASC. In Episode 8, on reunion, Main’s notes state “as Mother leans towards her, talking, pleased smile, but then it takes on a strange element.” The notes report that “Mother caresses her leg, Child says ‘no’, Child’s face twists, very peculiar expression. The increase in odd facial expressions on touch indicates a possible autistic response.” However, Main does states that “there are a lot of very odd faces in this episode, which were rare before” which raises the possibility of disorganization since the context suggests that the behaviors are not pervasive, but specific to attachment-relevant interactions with the caregiver. It is not clear. (Although outside of the SSP and therefore technically uncodable, when the Father enters at the end, Main notes that the child has a “broad smile with a fear element.” In this reunion with father, Main wonders whether the conflict between approach and avoidance producing the smile with a fear element is caused by autism: It could be that autism is causing the child to be worried by proximity with the caregiver, producing phenotypic resemblance to disorganized attachment.)

If so then ASC here would be, technically, causing disorganization of the attachment system. But this may or may not serve as a marker of the history of the child’s attachment relationships or have the predictive significance associated with disorganized attachment in other samples, upon which its validity as a classification is regarded to rest by developmental scientists. The child may not necessarily be afraid or worried by the caregiver, but instead worried by the unwanted proximity they anticipate will be forced on them as part of expected child–caregiver routine. Indeed, Main suspects that some parents may have “pre-trained” their child to act affectionately on reunion, when their inclination—influenced or not influenced by ASC—might have otherwise not inclined them to do so. This can be seen in Main’s observations on Case 5. Main notes that when the mother and child exchange saying “I love you,” they do so in voices that “suggest she says this a lot, or it is part of a routine.” In her summary of this case, Main says “we again have the problem of some training in affectionate behavior,” indicating this is not an isolated case in this sample (though it is also likely not an isolated case in general as typically developing infants may well also be taught artificial or routinized sharing of affection).