Attachment Security in Autism: A Longitudinal Study
Introduction
Early clinical observations suggested that autism involves a failure to develop typical attachments.
However, empirical research indicates that children with autism do form attachments, and a significant portion develop secure attachments.
A key research task is understanding how children with autism form secure attachments, given their differences in social understanding and motivation.
This study investigates the correlates and implications of attachment security in children with autism.
Goals of the Study
Investigate the correlates of attachment organization quality in children with autism.
Maternal sensitivity.
Child cognitive and social skills.
Examine the implications of attachment security for subsequent development.
Focus on children’s response to another person’s apparent distress.
Attachment Organization in Autism
Early studies show that children with autism exhibit discriminative attachment behaviors.
Looks, touches, and vocalizations directed towards caregivers.
Proximity-seeking upon reunion after separation.
Preschool-age children with autism use caregivers as a secure base and haven of safety.
Toddlers with autism show distress and search behaviors upon separation from parents.
Studies using the Ainsworth Strange Situation procedure classify 40-60% of preschool children with autism as securely attached.
Challenges in Assessing Attachment Security
Differentiating stereotyped and repetitive behaviors from relationally based disorganization is a challenge.
Disorganized attachment is coded when behaviors are not explicable in other terms or stem from conflict related to alarm caused by the caregiver.
In the first study to include the D classification, children classified with the Ainsworth tripartite system also showed behaviors in the Main and Solomon indices.
Children classified as alternate secure showed no signs of disorganization apart from repetitive behaviors, suggesting these behaviors were not relationally based.
Mothers of alternate secure children were rated as more sensitive.
Differentiating Disorganization from Autistic Stereotypies
Empirical evidence suggests attachment disorganization can be differentiated from autistic stereotypies.
Willemsen-Swinkels et al. (2000) first assigned a secure/avoidant/resistant classification, then classified each child as disorganized or not, excluding stereotyped behaviors.
Disorganized children showed greater average heart rate changes during separation and reunion.
These findings suggest the disorganized attachment classification can be assigned with expectable correlates when autistic stereotypies are excluded.
Maternal Behavior and Child Abilities as Correlates
Mothers of securely attached preschoolers with autism are rated as more sensitively responsive.
One study did not find a link between sensitivity and attachment security in toddlers with autism, possibly due to methodological differences.
Koren-Karie et al. (2009) reported that the relationship between maternal sensitivity and attachment security holds after controlling for children’s developmental functioning.
Siller et al. (2014) found that an intervention improving caregiver sensitivity led to increased attachment behaviors.
Parental Insightfulness and Attachment
Oppenheim et al. (2009) found that a parent’s insightfulness into their child’s experience increases the likelihood of secure attachment.
This has been demonstrated for typically developing children (Koren-Karie, Oppenheim, Dolev, Sher, & Etzion-Carasso, 2002) and children with medical diagnoses (Marvin & Pianta, 1996).
The link between maternal insightfulness and attachment security in autism is mediated by maternal sensitivity.
Disorganized Attachment and Maternal Sensitivity
Organized insecurity (avoidant and resistant) and disorganization are often independent.
Disorganized/secure (D/B) behavior can emerge when a caregiver is predominantly sensitive but occasionally engages in alarming behavior.
Van IJzendoorn, Schuengel, and Bakermans-Kranenburg (1999) found no significant relationship between disorganization and maternal sensitivity.
Child Cognitive and Social Skills
Securely attached children with autism have better receptive language and play skills.
They make more social initiations during play with caregivers.
They are more responsive to joint attention bids and direct more requesting bids.
Links between severity of autism symptoms and attachment security are equivocal.
Disorganization is linked with intellectual disability in autism.
Meta-Analysis Findings
De Wolff and Van IJzendoorn (1997) found a moderate association (r = .24) between sensitivity and secure attachment.
The association was weaker in clinical samples.
Caregiver sensitivity must be considered within the context of parent-infant interactive processes.
Developmental Consequences of Attachment Security
Research is needed to investigate whether secure attachments predict subsequent socio-emotional functioning.
Securely attached children are rated as more socially competent, empathic, and popular.
Increased empathic responsiveness has been documented in secure children.
Current Study - Empathy Assessment
The study assessed changes in children’s affect and looking time to an examiner expressing distress.
This paradigm elicits empathic responses from typically developing infants and children with autism.
Early behavioral responses to distress are linked with later empathy expressions.
Current Study - Goals
Examine the continuity between attachment quality and later empathic responding.
Hypothesis: Securely attached children would be rated as more empathically responsive.
Examine cognitive-linguistic correlates of attachment security.
Hypothesis: Securely attached children would have better language skills.
Examine the role of caregiver sensitivity in the development of attachment.
Hypotheses
Securely attached children would be more likely to initiate social interactions and be more responsive to caregiver bids.
Methods - Participants
Participants were recruited through the UCLA Autism Evaluation Clinic (1997-2000).
40 families were enrolled; 30 seen in the Ainsworth Strange Situation Procedure.
Clinical diagnoses were confirmed using ADI-R and ADOS-G.
Sample: 23 boys, 7 girls; mean age 47 months (SD = 9).
Average NVMA: 31 months (SD = 14); average language age: 21 months (SD = 13).
Follow-up NVMA: 40 months (SD = 21); follow-up language age: 27 months (SD = 17).
Sample was heterogeneous in terms of functioning.
28 children met criteria for autism; 2 for PDD-NOS (likely ASD in DSM-5).
Methods - Procedures
Initial assessment: two laboratory sessions.
Measures of language and cognitive ability.
Assessment of empathy.
Ainsworth’s Strange Situation Procedure.
Third session: home, child-caregiver interactions videotaped.
Follow-up: approximately 1 year later (mean interval = 12.3 months, SD = 1.5 months).
All assessments re-administered except for the strange situation.
Methods - Strange Situation Procedure
Eight standard episodes of increasing stress
Exposure to unfamiliar room and a stranger
Caregiver presence and absence
Two separations and reunions with the parent
Classification based on reaction to reunions with the parent
Proximity seeking
Contact maintenance
Avoidance
Resistance
Videotaped with cameras in three corners of the playroom.
Methods - Classification of Attachment
Erik Hesse and Mary Main classified attachment organization.
Ainsworth’s system (Ainsworth et al., 1978).
Infant disorganized/disoriented classification (Main & Solomon, 1990).
6-year-old system for coding (formerly disorganized) children as controlling (Main & Cassidy, 1988).
Hesse and Main had experience classifying attachment in children with autism.
Both coded each child’s strange situation independently; disagreements resolved through discussion.
Coding Considerations for Autism
Observers should consider the individual’s baseline behavior to differentiate between disorganized behavior and typical autistic behaviors.
Criterion: whether behaviors were shown in connection with the mother or indiscriminately.
Stereotypic hand flapping, squealing, or perseverative treatment of objects were excluded.
D-Autism: behaviors readily explicable in terms of autism.
D-Attachment: behaviors explicable in terms of breakdowns in the attachment system.
Some stereotypic behaviors contributed to a D-Attachment score if they occurred on reunion and interrupted the child’s approach.
Behaviors like twirling and tip toe walking were excluded as unique to autistic children.
Methods - Parent-Child Interaction
Parents were told to interact with their children as they typically would during play.
Videotaped during a separate visit to the families’ homes.
Unstructured play (15 min).
Structured play (15 min).
Ratings of parental sensitivity were undertaken across the 30 min of play interaction by two independents observers, blind to the child’s attachment classification, and using Ainsworth’s original sensitivity–insensitivity scale (Ainsworth, Bell, & Stayton, 1974; Ainsworth et al., 1978).
Interclass correlation: 0.83.
Mother–Child Rating Scale (Crawley & Spiker, 1982) was used to rate the quality of the children’s social initiations and social responsiveness, both rated on a 5-point Likert scale.
Interclass correlations: 0.91 for social initiative and 0.89 for social responsivity.
Methods - Assessments of Cognitive Skills
Nonverbal cognitive abilities were assessed using two subscales of the Mullen Scales of Early Learning (MSEL; Mullen, 1995).
Visual Reception Scale.
Fine motor scale.
Age equivalents averaged to generate NVMA.
For one child, NVMA was based on two nonverbal subscales of the Stanford–Binet Intelligence Scale (Pattern Analysis and Bead Memory; Thorndike, 1972).
Three different language tests were administered, based on the child’s functional level:
Reynell Developmental Language Scales (RDLS; Reynell, 1983).
Childhood Evaluation of Language Fundamentals – Revised (CELF-R; Semel, Wiig, & Secord, 1987).
MSEL (Mullen, 1995; Expressive and Receptive Language Scales).
Age equivalents for receptive and expressive language skills were averaged to compute an overall language age equivalent.
Methods - Assessment of Empathy
Children’s responsiveness to the examiner’s display of apparent distress was observed using a structured empathy task.
Child and examiner seated at a table; examiner pretends to hurt her finger.
Examiner displays vocal and facial expression of distress for 30 s, followed by 10 s of neutral affect.
The child’s degree of interest and concern was coded using a 6-point rating scale.
Interclass correlation for ratings of children’s empathy was 0.86, and for the examiner’s intensity and quality of affective display was 1.00.
*
Results - Attachment Organization
All children showed indications of attachment in the Strange Situation.
Ainsworth classification:
17 secure.
12 insecure.
Main and Solomon indices:
7 primarily secure (B).
5 relationally disorganized (D-Attachment).
17 D-Autism.
One child was coded as “cannot classify” and included in the insecure group.
Small sample size precluded three-way (ABC) and four-way (ABCD) analyses.
Children were divided into two groups: secure and organized insecure/disorganized.
Final sample: 13 secure (45%) and 16 insecure (55%) children.
Results - Cognitive Correlates
Four independent samples t-tests compared secure and insecure classifications for NVMA and language skills.
Bonferroni adjustment: adjusted alpha level of 0.013.
Children with a secure attachment classification exhibited significantly higher language skills at intake t(27) = 3.45, p = .005 and follow-up t(27) = 3.19, p = .008.
The two groups did not differ regarding NVMA at intake t(27) = 2.76, p = .02 or follow-up t(27) = 1.85, p = .08.
However, a trend is evident in the results, and the secure children at initial assessment had a higher NVMA than the other infants even at the 1 year follow-up.
Results - Parent-Child Interaction
Maternal sensitivity was correlated with children’s language skills r(29) = .42, p = .02.
ANCOVA was used to analyze the association between sensitivity and attachment classification, with language abilities as the covariate.
Maternal sensitivity ratings for the securely attached group were significantly higher than maternal sensitivity ratings for the insecurely attached group F(1,26) = 12.15, p = .002.
When disorganization was taken into account, caregiver sensitivity did not differ between groups F(1,26) = 3.67, p = .12.
Securely attached children were significantly more likely than the others to initiate interactions with their parent F(1,26) = 9.01, p = .006, partial eta squared = .26.
Securely attached children were also significantly more responsive to the parents’ social initiations F(1,26) = 9.35, p = .005, partial eta squared = .27.
Results - Children's Empathy
Children’s empathy ratings at follow-up were correlated with intake ratings r(29) = .59, p = .001 and initial language abilities r(29) = .64, p = .001.
A one-way repeated measures ANCOVA was used to examine the unique contribution of attachment security, with time as a within-subject factor, attachment security as a between-subjects factor, and initial language skill as the covariate.
The ANCOVA yielded a significant time by group interaction F(1,25) = 8.88, p = .006, partial η2 = .26.
Empathy ratings for the securely attached group at follow-up were significantly higher than their ratings at intake F(1,25) = 13.07, p = .001, whereas there was no change for the other children F(1,25) = .70, p = .41.
Discussion
The study provides evidence that the sequelae of attachment security in autism may be similar to those reported for typically developing children.
Children classified as securely attached were judged as more empathic during a follow-up.
Maternal sensitivity was associated with attachment security when excluding disorganization.
Securely attached children had higher language abilities and directed more initiations to caregivers.
Validity of Attachment Classifications
Need to evaluate whether attachment classifications have the same developmental sequelae for children with autism as found in normative samples.
Securely attached infants may be better attuned to others’ emotional needs.
Securely attached children were judged to be more empathic during the follow-up visit.
Attachment security was linked to gains in empathy above and beyond the contribution of language.
Dynamics Underlying Attachment Organization
Associations were found between attachment classification and children’s and parents’ behavior in a separate social interaction.
Parents of children classified as securely attached were more sensitive than parents of children classified as insecure
Limitations and Future Research
Caregiver sensitivity and attachment security were assessed concurrently, precluding causal inferences.
A longitudinal study of infants at risk for autism is needed.
Longer-term follow-up is needed to establish the causal role of attachment security.
Modest sample size precluded examining correlates and consequences of individual differences in attachment security at finer resolutions.
Researchers may also wish to examine attachment in the home setting and to interview the parent (or other primary caregiver) to obtain more specific information regarding which behaviors may be attachment related for the child with autism and which may be behaviors exhibited in the child’s general functioning.
Conclusion
The researchers were able to find a number of different correlations from the experiment performed.
They hope that future researchers will examine the limitations of their experiment and improve upon it.
Early clinical observations suggested that autism involves a failure to develop typical attachments, which are crucial for social and emotional development.
Empirical research, however, indicates that children with autism do form attachments, and a significant portion of them develop secure attachments. This challenges earlier assumptions and highlights the need for a deeper understanding of attachment processes in autism.
A key research task is understanding how children with autism form secure attachments, given their differences in social understanding and motivation. These differences may influence how they interact with caregivers and develop attachment bonds.
This study investigates the correlates and implications of attachment security in children with autism, aiming to identify factors that contribute to secure attachment and its impact on development.
Goals of the Study
Investigate the correlates of attachment organization quality in children with autism.
Maternal sensitivity: Examining how a mother's responsiveness and attunement to her child's needs affect attachment security.
Child cognitive and social skills: Assessing the relationship between a child's cognitive abilities, social skills, and attachment security.
Examine the implications of attachment security for subsequent development.
Focus on children’s response to another person’s apparent distress: Investigating whether secure attachment is associated with increased empathic responding in children with autism.
Attachment Organization in Autism
Early studies show that children with autism exhibit discriminative attachment behaviors, indicating they can differentiate and respond differently to caregivers compared to strangers.
Looks, touches, and vocalizations directed towards caregivers: Observing how children with autism use these behaviors to engage with and seek comfort from their caregivers.
Proximity-seeking upon reunion after separation: Assessing whether children with autism seek closeness to their caregivers after a period of separation, a key indicator of attachment.
Preschool-age children with autism use caregivers as a secure base and haven of safety, exploring their environment while maintaining a connection with their caregivers.
Toddlers with autism show distress and search behaviors upon separation from parents, indicating they experience separation anxiety and seek to be reunited with their caregivers.
Studies using the Ainsworth Strange Situation procedure classify 40-60% of preschool children with autism as securely attached, suggesting that a significant proportion of children with autism can form secure attachment bonds.
Challenges in Assessing Attachment Security
Differentiating stereotyped and repetitive behaviors from relationally based disorganization is a challenge, as these behaviors can mask or mimic attachment-related behaviors.
Disorganized attachment is coded when behaviors are not explicable in other terms or stem from conflict related to alarm caused by the caregiver, indicating a breakdown in the child's attachment system.
In the first study to include the D classification, children classified with the Ainsworth tripartite system also showed behaviors in the Main and Solomon indices, providing further evidence for the validity of the D classification.
Children classified as alternate secure showed no signs of disorganization apart from repetitive behaviors, suggesting these behaviors were not relationally based and did not indicate attachment insecurity.
Mothers of alternate secure children were rated as more sensitive, indicating that maternal sensitivity may play a role in distinguishing between disorganized and alternate secure attachment patterns.
Differentiating Disorganization from Autistic Stereotypies
Empirical evidence suggests attachment disorganization can be differentiated from autistic stereotypies, allowing for a more accurate assessment of attachment security in children with autism.
Willemsen-Swinkels et al. (2000) first assigned a secure/avoidant/resistant classification, then classified each child as disorganized or not, excluding stereotyped behaviors, providing a method for distinguishing between attachment disorganization and autistic stereotypies.
Disorganized children showed greater average heart rate changes during separation and reunion, suggesting that attachment disorganization is associated with physiological stress responses.
These findings suggest the disorganized attachment classification can be assigned with expectable correlates when autistic stereotypies are excluded, improving the validity of attachment assessments in autism.
Maternal Behavior and Child Abilities as Correlates
Mothers of securely attached preschoolers with autism are rated as more sensitively responsive, indicating that maternal sensitivity is an important factor in promoting secure attachment in children with autism.
One study did not find a link between sensitivity and attachment security in toddlers with autism, possibly due to methodological differences, such as the age of the children or the specific measures used.
Koren-Karie et al. (2009) reported that the relationship between maternal sensitivity and attachment security holds after controlling for children’s developmental functioning, suggesting that maternal sensitivity has a unique contribution to attachment security, independent of a child's developmental level.
Siller et al. (2014) found that an intervention improving caregiver sensitivity led to increased attachment behaviors, providing evidence that interventions targeting maternal sensitivity can improve attachment security in children with autism.
Parental Insightfulness and Attachment
Oppenheim et al. (2009) found that a parent’s insightfulness into their child’s experience increases the likelihood of secure attachment.
This has been demonstrated for typically developing children (Koren-Karie, Oppenheim, Dolev, Sher, & Etzion-Carasso, 2002) and children with medical diagnoses (Marvin & Pianta, 1996).
The link between maternal insightfulness and attachment security in autism is mediated by maternal sensitivity, suggesting that insightfulness may influence attachment security by promoting sensitive parenting behaviors.
Disorganized Attachment and Maternal Sensitivity
Organized insecurity (avoidant and resistant) and disorganization are often independent, suggesting that different factors may contribute to these different forms of attachment insecurity.
Disorganized/secure (D/B) behavior can emerge when a caregiver is predominantly sensitive but occasionally engages in alarming behavior, indicating that even sensitive caregivers can sometimes exhibit behaviors that lead to attachment disorganization.
Van IJzendoorn, Schuengel, and Bakermans-Kranenburg (1999) found no significant relationship between disorganization and maternal sensitivity, suggesting that factors other than maternal sensitivity may play a more significant role in attachment disorganization.
Child Cognitive and Social Skills
Securely attached children with autism have better receptive language and play skills, indicating that secure attachment may promote cognitive and social development.
They make more social initiations during play with caregivers, suggesting that secure attachment may foster social engagement and interaction skills.
They are more responsive to joint attention bids and direct more requesting bids, indicating that secure attachment may enhance communication and social understanding.
Links between severity of autism symptoms and attachment security are equivocal, suggesting that the relationship between autism symptoms and attachment security is complex and may depend on other factors.
Disorganization is linked with intellectual disability in autism, indicating that attachment disorganization may be associated with cognitive impairments.
Meta-Analysis Findings
De Wolff and Van IJzendoorn (1997) found a moderate association (r = .24) between sensitivity and secure attachment, highlighting the importance of caregiver sensitivity in promoting secure attachment.
The association was weaker in clinical samples, suggesting that the relationship between sensitivity and secure attachment may be more complex in clinical populations like children with autism.
Caregiver sensitivity must be considered within the context of parent-infant interactive processes, emphasizing the importance of understanding the dynamic interplay between caregiver and child in shaping attachment security.
Developmental Consequences of Attachment Security
Research is needed to investigate whether secure attachments predict subsequent socio-emotional functioning, as this has important implications for understanding the long-term impact of attachment security on development.
Securely attached children are rated as more socially competent, empathic, and popular, indicating that secure attachment may promote positive social and emotional development.
Increased empathic responsiveness has been documented in secure children, suggesting that secure attachment may foster empathy and social understanding.
Current Study - Empathy Assessment
The study assessed changes in children’s affect and looking time to an examiner expressing distress, providing insights into their empathic responses.
This paradigm elicits empathic responses from typically developing infants and children with autism, making it a useful tool for studying empathy in autism.
Early behavioral responses to distress are linked with later empathy expressions, suggesting that early empathic responses may predict later social and emotional development.
Current Study - Goals
Examine the continuity between attachment quality and later empathic responding, investigating whether early attachment security is associated with later empathy.
Hypothesis: Securely attached children would be rated as more empathically responsive, predicting a positive association between attachment security and empathy.
Examine cognitive-linguistic correlates of attachment security, exploring the relationship between attachment security and cognitive and language skills.
Hypothesis: Securely attached children would have better language skills, predicting a positive association between attachment security and language development.
Examine the role of caregiver sensitivity in the development of attachment, investigating whether caregiver sensitivity is associated with attachment security.
Hypotheses
Securely attached children would be more likely to initiate social interactions and be more responsive to caregiver bids, predicting that secure attachment is associated with increased social engagement and responsiveness.
Methods - Participants
Participants were recruited through the UCLA Autism Evaluation Clinic (1997-2000), ensuring a sample of children with autism.
40 families were enrolled; 30 seen in the Ainsworth Strange Situation Procedure, providing a sufficient sample size for attachment assessment.
Clinical diagnoses were confirmed using ADI-R and ADOS-G, ensuring the accuracy of the autism diagnoses.
Sample: 23 boys, 7 girls; mean age 47 months (SD = 9), providing demographic information about the sample.
Average NVMA: 31 months (SD = 14); average language age: 21 months (SD = 13), providing information about the cognitive and language abilities of the sample.
Follow-up NVMA: 40 months (SD = 21); follow-up language age: 27 months (SD = 17), providing information about the cognitive and language abilities of the sample at follow-up.
Sample was heterogeneous in terms of functioning, ensuring a diverse representation of children with autism.
28 children met criteria for autism; 2 for PDD-NOS (likely ASD in DSM-5), providing information about the specific diagnoses of the participants.
Methods - Procedures
Initial assessment: two laboratory sessions.
Measures of language and cognitive ability: Assessing the children's cognitive and language skills.
Assessment of empathy: Evaluating the children's empathic responses.
Ainsworth’s Strange Situation Procedure: Assessing the children's attachment security.
Third session: home, child-caregiver interactions videotaped, providing observational data on parent-child interactions.
Follow-up: approximately 1 year later (mean interval = 12.3 months, SD = 1.5 months).
All assessments re-administered except for the strange situation: Assessing changes in cognitive, language, and empathic skills over time.
Methods - Strange Situation Procedure
Eight standard episodes of increasing stress.
Exposure to unfamiliar room and a stranger: Assessing the child's response to novelty and social interaction with a stranger.
Caregiver presence and absence: Assessing the child's attachment behavior in response to the caregiver's presence and absence.
Two separations and reunions with the parent: Assessing the child's separation anxiety and reunion behaviors, key indicators of attachment security.
Classification based on reaction to reunions with the parent.
Proximity seeking: Assessing whether the child seeks closeness to the caregiver upon reunion.
Contact maintenance: Assessing whether the child maintains contact with the caregiver upon reunion.
Avoidance: Assessing whether the child avoids the caregiver upon reunion.
Resistance: Assessing whether the child resists contact with the caregiver upon reunion.
Videotaped with cameras in three corners of the playroom, allowing for detailed coding of attachment behaviors.
Methods - Classification of Attachment
Erik Hesse and Mary Main classified attachment organization, ensuring expertise in attachment coding.
Ainsworth’s system (Ainsworth et al., 1978): Using the standard Ainsworth system for classifying attachment patterns.
Infant disorganized/disoriented classification (Main & Solomon, 1990): Assessing attachment disorganization in the children.
6-year-old system for coding (formerly disorganized) children as controlling (Main & Cassidy, 1988): Assessing controlling behaviors in children previously classified as disorganized.
Hesse and Main had experience classifying attachment in children with autism, ensuring their ability to accurately code attachment in this population.
Both coded each child’s strange situation independently; disagreements resolved through discussion, ensuring reliability of the attachment classifications.
Coding Considerations for Autism
Observers should consider the individual’s baseline behavior to differentiate between disorganized behavior and typical autistic behaviors, ensuring the accurate coding of attachment disorganization.
Criterion: whether behaviors were shown in connection with the mother or indiscriminately, helping to distinguish between attachment-related behaviors and autistic stereotypies.
Stereotypic hand flapping, squealing, or perseverative treatment of objects were excluded, avoiding misclassification of autistic behaviors as attachment disorganization.
D-Autism: behaviors readily explicable in terms of autism, classifying behaviors that are primarily related to autism.
D-Attachment: behaviors explicable in terms of breakdowns in the attachment system, classifying behaviors that are primarily related to attachment disorganization.
Some stereotypic behaviors contributed to a D-Attachment score if they occurred on reunion and interrupted the child’s approach, recognizing that certain stereotypies can interfere with attachment behaviors.
Behaviors like twirling and tip toe walking were excluded as unique to autistic children, avoiding misclassification of unique autistic behaviors as attachment disorganization.
Methods - Parent-Child Interaction
Parents were told to interact with their children as they typically would during play, ensuring naturalistic observation of parent-child interactions.
Videotaped during a separate visit to the families’ homes.
Unstructured play (15 min): Observing spontaneous parent-child interactions.
Structured play (15 min): Observing parent-child interactions during a structured activity.
Ratings of parental sensitivity were undertaken across the 30 min of play interaction by two independents observers, blind to the child’s attachment classification, and using Ainsworth’s original sensitivity–insensitivity scale (Ainsworth, Bell, & Stayton, 1974; Ainsworth et al., 1978).
Interclass correlation: 0.83, ensuring reliability of the sensitivity ratings.
Mother–Child Rating Scale (Crawley & Spiker, 1982) was used to rate the quality of the children’s social initiations and social responsiveness, both rated on a 5-point Likert scale.
Interclass correlations: 0.91 for social initiative and 0.89 for social responsivity, ensuring reliability of the social initiative and responsivity ratings.
Methods - Assessments of Cognitive Skills
Nonverbal cognitive abilities were assessed using two subscales of the Mullen Scales of Early Learning (MSEL; Mullen, 1995).
Visual Reception Scale: Assessing visual processing skills.
Fine motor scale: Assessing fine motor skills.
Age equivalents averaged to generate NVMA: Providing a measure of nonverbal cognitive ability.
For one child, NVMA was based on two nonverbal subscales of the Stanford–Binet Intelligence Scale (Pattern Analysis and Bead Memory; Thorndike, 1972), providing an alternative measure of nonverbal cognitive ability.
Three different language tests were administered, based on the child’s functional level.
Reynell Developmental Language Scales (RDLS; Reynell, 1983): Assessing language skills in children with lower functional levels.
Childhood Evaluation of Language Fundamentals – Revised (CELF-R; Semel, Wiig, & Secord, 1987): Assessing language skills in children with higher functional levels.
MSEL (Mullen, 1995; Expressive and Receptive Language Scales): Assessing language skills across a range of functional levels.
Age equivalents for receptive and expressive language skills were averaged to compute an overall language age equivalent, providing a comprehensive measure of language ability.
Methods - Assessment of Empathy
Children’s responsiveness to the examiner’s display of apparent distress was observed using a structured empathy task, providing a standardized assessment of empathy.
Child and examiner seated at a table; examiner pretends to hurt her finger, creating a scenario to elicit empathic responses.
Examiner displays vocal and facial expression of distress for 30 s, followed by 10 s of neutral affect, providing a controlled stimulus for assessing empathy.
The child’s degree of interest and concern was coded using a 6-point rating scale, providing a quantitative measure of empathy.
Interclass correlation for ratings of children’s empathy was 0.86, and for the examiner’s intensity and quality of affective display was 1.00., ensuring reliability of the empathy ratings and standardization of the distress display.
Results - Attachment Organization
All children showed indications of attachment in the Strange Situation, confirming that children with autism do form attachments.
Ainsworth classification.
17 secure: Identifying the number of children classified as securely attached.
12 insecure: Identifying the number of children classified as insecurely attached.
Main and Solomon indices.
7 primarily secure (B): Identifying the number of children classified as primarily secure according to the Main and Solomon indices.
5 relationally disorganized (D-Attachment): Identifying the number of children classified as relationally disorganized.
17 D-Autism: Identifying the number of children classified as D-Autism.
One child was coded as “cannot classify” and included in the insecure group, accounting for all participants in the analysis.
Small sample size precluded three-way (ABC) and four-way (ABCD) analyses, limiting the statistical analyses that could be performed.
Children were divided into two groups: secure and organized insecure/disorganized, simplifying the analysis due to the small sample size.
Final sample: 13 secure (45%) and 16 insecure (55%) children, providing the final group sizes for analysis.
Results - Cognitive Correlates
Four independent samples t-tests compared secure and insecure classifications for NVMA and language skills, assessing the relationship between attachment security and cognitive abilities.
Bonferroni adjustment: adjusted alpha level of 0.013, controlling for multiple comparisons.
Children with a secure attachment classification exhibited significantly higher language skills at intake t(27) = 3.45, p = .005 and follow-up t(27) = 3.19, p = .008, indicating a positive association between secure attachment and language development.
The two groups did not differ regarding NVMA at intake t(27) = 2.76, p = .02 or follow-up t(27) = 1.85, p = .08, suggesting no significant relationship between attachment security and nonverbal cognitive abilities.
However, a trend is evident in the results, and the secure children at initial assessment had a higher NVMA than the other infants even at the 1 year follow-up, suggesting a possible trend towards higher nonverbal cognitive abilities in the securely attached group.
Results - Parent-Child Interaction
Maternal sensitivity was correlated with children’s language skills r(29) = .42, p = .02, indicating a positive association between maternal sensitivity and language development.
ANCOVA was used to analyze the association between sensitivity and attachment classification, with language abilities as the covariate, controlling for the influence of language abilities on the relationship between sensitivity and attachment security.
Maternal sensitivity ratings for the securely attached group were significantly higher than maternal sensitivity ratings for the insecurely attached group F(1,26) = 12.15, p = .002, indicating that secure attachment is associated with higher maternal sensitivity.
When disorganization was taken into account, caregiver sensitivity did not differ between groups F(1,26) = 3.67, p = .12, suggesting that the relationship between caregiver sensitivity and attachment security may be more complex when considering attachment disorganization.
Securely attached children were significantly more likely than the others to initiate interactions with their parent F(1,26) = 9.01, p = .006, partial eta squared = .26, indicating that secure attachment is associated with increased social initiation.
Securely attached children were also significantly more responsive to the parents’ social initiations F(1,26) = 9.35, p = .005, partial eta squared = .27, indicating that secure attachment is associated with increased social responsiveness.
Results - Children's Empathy
Children’s empathy ratings at follow-up were correlated with intake ratings r(29) = .59, p = .001 and initial language abilities r(29) = .64, p = .001, indicating that empathy is associated with both prior empathy and language abilities.
A one-way repeated measures ANCOVA was used to examine the unique contribution of attachment security, with time as a within-subject factor, attachment security as a between-subjects factor, and initial language skill as the covariate, controlling for the influence of language abilities on the relationship between attachment security and empathy over time.
The ANCOVA yielded a significant time by group interaction F(1,25) = 8.88, p = .006, partial η2 = .26, indicating that the change in empathy over time differs between the secure and insecure groups.
Empathy ratings for the securely attached group at follow-up were significantly higher than their ratings at intake F(1,25) = 13.07, p = .001, whereas there was no change for the other children F(1,25) = .70, p = .41, indicating that secure attachment is associated with an increase in empathy over time.
Discussion
The study provides evidence that the sequelae of attachment security in autism may be similar to those reported for typically developing children, suggesting that the benefits of secure attachment may generalize across populations.
Children classified as securely attached were judged as more empathic during a follow-up, indicating that secure attachment may promote empathy in children with autism.
Maternal sensitivity was associated with attachment security when excluding disorganization, highlighting the importance of maternal sensitivity in promoting secure attachment.
Securely attached children had higher language abilities and directed more initiations to caregivers, indicating that secure attachment may be associated with improved language skills and social engagement.
Validity of Attachment Classifications
Need to evaluate whether attachment classifications have the same developmental sequelae for children with autism as found in normative samples, ensuring that attachment classifications are valid and meaningful in this population.
Securely attached infants may be better attuned to others’ emotional needs, suggesting a potential mechanism through which secure attachment promotes empathy.
Securely attached children were judged to be more empathic during the follow-up visit, providing evidence for the link between attachment security and empathy.
Attachment security was linked to gains in empathy above and beyond the contribution of language, suggesting that attachment security has a unique contribution to empathy, independent of language abilities.
Dynamics Underlying Attachment Organization
Associations were found between attachment classification and children’s and parents’ behavior in a separate social interaction, providing evidence for the behavioral correlates of attachment security.
Parents of children classified as securely attached were more sensitive than parents of children classified as insecure, highlighting the role of parental sensitivity in shaping attachment security.
Limitations and Future Research
Caregiver sensitivity and attachment security were assessed concurrently, precluding causal inferences, limiting the ability to determine the direction of the relationship between caregiver sensitivity and attachment security.
A longitudinal study of infants at risk for autism is needed, allowing for the examination of the development of attachment security over time.
Longer-term follow-up is needed to establish the causal role of attachment security, determining the long-term impact of attachment security on development.
Modest sample size precluded examining correlates and consequences of individual differences in attachment security at finer resolutions, limiting the ability to detect more subtle relationships between attachment security and other variables.
Researchers may also wish to examine attachment in the home setting and to interview the parent (or other primary caregiver) to obtain more specific information regarding which behaviors may be attachment related for the child with autism and which may be behaviors exhibited in the child’s general functioning, providing a more comprehensive understanding of attachment-related behaviors in children with autism.
Conclusion
The researchers were able to find a number of different correlations from the experiment performed, providing valuable insights into attachment security in children with autism.
They hope that future researchers will examine the limitations of their experiment and improve upon it, advancing the field of attachment research in autism. - They hope that, armed with this data,