observed in typical development or children with general developmental delays. Recent
research by van IJzendoorn et al. (2007) also demonstrated that the attachment security of
children with high-functioning autism was not related to their parents’ sensitivity (in
contrast to the findings in typically-developing children), but that children with better
social functioning were more likely to be securely attached.
A meta-analysis (Rutgers, Bakermans-Kranenburg, van IJzendoorn & van
Berckelaer-Onnes, 2004) showed that children with autism and cognitive delay are less
likely to be classified as securely attached than typically-developing children. However,
children with high-functioning autism (HFA; autism spectrum disorder with IQ in the
average range or above) have normal levels of attachment security. This meta-analytic
result is unusual because there is little evidence that cognitive delay reduces attachment
security in children with other developmental disorders (Dissanayake & Crossley, 1996,
1997; van IJzendoorn, Goldberg, Kroonenberg & Frenkel, 1992; Willemsen-Swinkels,
Bakermans-Kranenburg, Buitelaar, van IJzendoorn & van Engeland, 2000). So why
should cognitive delay reduce attachment security in children with autism? One
suggestion put forward by several authors is that autism in itself may disrupt the formation
of secure attachments, but children with higher cognitive abilities are somehow able to
compensate for this disruption. This suggests that, although children with HFA are
successfully forming secure attachments, they may be doing so in a different way from
other children, using effortful cognitive strategies to compensate for difficulties in
automatic social and emotional responsiveness (Dissanayake & Sigman, 2001).
Understanding mental states in autism
The best-established measure of attachment in adulthood is the Adult Attachment
Interview (AAI; George, Kaplan & Main, 1996). It is essential in the context of this