Attachment in adults with autism has, to our knowledge, not been investigated before. It is
therefore necessary to rely on what is known about attachment in children with autism to
provide an empirical background to the current study. The earliest theories of autism
linked it directly with attachment, either claiming that inappropriate parenting actually
caused or exacerbated autism (Bettelheim, 1967; Mahler, 1968; Tinbergen & Tinbergen,
1972) or, conversely, that autism resulted in a fundamental inability to form attachments
(Kanner, 1943). As recently as DSM-III (APA, 1980), the mandatory diagnostic criteria
included “a pervasive lack of responsiveness to other people” (p.89), implying the absence
of any form of attachment. These ideas were based on the general impression that those
with autism are not particularly interested in other people, and often actively avoid
interactions and physical contact even when they are distressed.
However, from the 1980s, researchers began using the Strange Situation
(Ainsworth, Blehar, Waters & Wall, 1978) to explore the attachments of children with
autism spectrum disorders in a more systematic way (a number of related disorders are
clustered with autism in the DSM-IV-TR (APA, 2000); for simplicity I will use the term
autism to cover all of these disorders often described as being on the autism spectrum). A
series of studies demonstrated that children with autism showed active preferences for
their caregivers over a stranger, that they were distressed by separations from the caregiver
and sought proximity when they were reunited (Capps, Sigman & Mundy, 1994; Field,
1987; Rogers, Ozonoff & Maslin-Cole, 1991, 1993; Shapiro, Sherman, Calamari & Koch,
1987; Sigman & Mundy, 1989; Sigman & Ungerer, 1984). Sometimes these things were
done in atypical ways, and sometimes unusual behaviour was observed that appeared to be
unrelated to attachment, but it seemed to have been demonstrated conclusively that some
children with autism were capable of demonstrating secure attachment. The studies were
inconsistent in whether they found attachment security at a comparable rate to that