Letters

Autism and Deficits in Attachment Behavior

See allHide authors and affiliations

Science  25 Feb 2005:
Vol. 307, Issue 5713, pp. 1201-1203
DOI: 10.1126/science.307.5713.1201

A. Moles et al. reported a “Deficit in attachment behavior in mice lacking the μ-opioid receptor gene” (Reports, 25 June 2004, p. 1983). They stated that their results “may indicate a molecular mechanism for diseases characterized by deficits in attachment behavior, such as autism,” an assumption recapitulated in Mary Beckman's news story “The mice that don't miss mom: love and the μ-opioid receptor” (News of the Week, 25 June 2004, p. 1888) and other news coverage.

However, the basis for D'Amato et al.'s speculation that autism is “characterized by deficits in attachment behavior” was a 25-year-old theoretical treatise (1), which subsequently failed to be supported by numerous empirical investigations. In 1984, Sigman and colleagues first demonstrated that young children with autism behaved no differently in their response after separation from and reunion with their primary caregiver than did children with other developmental disabilities (2). Following Sigman's seminal study, every laboratory experiment investigating attachment behavior with young children with autism has replicated the initial finding, demonstrating unambiguously that children with autism are as securely attached to their mothers as are their peers (38).

Given the striking lack of empirical evidence to support D'Amato and colleagues' speculation that their results are relevant to autism, we urge considerably greater caution.

References

  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.

Response

Gernsbacher and colleagues claim that our findings are not relevant to autism because autism is not characterized by deficits in mother-child attachment. To support their argument, these authors cite several studies showing that children with autism can form a secure attachment with their mothers.

Mice lacking the μ-opioid receptor gene are not a model of autism or a model of any other human clinical syndrome as described in current psychiatric classification systems. Rather, our findings could be useful for identifying a biological pathophysiology common to a variety of conditions that are currently classified in very different categories of psychiatric nosography. Current classification of psychiatric disorders is not based on measures of the underlying genetic or biological pathophysiology of the disorders (1). As a consequence, diagnostic categories often include heterogeneous populations, and their classification is not of great help in deciding if a core neurobehavioral deficit is present or not in a psychopathological condition.

The question of whether autistic children are able to form an attachment bond (secure or insecure) with their mothers cannot be answered unequivocally, as indicated by a recent review by Rutgers et al. (2). That a substantial proportion of autistic children (as diagnosed by current classification criteria) may develop secure attachments is not in contradiction with the clinical notion that many children and adults with autism show a marked reduction in the desire to engage others socially and that the opioid system could be implicated in the pathogenesis of these social deficits (3). Autism is not only a heterogeneous condition but is also best described as a continuum rather than as a category (especially its core component of defective reciprocal social behavior) (4).

In conclusion, we believe that our findings in mice lacking the μ-opioid receptor gene should be interpreted as an attempt to model a biological dysfunction that can be present in some (but not in all) cases of autism, as well as in other psychiatric conditions that have in common a deficit in the capacity to experience affiliative reward (5). Such conditions might include schizophrenia spectrum disorders and primary psychopathy (6).

References

  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.

Navigate This Article