difficult to classify, it can be assigned two or three alternative categories. Participants who
display a marked and unusual mixture of contradictory mental states are assigned to
“cannot classify”, regarded as an abnormal category on the assumption that adults
generally have a single integrated strategy or state of mind with respect to thinking and
talking about early attachments. In this case, a ‘forced’ classification into the most
appropriate organised category or categories is given, secondary to “cannot classify”.
Participants who show major lapses in monitoring of reasoning or discourse when
discussing traumatic events are assigned to a fourth category, “unresolved/disorganised”.
They are also given a best fitting alternative “organised” category.
All AAI transcripts in this study were scored on five probable experience scales
and eight state of mind scales. The control interviews were coded by an AAI coder trained
and certified as reliable (MT). Coding of the interviews with ASD participants was
divided between MT and another reliable coder. Possible scores on all scales range from 1
(little evidence of the characteristic being scored) to 9 (extreme evidence). In some cases
it is not possible to assign a scale score, particularly for experience scales, if the
participant does not report many specific memories. Of particular interest in the current
study are the two coherence scales, high scores on which are associated with the
secure/autonomous category: coherence of mind and coherence of transcript. The scoring
criteria describe coherence as “connectedness of thought, such that the parts of the
discourse are clearly related, form a logical whole, or are suitable or suited and adapted to
context” (Main & Goldwyn, 1998, p.44). Both coherence scores are strongly related to
overall attachment security, as they are heavily weighted in assigning secure status. All
AAI transcripts were also scored for reflective function (RF), which is the
operationalisation of mentalising specifically within the context of attachment. The RF
criteria were developed by Fonagy et al. (1991; 1997) and scores range from -1 to 9, with