The study reported here investigated how parents’ antenatal representations of
attachment experiences related to their MM during infant–parent interactions at 6
months postpartum, and to subsequent attachment security at 12 (mother–infant) and
15 (father–infant) months. Fathers with autonomous AAI classifications produced
proportionately more appropriate mind-related comments than their non-autonomous
group counterparts. No significant differences were found between autonomous and
non-autonomous group mothers, although the effect size for the relation between AAI
classification and inappropriate mind-related comments was medium, suggesting that
autonomous group mothers produced proportionately fewer inappropriate comments.
Parental RF during the AAI was related to MM in on-line infant–parent interactions in
both mothers and fathers. There was a negative correlation between RF and
inappropriate mind-related comments in mothers, but in fathers, there was a positive
correlation between RF and appropriate mind-related comments. Thus, mothers who
scored more highly on RF produced proportionately fewer inappropriate comments
about their 6-month-olds’ internal states, whereas fathers with higher RF scores
produced proportionately more appropriate mind-related comments.
The study reported here also investigated concordance in MM between
partners. There was some suggestion of concordance in partners’ proportional use of
appropriate mind-related comments (although this relation failed to reach
significance), but no concordance was seen for inappropriate mind-related comments.
The lack of significant concordance in MM between partners provides further support
for Meins et al.’s (2005) argument that MM is a relationship-specific construct.
A number of other findings are worthy of note. First, we replicated Fonagy et
al.’s (1991) finding that autonomous parents attained higher RF scores than their non-
Attachment and Mind-Mindedness 18
autonomous counterparts. Second, no concordance was found between partners’
dichotomous AAI classifications, replicating the results of Steele et al. (1996) on a
separate British sample. Third, partners’ RF scores did not concord with one another.
Fourth, comparing mothers and fathers as a whole, parental gender had no effect on
AAI classification, RF, or proportional use of appropriate mind-related comments, but
fathers produced proportionately more inappropriate mind-related comments than did
mothers. Fifth, fathers’ proportional scores for appropriate and inappropriate mindrelated
comments were positively correlated, whereas these scores were unrelated for
mothers. This lack of association in mothers replicates the findings of Meins et al.
(2002). Finally, we largely replicated previously observed relations between infant–
parent attachment security and (a) parental AAI classification (e.g., Fonagy et al.,
1991; Steele et al., 1996) and (b) MM (Lundy, 2003; Meins et al., 2001).
Before considering potential interpretations of these results, it is important to
note that it would be wise to treat the relations reported here as preliminary until they
have been replicated in a larger sample. That said, we addressed concerns over power
by reporting effect sizes, and the study is noteworthy in being the first to investigate
relations between parents’ attachment representations, MM, and infant–parent
attachment security in a sample of biologically-related parents and children. When
MM was assessed in such dyads using an age-appropriate measure, the relations
observed with parents’ AAI classification and RF were positive. Our findings
therefore contrast with those of Bernier and Dozier (2003) who found negative
associations between maternal MM and attachment security in foster families. The
results of our study do, though, support Bernier and Dozier’s (2003) argument that
their counter-intuitive findings arose from using a measure of MM that was not ageappropriate.
Attachment and Mind-Mindedness 19
The finding that the relation between parental AAI classification and MM is
stronger in fathers than in mothers intriguingly parallels the pattern of findings seen in
relations between AAI and parental sensitivity. Van IJzendoorn’s (1995) metaanalysis
showed that paternal AAI classification was more strongly related to fathers’
sensitivity compared with the relation between maternal AAI and sensitivity. Thus,
both in terms of responding sensitively and commenting appropriately on internal
states, fathers’ interactions with their infants are more strongly governed by
representations of their own attachment experiences than are infant–mother
interactions. Why might mothers be less influenced by their representations of
childhood attachment than fathers? One possibility is that mothers are likely to be
exposed to considerably more advice and information on how to be an effective and
responsive parent, and will therefore have a greater wealth of material on which to
draw. Alternatively, mothers tend to spend more time with their infants in the early
postpartum months, resulting in more opportunities to learn about their infants’ likes,
dislikes, interests, and moods. The latter alternative is thus consistent with the
argument, mentioned above, that MM is a relationship–specific quality and requires
extensive previous knowledge of the infant (Meins et al., 2005).
It is interesting to consider why RF was related to different indices of MM
depending on parental gender. Although the observed gender specificity of these
relations may be simply a result of our small sample size, it may be that there are
genuine differences between mothers and fathers with respect to how they engage in
mind-minded interaction with their infants. The fact that fathers were more likely than
mothers to produce inappropriate comments, coupled with the positive correlation
between appropriate and inappropriate mind-related comments seen only for fathers,
highlights differences between parents. These findings suggest that fathers may be
Attachment and Mind-Mindedness 20
less skilled than mothers in accurately reading their infants’ internal states. Future
research should attempt to verify whether such differences between mothers and
fathers are found in a larger sample of families more representative of the population
as a whole.
Finally, we return to the issue with which we began: intergenerational transfer
of security. Given our small sample, it was not possible to perform a full statistical
test of whether MM could mediate the relation between parents’ antenatal attachment
representations and infant–parent attachment security. However, the preliminary
analyses reported here suggest two main pathways from AAI to strange situation
classification: (a) 100% of infants whose mother or father was both autonomous and
high in MM were securely attached; and (b) 80% of infants whose mothers were nonautonomous
and low in MM, and 100% of infants whose fathers showed this pattern,
were insecurely attached. Although these findings cannot speak to the issue of
whether MM mediates the relation between parental and infant attachment, they do
show that certain combinations of AAI classification and level of MM appear to be
powerful predictors of infant–parent patterns of attachment.
The descriptive pathways from AAI classification to infant–parent attachment
security via parental MM also highlight two further points of interest. First, one could
argue that a lack of MM on the mother’s or father’s behalf is relatively unimportant if
the parent is autonomous, since 4 out of 5 of the attachment relationships in which
there was a combination of autonomous representation and low MM were secure.
Second, having a parent who manages to be mind-minded despite being nonautonomous
appears to give the dyad an improved chance of establishing a secure
attachment relationship than if a parental non-autonomous classification were coupled
with a lack of MM. Two of the 4 attachment relationships where the non-autonomous
Attachment and Mind-Mindedness 21
parents scored high in MM were secure, whereas 13 out of 14 attachment
relationships with a non-autonomous parent who scored low in MM were insecure. If
these findings can be replicated on a larger scale, they will provide important
information on potential intervention procedures that might help increase the chances
of an infant forming a secure attachment relationship despite the parent having nonautonomous
attachment representations. The findings of this preliminary study thus
provide suggestive evidence that the unique property of MM as a construct at the
interface between representation and behavior may enable it to help bridge the
transmission gap.