Inge Bretherton
Human Development & Family Studies, University of Wisconsin-Madison, Wisconsin, USA
This paper provides a brief history of attachment research on fathers as a backdrop against which the other contributions to this volume can be viewed. Empirical research on childâfather attachment progressed in four phases and began before Bowlby in 1969 published the first volume of his attachment trilogy. During each phase a different set of questions emerged. Initially, researchers compared fathers to mothers as potential attachment figures. More recent studies emphasize the notion that mothers' and fathers' roles as attachment figures and their influences on child outcomes may be different and complementary. At the same time, calls for a family approach to attachment studies are increasing.
Introduction
In his first still tentative formulation of attachment theory, Bowlby (1958) made no mention of fathers as potential attachment figures. His primary aim was to âdebunkâ the notion held by his psychoanalytic colleagues that infants love their mothers because mothers provide oral gratification. Instead, taking his cue from the emerging ethological literature on parentâoffspring bonds in birds and nonhuman primates, Bowlby proposed an evolutionary explanation of human infantâmother attachment. Attachment behaviour, he theorised, is adaptive because it âmaintains a younger more vulnerable individual in more or less close proximity to another discriminated and stronger individualâ (Bowlby, 1979, p. 129) who can provide protection when needed. Bowlby invented the term âmonotropyâ to designate infants' tendency to direct attachment behaviours to one specific individual (the mother or mothering person).
Bowlby's thinking about the role of fathers as attachment figures evolved over time in line with the publication of relevant research findings. Although considerably more attention in the attachment field has been given to mothers, interest in fathers emerged very early in the development of attachment theory. To provide a historical background for the articles in this special issue, I delineate what I see as four phases in father attachment research. Each phase was dominated by specific questions and was associated with changes in the conceptualisation of childâfather attachment by Bowlby and others. Because of scant research on studies of father-attachment in different family structures and father-attachment in non-industrialised societies these important topics are not included in this review.
Two related research questions were asked during Phase 1: âCan fathers serve as attachment figures, and if so, is fathers' role secondary or equal to mothers'?â The answer to the first question was a clear-cut yes, but the question on fathers' place in a hierarchy of attachment figures was not fully resolved. In Phase 2, research continued to focus on fathers' place in an attachment hierarchy, but towards the end of the phase this topic was left aside in favour of a new question that signalled the transition to Phase 3. Two questions dominated Phase 3: âWhat is the comparative quality (security) of an infant's attachment to mother and father?â The second question that dominated Phase 3 was: âAre intergenerational relationship qualities independently transmitted from father and mother to child?â Interest in intergenerational issues was precipitated by the development of representational assessments of attachment for parents and children. Initial findings from longitudinal research with both parents suggested that the quality of infantâmother and childâmother attachment had the greatest or even sole impact on children's subsequent socio-emotional development, but this view changed during the next phase of father attachment studies. In Phase 4, results from longitudinal investigations in the USA, Germany, Great Britain and Israel that had followed children's development from birth to adolescence and young adulthood caused researchers to ask more nuanced questions: âAre the developmental outcomes of father and mother attachment different, even if both relationships were secure in infancy and childhood?â and âAre the same assessments appropriate for the study of attachment to mother and father?â At the same time, there were calls for a family perspective on attachment relationships.
Phase 1
Findings from the first two empirical studies relevant to infantâfather attachment became available while Bowlby was working on his second, more elaborated version of attachment theory (published in 1969). These studies were conducted by individuals who had been members of Bowlby's research team at the Tavistock Clinic in London during the 1950s. Both documented infants' attachments to family members other than the mother, and led Bowlby to rework his earlier definition of âmonotropyâ.
The first study (Ainsworth, 1963, 1967) was based on naturalistic home observations of 26 motherâinfant pairs in Uganda. Ainsworth carefully described the sequential emergence of attachment behaviours that infants around six months of age began to direct towards their mothers in preference to others (e.g. separation protest, retreating and clinging to mother when afraid, greeting mother after an absence, calming when held by mother, following mother when she left the room and using her as a base for exploration). Ainsworth found that differential mother-directed responses were fairly quickly followed by differential attachment behaviour towards a small number of other figures, including father, grandmother, co-wife, even sibling. Attachment to fathers seemed to be especially common, even in babies who did not see their fathers often. One of the 26 infants showed attachment behaviour exclusively to the father and three others were said to prefer the father as attachment figure over the mother. However, in most cases the mother was preferred, and this was especially striking when babies were tired, hungry or ill. At the same time, babies were likely to accept comforting from specific secondary figures if the mother was not present. Ainsworth summed up her findings with the following statement:
It is clear that infants during their first year of life may establish several attachments and a complex set of interpersonal relations. Our stereotype of the infant developing an attachment to the mother and to the mother alone during the first year is not borne out by these observations, despite evidence that when the chips are down the attachment to the mother usually seems to be the focal one. (1967, p. 356)
The second relevant study was conducted by Schaffer and Emerson (1964) in Scotland. Their objective was to ascertain the onset and describe the further development of separation anxiety directed to specific âattachment objectsâ across infants' first year of life. Findings were based on monthly home interviews of 60 mothers, supplemented by informal observations. Whereas a majority of the Scottish mothers reported that their infants exhibited separation anxiety to them first, most of these infants soon adopted a small number of additional attachment objects. Using the intensity of separation protest (e.g. a âfull-bloodedâ cry versus a whimper or a moan) as the criterion, Schaffer and Emerson identified the mother as the âprincipal attachment objectâ for 80% of the infants during the month when directed separation protest was first reported. However, by 18 months only half of the mothers were still characterised as infants' sole principal objects. In many of the remaining families, mothers and fathers were said to fill the âprincipalâ role jointly. In 10 of the 60 families, the father was identified as 18-month-olds' sole principal object. As had also been the case in Uganda (Ainsworth, 1967), responsiveness rather than physical care or time spent with the infant seemed to guide the infants' choice of âprincipal objectâ.
Bowlby reviewed these two studies in considerable detail in his 1969 volume (pp. 199â204, 306â308), and adopted Schaffer and Emerson's notion of an attachment hierarchy with mother at the apex. He retained the term monotropy (p. 209), but redefined it to denote infants' tendency to seek out a principal attachment figure (if present) in preference to a small number of subsidiary figures. At the same time, Bowlby seemed to discount Schaffer and Emerson's rankings of fathers and mothers as co-principal attachment figures in favour of Ainsworth's report that infantâmother attachment was the focal one in most cases. He contended that the two studies had failed to distinguish between fathers as preferred attachment figures and preferred playmates. Children seek playmates, he said (1969, p. 307), when they are in good spirits and know where their attachment figure can be found, but seek the attachment figure when they are tired, hungry, ill, alarmed or uncertain of the figure's whereabouts. Possibly, paternal playmates might have been miscounted as principal rather than subsidiary attachment figures, confounding the reported findings. In addition, Bowlby expressed reservations about Schaffer and Emerson's use of intense separation anxiety as a criterion for identifying the principal figure, because Ainsworth had discovered intense separation anxiety to be a hallmark of insecure infantâmother relationships in her Ganda study (see Bowlby, 1969, pp. 304â309).
Phase 2
Given the prominence Bowlby accorded to the Phase 1 studies in his 1969 reformulation of attachment theory, it is somewhat surprising that subsequent father-attachment research progressed only in fits and starts. One reason may be that the attention of developmental researchers was diverted by contentious findings obtained with the Strange Situation Procedure (SSP; Ainsworth & Wittig, 1969), a laboratory assessment of infantâmother attachment. The SSP consists of a 20-minute sequence of episodes during which one-year-olds encounter a stranger and experience two separations from and reunions with the mother. Patterns of 12-month-olds' attachment behaviour to the mother, especially during the two SSP reunions, turned out to be significantly correlated with prior ratings of mothers' observed sensitivity and responsiveness in several caregiving contexts at home (e.g. crying, feeding, holding and face-to-face play) throughout infants' first year of life (Ainsworth, Bell, & Stayton, 1974; see also Ainsworth, Blehar, Waters, & Wall, 1978). Because these observations were conducted while fathers were at work, fatherâinfant interactions were not assessed. Most of the mothers, as usual during the 1960s, were not employed outside the home.
The initial SSP findings led to an enormous upsurge of infantâmother attachment studies during the 1970s. Many of these had the goal of confirming or disconfirming the SSP's validity as a short-cut assessment of infantâmother attachment quality, but there were no immediate attempts to use the SSP to investigate the security of father attachment. Instead, a handful of researchers designed experimental laboratory procedures with multiple trials to compare infant attachment behaviour to mother, father and stranger (e.g. Cohen & Campos, 1974; Kotelchuck, 1972). The pace of father-attachment research quickened when Michael Lamb began a series of systematic studies to resolve issues raised by the attachment hierarchy hypothesis. In his first study, Lamb (1976) found that infants showed similar levels of approach and contact-seeking to mothers and fathers during a series of laboratory episodes in which either one or both parents were present, but fathers were the recipients of more affiliative behaviours (smiling, vocalising, proffering toys) than mothers. However, when a stranger joined the two parents, âthere was a dramatic transformation of the infants' behaviourâ, as infants promptly shifted attachment and affiliative behaviour towards the mother (Lamb, 1976, p. 242). As in the two Phase 1 studies, stress heightened the visibility of mothers as principal attachment figures.
Next, Lamb (1977a) conducted unstructured home observations when infants were 7â13 months old. At four age-points, infants approached and sought physical contact equally from both parents, although mothers held infants more for caregiving routines and fathers held them more for play. As in the laboratory, infants directed more affiliative behaviour towards fathers. During a similar study with 15â24-month-olds, toddlers showed both more affiliative and more attachment behaviour to the father than the mother (Lamb, 1977b). In addition, boys were âin proximity of, approached, and fussed to, their fathers more than the girls while the latter were in proximity of, and fussed to, their mothers more than the boysâ (p. 643). Finally, fathers were more active than mothers, particularly with their sons. Based on these findings, Lamb argued that fatherâinfant and motherâinfant relationships may involve different kinds of experiences for infants, resulting in differential influences on children's personality development from infancy onward. This topic was not broached again by other attachment researchers until Phase 4. Note that Lamb did not distinguish infants' approaches to a parent for play from approaches for comforting, hence the preference for the father during home observations may, in part, be due to his role as play figure. In his fourth study (1978), Lamb asked a new question that signalled the transition to Phase 3.
Phase 3
Rather than continuing to evaluate the relative frequency of infants' attachment and affiliative behaviour to mother and father under stress and non-stress conditions, Lamb decided to find out whether the secure, avoidant and ambivalent SSP patterns of motherâinfant attachment that Ainsworth and Wittig (1969) had identified in the Baltimore study could be replicated with fathers. Whereas group comparisons revealed no motherâfather difference in the distribution of the three SSP patterns at 12 months, many infants who were judged insecure with the mother were classified as secure with the father and vice versa. This challenged Ainsworth's (1967, p. 356) assumption that insecure attachment to the mother might prevent an infant from developing attachments to other figures.
Main and Weston (1981) replicated Lamb's comparative SSP study with similar results. Finding no systematic match of infantâfather with infantâmother SSP patterns, the authors concluded that the SSP assesses the quality of an infant's distinct relationship to each parent, not infant personality or temperament as some had proposed. Main and Weston also reported that infants with two secure SSPs responded most favourably to a stranger's friendly invitation to play ball (in the presence of the mother). Infants with two insecure SSPs responded least favourably while those with discordant (one secure, one insecure) SSPs received intermediate ratings. Taken singly, the classification with mother was a stronger predictor, however.
These findings induced Bowlby to add a sentence to the concluding chapter of the second edition of Attachment (1982), not present in the 1969 edition. It underscores the similarity of mothers' and fathers' function as attachment figures:
A young child's experience of an encouraging, supportive, and cooperative mother, and a little later father, gives him a sense of worth, a belief in the helpfulness of others, and a favorable model on which to build future relationships ⊠By enabling him to explore his environment with confidence, and to deal with it effectively, such experience also promotes his sense of competence. (Bowlby, 1982, p. 378)
In the mid-1980s, a new twist was added to the comparative study of mother- and father-attachment relationships as new measures of attachment at the representational level became available. Before reviewing these studies, I will interject a brief review of Bowlby's theorising about the role of representation in attachment relationships that had already been laid out in the second volume of his attachment trilogy, Separation (1973, p. 203). Here Bowlby postulated that a child builds representatio...