Patterns of Attachment: A Psychological Study of the Strange Situation - Mary D. Salter Ainsworth 2015
Discussion of Normative Issues
Despite our conviction that normative trends in behavior across episodes and that findings pertinent to individual differences in attachment and attachment behaviors throw light upon each other, it is convenient in our final discussion to focus first on one and then the other. In this chapter we discuss issues related to normative trends, and in Chapter 14 we turn to issues pertinent to individual differences. Nevertheless, where relevant, we refer to individual differences when discussing normative issues, and similarly we must on occasion refer back to normative trends when discussing individual differences.
Let us first consider the three sets of phenomena that the strange situation was primarily devised to examine: use of the mother as a “secure base” from which to explore, responses to a stranger, and responses to separation from the mother. In conjunction with these three topics, it is also convenient to consider the following: use of the mother as a “secure haven,” differential behavior to mother versus stranger, and the issue of interchangeability of attachment figures. Later we consider a variety of other issues: activation and termination of attachment behavior, the effect of the intensity of activation of the attachment system on the behaviors elicited, and the interplay between attachment behavior and other behavioral systems.
Exploratory Behavior and the Secure-Base Phenomenon
An unfamiliar or strange situation might be expected to activate three behavioral systems in varying degrees of strength: exploratory behavior, wary/fearful behavior, and attachment behavior. Exploratory behavior is antithetical to attachment behavior in that it leads the infant toward interesting features of his environment and thus usually away from the attachment figure. If, however, the baby is alarmed, attachment behavior as well as wary/fearful behavior tends to be activated (Bowlby, 1969, 1973), and commonly (although not invariably) these two systems work in concert. Thus behavior that promotes proximity to the attachment figure also tends to lead the baby away from the alarming stimulus or at least to reduce its impact (see Chapter 4). To the extent that exploratory behavior is activated more strongly than the other two systems in combination, a child could be expected to explore the new environment. Should the unfamiliar situation activate wariness/fear more strongly than the exploratory system, a child would not be expected to explore. Rather, he would be expected to direct attachment behavior toward his mother, for attachment behavior tends to be activated under circumstances of alarm. Episode 2 was intended to provide strong instigation to exploratory behavior through a massed array of toys. With the mother present and with the toys the most salient aspect of the unfamiliar environment, it was expected that the typical 1-year-old would approach the toys with little delay, rather than approach or signal his mother. The findings support these expectations. On the other hand, when the mother is absent (in Episodes 4, 6, and 7) nearly all infants explored much less than they had in Episode 2. This had been predicted to occur because attachment behavior is strongly activated under circumstances when the attachment figure is inaccessible and/or unresponsive. In the strange situation the mother’s departures and brief absences provided strong enough activation of attachment behavior to override even the strongly activated exploratory system. This is an aspect of the phenomenon that we have referred to as “using the mother as a secure base from which to explore” (Ainsworth, 1963, 1967).
We consider it important to view this phenomenon as reflecting the relative strength of activation of the relevant behavioral systems. It does not imply that the physical environment did not activate wariness/fear at all—merely that its novel features activated exploratory behavior more strongly. Similarly, it does not imply that the attachment system was altogether inactive in Episode 2 but merely that it was relatively less intensely activated. In the separation episodes, on the other hand, the fact that exploratory behavior tended to be overridden by attachment behavior does not imply that the exploratory system was totally deactivated. At such a time that attachment behavior was terminated (or sufficiently reduced in intensity), one might expect exploratory behavior to reemerge, as indeed it did to some extent in many infants in the reunion episodes. Furthermore, one ought to expect individual differences in the relative strengths with which the relevant behavioral systems are activated even in this standardized situation.
Cohen (1974) objected to a comparison of behavior in the mother-present and mother-absent conditions as demonstration of the secure-base phenomenon. She pointed out that separated infants often cry and that a crying baby does not explore; crying inhibits exploration. This is essentially another way of saying that when attachment behavior is activated strongly enough, it overrides exploratory behavior. When a baby uses his mother as a secure base from which to explore, his attachment behavior is not activated strongly enough to interfere with exploration; the mother’s presence is one of the conditions that operate to keep attachment behavior at low intensity.
Nevertheless it is conceivable that the mother’s mere presence might not be enough under some circumstances or in the case of some individuals to hold attachment behavior down to a lower level of activation than that of exploratory behavior. This was the case with some of the Ganda infants who were observed at home (Ainsworth, 1967). There are three points of differences that must be considered when comparing the Ganda and Baltimore samples. The first, and perhaps least important, is that the Ganda infants were observed in the familiar home environment, where there was no massive instigation to exploration provided by a novel array of toys. Second, the Ganda infants tended to be more intensely afraid of strangers, and the relatively unfamiliar observers were present throughout the observations. Third, some of the Ganda 1-year olds who seemed least able to use the mother as a secure base from which to explore had had their relationship with her disrupted by recent weaning from the breast (Ainsworth, 1967, pp. 456—457). These conditions, separately or together, might be expected to lead to relatively more intense activation of attachment behavior and/or relatively less intense activation of exploratory behavior. Individual differences are evident both in the home behavior of Ganda infants and in the strange-situation behavior of our Baltimore sample. These are not discussed here, except to point out that distress in Episode 2 emerged as a conspicuous feature in discriminating Group-C from non-C babies. It could be argued that a baby who is distressed, even minimally, in Episode 2 and who is presumably alarmed by the unfamiliar but benign situation that does not alarm the majority of other children does not experience his mother as a secure base.
Cohen’s (1974) discomfort with the concept of the secure base might be alleviated by the use Sroufe and Waters (1977b) made of heart-rate measures in conjunction with strange-situation behavior. Exploratory behavior was typically associated with heart-rate decelerations. All infants showed heart-rate acceleration in the separation episodes, including those who did not cry, and thus presumably including those who maintained some exploration in the separation episodes, albeit at a lower level than in Episode 2. Full recovery of exploratory behavior in the reunion episodes is associated with the disappearance of the acceleration characteristic of separation distress; and a return to basal tonic heart-rate levels is associated with the episodic decelerations characteristic of exploratory behavior. These findings throw light on the significance of both the lack of conspicuous distress in some children and the exploratory behavior of infants who avoid their mothers in the reunion episodes. With the convergent use of heart-rate and behavioral measures, we may become attuned to subtle behavioral clues that indicate when a baby is using his mother as a secure base from which to explore and when anxiety underlies his exploratory activity. We shall return to this point in the next chapter in the context of individual differences in the attachment relationship.
There is more to the secure-base phenomenon, however, than exploring when the mother is present and diminishing exploration when she is absent. Our previous discussion of secure-base behavior among the Ganda (Ainsworth, 1967) specified that the infant who explored away from his mother nevertheless monitored her whereabouts by glancing at her from time to time, perhaps occasionally interacted with her across a distance, and returned to her after a while, perhaps clambering up for a moment’s close contact before making another foray away from her. Harlow’s (1961) rhesus infants were briefly separated from their surrogate mothers before being placed in an open field, and hence initially rushed to the surrogate. After a few moments, however, their behavior in the open-field situation matches our description of Ganda infants very well.
Bowlby’s (1969) concept of set-goal is highly relevant to this second aspect of the secure-base phenomenon. According to his theory, attachment behavior is activated when the distance between an infant and his attachment figure exceeds a certain point, whether it was the movement of the mother or of the infant that increased the distance between them. Consequently a baby, having moved away to explore, would be prompted to return to the attachment figure after exceeding the distance specified by his set-goal. The set-goal differs from time to time, however, according to circumstance. Two researches have been relevant to the concept of a spatial set-goal. Anderson (1972) observed that toddlers in a London park kept within a distance of approximately 200 feet from their mothers. They tended to move away from the mother slowly, with intermittent stops, but to return to her quickly from time to time on their own initiative. During a typical sortie away from his mother, the child looked at her occasionally, and when he returned he tended to make physical contact with her. Rheingold and Eckerman (1970) observed young children aged from 1 to 5 years in a large yard. The mean farthest distance that the 1-year-olds went from their mothers was 23 feet; for 2-year-olds, 50 feet; and for 3-year-olds, 57 feet. Both these researches imply that the strange-situation room (and indeed the usual room at home) tends to be too small for an infant to exceed his spatial set-goal while still remaining in the room. Consequently, in Episode 2 it was a reasonable finding that few infants returned to their mothers after beginning to explore, because they perforce remained well within the limits of their spatial set-goals for relatively nonalarming situations.
The baby’s perception of his mother’s relative accessibility or inaccessibility also may alter the set-goal. Carr, Dabbs, and Carr (1975) have shown that when the mother is faced away from the child or behind a screen proximity-seeking behavior much more frequently overrides exploratory behavior than when the mother is both visible and faced toward the child. Apparently a turned back, in addition to being a physical barrier to vision, gives the impression of relative inaccessibility. Individual differences in long-term interaction with the mother may also make for individual differences in the baby’s perception of his mother as accessible and responsive, and thus influence the set-goal for proximity maintenance.
Bowlby (1969) suggested that there was also a temporal set-goal, so that, having been at some distance from his mother for a certain time specified by the set-goal, a child tends to return to his mother, whether or not he had exceeded the spatial set-goal. The only research with relevance to this concept is by Brooks and Lewis (1974). They divided a 15-minute free-play session (comparable to our Episode 2 except for duration and for number and placement of toys) into five 3-minute periods. Proximity to or touching the mother was of much shorter duration during the first 6 minutes than during the last 6. One possible interpretation of their findings is that the temporal set-goal was not exceeded by most infants during the first 6 minutes, but that attachment behavior was later activated. Brooks and Lewis do not report behavioral sequences, however. If a return to the mother tended to be succeeded by further exploration, this would fit with the set-goal interpretation.
Brooks and Lewis suggest another explanation—namely, that the novel features of the toys were exhausted during the earlier periods of the session. According to our paradigm, this would imply that the toys initially activated the exploratory system more strongly than the attachment system, but that as the novelty wore off exploratory behavior became weaker until it was no longer more strongly activated than the attachment system. Under these circumstances the infant might be expected to return to the mother and to remain close to her rather than resume exploration. Weight for this explanation is provided by previous research by Rheingold and Eckerman (1970), who found that how quickly 10-month-old infants left their mothers to enter a new environment (an adjoining room), how far they went, and how long they stayed away were influenced by the number and location of the toys that had been placed in it. Further evidence is provided by Willemsen and associates (1974), who found a differing amount of exploratory behavior in the strange-situation in two sessions differing in the degree to which they were judged to be “interesting.”
All of these considerations suggest that the relative balance between exploratory and attachment behavior, and thus the way in which an infant uses his mother as a secure base from which to explore, are influenced by a variety of circumstances—including the size of the room; the length of the observational session; the nature, diversity, and complexity of the stimuli that activate and maintain exploration; the orientation and behavior of the mother, as well as the internal condition of the infant (e.g., whether tired, hungry, or ill); and the influence his previous experience has on his expectations of his mother’s accessibility and responsiveness. In Episode 2 of our strange situation, the combination of the following circumstances seems to have kept the balance tipped for most infants in the direction of exploratory spatial set-goals for proximity to the mother were not exceeded), and brief duration of the episode (so that most infants’ temporal set-goals were not exceeded). Consequently, the main evidence from the strange situation that supports the concept of an infant’s using his mother as a secure base from which to explore necessarily rests on the comparison of the mother-present and mother-absent conditions. Further research on the spatial and temporal set-goal aspects of the secure-base phenomenon is obviously needed.
Responses to a Stranger
Wariness/Fear Versus Friendliness. The design of the strange situation implied that the entrance of a stranger into the unfamiliar environment would be more alarming to most infants than the unfamiliarity of the strange environment itself. Thus it was expected that the wariness/fear system would be more strongly activated in more infants in Episode 3 than in Episode 2. This expectation was confirmed by the fact that more infants cried in Episode 3 and that more gained proximity to the mother. Furthermore, only 3% spontaneously and fully approached the stranger in Episode 3, whereas 80% spontaneously and fully approached the toys in Episode 2. Indeed only 4% of the sample failed to show some sign of wariness in Episode 3 (Bretherton & Ainsworth, 1974). Nevertheless, we found that fear of stranger was neither as intense nor as ubiquitous as Spitz (e.g., 1965) implied. In Episode 3 of the strange situation, very few 1-year-olds cried when the stranger entered, and less than one-quarter of the total sample showed distress at any time during the episode. Even less distress was reported by Rheingold and Eckerman (1973) in their observations of somewhat younger infants in a laboratory setting. Indeed their findings highlighted friendly behavior to a friendly stranger.
Our findings provide some support for Rheingold and Eckerman. In our sample 89% of the subjects showed friendly behavior toward the stranger in some form or degree in Episode 3. In only 30%, however, was such behavior more conspicuous than wary behavior. Both friendly and wary behaviors tended to be mild rather than intense in our strange situation. Most babies smiled at the stranger or accepted a toy that she offered them, but very few approached her spontaneously or actually entered into interactive play with her during Episode 3. Eighty-five percent of the sample showed signs that both friendly and wary/fearful behavior were activated by the stranger.
In using “eight-month anxiety” as the criterion that “true object relations” (i.e., attachment) have been achieved, Spitz implied that stranger anxiety is an essential milestone in normal social development. However, he tested for stranger anxiety with the mother absent, thus confounding it with separation anxiety. Indeed he interpreted fear of strangers as a manifestation of underlying separation anxiety.
Bowlby (1969) also implied that alarm when faced by unfamiliar situations and persons was to be expected among infants in the second half of the first year, although he acknowledged that novelty also aroused exploratory interest and that the presence of the mother tended to dampen the intensity of alarm. He later identified the strange as a natural clue of danger, emerging without any necessary conditioning experience, although not until the infant had accumulated enough experience with the familiar to be able to distinguish it from the unfamiliar (Bowlby, 1973).
Our sample of Ganda infants (Ainsworth, 1967) so regularly showed fear of strangers that it was judged to be a normal phase of development, although one that came somewhat later than the phase during which an infant’s active initiative in seeking proximity and interaction with his mother first made it clear that he had become attached to her. In most instances these Ganda infants showed more intense fear of strangers than did our sample of white, middle-class infants observed at home (Ainsworth, 1977). According to Konner’s (1972) report, infants and young children of the Zhun/twa Bushmen show much more intense and persistent fear of strangers than either the Ganda or American infants whom we have observed.
How may one reconcile the apparent discrepancies between theory and findings, and among various sets of findings? Several variables seem to affect the intensity of activation of wary and/or fearful behavior in the presence of a specified stranger or class of strangers: the characteristics and behavior of the stranger, length of exposure to this stranger, previous experience with strangers or its lack, the context in which the stranger is encountered, and the degree of anxiety characteristic of the infant’s attachment to his mother.
Some strangers may be more strange than others, in that their characteristics depart more widely from the characteristics of persons with whom the child is familiar. Thus a child chiefly familiar with female adults may find male strangers more alarming than females. Ganda infants may have found the skin color and dress of the visiting European especially strange (Ainsworth, 1967). The behavior of one stranger may be more alarming than the behavior of another. In most systematic studies of infants’ responses to strangers, there is an attempt to control for this variable, either by using more than one stranger or by strictly controlling the behavior of the stranger through instructions. Furthermore some studies (such as ours) attempt to have the stranger behave in what is assumed to be a minimally alarming way, so that any fear reaction may be attributed to unfamiliarity.
A stranger can remain a complete stranger for only a limited period of time. Thus Bretherton (1978) demonstrated that 1-year-olds’ wary behavior declined and friendly behavior increased over a period of 8-minutes’ exposure to a friendly stranger who attempted to interact with the baby through the intermediary of toys. Rheingold and Eckerman (1973) provided a period of familiarization with the stranger before systematically observing babies’ responses to her—a fact that undoubtedly had some influence on their findings of infrequency of full-blown fear responses.
Children reared under widely differing conditions of opportunity to encounter strangers may differ in the intensity of wary/fearful behavior with which they respond to strangers. Our sample of middle-class American babies was taken to supermarkets, restaurants, and pediatricians’ offices and experienced the visits to their homes of delivery men, meter readers, postmen, and appliance-repairers (to say nothing of babysitters). This relative familiarity with people in the general class of strangers may well have accounted at least in part for the fact that they found strangers less alarming than did the infants of the Ganda sample reared in villages with much less exposure to strangers (Ainsworth, 1977). Ganda infants, however, experienced visits from unfamiliar friends and relatives of their parents, whereas Konner’s Zhun/twa Bushmen very rarely encountered persons other than members of the small group of families to whom they were exposed daily. It seems likely that the extent of a child’s previous experience with unfamiliar people plays a role in determining his response to strangers, even though minor differences in such experience within a sample of children reared similarly (e.g., Bronson, 1972) may not correlate significantly with range of response to strangers.
The context in which a stranger is encountered has been demonstrated to influence infants’ responses to him, as Sroufe, Waters, and Matas (1974) have pointed out. Thus Morgan and Ricciuti (1969) showed that a baby held on his mother’s lap was more likely to respond positively to a stranger’s advances than one seated some feet away from his mother. Bowlby (1973) suggested that natural clues to danger, including the strange, may be more alarming under conditions when a child is also anxious because his mother is inaccessible. It must be recalled that Spitz tested for stranger anxiety when the mother was absent, whereas the mother was present throughout the sequence of episodes observed by Rheingold and Eckerman (1973).
Finally some individual children are chronically anxious about their mothers’ accessibility and responsiveness and are thus identified as anxiously attached. Such children may be expected to show more intense fear of strangers than children who are securely attached. The present study yields evidence pertinent to this point; Group-C infants are more likely to cry and to approach their mothers in Episode 3 than are the securely-attached infants of Group B. Ainsworth (1967) reported that some Ganda infants whose attachment to the mother had been made anxious by recent weaning from the breast demonstrated much more intense fear of strangers than they had before weaning. Any sample in which anxiously attached children are especially numerous might be expected to show more intense fear of strangers than a more “normal” sample. In this context it is pertinent to point out that Spitz’s core sample consisted of infants of delinquent mothers in a correctional institution. It is unlikely that such a setting could provide optimum conditions for the development of a secure attachment of infant to mother, especially because the conditions of such an institution do not permit mothers to be regularly accessible to their infants (Arsenian, 1943). Blehar (1974) found that 30- and 40-month-old day-care children showed significantly more avoidance of the stranger in the strange situation than their home-reared controls; she attributed this to anxious attachment fostered by their very long daily separations.
These considerations go a long way toward accounting for the apparent discrepancies between theory and findings, and among findings of different studies; but there is also the matter of the indicators used in the identification of wariness/fear aroused by strangers. Rheingold and Eckerman (1973) recorded crying and approach (retreat) to the mother as signs of fearfulness, but did not record any of the more subtle indices of wariness that we (Bretherton & Ainsworth, 1974) employed. Whereas Spitz, too, focused on crying and withdrawal, he also mentioned looking and turning away as behaviors characteristic of stranger anxiety. Waters, Matas, and Sroufe (1975) employed a number of low-intensity indicators of wariness, such as gaze aversion, and found these as well as crying to be associated with heart-rate acceleration.
In conclusion, we suggest that the response of an infant to strangers can best be understood as determined by the relative strength of potentially conflicting behavior systems. On the one hand, we accept the proposition that the strange and strangers tend to activate wary/fearful behavior. On the other hand, it has been well demonstrated that exploratory behavior may be activated by the strange and lead to approach and investigation. Both sets of responses may be viewed as adaptive in an evolutionary sense. In any species whose behavior is not largely determined by fixed action patterns, it is of obvious advantage for the young to explore the environment and to learn ways of coping with it. Yet because such exploration may well be hazardous for the inexperienced infant, it is also of advantage for wary/fearful behavior to be activated by the strange—particularly in the absence of the mother, who can ordinarily provide protection. Furthermore, it seems likely that in social species such as man there is a survival advantage to friendly behavior toward conspecifics other than the mother or other attachment figures. Indeed there is more evidence of friendly behavior toward strangers among young humans than there is of exploratory behavior activated by strangers. Waters, Matas, and Sroufe (1975) suggest that even gaze aversion serves an adaptive function when an infant encounters a stranger. Such behavior may be analogous to the “cut-off” behaviors described by Chance (1962), when permitting the child to remain in proximity to the stranger rather than moving away, and thus enabling him a few moments later to entertain the possibility of friendly interaction. Waters and associates suggest that through such wary, cutoff behaviors the infant modulates arousal and prevents the disorganization of acute distress and crying, which would greatly delay the process of becoming acquainted.
In any situation, it seems reasonable to suppose that the response to an unfamiliar person will be determined by the interplay between the wary/fearful system and the exploratory and/or affiliative (friendly) systems (and, indeed, by the conflict among them). (Conflict behavior is discussed in more detail later.) It is reasonable to suppose that the relative strengths of the several behavior systems activated by strange conspecifics depend on a variety of circumstances, including those discussed above—environmental and behavioral context and experience with both attachment figures and strangers. Indeed there is every reason to believe that the state of activation of the attachment system is among the most important of the contextual variables and enters into the interplay among the other relevant systems in determining a baby’s response to a stranger.
Use of the Mother as a “Secure Haven.” Because attachment behavior tends to be activated in an alarming situation, it was expected that when the wariness/fear system is more strongly activated than other relevant systems (such as exploratory or friendly behavior), the attachment system would simultaneously be activated so that the baby would tend to move away from the stranger and toward the attachment figure. It was not necessarily expected that he would remain close to his mother, however. Harlow (1961) showed that once an infant monkey had fled to his surrogate mother and clung, he thereby gained courage to approach a fear-producing stimulus object and even to explore it. In Episode 3 of the strange situation, a substantial minority of infants were sufficiently alarmed so that strong attachment behavior was activated; these infants moved away from the stranger into close proximity to or actual bodily contact with the mother. This movement can legitimately be identified as “retreat to the mother,” for these infants did not approach the mother intending to interact with her; on the contrary they tended, as soon as they had reached her, to turn back to stare or even smile at the stranger from the secure haven provided by the mother.
Although 96% of the sample showed some wary behavior, by no means all of these wary infants approached their mothers. It seems likely that the mother’s mere presence in the same room provided a “secure haven” for many, for she was sufficiently close that most babies probably felt confident of her accessibility. They could reach her if they wished or at least signal her to come closer; her mere presence provided security enough that attachment behavior was not strongly activated.
It might be argued that the concept of the mother as a secure haven is essentially the same as the concept of her as a secure base from which the child can venture forth to explore. There is no doubt that the two concepts are similar. Nevertheless it seems desirable to retain a distinction between them. When a baby uses the mother as a secure base from which to explore, there is no necessary implication of wariness/fear. Whatever fear might have been evoked by an unfamiliar stimulus object or situation is overridden by the security provided by the mother’s presence. On the other hand, when a baby seeks to come or to remain in proximity to his mother as a secure haven, the implication is that he is at least somewhat alarmed. If, subsequently, his alarm is sufficiently moderated by proximity to this attachment figure so that he can venture forth to explore, as Harlow’s infant monkeys did, then the attachment figure shifts from being a secure haven to being a secure base from which to explore. The implication is that proximity to it has reduced the activation of the wariness/fear system to a level of strength lesser than the activation of the exploratory system attributable to novel features of the alarming stimulus object.
Perhaps we also need a concept of the attachment figure as a secure base from which to make a friendly approach to a conspecific; for infants who, in the presence of their mothers, approach an amiable stranger or even accept her overtures do not seem impelled to explore the stranger as much as to interact with her. In the last minute of Episode 3 of the strange situation, however, a baby’s approach to or acceptance of the stranger seems to combine friendly and exploratory behavior, because the stranger attempts to evoke interaction by inviting the baby to play with a toy that she offers him.
Differential Behavior to Mother Versus Stranger. Although proximity-promoting (attachment) behaviors tend to be nondifferential in an infant’s earliest weeks or months, it may be expected that as he learns to discriminate his attachment figure(s) from others, his attachment behaviors would become increasingly differentiated, in that they would be directed more frequently, more readily, and/or more intensely toward the persons to whom he is becoming or has become attached than toward others. We have proposed (Ainsworth, 1963, 1964, 1967) a variety of “patterns of attachment behavior” as a set of criteria for judging when and to whom a baby had become attached. Implicit in these criteria is that the behavior would be manifested differentially (i.e., selectively or preferentially) toward an attachment figure in comparison to others.1
Others (e.g., Cohen, 1974; Feldman & Ingham, 1975) have assumed that we proposed attachment behaviors as measures of attachment, implying that the more strongly (and/or more frequently or longer) the behavior is manifested the stronger the attachment might be judged to be. Before identifying a behavior as an attachment behavior, therefore, they argue, it must be demonstrated that the behavior in question is indeed differential to attachment figures in comparison with others to whom the infant is not attached. Specifically, Feldman and Ingham, as well as Cohen, have criticized the strange situation as a “measure of attachment,” because the comparison figure is a stranger rather than a nonattachment figure with whom the child has some familiarity.
Although we consider it useful to consider the differentiality of attachment behaviors as criteria for ascertaining the onset of attachment, the ethological-evolutionary theory of attachment that serves as our framework does not define attachment behavior in terms of either exclusiveness or relative strength or frequency of manifestation toward an attachment figure. Furthermore, as we have pointed out before (e.g., Ainsworth, 1967, 1972; Ainsworth & Bell, 1970), there are grave flaws in the assumption that the strength with which attachment behavior is manifested reflects the strength of any underlying attachment relationship. (This point is discussed more fully later in this chapter.) Finally, we did not intend the strange situation to provide the basis for identification or “validation” of behaviors as attachment behaviors, nor indeed to measure the strength of an infant’s attachment to his mother.
Attachment behavior is defined as a class of behaviors that have the predictable outcome of gaining or maintaining proximity to a caregiver or later to an attachment figure. They are conceived as playing an important role in the development of attachment and in mediating the attachment once it has been formed. At least some of these behaviors—perhaps, indeed, all of them—may sometimes serve behavioral systems other than the attachment system, as Tracy, Lamb, and Ainsworth (1976) have argued in the case of locomotor-approach behavior. It is only with consideration of the context—both environmental and behavioral—in which the behavior appears that we can assert that the behavior in question is operating in the service of the attachment system or in the service of some other behavioral system at the time it is observed. To demand that the label “attachment behavior” be reserved for a discrete action that is displayed exclusively, or even more frequently, toward an attachment figure rather than toward others is to distort our understanding of the function of attachment behavior.
Nevertheless, a study of the selectiveness with which proximity-promoting behaviors are displayed toward different figures in various contexts throws light on what we consider to be a more fundamental task of research into the development of attachment—namely, the ways in which such behaviors become organized together to serve the attachment system and the ways in which the attachment system itself interacts with other behavioral systems. In this context let us consider the degree of differentiality shown by various behaviors in the strange situation. It must be kept in mind, however, that the sequence of episodes in the strange situation was designed to progressively intensify proximity-promoting behavior directed toward the attachment figure, and that the only figure with whom the attachment figure can be compared is a stranger—circumstances that facilitate maximum differentiality.
It is therefore of no great import to find that in the strange situation a number of proximity-promoting behaviors are sharply differential to the mother in comparison with the stranger, and thus identifiable as serving the attachment system rather than another behavioral system. These include approaching, touching, clinging, resisting release, and all of the behaviors that are comprehended in the measures of proximity/contact seeking and contact maintaining. They also include following and some (but not all) greeting behaviors. Furthermore, vocalizing was found to occur more frequently when the infant was alone with his mother than when the stranger was present. Of these approach, following, and touching in the home environment were not strikingly differential to the mother in comparison with a relatively unfamiliar figure (Tracy, Lamb, & Ainsworth, 1976). There it was found that approach was strongly differential to the mother only when it was acompanied by distress or when it ended in a pick-up appeal, and touching only when in the context of a pick-up appeal. Stayton, Ainsworth, and Main (1973), in a similar study of home behavior, were not able under the circumstances of their analysis to identify the contextual features of differential following.
It is of more interest in regard to strange-situation behavior to highlight behaviors that did not emerge as differential to the mother—namely, smiling and looking. Indeed, smiling was more frequent to the stranger than to the mother, although a measure of intensity of smiling favored the mother (Bretherton & Ainsworth, 1974). Similar findings for greeting behavior at home were reported by Stayton, Ainsworth, and Main (1973) and for face-to-face behavior by Blehar, Lieberman, and Ainsworth (1977). Thus it is apparent that smiling not only serves the attachment system but also serves in sociable interaction with persons who are clearly not attachment figures. We suggest that it may also occur as a propitiatory behavior when wariness/fear is activated at a relatively low degree of intensity (Bretherton & Ainsworth, 1974).
A moment’s consideration suggests that looking, too, must serve many systems other than the attachment system. Certainly in Episode 3 of the strange situation, infants looked much more frequently at the stranger than at the mother—whether this was in the service of exploratory behavior, wary/fearful behavior, or sociability. Furthermore, our composite measure of distance interaction (comparable to the smile-speak-show measure of Maccoby and Feldman, 1972) was stronger in regard to interaction with the stranger than in interaction with the mother. Although a child—or an adult, for that matter—may interact with an attachment figure across a distance, and this may serve the attachment system under conditions of low-intensity activation, it is clear that all the various modes of interaction across a distance commonly serve other purposes as well. It would plainly be a mistake to focus entirely on such behaviors when studying attachment and attachment behavior.
It might have been predicted that avoidant and resistant behaviors—which are antithetical to proximity promotion—would be directed more frequently and more intensely to the stranger than to the mother. This did not prove to be the case. Both are relatively uncommon behaviors, even at low intensity. In the case of avoidance there was effectively no overlap between those who avoided the mother and those who avoided the stranger. In the case of resistance, however, those who resisted the mother also tended to resist the stranger, and vice versa.
Finally, it is of interest to compare mother and stranger in regard to the termination of attachment behavior once it had been activated at high intensity, as it had been for most infants, especially in the second separation episode. Whereas most infants who had been distressed during separation calmed quickly when reunited with their mothers in Episodes 5 and 8, the stranger’s return in Episode 7 reduced the distress of very few. Although some infants who were distressed by separation stopped crying when the mother merely entered the room, for most attachment behavior was terminated only by close bodily contact for varying lengths of time. Close bodily contact with the stranger in Episode 7, however, scarcely reduced the intensity of activation of attachment behavior—and the stranger offered such contact to all who continued to be distressed after she returned.
Responses to Separation
Crying when briefly separated from the mother in the unfamiliar environment of the laboratory was not as ubiquitous as anticipated. Barely half of the babies cried during the first separation, although over three-quarters did so during the second. To judge from the babies of Sample 1, for whom extensive longitudinal data were available, failure to show separation distress in the strange situation may not be interpreted to mean that an infant has not become attached to his mother. Group-A babies, for example, showed relatively frequent separation protest at home even though they showed little or no distress in the strange situation. Furthermore, we have long believed that failure to show distress in very brief or everday separations at home is an undependable criterion of attachment—or the lack thereof (Ainsworth, 1963, 1967, 1972, 1973).
Nevertheless there is evidence that separation in the strange situation activated the attachment-behavioral system strongly enough in most infants that it competed with—and in many infants overrode—exploratory behavior, which had been so strongly activated in Episode 2. Furthermore, as mentioned earlier, Sroufe and Waters (1977b), using both behavioral and heart-rate measures, reported: “It appears that virtually all 1-year-olds are stressed by separation in the ’strange situation,’ but that infants differ in the behavioral resources available to cope with the stress (i.e., the degree of distress is not the only or necessarily the crucial variable).”
As reported in Chapter 10, studies of 2- to 4-year-olds (Blehar, 1974; Marvin, 1977; and Maccoby & Feldman, 1972) showed that overt distress as evidenced by crying declines with age. Whereas 2-year-olds still tend to be distressed in the separation episodes, 2½-year-olds cry less frequently, and 3-and 4-year-olds do so little or not at all. Blehar’s study suggests that 3-year-olds who cry in these brief separations may have been sensitized to separation anxiety by long, frequently daily separations. Marvin suggests that 4-year-olds who cry may do so not so much because they are distressed by separation as through frustration in not having persuaded the mother either to stay or to take them along—that is, frustration in having been unsuccessful in arriving at a mutual adaptation of “plans.” Common sense experience suggests that as they grow older, children can sustain increasingly longer periods of separation from attachment figures without distress. Undoubtedly, increasing cognitive capacities enable a child to better understand the circumstances of even involuntary separations, and to have established well-based expectations that mother will return soon, unless some traumatic separation experience has led them to doubt that she will.
The miniscule separations in the strange situation, as well as little everyday separations occasioned by mother’s leaving the room in the home environment, constitute situations much less anxiety provoking than “major” separations. A “major” separation might be defined as an involuntary separation either of very long duration (or permanent) or one that is at least long enough to greatly exceed a young child’s expectations of the likely period of time that must elapse before reunion. Such major separations are the more anxiety provoking if he has no information about or understanding of the reasons for the separation or the conditions of its termination, or if he is at the same time separated from other attachment figures, especially if he has no opportunity for interaction with any figure who might substitute for his absent attachment figures.
At home, an infant may cheerfully accept separation because his past experience has engendered confidence that either his mother will return soon or that he can gain access to her if he wants to (Stayton & Ainsworth, 1973). Consequently some children are trusting at first, and only gradually become distressed when the mother’s absence is longer than expected, or when their efforts to reach her are frustrated. Such confident anticipation may be expected to be characteristic of the baby who is securely attached. Some secure infants apparently carry their confident expectations over to an unfamiliar environment. This, we believe, accounts for the fact that some Group-B infants in Episode 4 are not apparently distressed at first, and only later begin to cry when their efforts to regain access to the mother are unavailing or when her absence is longer than expected. Indeed, some weather the entire first separation episode without distress, only to become distressed during the second separation when, it seems reasonable to suppose, their confidence in the mother’s accessibility in this environment has been shaken.
Brief separations from the mother in an unfamiliar environment may evoke no distress if the baby is left with another attachment figure (e.g., Kotelchuck, 1972), and indeed the same phenomenon may be observed at home (Ainsworth, 1967). Even a relatively brief (i.e., 8-minute) previous play interaction with a stranger may diminish the distress manifested when the mother departs leaving the baby with a stranger (Bretherton, 1978). Also, in major separations, the presence of responsive substitute parent figures greatly reduces the distress occasioned by temporary loss of parents (Robertson & Robertson, 1971).
Whether a separation is voluntary or involuntary can strongly affect a child’s initial responses to it. Infants and young children may quite cheerfully leave an attachment figure to explore elsewhere (Ainsworth, 1967; Rheingold & Eckerman, 1970) and show no distress if they are not prevented from an equally voluntary return to that figure. Nurses in hospitals often show an intuitive knowledge of this phenomenon; they know that many young children may be lured away from mother and will only later become distressed when they are prevented from returning to her.
Previous separation experiences may, however, make an infant or young child all the more alert to the likelihood of separation in a given situation. A child who has once been fooled by a nurse will not be so trusting a second time when invited to leave his mother to do something “interesting.” Similarly, a child with a history of major (or even a series of seemingly minor) separation experiences is not likely to be as trusting as a child who has had no previous unhappy separations. Thus, for example, one child in Sample 1 could not tolerate separation in the strange situation. Throughout the first year he had been left by his working mother with a responsive housekeeper. Until he was about 10 months old, he accepted his mother’s departures in the morning, but then began to protest them. In the strange situation, the moment his mother got up to go at the end of Episode 3 he was undone. Ganda infants (Ainsworth, 1977) showed more intense distress in everyday separation situations at home than did the American babies of our Sample 1. Most of them had been left with other caregivers every day for 4 hours or more while their mothers worked in the garden, whereas when the mother was at home she tended to take the baby with her as she moved from room to room. It would seem that when the Ganda mother did leave the baby behind, this signified to him a much longer absence than that expected by most of our American sample babies when the mother left the room. Similarly, we found (Blehar, 1974) that children in full-time day care, having been previously home reared, showed significantly more distress in the separation episodes of the strange situation than home-reared age peers—a finding that may be due to their having become sensitized to separation by their frequent, long absences from home. On the other hand, it would seem likely that these same day-care children might have left the mother’s side voluntarily in order to approach other children when introduced to a new play group, as Ricciuti (1974) found with a sample of young children who had been reared in a day-care center.
The Interchangeability of Attachment Figures. Because of contemporary interest in the role of the father, we have often been chided because we have used the mother as representative of the class of attachment figures in our strange-situation research. We do not consider the mother as the only figure who can provide a secure base from which to explore or, indeed, whose departure could arouse distress. Three sets of studies (viz., Cohen & Campos, 1974; Kotelchuck, 1972; and Lamb, 1976c), which have used modifications of our strange-situation procedure, have demonstrated that a baby can tolerate the departure of either mother or father without distress, as long as he still has the other figure available to him. Kotelchuck also (Kotelchuck, Zelazo, Kagan, & Spelke, 1975; Ross, Kagan, Zelazo, & Kotelchuck, 1975; and Spelke, Zelazo, Kagan, & Kotelchuck, 1973) has shown, in contrast, that when either figure departed leaving a 1-year-old with a stranger, he tended to protest the separation and to decrease his exploratory behavior.2 Whereas these studies used situations in which response to the departures of mother and father could be compared in the same children in the same laboratory session, Willemsen and associates (1974) and Lamb (1978) compared the responses of the same children in two sessions of the strange situation, once accompanied by the mother and once by the father. Feldman and Ingham (1975) used two groups of children but only one session for a similar comparison. (See Chapter 9.) The findings of all three studies showed minimal differences in responses to the two figures.
Arsenian (1943) investigated the behavior of young children under conditions of presence and absence of a mother figure in a laboratory situation. The children had been reared with their mothers in a facility for delinquent women. If the mother was not available to participate in the study, her role was assumed by a substitute caregiver who was responsible for the child in the mother’s absence. The children explored when the substitute figure was present and protested her absence, just as did the children whose mothers were available for the experiment. These findings are quite in line with our Ganda findings (Ainsworth, 1967) that infants tended not to cry when the mother departed if another attachment figure remained in the same room.
Such findings have sometimes been interpreted as a refutation of Bowlby’s (1958, 1969) concept of “monotropy,” in that they demonstrate that an infant or young child is not necessarily attached solely to his mother. Bowlby, however, did not mean that there could be only one attachment figure, but implied that there was one principal attachment figure, to whom the others were secondary. This implies a hierarchy of attachment figures. Presumably for some purposes and in some situations, attachment figures can be interchangeable—as they tended to be in the studies just cited. But for other purposes and in other situations (e.g., when ill or fatigued), the principal attachment figure would be required, or at least preferred.
There has been very little research relevant to this issue. Lamb, however, in a series of recent articles, has addressed himself specifically to the question of whether, under what circumstances, through which attachment behaviors, and at what ages a child will express preference for his mother over his father or vice versa (e.g., Lamb, 1976a, 1976b, 1976c, 1977a, 1977b, 1977c). In summary, his findings indicate that under some circumstances infants express preferences for the father, but under other circumstances the preference is clearly for the mother. Furthermore, he has made a distinction similar to ours between behaviors that emerge with high-intensity activation of the attachment system, for which he reserves the term “attachment behaviors,” and behaviors that emerge with low-intensity activation, which he terms “affiliative behaviors.” At home, at age points from 7 months until at least 2 years of age, infants show preference for their fathers through “affiliative” behaviors in stress-free situations, whereas in stressful situations (including a situation as minimally stressful as the presence of a stranger in an unfamiliar laboratory situation) the preference tends to shift to the mother.
On the whole, Lamb confined himself to an examination of normative trends, although in one paper (1977b) he looked at individual differences.3 In some cases it seems likely that an infant might fairly consistently prefer one figure to the other under both stressful and nonstressful circumstances, as for example when the father has played the maternal role, and/or the relationship with the mother is extremely disturbed, or when the father has estranged himself from the family. These exceptional cases, as well as the more usual trends reported by Lamb, are not incompatible with Bowlby’s concept of monotropy—or, as more accessibly worded, his concept of a hierarchy of attachment figures—for he would identify the principal attachment figure as the one preferred under stress. Lamb’s main point, however, is that although the relationship of a child to his mother is likely to differ qualitatively from his relationship to his father, both attachments are significant determinants of social and personality development.
It is possible that in the absence of any attachment figure, the child may direct even high-intensity attachment behavior to someone else. Both Maccoby and Jacklin (1973) and Rosenthal (1967) have shown that children in fear-arousing situations may direct attachment behavior toward an unfamiliar figure when the mother is not available, even though the intensity of such behavior may be somewhat muted in comparison with the behavior of children who faced the same situation with an attachment figure present. Fleener (1973) suggested that he had experimentally produced attachment within 3 days during which a young child was separated from his mother during the daytime and played with and cared for by a research assistant. At the end of the experimental period the subjects demonstrated differentiality of approach to the familiar surrogate in preference to another research assistant who was unfamiliar to the child, and manifested distress when the familiar person left but not when the unfamiliar person did. Nevertheless, when a subsample of children was faced with preference for the mother (or distress when separated from her) in contrast with the surrogate, it was clear that the mother was the preferred figure.
Can such temporary relationships be classed as attachments? There is no doubt that they can play an extremely useful role during periods of separation from other figures with whom the child has more enduring attachment relationships, as the Robertsons have plainly demonstrated (Robertson & Robertson, 1971). Perhaps it is best to consider them as incipient attachments. Should circumstances permit a continuing relationship, they might well be consolidated as attachments, but when circumstances make the relationship of short duration they do not become sufficiently well consolidated to endure. However, should such relationships become well enough consolidated that they result in distress when the child is returned to his original attachment figures, even though the distress may be of relatively short duration and even though the child may reestablish his original attachment relationships without undue delay, surely one could identify the attachment to the temporary surrogate as more than merely incipient.
Other Normative Findings
Although the strange situation was originally designed to investigate the secure-base phenomenon, response to strangers, and response to separation in an unfamiliar environment, its yield of findings has by no means been limited to these three areas of interest. Without doubt, the most interesting additional findings pertain to individual differences, which are discussed in the next chapter; but some of them may be classed as “normative”—specifically, findings pertaining to the activation and termination of attachment behavior, shifts in specific attachment behaviors that emerge at different levels of activation of the system, and the interplay between attachment behavior and other behavioral systems, including conflictful interplay.
Activation of Attachment Behavior. Of the several manipulations intended to activate attachment behavior, the one that activated it at highest intensity was separation from the mother. The presence of a stranger in Episode 3 effected observable activation of attachment behavior in some infants, but these were in a minority. The unfamiliar environment per se was the least effective, even though Sroufe and Waters (1977b) observed small-magnitude tonic heart-rate increases, which they interpreted as signs of wariness, upon first entering the room in the introductory Episode 1. Of course, this generalization is weakened by the fact that these three manipulations occurred in a fixed order. We do not, however, believe that an experiment that attempted to control for order effects would throw significant additional light on the relative effectiveness of the three relevant conditions in activating attachment behavior.
A comparison between responses to brief separation at home (Stayton, Ainsworth, & Main, 1973) strongly suggests that the unfamiliar features of the physical and social environment interacted with brief separations to create what Bowlby (1973) identified as a compound situation. Alarm occasioned by the strange environment (however minimal) and by the stranger interacted with the anxiety occasioned by the inaccessibility of the attachment figure in the separation episodes to activate attachment behavior at a higher level of intensity than would have been occasioned by any of the three sets of stimulus variables separately—indeed at a substantially higher level than their separate intensities summed. Furthermore, as Cohen (1974) has also suggested, cumulative stress has its effects. Once attachment behavior has been activated at high intensity in one episode, it does not subside to what might be conceived as “base-level” in the next episode, but rather tends to remain at a relatively high level, to be intensified easily by the next instigation. Or, as Cohen would put it (and as Fleener and Cairns, 1970, found), once a baby begins to cry in one episode the likelihood that he will cry again in later episodes is significantly increased. Finally, a strange situation by necessity is undertaken in an unfamiliar milieu. This background unfamiliarity would enter into every episode, whether the mother and/or stranger are present or absent.
We did not intentionally introduce any (nonseparation) conditions other than the merely unfamiliar that might be expected to activate attachment behavior. Indeed we discarded three babies from our analyses in which such conditions existed—two whose mothers brought them to the laboratory despite the fact that they were ill, and one who was intensely disturbed by the whirr of a fan in the experimental room. We did not control in advance either for individual differences in expectations relevant to brief separations or strangers, or for individual differences in the quality of the attachment relationship itself. Both undoubtedly accounted for individual differences in strange-situation behavior, but qualitative differences in attachment clearly emerged as the crucial variable. Furthermore, as mentioned previously, the limitations provided by the spatial features of the experimental room, together with the brief duration of the episodes, probably prevented us in most instances from observing the activation of attachment behavior attributable to a child’s exceeding a spatial or temporal set-goal.
Our firmest conclusion, therefore, is that the combination of an unfamiliar physical environment, a stranger, and separation from the mother provides very strong instigation to attachment behavior.
Behaviors at Different Levels of Activation of the Attachment System. Because it is evident that attachment behavior was activated at different levels of intensity from one strange-situation episode to another, an examination of behavioral changes across episodes throws light on the specific attachment behaviors characteristic of different intensities of activation of the system. At low levels of intensity of activation, as in Episode 2, some infants manifested no attachment behaviors except an occasional smile or vocalization directed toward the mother. As we previously suggested, the set-goal for proximity for most infants was set at wider limits than would be exceeded in the episode. Because his mother seemed settled in her chair, there was little need for the infant even to check her whereabouts through occasional glances. Such behavior is entirely characteristic of what happens much of the time at home, when both mother and baby are occupied with their own activities and each is confident of the whereabouts of the other.
A few infants alternated in Episode 2 between exploratory play and initiating interaction with the mother, although such behavior seemed more characteristic of the young children observed by Maccoby and Feldman (1972) and Marvin (1977). At home, however, when the mother was free to initiate interaction herself, and also probably able to respond more naturally to infant initiations, episodes of interaction—whether playful, tender, or merely sociable—occur intermittently.
When attachment behavior is activated at somewhat higher intensity—whether because of a spatial or temporal set-goal being exceeded or because of a mildly alarming environmental stimulus—active proximity seeking is likely. The child approaches his attachment figure and may make contact, perhaps only momentarily and without apparent urgency. In our sample a few children behaved in this way toward the end of Episode 2, but more in Episode 3.
Still higher levels of intensity of activation seem to change the set-goal itself so that mere proximity is no longer sufficient and close bodily contact is required. Such a shift may be occasioned by an alarming environmental stimulus, or by separation or the threat thereof. Under these circumstances approach is likely to be quicker and accompanied or ended by active attempts to achieve close contact or by signals such as reaching and crying. Indeed crying may supplant active approach. If the intensity of activation is extremely high, an infant may abandon active behavior in favor of full-blown crying, as seemed to be the case with some infants who actively sought proximity and contact in Episode 5 but who abandoned themselves to intense signaling in Episode 8.
If, as in separation situations, the mother is not accessible when attachment behavior is intensely activated, there may be either an active attempt to follow her, or crying, or a combination of both. The frustration implicit in unsuccessful efforts to regain the absent mother can be expected to arouse attachment behavior to a still higher level of intensity, so that when the mother returns her mere presence is unlikely to terminate the behavior. Indeed attachment behavior at the moment of reunion and for a short time afterwards is likely still to be intense.
Thus, a variety of different attachment behaviors have the same predictable outcome—namely, gaining or maintaining proximity to an attachment figure or, under higher degrees of activation, gaining or maintaining bodily contact. One infant may cry and reach and thus induce his mother to come and pick him up, whereas another may make a beeline toward her and then clamber up on her. In this sense attachment behaviors are interchangeable to a degree. Because of this we had from the beginning more confidence in our scaled measures of proximity/contact seeking and of contact maintaining than in discrete measures of the frequency of the component behaviors that enter into these measures (e.g., approaching, crying, reaching, touching, and the like).
The difference between behaviors commonly emitted at low intensity of activation of the attachment system and those emitted at high intensity is so great that it seems useful to consider them as two subclasses of the general class of attachment behavior. Both promote proximity but they do so in different ways and under different circumstances. Bowlby (1969) proposed that attachment behavior has evolved because its predictable outcome of proximity of infant to mother favors protection and hence survival. High-intensity behaviors are easy to perceive as promoting a protective function. Whether because he is alarmed or because he is anxious about being separated from his mother, a child who does his best to get close to his mother may even have gaining protection as his conscious intent, although we do not posit that this is necessarily so. Low-intensity behaviors seem to serve the function of attachment behavior more indirectly.
Bowlby described how smiling and vocalization tend to attract the caregiver to the infant and to induce him or her to linger close by. These, and enthusiastic greeting behaviors, do much to evoke from the adult the parental behaviors that are the reciprocal of attachment behaviors, even though they are emitted under circumstances that are not crucial for the infant’s survival, as indeed Sroufe and Waters (1977a) have suggested. Crying might be described as implying: “Come, I need you desperately!” Smiling and vocalizing might be interpreted as implying: “Stick around, I enjoy your company!” To the extent that the caregiver does remain reasonably close, for whatever reason, the protective function of attachment behavior is served. Furthermore, the reciprocal bond of caregiver to infant tends to be cemented thereby. Even after the infant becomes increasingly capable of maintaining proximity through his own efforts, and relies less on signals to attract people to come to him and to “stick around,” a good proportion of the behavior he directs toward attachment figures is low-intensity attachment behavior. Smiling becomes supplemented by many nuances of facial expression and gesture, and vocalization begins to shade into the early stages of language. These forms of communication support interaction with attachment figures, and there is no doubt that they continue to maintain the attachment bond. Therefore, they indirectly continue to serve the biological function of this bond, even though they seem on the face of it to have little to do with protection.
Another distinction between low- and high-intensity attachment behavior is that the components of low-intensity behavior are not reserved for attachment figures. As we pointed out earlier, smiling frequency is scarcely differential to the mother in contrast with relatively unfamiliar figures even after the infant has learned to discriminate between them. Vocalization and other modes of nonverbal communication are surely used in interaction with nonattachment figures, even though they tend to be differential to attachment figures longer than smiling is. Even approach, as Tracy, Lamb, and Ainsworth (1976) have shown, is barely differential in frequency to the mother in contrast with a relatively unfamiliar visitor in the last quarter of the first year. What remains sharply differential is high-intensity attachment behavior—the approach when crying, the approach to be picked up, and, in general, behavior directed toward achieving and maintaining close bodily contact, whether it be signaling or a more active initiative.
A comprehensive study of attachment must include observation of both high- and low-intensity behaviors, as well as those that might be identified as characteristic of intermediate levels of activation. The strange situation comprehends the gamut from low to high intensity, but obviously focuses on the high-intensity behaviors. It is only because the infants in our samples, especially in Sample 1, were also observed under more extensive circumstances of low-intensity activation of attachment behavior that we have been able to appreciate the vast differences attributable to the situational context. The issues raised here have profound implications for the study of stability of individual differences in behavior at different times and in different kinds of situations, but these implications are discussed in a later section.
Finally, we do not consider measures of the strength of proximity and contact seeking—let alone measures of the frequency of smiling, vocalization, or looking—as measures of the strength of attachment. The very fact that there is such a shift in the nature and intensity of attachment behavior under different conditions and levels of activation suggests that the strength of attachment behavior reflects the situational intensity of activation rather than some postulated underlying strength of the bond between infant and attachment figure. We have said this repeatedly (Ainsworth, 1963, 1967, 1972, 1973; Ainsworth & Bell, 1970) but have had difficulty in convincing others—so ingrained in various current psychological paradigms is the notion that any construct such as attachment must have a high—low dimension of strength or intensity. Without wanting to imply that attachment is a drive—which it is not—it is useful to compare it with the hunger “drive.” The more that is known about hunger, the more evident it becomes that there are many conditions, both internal and external, that influence the strength of food-seeking behavior and of food intake at any given time. Even hypo- and hyperphagia can be accounted for in terms of these conditions. It is no longer useful—if, indeed, it ever was—to think of strength of hunger drive as a significant dimension of an individual’s personality.
Termination of Attachment Behavior. The conditions for termination of attachment behavior activated at a high level of intensity differ from those sufficient to terminate low-intensity behavior. Most of the relevant facts are implicit in our previous discussion. When attachment behavior is strongly activated, it is likely that only close bodily contact with an attachment figure will terminate it. Indeed, in the reunion episodes, especially in Episode 8, infants who had been distressed by separation needed to be picked up by their mothers before they were soothed, but most of them were quickly soothed by close bodily contact. This did not necessarily imply that attachment behavior was terminated, because a significant number of babies resisted or protested release if the mother attempted to put them down soon after they had been picked up. Apparently some prolongation of close bodily contact was necessary to terminate attachment behavior to the extent that exploratory behavior could again take over, even though it was unusual for this prolongation to involve more than a minute or two.
The concept of set-goal suggests that, unless the set-goal itself has shifted, the mere restoration of the limits of proximity in spatial terms may terminate attachment behavior. Thus, for some infants, perhaps especially those older than the 1-year-olds of our sample, the mother’s return to the room seems to be sufficient, although this might not be the case if the mother did not acknowledge or greet the child when she returned. Still others seem to require closer proximity to the mother than before, as though their set-goal had narrowed, even though it did not shift to requiring close bodily contact.
Interplay Between Attachment Behavior and Other Behavioral Systems. The chief behavioral systems activated in the strange situation appear to be: exploratory behavior, wary/fearful behavior, attachment behavior, sociable behavior, and angry/resistant behavior. Exploratory behavior is antithetical to attachment behavior in the sense that approach to the toys decreases proximity to the mother, although it is not uncommon for toddlers to compromise by bringing the toys closer to the mother and playing with them there, or by involving the mother in play. We are assuming that wary and fearful behaviors are manifestations of the same behavioral system, with wariness resulting from low-intensity activation, and fearfulness from activation at higher levels of intensity. At all levels of intensity of activation, however, wary/fearful behavior is antithetical to exploratory behavior and sociable behavior in that it militates against approach to and manipulation of (or interaction with) those features of the unfamiliar physical or social environment that have activated the wariness/fear system. Wary/fearful behavior is not antithetical to attachment behavior, and at levels of activation beyond mild and brief wariness, it is usually congruent with attachment behavior. Indeed the same stimulus situation that activates wariness/fear at a moderate to high level of intensity tends simultaneously to activate or to intensify attachment behavior. The child is likely to move away from the alarming stimulus and/or toward the attachment figure, provided that one is accessible; or he may signal to the attachment figure, perhaps by crying.
We are distinguishing between attachment behavior, directed toward the mother as representing the class of figures to whom an infant has become attached, and sociable or friendly behavior, directed toward the stranger as representing the class of figures to whom he has not become attached. These two systems of behavior are antithetical in the sense that when a child directing behavior toward one figure, he usually can not simultaneously direct behavior toward the other. As we suggested earlier, the particular behaviors that serve the attachment system may also enter into sociable behavior with a nonattachment figure, especially low-intensity attachment behaviors. High-intensity attachment behaviors tend to be reserved for the attachment figure.
When two antithetical systems are activated simultaneously, they may be said to be in conflict. This conflict may not be readily apparent if one system is activated at a level of much greater intensity than the other; the more strongly activated system tends to determine the overt behavior. The other system may not become manifest in behavior until either the overriding behavior is terminated (or becomes less strongly activated) or some shift in the situation increases the activation of the system until it overrides the behavior of the previously stronger system. When two conflicting systems are more nearly equal in level of activation, there may be alternation of behaviors, “compromise” behaviors in which behavioral elements of both systems are combined, or intention movements or other fragmentary behavioral representatives of one or the other system. Furthermore, the behavior activated by one stimulus object may be redirected toward another that is not involved in the conflict—as in “displacement,” as the psychoanalysts would label it. Finally, overt behavior may be determined by a third system, which is also at a moderate level of activation, although at not as high a level as the two conflicting systems that tend to block each other—a phenomenon that ethologists label “displacement behavior.”
The interplay between exploratory, wary/fearful, and attachment behavior may be seen in Episode 2; certain aspects of this have already been discussed relevant to the use of the mother as a secure base from which to explore. For most infants exploratory behavior overrode both wary/fearful and attachment behavior, either with or without a brief delay during which one may presume wary behavior was dominant at first but soon weakened when the child perceived nothing really alarming about the toys that he was also stimulated to approach. A few infants approached the mother toward the end of the episode, perhaps because their temporal set-goal for this situation had been exceeded, thus intensifying attachment behavior enough to at least temporarily override exploratory behavior.
The interplay between wary/fearful, sociable, and attachment behavior may best be seen in Episode 3—exploratory behavior directed toward the toys having been overriden in most infants by some combination of the other three behavioral systems that were activated (or intensified) by the entrance of the stranger. Most infants could be described as being in a state of conflict between wary/fearful behavior and sociable behavior, both of which were activated by the stranger. Some behavior could be interpreted as expressing both of the conflicting systems simultaneously—for example, coy behavior, intention movements, and tentative responses to the stranger’s offer of a toy. In other behavior, the competing tendencies alternated, as for example when approach to the stranger was followed immediately by rapid movement away from her usually toward the mother. In the latter case, attachment behavior was clearly involved in the conflict, as it was also in instances in which the infant, wary/fearful of the stranger, retreated to the mother as a secure haven from which vantage point he turned to examine the stranger, still wary of her. Attachment behavior was eventually overridden by sociable behavior (or possibly by a combination of sociable and exploratory behavior) in most infants, who were attracted away from the mother by the stranger’s inviting him to play with the toy she offered. Wary behavior continued to conflict with exploratory and/or sociable behavior, however, for few infants in Episode 3 did more than tentatively reach toward the stranger’s toy.
In the separation episodes, attachment and exploratory behaviors were in some conflict. Attachment behavior was intensified by the mother’s departure and/or continuing absence. In many children attachment behavior was activated so strongly that it quite overrode exploratory behavior; the child explored little but cried, or searched, or did both in an attempt to regain his absent mother—especially in the second separation Episodes 6 and 7. In Episodes 4 and 7 attachment and sociable behavior were also in conflict, with wary/fearful behavior also likely to have been involved.
Especially in Episode 7 attachment behavior, possibly supported by the wariness/fear system, overrode sociable behavior in most cases. The conflict of most interest occurred in infants whose attachment behavior was activated to such a pitch that they seemed about to accept the stranger as a substitute attachment figure; yet in nearly all cases an approach or a signal for contact was succeeded by resistance or avoidance. The conflict in such cases seemed to be an approach—avoidance conflict with the stranger as focus. It could be hypothesized that attachment behavior was manifest in the approach but that wariness/fear and/or anger (prompted by the fact that the stranger was not an attachment figure) were manifest in the resistance or avoidance.
Conflicts in the reunion episodes are of particular interest. To be sure, conflict seemed at a minimum in Group-B babies. Attachment behavior, having been activated at high intensity by separation (especially by the second separation), overrode all other behavioral systems when the mother returned, and for varying periods afterward until attachment behavior was terminated—or at least sufficiently reduced in intensity so that (perhaps with the mother’s cooperation) exploratory behavior was again activated. But in A and C babies, conflict was evident even in the reunion episodes. We should like to defer discussion of these instances of conflict to the context of individual differences, because they are at the nub of the whole issue of individual differences.
Shortcomings of Our Normative Research
The strange-situation procedure has proved to be useful far beyond our initial expectations. Even after the thorough analysis of our findings that we are reporting here, there is very little in our procedure that we would like to change. In regard to normative behavior, however, a few refinements of the procedure would make it a more powerful instrument for future research. Begun in 1964, our use of the strange situation antedated most ethological studies of facial expression, postural orientation, and gesture. Although the most obvious of these did not escape our observers’ attention, there is much in the findings of ethological research on nonverbal communication in infants and young children that would have enriched the narrative records of our observers had they been trained to take account of them. In part this implies desirable further training in observation, but in part it merely implies a consistent vocabulary in terms of which observation can be reported. Without a lexicon of such communicative behavior (e.g., Blurton Jones, 1972; Brannigan & Humphries, 1972; McGrew, 1972), it is difficult for an observer to quickly put into words a description of the facial expression, posture, and gesture that he sees.
One of the categories of behavior that our observers sometimes reported was “tension movements,” including fingering clothing, repetitive movements such as pulling at an ear lobe, tense movements such as hunching the shoulders, putting the hands behind the neck and tensely cocking the head, and so on. It was our clear impression that such tension movements signified stress, both because they tended to occur chiefly in the separation episodes and because they tended to be prodromal to crying. Indeed, our hypothesis is that they occur when a child is attempting to control crying, for they tend to vanish if and when crying breaks through. Sometimes our observers noted such behavior clearly, characterizing it as a “tension movement;” sometimes they noted such behavior in purely descriptive terms, which made it difficult to identify as reflecting tension; and sometimes undoubtedly they did not mention it at all. In any event, such behaviors were not reported frequently enough, clearly enough, or in enough detail for us to be able to code tension movements or to devise a measure thereof.
Another closely related class of behavior that is not represented in our analyses is “displacement behavior.” As ethologically defined, this behavior occurs when two other strongly activated behavioral systems conflict, effectively blocking the full expression of either; then a third moderately activated behavior (usually commonly appearing in the individual’s repertoire) may find expression, this being the displacement behavior. We suspect that some exploratory behavior, especially in the separation and reunion episodes, operated as a displacement behavior. We had the impression, for example, that some Group-A babies (who neither evinced separation distress nor sought to be close to their mothers upon reunion) “explored” in a hyperactive way in such episodes, showing no investigative interest in the objects that they were either manipulating or moving toward, but rather banging them about repetitively or throwing and retrieving them repeatedly. Such an impression was very rarely recorded by the observers, however, who tended to confine themselves to a descriptive account of the locomotor and manipulative movements. Thus, although we have a hunch that displacement exploration might be qualitatively distinguished from more genuine investigative exploration, it is impossible to make the distinction in our present data in any systematic way. Another type of displacement behavior may have been thumb- and finger-sucking (which in Episode 7 was especially characteristic of Group-A infants).
Heart-rate records, such as those used by Sroufe and Waters (1977b), might very well have been a useful “convergent measure” to enable us to identify both tension movements and displacement behavior. Indeed, such a measure used in one crucial strange-situation study might lead to the sharpening of the behavioral criteria of tension and displacement to the extent that subsequent studies could rely solely on detailed behavioral records.
Finally, no videotape equipment was available to us throughout our strange-situation research; had it been available we would undoubtedly have used it. Certainly for a record of facial expression, postural orientation, gesture, resistance, avoidance, and tension movements, it would have enabled us later to retrieve detail not included in the observers’ narrative accounts.
There are advantages and disadvantages to each kind of record. Narrative records depend for their usefulness on the observational and narration skills of the observer-narrators. Transcribing the dictated narratives is time consuming if done by research personnel (and expensive if a typist is employed), the more so if two or more observer-narrators are used, as of course they should be. Once transcribed, however, dictated narratives require relatively little time to examine for coding and scoring. Furthermore the transcriptions are permanent records that can be stored, available for reexamination. New variables can emerge from dictated narratives—and indeed in our research all of our measures of interactive behavior did so emerge—through reading the narratives over and over again, an enterprise that requires a fraction of the time required for viewing videotapes repeatedly.
Videotapes require a skilled cameraman who is alert to precisely what is of critical interest for the analysis of strange-situation behavior. A single camera may not be able to pick up behavior of both child and relevant adult when they are at a distance from each other, as well as from the observer-narrator. If one can use a split-screen technique, this disadvantage may be overcome by having two or three cameras focusing on the action from different vantage points and/or with different persons as targets. This requires a skilled and knowledgable person monitoring all cameras and combining their records to the best advantage. Coding and scoring require the same skill of the observer-coders as the other method requires of the observer-narrators, but training toward acquiring such skill is easier than training in the “live” situation. Coding and scoring are infinitely more time consuming for videotapes than for transcribed narrative records. Repeated viewing is possible, however, and through this it is entirely possible that new variables can emerge—indeed, with more confidence than for transcribed narratives. Videotape records may be permanent and thus available for reexamination; on the other hand, this is expensive, and often the tapes must be erased after the initial coding, thus eliminating the possibility of later reexamination. Finally, for better or for worse, videotape records can substantially alter the measures that can be derived. In lieu of frequency per 15-second interval measures, precise measures of duration of a behavior can be obtained, although this would be most useful for measures that are not very important components for our classificatory system. In the case of our interactive measures, an overwhelming amount of detail can be provided by videotape records, which might lead the investigator to short-cut methods of scoring and classification.
Only one of us (EW) has used videotape records of strange-situation behavior, supplementing them with transcriptions of dictated narratives. He has come to rely mostly on the videotape record, in the conviction that a good videotape record can comprehend all of the essentials of a dictated narrative, as well as provide more detail that may be very useful for resolving discrepancies in scoring and for elucidating behaviors not at present included in our scoring system. Otherwise, all of the studies reported or reviewed here (except Lamb, 1978) have relied solely on dictated narratives for recording the strange situation. The choice of methods of recording, we believe, can be left up to the individual investigator who wishes to use our strange-situation procedure, in accordance with his own preference, experience, and resources. The fact that our procedure was founded on dictated narrative records rather than on videotape recording should not discourage those who have come to rely on videotaping. In our opinion the dictated narratives make it easier to use our procedures and to make use of our findings, but the videotape records facilitate the extension of our work toward a more detailed and subtle understanding of attachment and attachment behavior.
1 There is general agreement that before a baby may be described as having become attached to a figure, he must discriminate that figure from others. It can only be demonstrated that such discrimination has been acquired through the fact that he behaves differently toward that figure than toward a comparison figure. There is, however, difference of opinion as to whether mere discrimination of and preference for a figure constitute attachment to that figure or whether attachment emerges first in a later stage of development. (See Chapter 1.) Even at such a later stage of development, however, the differentiality with which attachment behaviors are manifested toward various figures serves as a useful criterion for the identification of those figures to whom an infant has become attached.
2 Kotelchuck found this to hold for infants in his 12-, 15-, 18-, and 21-month-old age groups, but not for those 6 or 9 months old. The latter two groups showed no crying that was clearly associated with the departure of a specific figure.
3 This analysis was based on his longitudinal observation of infants in the relatively nonstressful environment of the home. Some, but by no means all, infants at 7 and 8 and 12 and 13 months showed clear preference for one parent over the other. During the second year of life, significant sex differences emerged, with boys preferring their fathers (p < .02). Although five of the six children who preferred their mothers were girls, some girls preferred their fathers. This analysis did not, however, examine preference in stressful environments.