Subgroups and Their Usefulness - Measures and Methods of Assessment

Patterns of Attachment: A Psychological Study of the Strange Situation - Mary D. Salter Ainsworth 2015

Subgroups and Their Usefulness
Measures and Methods of Assessment


The findings so far reported have tended to omit reference to the subgroups of the strange-situation classificatory system. None of the three major groups—A, B, and C—is large enough to support a discriminant-function analysis for its component subgroups. Within Sample 1, for which measures of infant and maternal behavior at home are available, the number of infants in each of the subgroups is too small to make it reasonable to expect significant differences among them. Therefore, it is difficult to assemble convincing evidence of the validity of the distinctions between subgroups that are made by the classificatory system. This being so, the reader may wish to conclude that such distinctions might well be ignored.

We are, however, reluctant to abandon subgroup distinctions at this stage of our knowledge, even though the accumulation of a large enough sample for the proper investigation of subgroups is a task for the future, probably contingent on the pooled findings of a number of independent investigators. Our reasons for reluctance are as follows. First, some of our findings with Sample 1 strongly suggest that infants in different subgroups within the same major classification may have had different kinds of experiences, and that these may make for different patterns of interaction with their attachment figures and for qualitatively different attachment relationships. Some evidence pertinent to this point is reported shortly. Meanwhile let us merely state our conviction that a more refined identification of “outcome” patterns offers more aid to the identification of relevant antecedent variables than does a cruder kind of identification.

Second, the specifications for classification into subgroups are more explicit than those for classification merely into one of the three major groups. Therefore, the use of the classificatory system is likely to be more reliable for classification even into the major groups if the judges in question first make (or attempt) a subgroup classification. At the same time it seems wise to think of any classificatory system as openended. It is inconceivable that any system based on a relatively small sample could comfortably accommodate all patterns represented in the total population from which the sample is drawn. We conceive of it as being relatively easy to extend the number of subgroups, if required, in order to comprehend new patterns encountered in further samplings, and indeed this was once done in our study in order to comprehend the infants now classed in Subgroup B4. So far our experience has encouraged us to believe that the three major classificatory groups can, in fact, cover the variety of patterns of infant—mother attachment relationships commonly encountered. This is so, we believe, because the specifications for the classification into subgroups have enabled us to abstract the specifications for the three major groups in ways that are more incisive than they would have been had subgroups not previously been identified.

Finally, we consider the subgroups to be the foundation of the classificatory system. It may not be relevant that the subgroups were identified first, in the process of grouping together strange-situation protocols that were maximally similar. Nevertheless, it was through examination of the similarities among members of each subgroup that our attention was first drawn to those variables whose patterning reflected important qualitative features of the infant—mother attachment relationship.

Because of these considerations, we proceed to present our findings relevant to subgroups.

Distribution of Infants Among Strange-Situation Subgroups

Table 26 shows the distribution of infants in each of the four samples and in the total sample among the subgroups. It may be seen that Subgroup B3 is the largest in the sample, and accounts for 42% of the total sample. We consider it to be the normative group, not merely because it is the largest, but also because, as we subsequently show, it is the subgroup whose members have had the most harmonious interaction with their mothers and who have established the most secure attachment relationships. The rest of the subgroups are small, especially B4, C1, and C2.

Subgroups and Maternal Behavior

Perhaps the major piece of evidence to date that has confirmed our decision to retain subgroup distinctions is the set of subgroup means for Sample 1 in regard to the fourth-quarter ratings of maternal behavior—first reported by Ainsworth, Bell, and Stayton (1971) and shown in Table 27. Here Subgroups B1 and B2 were combined, because there was only one infant in B1, and her mother’s scores were similar to those of the B2 mothers.

TABLE 26 Distribution by Sample of Infants Among Strange-Situation Subgroups

Table 26 Distribution by Sample of Infants Among Strange-Situation Subgroups

TABLE 27 Subgroup Means on Scales of Maternal Behavior in the Fourth Quarter

Table 27 Subgroup Means on Scales of Maternal Behavior in the Fourth Quarter

On each of the four scales, the mothers of B3 babies received the highest mean score, and one each of the mothers of the B1/B2 babies came next. This consistency is reflected in a coefficient of concordance significant at the level beyond .01. All of the mothers of the A and C infants were conspicuously insensitive to infant signals and communications. Group-A mothers, especially A1 mothers, were more rejecting than C mothers. A2 and C2 mothers were the most inaccessible and ignoring. A1 mothers were the most interfering. We consider these differences so important that we discuss them in more detail.

The mothers of B3 babies are sensitively responsive to the baby’s signals and capable of perceiving things from his point of view. The B3 mother views her baby as a separate individual; she respects his activity-in-progress and therefore avoids interrupting him. She accepts the intermittent attachment behavior that leads him to her.

The mothers of B1 and B2 babies may be described as inconsistently sensitive in their responsiveness. For various reasons they were also inconsistently accessible to the baby, sometimes giving him much attention and sometimes ignoring him. During the periods of attention, the four mothers in these subgroups were somewhat interfering, tending to interrupt exploratory play. In two cases there was clear mismatch in regard to timing of desire for close bodily contact; the mothers sometimes interrupted the baby to cuddle him when he was interested in play, only to rebuff him later when he himself sought contact.

The mothers of A1 babies were more rejecting and interfering than the mothers of the babies of any other subgroup. They were quite unable to see things from the baby’s point of view or to respect his autonomy. They did not so much ignore the baby’s communications as they discounted them as relevant guidelines, and thus were very arbitrary in their interventions.

The mothers of A2 babies were inaccessible for prolonged periods. They were bored with the maternal role and found other interests to occupy them both at home and away from home. When they entered the room in which the baby was, they were usually so preoccupied with their own thoughts or activities that they failed to acknowledge his presence. They tended to respond only to very strong and persistent signals from the baby. Because they paid the baby so little attention, they were infrequently interfering, although they could not be described as cooperative or “codetermining.” They seemed to reject the baby along with the maternal role.

The mothers of C2 babies were also highly inaccessible and ignoring. Unlike the A2 mothers, however, they had a strong investment in the maternal role. They were severely disturbed women. Multiple demands on their time provoked fragmented behavior. To hold themselves together, they often had to ignore the baby and “tune out” his crying. Especially in the first quarter, they left the baby alone to cry for prolonged periods. When the C2 mother did intervene, she did so arbitrarily, even though the intervention itself might be pleasant. Because the C2 infant could rarely experience a consequence contingent on his own behavior, it is not surprising that he behaved very passively both in the strange situation and at home, whereas A2 infants, whose strong persistent signals eventually evoked a response, developed more active modes of coping.

The mothers of the two C1 babies were disparate, except that both were highly insensitive to infant signals. One was highly interfering, although not rejecting. She continually interrupted her daughter to train her, to show off her accomplishments, or merely because she herself felt like playing with the baby or showing her affection. The other C1 mother was compulsive, preoccupied, and generally unresponsive to all but emergency signals. On the other hand, she scored mid-scale in cooperation—interference, being excessively controlling in the feeding situation but not otherwise. Although both little girls behaved similarly in the strange situation, the background of mother—infant interaction differed, and to a much greater extent than in the case of any other subgroup.

To this account we can add the findings for Main’s four variables, shown in Table 28. Again it is clear that B3 mothers emerge with the best ratings on the whole—in this case, the lowest ratings. They neither provided their babies with unpleasant experiences when in close bodily contact nor did they themselves find contact aversive. They showed no substantial lack of emotional expressiveness and were not rigid or compulsive

As shown in chapter 8, Main’s hypotheses were supported by the high scores Group-A mothers received on these four variables, but there are interesting individual and subgroup differences. A1 mothers were high in aversion to physical contact and in providing the baby with unpleasant experiences in contact. The two A2 mothers were similar except that one of them was judged not to have provided the baby with unpleasant physical-contact experiences, which of course lowered the A2 mean. Both A1 and A2 subgroups have relatively high means on lack of emotional expressiveness, which Main linked with suppressed or repressed anger. The A1 mean would have been even higher had it not been for one mother whose anger toward the baby was quite overtly expressed.

For one variable—aversion to physical contact—the B1/B2 mean is higher than those of any but the two Group-A subgroups, and it is also relatively high in lack of emotional expressiveness. One B2 and the one B1 mother suffered from postpartum disturbance during the baby’s first quarter-year—a disturbance that left them less responsive to their babies than they might otherwise have been, and indeed less responsive than they later became. This, we believe, is why they both scored relatively high in lack of emotional expressiveness. In addition, the B1 mother received a high rating on aversion to physical contact.

TABLE 28 Subgroup Means on Further Scales of Maternal Behaviora

Table 28 Subgroup Means on Further Scales of Maternal Behaviora

As Main hypothesized, Group-C mothers differ from Group-A mothers in regard to three of the four variables here under consideration. Indeed, most of the mothers—and it must be recalled that there were only four—resembled Group-B mothers on the whole. None of them showed any substantial aversion to close bodily contact. One C2 mother gave the baby highly unpleasant physical-contact experiences, but this was for reasons quite different from those that seemed pertinent with A1 mothers—reasons associated with ignoring and neglect rather than interference, anger, or rejection. One C1 mother was very bland and hence received a high score for lack of emotional expressiveness. C2 mothers were more rigid and compulsive than C1 mothers, and indeed received the same mean rating as A2 mothers. Their compulsiveness seemed to be a mode of coping with their tendencies to become fragmented when exposed to any degree of stress, and was less consistently shown than the rigidity—compulsiveness of Group-A mothers. However, only research with future samples that include larger numbers of C1 and C2 babies can clarify our understanding of the patterns of maternal behavior associated with these subgroups. Meanwhile the evidence from our small sample of Group-C mothers suggests that they differ from Group-A mothers in regard to Main’s four variables.

We have also examined subgroup differences in regard to the fourth-quarter measures of maternal behavior derived from the coding of the narrative reports. We do not present a table comparable to Tables 25 and 26, because for the most part the subgroups tend to reflect the trends characteristic of the three main groups, A, B, and C. However, we do report a few subgroup differences that override group trends, but only in instances in which there was minimal variation within the subgroups themselves.

A2 mothers, much less frequently than the mothers of other subgroups, including A1, acknowledged their babies upon returning from an absence from the room. C1 mothers, more frequently than those of any other subgroup, picked up their babies primarily in order to play with them. Most of the other notable differences pertain to maternal interference. Although on the whole B1/B2 mothers resemble B3 mothers more closely than the mothers of any of the A or C subgroups, they more frequently picked up their babies in an abrupt and interfering way than did B3 mothers, and they were also more often inept in their holding. C1 mothers, more frequently than others, and especially more frequently than C2 mothers, intervened physically to back up their verbal commands to the baby. Both B1/B2 and C1 mothers issued more frequent verbal commands than mothers of infants in other subgroups. A1 mothers, however, were the most interfering in the context of physical contact.

Finally, let us consider three studies that offer evidence to justify the distinctions between B1/B2 and B3 subgroups in light of maternal behavior. In Chapter 9 we reported the findings of Tolan (1975) and Tomasini (1975). In their analyses of the behavior of the mothers of toddlers in Main’s (1973) laboratory play session, they distinguished the behavior of three groups—B3, B1/B2, and A/C—as representing three degrees of security—insecurity in the attachment relationship. Tolan found significant correlations between maternal facial expression and degree of security—insecurity of the infant’s attachment to his mother, and Tomasini found significant correlations of maternal sensitivity—insensitivity and acceptance—rejection with security—insecurity of attachment. We assume that the intermediate degree of security—insecurity provided by the B1/B2 subgroups contributed to the magnitude of the correlations reported by these investigators. Blehar, Lieberman, and Ainsworth (1977), in their study of early face-to-face interaction, also considered the same three groups, with B1/B2 intermediate between B3 and A/C. Only one difference emerged as significant for either maternal or infant behavior—namely, B3 mothers were more likely than B1/B2 mothers to be contingent in their pacing of their behavior to mesh with infant behavior. Nevertheless, for most of the measures—pertaining both to mother and infant—the B1/B2 subgroup was truly an intermediate group. Especially noteworthy is that it differed from the B3 subgroup.

Thus, even though the subgroups in Sample 1 are too small to support estimates of significance of intersubgroup differences, the observable differences are impressive in that they suggest different forms of mother—infant interaction that tie into different dynamic processes affecting infant development. Our accounts of these processes must remain tentative, however, until they can be tested in other samples. Nevertheless, subgroup differences do suggest different processes, and this we consider to be a potent justification for continuing to distinguish subgroups.

Subgroups and the Attachment—Exploration Balance at Home

Ainsworth, Bell, and Stayton (1971) also reported on the congruence between subgroup classification of strange-situation behavior and classification of Sample-1 infants in terms of their behavior at home. The classificatory system was geared toward an assessment of the balance between attachment behavior and exploratory behavior in the home environment. Five groups were identified, and their behavior may be summarized as follows:

Group-I infants showed an optimum balance between attachment and exploratory behavior. They used the mother as a secure base from which to explore the world. The Group-I infant could freely move away from his mother to explore, but he would keep track of her whereabouts, and from time to time would gravitate back to her again, perhaps to gain physical contact with her, if only briefly.

Group-II infants at times showed a disturbance of this balance. This disturbance seemed to be in reaction to maternal behavior and to represent a mismatch between mothers’ and infants’ wishes for contact. Thus, the Group-II infant sometimes behaved precisely as a Group-I infant would; at other times he avoided his mother; at still other times he sought contact with her more anxiously than Group I-infants usually do.

Group-III infants tended to display active exploratory behavior with much less frequent or intense interest in either proximity or in close bodily contact with the mother than is characteristic of either Group-I or Group-II infants.

Group-IV infants seemed conflicted about proximity to and close bodily contact with the mother. The Group-IV baby did explore, although perhaps more briefly than infants of Groups I, II, and III. He was anxious about his mother’s whereabouts and actively attempted to keep in proximity to her. On occasion he actively sought contact with her; but he did not seem to find pleasure in it once achieved, and indeed may have been markedly ambivalent about it.

Group-V infants tended to be passive either in attachment behavior or in exploration, or in both—some only intermittently and others more consistently. Stereotyped, repetitive, autoerotic activities were more frequent in this group than in any other.

We are now not altogether satisfied with this classificatory system and we believe that it can be improved through revision in light of data analysis that has subsequently been completed—a revision that we will not attempt to gear primarily to the concept of attachment—exploration balance. Nevertheless, a comparison between the strange-situation classification into subgroups and this present classification yields highly suggestive results. (See Table 29.) Let us discuss the congruencies and discrepancies between the two classifications. In fact, there are only four instances in which the members of a strange-situation subgroup do not all fall into the same attachment-exploration balance classificatory group. In the case of three of these, mother—infant interaction had deteriorated during the last few months of the first year. The classification in terms of home behavior seemed to better reflect this deterioration, whereas the strange-situation classification seemed to better reflect an earlier state of affairs.

It may be seen from Table 29 that Subgroup B3 is nearly coincident with Group I. One B3 infant, however, was classified in Group II in regard to home behavior. The mother was conspicuously sensitive to her baby’s signals throughout the first 9 months, but then became anxious and depressed because of marital difficulties that suddenly surfaced. Consequently, in the fourth quarter this mother alternated between responsiveness to the baby and impatient rebuff.

B1/B2 infants tend to fall in Group II. Both A2 babies, as well as one B2, were classified in Group III—the groups that emphasized exploratory behavior at the expense of proximity seeking. The exceptional B2 baby enjoyed much interaction with his mother during his first 5 months or so. At this point a family crisis put the mother under such strain that she could not tolerate the baby’s recently acquired and active seeking both for interaction and for floor freedom. She confined him in his crib alone in his room for most of the day, and thus was conspicuous for inaccessibility and ignoring. It is therefore not surprising that at home he behaved like the A2 babies whose mothers were also highly ignoring.

TABLE 29 Classification of Strange-Situation Behavior and Classification of Attachment—Exploration Balance in Behavior at Homea

Table 29 Classification of Strange-Situation Behavior and Classification of Attachment—Exploration Balance in Behavior at Homea

Most A1 infants, as well as one C1 baby, fall into the conflicted Group IV. Finally, Group-V, which may be the most disturbed of all, included both C2 infants as well as one each from A1 and C1. The split of the C1 subgroups between Groups IV and V should arouse no surprise, for it has been acknowledged from the beginning that the two Sample-1 C1 mother—infant dyads were disparate. The A1 baby classified in Group V is worthy of mention, however. Although his mother had been both rejecting and ignoring from the beginning, she became in addition anxious and inconsistent during the fourth quarter, again presumably because of severe marital difficulties. Because of disruption in the home, the baby underwent several separation experiences, and the marriage broke up soon afterwards.

In brief, these findings indicate that classification in terms of strange-situation behavior is strikingly congruent with classification in terms of behavior at home in the fourth quarter, even though the former may be “phenotypically” different from the latter. Perhaps more important, especially for our present consideration of the value of retaining subgroup distinctions, it is clear that B3 babies behave differently from B1/B2 babies at home, and that A2 babies are strikingly different from A1 babies. Furthermore, the passivity of C2 infants is borne out in their home behavior.

Subgroups and Other Measures of Infant Behavior at Home

In regard to the infant-behavioral measures derived from the coding of the narrative reports, there are a few subgroup means that deviate from the general trend of the group means, although we do not present a table comparable to Tables 16 and 17. It is clear that B1/B2 infants differed from B3 infants in regard to some measures of behavior relevant to close bodily contact; they were less frequently positive (and more frequently negative) in their response to contact, although they were more often positive than Group-A infants. They also more frequently initiated being put down than any other subgroup except C1. These findings are consistent with the fact that B1/B2 infants tended to fall into Group II of the home-behavior classification—the group in which there tended to be a mismatch between mother’s and infant’s timing of desire for close bodily contact. In addition, B1/B2 infants were found by Blehar, Lieberman, and Ainsworth (1977) to be intermediate between B3 and A/C infants in regard to a number of features of behavior in face-to-face situations between 6 and 15 weeks of age.

C1 infants were more often negative to being picked up than the infants of any other subgroup, and less often responded positively to being put down. In these respects they differed strikingly from C2 infants. It would appear that C1 infants were overtly ambivalent about physical contact with their mothers—as indeed they were also in the strange situation—even though they showed much less evidence of approach—avoidance conflict than A babies.

The passivity of C2 infants is reflected in a number of behavioral measures. Less often than any other subgroup did they initiate either being picked up or being put down, and less often did they follow their mothers when they left the room. They received a zero score on compliance to mother’s commands, but the major discriminator here is that they moved about so little that their mothers issued few or no commands. Perhaps related to their passivity is the fact that the C2 infants received lower DQs in the fourth quarter than the infants of any other subgroup, although A1 babies also lagged in this respect.

Finally, A1 infants more-frequently followed when the mother left the room than did babies of any other subgroup, whereas A2 infants did so less frequently than Group-B infants. More frequently than A2 infants, A1 babies cried or gave mixed greetings when the mother returned from an absence.

In summary, even though the subgroups are too small to assess statistical significance of differences between them, there is suggestive evidence that intragroup subgroup differences in the strange situation correspond to differences in behavior at home. This being so, we have somewhat more justification than we otherwise would have for examining strange-situation behavior for evidence of differences between subgroups.

Subgroup Differences in Interaction with the Mother in the Strange Situation

Figure 11 shows the mean scores of each of the subgroups for each of 5 measures of interactive behavior in each of the episodes from 2 to 8. (Appendix VI, Table 33, gives the means and standard deviations for the subgroups, based on the total sample of 106.)

Figure 11 Means of subgroups on five measures of interactive behavior directed toward the mother.

FIGURE 11 Means of subgroups on five measures of interactive behavior directed toward the mother.

It may be seen at a glance that the differences between A1 and A2 infants are confined to the reunion episodes, and that these differences are clearly in line with the criteria for classification. A2 infants, in comparison with A1 infants, show stronger proximity seeking and somewhat weaker avoidance in the reunion episodes. In addition, in Episode 8 (when A2 but not A1 infants were picked up by their mothers) they showed some contact maintaining and resistance.

Subgroup B1 is of particular interest as a group intermediate between Group A and the rest of Group B. This subgroup was originally included in Group B because of the positive reaction the infants showed to the mother’s return in the reunion episodes. This was specified as a positive, undelayed greeting to the mother when she returned, followed by little or no avoidance and by interaction with her across a distance, even though the baby did not seek proximity or contact. Figure 11 reflects these specifications, except that there is a moderate amount of avoidance shown in the reunion episodes—in Episode 5, it is almost as much as that shown by Subgroup A2. Furthermore although the distance-interaction measure was introduced specifically because of this subgroup, even it does not adequately take into account greeting behavior. In retrospect, we are inclined to conclude that some infants were classified in Subgroup B1 who might more appropriately have been classified in Group A, probably because the avoidance behavior (which is the hallmark of Group A) was mistakenly given lesser weight than the pleasant distance interaction characteristic of these babies. If this is indeed true, then Connell and Rosenberg (1974), as a preliminary to their multiple discriminant function analysis based on our data, had some justifiable grounds for excluding B1 infants from Group B—although surely only a minority of our B1 infants could have been misclassified.

Subgroup B2 matches the specifications very well. Up to Episode 8 they are indistinguishable from B1 infants, but then they show strong proximity-seeking and contact-maintaining behavior with a minimum of avoidance. In Episode 8 they resemble B3 infants more strongly than they do B1 infants.

The behaviors of B3 infants very closely match the criteria for classification. They are particularly conspicuous for seeking to gain proximity to and contact with the mother in the reunion episodes and to maintain contact once achieved, especially in Episode 8. What especially distinguishes them from the other subgroups, however, is their almost complete lack of avoidant and resistant behavior.

Subgroup C1 differs from B3 in a number of respects, chiefly consistent with criteria for classification. The major difference, however, is that C1 infants show strong resistance in the reunion episodes, whereas B3 infants show little or none. The major surprise is that C1 infants were scored as high as they were in avoidance in Episode 8, although the mean is substantially less than that of Group-A infants. It is possible that one or two infants classified as C1 primarily because of strong resistant behavior might better have been classified in Group A. Indeed, the infant who appears as A0 in Table 24 (i.e., unclassified as to subgroup) was originally put into Group C and then reclassified. As a result of the MDFA, with its great emphasis on avoidance behavior following reunion with the mother as the main discriminant between A and non-A behavior, we are inclined in retrospect to give more emphasis to strong avoidance behavior than to the combination of proximity-seeking, contact-maintaining, and resistant behaviors; thus, if all of these behaviors appear, the strong avoidance would call for classification in Group A.

A cardinal characteristic of C2 infants was specified as passivity. Because our scoring of interactive behavior (except for distance interaction) reserved the highest scores for behavior showing strong, active initiative, it might be expected that C2 infants would score lower on these measures than C1 infants—and indeed this is borne out in Figure 11. Nevertheless it is clear that C2 infants resemble C1 infants much more closely than they resemble infants of any other subgroup, except for B4.

Subgroup B4, although the smallest in the sample, is of special interest. It was introduced into the classificatory system because Bell (1970) found three infants in her sample of 33 who seemed “wholly preoccupied with their mothers throughout the strange situation.” They were clearly more anxious in the infant—mother attachment relationship than other Group-B babies. From what information Bell could gather in the course of home visits undertaken for other purposes, it appeared that these infants had all recently undergone an anxiety-provoking experience, such as major separation, but that both mother and infant were positively oriented to each other, and appeared to be in the process of mending their relationship. It is clear from Figure 11 that B4 is intermediate between B3 and Group C and is closer in the pattern of scores to C2 than to C1. In the reunion episodes, B4 babies resemble Group-C babies, except that they show less resistance to the mother than do C1 infants, and more active proximity-seeking behavior than C2 infants.

Finally, it is clear that distance interaction with the mother is less differentiating among groups and subgroups than any of the other four interactive variables. It appears that for 1-year-olds in an unfamiliar situation, with cumulatively increasing instigation to high intensity attachment behavior, behaviors relevant to proximity and contact (both positive and negative) are more conspicuous and more differentiating than the so-called distal behaviors.

In summary, the finding for five behavioral measures of infant initiative in interaction with the mother match the criteria for classification into subgroups very well on the whole, with but few exceptions.

Group and Subgroup Differences in Greeting the Mother Upon Reunion

The baby’s initial greeting to his mother when she entered in the reunion episodes was specifically represented in the criteria for classification, but was not scored separately from the interactive behaviors that characterized each reunion episode as a whole. Consequently, greeting behaviors were not represented in the multiple discriminant function analysis. Because positive greeting (without proximity seeking) was especially crucial in the identification of Subgroup B1, as one of the behaviors that were specified to distinguish B1 from Group-A infants, a descriptive ethological analysis was undertaken of responses to the mother during the first 15 seconds of each of the reunion episodes. The findings of that analysis are summarized here.

Group-A infants were conspicuous for the relative absence of proximity-seeking greetings; if they greeted the mother at all, it tended to be with a mere smile or vocalization, and many did not greet her. In Episode 5, 35% greeted her with a smile or vocalization across a distance, whereas only 9% approached her. Most conspicuous, however, was failure to greet; 45% merely looked at the mother and 9% did not even look. In. Episode 8 these avoidant tendencies seemed even stronger; 59% merely looked, and again 9% did not even look. Only 18% greeted the mother across a distance, and only 9% made even a partial approach.

The B1 infants could not be distinguished from B2 infants by their behavior in the first 15 seconds of the reunion episodes. Of all subgroups B1/B2 babies were the most conspicuous for smiling and vocalizing in greeting across a distance; 45% did so in Episode 5 without any proximity-seeking behavior, although fewer did so in Episode 8. In Episode 5, 29% reached, leaned toward, or approached the mother; in Episode 8, more (43%) did so, suggesting that their attachment behavior was more intensely activated by the second separation than by the first. In contrast with Group A, only 14% of B1/B2 infants merely looked at the mother without greeting her in Episode 5, and 23% did so in Episode 8. None ignored the mother altogether. After the first 15 seconds were over, B2 babies, especially in Episode 8, showed proximity-seeking behavior, whereas B1 infants did not. On the other hand, more B1 than A2 babies greeted their mothers across a distance, and fewer subsequently mingled proximity-seeking with avoidant behaviors. The hallmark of B1 infants, therefore, is not so much in the fact that they greet the mother with a smile or a vocalization rather than with proximity-seeking behavior (or with a cry), but that they tend subsequently not to follow the initial greeting with either proximity-seeking or avoidant behavior.

The majority of B3 infants clearly showed a desire for close bodily contact or at least for increased proximity. In Episode 5, 33% approached the mother as soon as she entered, and 43% more reached or leaned toward her. Only 11% were content merely to smile or to vocalize across a distance. In Episode 8, 36% approached the mother, 40% reached for a pick-up, and 20% were so distressed that they merely signaled their desire for contact by renewing the intensity of their crying. Thus, in the second reunion episode 96% indicated a desire for closeness in their initial greeting. Although 4% merely greeted the mother across a distance, even in Episode 8, no B3 baby in either episode failed to look up when the mother entered.

The four B4 infants were less active than B3 infants in their greetings. In Episode 5, three continued or increased crying, and only one reached toward the mother for a pick-up. Their relative passivity continued in Episode 8. None gave smiles or vocalizations across a distance, but none failed to greet.

Only one Group-C infant greeted the mother merely with a smile in Episode 5, and none failed to give some kind of greeting in either episode. In Episode 5, only 3 (12%) approached the mother when she entered, and 2 of those were C1 babies. Nevertheless all but one (93%) indicated a desire for contact or proximity because they either approached the mother, or reached toward her, or cried. In Episode 8, 14% approached and 36% reached toward the mother, but 50% merely cried. It appears that they were more severely distressed than B3 infants, and less capable of active proximity seeking in their initial greeting responses. In this they resembled B4 infants.

Thus, in summary, initial responses to the mother’s return seem useful for distinguishing among subgroups. They are particularly useful in distinguishing B1 and B2 babies from Group-A babies on one hand and from B3 babies on the other hand, and for distinguishing B3 from both B4 and Group C.

Subgroup Differences in Other Behaviors

Subgroup Differences in Interaction With the Stranger. A detailed analysis was conducted, but only a summary of findings is reported here. Subgroup means and standard deviations are shown in Appendix VI, Table 34.

Subgroup B1 showed less distance interaction with the stranger than Group A in Episode 7, and whereas Group-A infants manifested no resistance, some B1 infants showed some resistance. Indeed, in regard to the means of all behaviors in all three episodes—3, 4, and 7—B1 babies closely resembled B2 and B3 babies.

Avoidance of the stranger occurred most frequently in Episode 3, while the mother was still present, but was conspicuous only among B4 and C1 babies, However, subgroup B4 can be distinguished from both C subgroups in terms of resistant behavior. C babies were conspicuous for resistance to the stranger in the separation episodes, whereas B4 babies showed almost none. On the whole, therefore, B4 infants resembled the infants of the other B subgroups more closely in their behavior to the stranger than the infants of Group C.

A1 and A2 babies cannot be distinguished by their behavior to the stranger, which featured a moderate amount of distance interaction. C2 infants resembled C1 infants, except that they showed little avoidance of the stranger in Episode 3, in contrast to the strong avoidance shown by C1, and that they did not show the marked resistance to the stranger in Episode 7 that is such a conspicuous feature of C1 behavior.

Subgroup Differences in Search Behavior. The means and standard deviations for each subgroup in regard to search behavior in the separation episodes are given in Appendix VI, Table 35. The findings are easily summarized. Subgroups B4 and C2 showed very weak search behavior even in Episode 6, when the babies of all other subgroups tended to show moderately strong to strong search.

Subgroup Differences in Exploratory Behavior. The means in Appendix VI, Table 36, show that exploratory behavior was maintained relatively well across all episodes by four subgroups A1, A2, B1, and, to a lesser extent, B2. B3 infants were slightly less exploratory, even in the preseparation episodes, than the infants of these four subgroups. The babies of three subgroups—B4, C1, and C2—explored less throughout than any of the other subgroups, and they explored almost not at all in the second separation and reunion episodes.

Subgroup Differences in Crying. Crying was very infrequent among the infants of subgroups A1, A2, B1, and B2, especially in the preseparation episodes and in the first separation and reunion episodes. Indeed it was only in Episode 6, when the baby was alone, that other than minimal crying occurred. (See Appendix VI, Table 36.) B3 babies cried little except in the separation episodes, and cried clearly more in the two second-separation episodes than in the first. Crying was much more characteristic of Subgroups B4, C1, and C2. Furthermore, these infants cried more in the reunion episodes than did the infants of other subgroups, suggesting that they were less readily soothed.

Summary. Exploratory behavior and crying scores differentiate between B1/B2, B3, and B4, but they distinguished neither B1/B2 from Group A nor B4 from Group C. Search behavior merely differentiates B4 and C2 babies from the rest. Interaction with the stranger, however, tended to distinguish B1/B2 from Group A and B4 from Group C, although it yielded little differentiation among the Group-B subgroups, or indeed between the Group-A subgroups. It did, however, yield some distinction between C1 and C2.


In this chapter we have assembled evidence pertinent to the issue of retaining the distinctions among subgroups in our classificatory system. The patterns of behavior reflected in the subgroup means of measures of interactive behavior and in initial greeting responses in the reunion episodes differentiate among the subgroups in close approximation to the specifications of the instructions for classifying into subgroups. Although crying, search, and exploratory behavior provide relatively little differentiation among most of the subgroups, findings match the specifications for classification in the few instances in which references were made to these behaviors. Responses to the stranger yielded some intersubgroup differences, but none had been specified by the instructions for classification. In general these findings demonstrate not only that the distinctions in the classificatory system could be objectified in quantitative measures, but also that the various judges who undertook the classifications could and did make the distinctions in accordance with specifications.

Although it is of obvious relevance to note ways in which the subgroups may be distinguished from one another in terms of strange-situation behavior this does not in itself demonstrate that the distinctions are worth making. Of all the evidence reviewed here, we attach most importance to that which shows that babies in different subgroups behave differently in interaction with their mothers in other situations, and especially in the natural environment of the home. This evidence is so far of limited extent, but it suggests that it is worth making distinctions between A1 and A2, between Group A and B1/B2, between B1/B2 and B3, and between C1 and C2. It does not help us with distinctions between B1 and B2, between B4 and B3, or between B4 and C2. A major difficulty is, of course, the size of the subgroups in any one sample, which is too small for an extensive and intensive investigation to be made of mother—infant interaction in other situations, especially longitudinal investigation of the development of such interaction. Indeed the whole issue of the usefulness of subgroups in our classificatory system is an empirical issue that depends on a number of replications of research that relates patterns of strange-situation behavior to behavior in other situations and/or to differences in mother—infant interaction.

The issue of Subgroups B1 and B4 presents somewhat different problems from other subgroup issues. These are clearly “borderline” subgroups—B1 clearly being intermediate between B2 and Group A, and B4 equally clearly being intermediate between B3 and Group C. Connell and Rosenberg (1974) suggested that B1 properly belonged in Group A, and B4 in Group C; they excluded both of these subgroups from their discriminant-function analysis on this account. This, we believe, is still an open question, which can be settled only by further research with behavior in other situations (especially mother—infant interaction at home) that provide external criteria. In the meantime, B1 babies also closely resemble B2 infants; combined into a group intermediate between B3 babies on the one hand and A/C babies on the other, they have been found to show clear-cut differences in mother—infant interaction. Even though much less is known about B4 infants, they are clearly less resistant to their mothers in the reunion episodes of the strange situation than the Group-C infants whom they otherwise resemble in many ways. In view of these considerations, we are inclined to keep both B1 and B4 in Group B until the issue is settled through further research.

Finally, for the reasons stated in the introduction to this chapter, we believe that retention of subgroup distinctions serves useful purposes in the present state of our knowledge. Foremost among these purposes is the forwarding of “etiological” research. Here we refer to research geared toward identifying patterns of antecedent variables that are associated with different “outcome” patterns. Some of these outcome patterns may suggest incipient pathology, but probably even more important is research into the antecedent variables associated with healthy outcomes. Indeed, we propose that the major heuristic value of the strange-situation procedure is that assessments based on it may themselves serve as criteria—a set of outcome criteria—through which the effects of different patterns of infant experience may be evaluated. The research reported in this volume—we refer especially to the research reported in Chapters 7, 8, and 9—suggests that the strange-situation classificatory groups are likely to prove even more useful as outcome criteria than the “behavioral category” scores implicit in our interactive behavioral variables, and certainly more useful than scores of separate, “discrete” attachment behaviors. Furthermore, we believe that the refinements of classification offered by distinctions among subgroups will in time prove even more useful than classification into the three major groups themselves.