Young children’s representation of people who are elsewhere—Or dead

Attachment Theory and Separation

In the 1950s, young children who were treated in London hospitals faced a strict regime of isolation. Parents were typically permitted to visit their children for only 1 or 2 hr per week. In some hospitals, they were not allowed to visit at all or permitted to see children only though a partition or when asleep. In an effort to record children’s reactions to such prolonged and rigorous separation from their parents, James Robertson made a documentary film of one such hospitalized child, titled A 2 year-old goes to hospital. The film was shown by Robertson at a meeting of the British Psychoanalytic Society in February 1952. One week later, at a further meeting of the Society, J. Bowlby et al. (1952) presented a paper discussing the film and offering an interpretation of the child’s distressed reaction to parental separation.

Because Bowlby and his colleagues did not invoke standard psychoanalytic concepts in their interpretation, the paper met with a stormy reception. Subsequently, critical commentaries appeared in The Psychoanalytic Study of the Child—the leading UK journal for the promulgation of psychoanalytic approaches to child development. However, one of these commentaries was constructive and thoughtful. Anna Freud pointed out that when a child is hospitalized, it is not easy to pinpoint the main trigger for the distress that the child might show. Admittedly, one plausible interpretation is that the child is distressed by the prolonged separation from his or her parents. However, the child is also faced with an unfamiliar physical environment, bearing little resemblance to home. In addition, the child is typically looked after by a rota of nursing staff and cannot seek reassurance from an alternative, familiar caregiver, such as a grandparent (Freud, 1960). In short, children’s distress during hospitalization might not be due solely to separation from the parents.

To better understand the causes of such distress, James Robertson, in collaboration with his wife Joyce Robertson, undertook a small intervention study with four children, each about to be separated from their mother (who was due to be hospitalized for a period). Joyce Robertson got to know each child ahead of the upcoming separation and the child then stayed at the Robertsons’ home during their mother’s absence. All four children coped quite well. Indeed, they did not show the 3-stage sequence of protest, despair, and detachment that Bowlby (1960) had described as the consequence of separation. Moreover, on being reunited with their mother, their bond with her was quickly re-established. There was little sign of the avoidance or detachment previously observed among some hospitalized children. The Robertsons concluded that the availability of a familiar, alternative caregiver forestalled the usual sequence of protest, despair, and detachment. Their interpretation was bolstered by the observation of a fifth child who was not looked after by the Robertsons but taken into group care, while his mother was hospitalized. This child did show considerable distress, both during the separation and afterwards on reunion with his mother. The Robertson and Robertson (1971) synthesized their findings in a report—Young Children in Brief Separation —in which they emphasized young children’s ability to cope with separation, provided they have access to an alternative, familiar caregiver.

Somewhat surprisingly, Bowlby was reluctant to accept the conclusions of his erstwhile collaborator and colleague. In his book Separation, Bowlby (1973) reviews the Robertsons’ findings but carefully underlines whatever intermittent upset each of the four children had displayed. Prior to its publication, an exchange of letters between Bowlby and Robertson betrays an increasing tension between the two men (Van der Horst & Van der Veer, 2009, p. 247). Robertson wrote: “I have long had reservations about the uses to which you have put my early institutional data and . . . my reservations are strengthened by the findings of our current project Young Children in Brief Separation.” Correspondence between the two men abruptly ceased in 1972 and the breach was never repaired (Van der Horst & Van der Veer, 2009).

I believe there are two good reasons for re-examining this disagreement. First, Bowlby’s view of the emotional consequences of separation has ended up dominating Attachment theory. For example, in the first paragraph of the opening chapter of the recent, 1000-page Handbook of Attachment, Jude Cassidy (2016), a leading attachment researcher, writes as follows: “Bowlby, along with his colleague James Robertson, observed that children experienced intense distress when separated from their mothers, even if they were fed and cared for by others. A predictable pattern emerged—one of angry protest followed by despair” (p. 3). Given its importance for Attachment theory, the disagreement between the two men with respect to the consequences of separation warrants discussion and investigation. Yet in this summary statement, the disagreement goes unmentioned. Indeed, Cassidy implies a nonexistent consensus between Bowlby and Robertson, despite the availability of persuasive evidence to the contrary.

A second reason for re-visiting the disagreement is that the young child’s reactions to being left by the mother in an unfamiliar environment, either alone or with a stranger, have been canonized as the primary index of attachment, as measured in the Strange Situation (Ainsworth & Bell, 1970). Yet the questions raised by Anna Freud and the subsequent findings of the Robertsons suggest that there are at least two important components to a child’s reaction to that situation: the departure of the mother but also the absence of any familiar caregiver. By implication, individual differences among infants observed in the Strange Situation might be due to: (1) individual differences in their reactions to their mother’s departure; or (2) individual differences in their reactions to the unavailability of an alternative caregiver; or (3) some combination of both factors. Yet the large body of research emphasizing the availability and responsiveness of the mother as a key determinant of behavior in the Strange Situation has focused on (1) to the neglect of (2) and (3).

If the presence versus absence of an alternative caregiver had no impact on children’s reactions to their mother’s departure, it could be argued that such misgivings about a foundational tool in Attachment theory may be safely set aside. For example, if infants respond to their mother’s departure with equivalent levels of distress whether or not an alternative caregiver remains available, interpretation of their behavior in the Strange Situation would be less problematic. However, a series of studies conducted in the 1970s showed that if the child is separated from one parent but left with the other parent, crying is rare and continued play is frequent. By contrast, if the child is left with a stranger, crying is frequent and play is disrupted (Lester et al., 1974; Ross et al., 1975; Spelke et al., 1973).

Additional food for thought has emerged from societies in which infants and children are looked after by a variety of caregivers. How do they respond to their mother’s departure and return? We can obtain an initial answer to this question by looking closely at life among the Aka, a group of foragers in the Central African Republic whose child-rearing practices have been carefully observed from the 1990s onward by Barry Hewlett and his colleagues. In general, infants of hunter-gatherers, including the Aka, spend most of the day and night in physical contact with caregivers, receive prompt attention from them when needed, are breast fed on demand, and rarely cry or fuss. However, caregiving is not provided exclusively by the mother but is shared. Not surprisingly, given this allocentric pattern of care, infants are almost always physically close to one or more caregivers. More specifically, in the course of the first year, whether in the first or the second 6 months of life, infants are rarely left alone. Meehan et al. (2017) confirm that this feature of early child care has proven stable across 20 years of study.

Aka encampments average around 25–35 people living in six to eight huts. With rare exceptions, food that results from a foraging or hunting expedition is shared among camp members and this pattern of sharing is echoed in the pattern of caregiving. More than 50% of Aka infants are breast fed by women in addition to their mothers. Indeed, alloparents carry out approximately one quarter of all caregiving and a wide age range of individuals serve as caregivers—juveniles, adults, and elders—both males and females.

Meehan and Hawkes (2013) offer a vivid, moment-to-moment documentation of the life of a young child raised within such an encampment. Each of 29 children—ranging from 2 weeks to 32 months—was followed for a total of 9 hr on a given day—from early morning to evening. Every 30 s, the observer noted on a score sheet which of several potential behaviors, such as fussing, or approaching a caregiver, or crawling into the lap of a caregiver, the child was engaged in. Collapsing across all the 29 children who were followed, more than 30,000 such observations were recorded. From these records, we probably know as much about a day in the life of an Aka toddler as we do about a day in the life of a toddler raised in Baltimore or Berlin.

As noted, an encampment includes a considerable number of individuals who provide care. So, on any given day, Meehan and Hawkes (2013) observed approximately 20 such individuals, sometimes more and sometimes less, caring for or interacting with a given child, with that number remaining relatively stable across the age range of children studied. However, care was not equally distributed. Even among the Aka, mothers are the primary caregivers especially during the first year, as indexed by the frequency with which children were in physical contact with her, or within easy access. That said, among 2- and 3-year-olds, maternal contact and access declined, whereas alloparental contact and access remained just as frequent. In other words, the primacy of the mother was less evident after the first year of life.

It would be easy to misconstrue how such collective parenting functions. It is not the case that infants and toddlers are always passive recipients of care from lots of different individuals. Especially from the age of 6 months, children exercise choice and discernment. To document this selectivity, Meehan and Hawkes (2013) kept track of the number of individuals to whom children directed attachment behaviors—such as approaching a caregiver, expressing a desire to be held via gestures or vocalizations, crawling into the lap of a caregiver, and so forth. On average, children directed attachment behaviors to six different caregivers—a subset, therefore, of the much larger number of individuals who offered care to them. Not surprisingly, the child’s mother was the target of most of these attachment bids—about two-thirds overall—but the remaining third were distributed across a variety of alloparents, including fathers, other adults, and other juveniles.

Although mothers received a good deal of help in child care, they were rarely absent altogether. They typically remained present within the encampment. Nevertheless, they did sometimes leave for short periods. Overall, they were absent for around 20 min on any given day, with most absences (80%) lasting 10 min or less. How did infants respond to these brief separations? If the earlier analysis is correct, granted the overall availability of familiar caregivers and granted infants’ demonstrable ability to direct attachment bids in a selective fashion, we can reasonably expect Aka infants to respond to their mother’s absence with equanimity. To underline the point explicitly, these naturally occurring absences should not be conflated with what happens in the Strange Situation. In the Strange Situation, the child is in an unfamiliar setting and is left either alone or with a stranger. Among the Aka, when the mother leaves, the child remains in a familiar setting in the presence of familiar caregivers. Indeed, children very rarely cried, either during or immediately after a separation. Fussing—with no tears—was a bit more frequent. But most separations—three quarters—passed off with neither tears nor fuss. In any case, there was invariably an alloparent nearby who could provide comfort if needed. Reunions were similarly calm. Indeed, only 30% of them provoked attachment bids by the child as indexed by attempts to get or stay close to the mother.

How should we explain this emotional tolerance for maternal absence? One possible interpretation, consistent with classic Attachment theory, is that most infants had sensitive mothers and trusted her to return promptly, especially since her absence was generally brief. However, no relationship was found between maternal sensitivity, on one hand, and children’s distress during separation, on the other. By contrast, Meehan and Hawkes (2013) found that there was a link between the behavior of alloparents and children’s distress. More specifically, children were less likely to fuss or cry during their mother’s absence if their alloparents were, on average, sensitive in their caregiving. Again, we see that a mother’s absence does not invariably lead to child distress, just as the Robertsons concluded, albeit in a different context. Instead, a key determinant of children’s emotional reactions to separation from their mother was the availability of sensitive caregivers who were familiar to them.

Looking back at the history of Attachment theory, it becomes clear that one of its central tools—the Strange Situation—is aptly named. First, children raised in communities where allomothering is standard practice, will virtually never have the experience of being left in a strange environment, whether alone or with a stranger. For them, the Strange Situation would be more than strange, it would be unprecedented. Indeed, for that reason Meehan and Hawkes (2013) opted—understandably—to study Aka children’s reactions to naturally occurring separations from their mother rather than attempting to recreate the standard, laboratory-based, Strange Situation. But even when infants are tested in the Strange Situation in the United States or Western Europe, we ought not to understate its strangeness. In a Western home, especially one in which mother and child often spend time alone, there is a reasonable likelihood that a 12-month-old will sometimes be left in one room while his or her mother goes to get something from another. In addition, many infants will be left to sleep in one room while parents sleep in another nearby. But such separations take place within an area whose layout will be familiar to infants, and where one or more caregivers remain within earshot, to be summoned if needed via vocalization or crying. In the Strange Situation, by contrast, these parameters do not obtain—the spatial environment is unfamiliar, the mother has left for an unknown location, and if she is summoned by the infant’s persistent distress calls, the procedure is halted, and the data are excluded. Hence, we should probably be cautious in extrapolating from the Strange Situation; its unfamiliarity may sometimes trigger distress and discombobulation rather than characteristic reactions to caregiver absence.

In sum, when separated from an attachment figure, toddlers do not always show distress. Their distress is much reduced if they are left with another, familiar caregiver or if they can seek out a familiar and sensitive caregiver. That conclusion holds for the more prolonged separations studied by the Robertsons and for the shorter-term separations that feature in experimental and observational studies or serve as a key component of the Strange Situation.

Talking about Missing Persons

The conclusions reached in the previous section raise an interesting but neglected question. When young children are separated from one familiar caregiver but remain in the care of another familiar caregiver, do they think and talk about the absent caregiver? Indeed, this question can be seen as part of a broader question about early language and cognitive development, namely when do children move beyond the here and now to start producing so-called decontextualized utterances? Most research on that topic has focused on temporal displacement—children’s ability to talk about past or future episodes. But that line of research leaves open the question of when and how children talk about spatial displacement and absence—phenomena that are spatially removed from the context of the utterance. In particular, it does not indicate whether infants and toddlers talk about people who are currently elsewhere and not part of an ongoing conversation.

As it happens, the Robertson and Robertson (1971) offer some clues: the two older children in their study (Thomas, aged 28 months and Kate, aged 29 months) did talk about their absent parents during their stay with the Robertsons. They made contact-related comments, for example, “I want my Mummy and my Daddy” as well as more emotionally neutral or reflective comments, for example, “I thinking about my rocking horse at home. Mummy says, ‘No. Mustn’t climb’; makes Mummy very cross.” But it is unclear from the Robertsons’ report if comments of either type were rare or frequent. More generally, children’s references to absent family members have not been systematically analyzed.

Keeping in mind the disagreement between Bowlby and Robertson, how will children who are separated from one caregiver but have an alternative caregiver available respond? On one hand, children might be concerned about the separation and produce many contact-related comments about the absent caregiver—in line with Bowlby’s broad conclusions about young child’s distress at separation from a caregiver and his interpretation of the findings reported by Robertson and Robertson (1971). On the other hand, the availability of an alternative caregiver might reduce or override any concerns about separation—so that contact-related comments about absent caregivers should be rare or nonexistent. Instead, more reflective comments might predominate.

Two recent studies begin to offer an answer to this question. Zhang and Harris (2022) studied the utterances of three Mandarin-speaking children (aged 20–40 months). Yang et al. (2021) studied the utterances of four English-speaking children (aged 24–59 months). In each case, in the context of observational studies of early language acquisition, children’s unstructured, naturalistic conversation had been recorded in their homes, while they were being looked after by one or more family members (e.g., a parent and/or grandparent) in the absence of other family members. Initial analysis of children’s utterances focused on how often (per hour) they referred to another person by name. These references were then divided into references to a person who was present during the conversation or absent from the conversation. Figure 1 shows the number of references per hour to absent versus present persons that were made by young Mandarin-speaking children (left-hand panel) and by young English-speaking children (right-hand panel).

OPEN IN VIEWER

Figure 1.

Number of References Per Hour to Absent Versus Present Persons.

Inspection of Figure 1 shows that young children mention a person who is present every 2–3 min and someone who is absent every 7–8 min. The rate of absent person references is similar among Mandarin- and English-speaking children and remains relatively stable among 2-, 3-, and 4-year-olds—unlike other aspects of displaced speech, which tend to increase with age.

However, it might be objected that children’s references to absent persons in the context of everyday conversations at home are mainly a consequence of comments or prompts by their conversation partners. Children might be responding to an interlocutor’s query (e.g., “Where’s Mommy?”), parroting a comment made by their interlocutor (“Daddy’s at work”) or responding to a prompt or reminder (e.g., a photograph of a grandparent). Accordingly, in a follow-up analysis, the frequency with which children made “fully spontaneous references” (i.e., references that had not been produced or prompted by their interlocutor in the preceding stretch of conversation) was calculated. Such fully spontaneous references occurred approximately each 15 min. Inspection of Figure 2 shows that this pattern emerged for both the Mandarin- and English-speaking children and remained relatively stable among 2-, 3-, and 4-year-olds.

OPEN IN VIEWER

Figure 2.

Number of Fully Spontaneous References Per Hour to Absent Persons.

To assess the function of such references to absent persons, especially absent family members, they were divided into two types—contact-related and contact-unrelated. In making a contact-related utterance, children voiced attachment-related concerns in one of three ways: (1) they expressed their desire for the absent person (e.g., “I want my Daddy”); (2) they repeatedly vocalized the absent person’s name (e.g., “Daddy, Daddy, Daddy”); or (3) they asked after the absent person’s whereabouts (e.g., “Where is Daddy?”). By contrast, in making contact-unrelated utterances, children produced reflective or emotionally neutral comments about the absent person rather than contact-related comments. Figure 3 shows the frequency of contact-related and contact-unrelated references to absent family members by young Mandarin-speaking children (left-hand panel) and by young English-speaking children (right-hand panel).

OPEN IN VIEWER

Figure 3.

Number of Contact-Related and Contact-Unrelated References Per Hour to Absent Family Members.

Inspection of Figure 3 reveals that contact-unrelated references to absent family members tended to be more frequent than contact-related comments with one exception. Among the youngest group of children, namely, Mandarin-speaking children aged 1;8–2;2 years, contact-related comments were more frequent. By contrast, among the oldest group of children, namely, English-speaking children aged 4;0–4;11 years, contact-related comments rarely occurred. When the data from the two studies are considered together, they suggest that the frequency of contact-related comments declines sharply with age, whereas the frequency of contact-unrelated comments remains stable with age.

Returning to the disagreement between Bowlby and Robertson, the following synthesis appears plausible. First, consistent with the findings of Robertson and Robertson (1971), children can cope quite well with the absence of an attachment figure. They do not display severe distress provided that another familiar attachment figure is available. Nevertheless, in line with Bowlby’s emphasis on the emotional sequelae of separation from a caregiver, the availability of an alternative attachment figure does not completely override all attachment-related concerns regarding an absent caregiver: 2-year-olds are prone to make contact-related comments about such an absent caregiver. Indeed, such comments are not infrequent. The youngest children (aged 1;8—2;2 years) studied by Zhang and Harris (2022) made such comments approximately every 10 min. However, by the age of 3 years, such contact-related comments are less frequent than more discursive or reflective comments and by 4 years they rarely occur at all. Finally, although it is tempting to assume that contact-unrelated comments serve a different function from contact-related comments—and are not linked to attachment concerns—it is worth keeping in mind an alternative possibility, namely, that such reflective conversation about absent caregivers can serve as a source of comfort or reassurance. We know that among adults, symbolic reminders of absent attachment figures—via either words or photographs—can provide comfort and reassurance (Master et al., 2009; Mikulincer et al., 2005; Selcuk et al., 2012). More broadly, children’s ability to mentally represent members of their immediate, social network—and to talk about those who are currently absent—might come to serve some of the functions that are ordinarily supplied—especially for younger children—by the physical presence of those individuals.

Suggestive evidence for the comfort provided by a symbolic reminder of an attachment figure was reported in an experimental study of the emotional effect of photographs. Passman and Longeway (1982) examined how toddlers, ranging from 20 to 30 months, responded when they were left alone with or without a photograph of their mother. An experimenter introduced children to a playroom, giving them either a clear and recognizable photograph of their mother or alternatively one that was blurred and unrecognizable. The child’s mother, who had stayed outside the room, said, “Goodbye, I’ll be back soon. You play with the toys.” Toddlers were then left alone in the playroom, but the door remained open so that they could leave the room to go look for her if they wished to. Children were observed for up to 5 min or until they left the playroom.

When left alone, children responded very differently to the clear versus blurred photograph. They held the clear photograph for a longer time and gazed at it more often. Children evidently recognized their mother because they often labeled the clear photograph—something they never did with the blurred photograph. The quality of the photograph also affected children’s behavior in the playroom. There was a good deal of individual variation in the time that children spent with the toys and in how much they moved around, but most of the children with the blurred photograph left the playroom well before the full 5 min was up—after about 1 min and 30 s on average. By contrast, less than half of the children with the clear photo left the playroom—and as a group they stayed for about 2½ min on average. By implication, the children who were given a photograph of their mother took some comfort from it—they were less likely to feel the need to quit the playroom in search of her.

To conclude this section, although the relatively slow development of children’s decontextualized language, especially with respect to past and future episodes, might be taken imply that young children will rarely think or talk about absent family members, recent findings show that such talk is quite frequent. Thus, around the second birthday, children make contact-related as well as more reflective comments. Among 3- and 4-year-olds, reflective comments predominate. In providing symbolic reminders of an absent caregiver, such comments may provide reassurance that the child’s social circle remains available and intact.

Children’s Conceptualization of Death

Granted that young children can remember and talk about absent members of their immediate social circle from an early age, how do they respond to the permanent absence that is entailed by death? Two lines of research indicate that children do not conceptualize death as a biological terminus that severs all ties to the deceased. First, not only bereaved adults but also bereaved children frequently display a continuing bond with an attachment figure who has died. Second, alongside their biological conception of death, many children endorse the possibility of some form of afterlife. I take up these two lines of research in turn.

In a longitudinal study of London widows who were interviewed several times during the first year of their bereavement, Parkes (1970) found that in the first month, most widows not only reported being preoccupied by thoughts of their dead husband, including memories that were accompanied by visualizations of him, but also reported that their husband felt near to them. Indeed, 1 year later, more than half continued to have a sense of his continuing presence. Parallel findings were obtained in a follow-up study of Boston widows (Glick et al., 1974). One year after their bereavement, many widows reported a continuing sense of their husband’s presence and of his watching over them. Indeed, many said that they deliberately invoked his presence when they were feeling uncertain or depressed. Building on such findings, researchers concluded, contrary to the conception of grief that dominated 20th-century research and clinical practice, that such continuing bonds—as they came to be called—were not pathological but typical. The bereaved do not relinquish their ties but frequently maintain some form of connection to loved ones who have died (Klass et al., 1996; Klass & Steffen, 2018).

Recent research has consolidated this claim. Continuing bonds among bereaved adults are manifest in a variety of ways, including: (1) keeping items associated with, or belonging to, the deceased; (2) reminiscing with others about the deceased; (3) endorsing the idea that despite, physical absence, the deceased remains “a loving presence”; (4) acknowledging the positive influence of the deceased on current identity; and (5) retaining and cherishing positive memories of the deceased (Field et al., 2003). Debate continues as to whether such bonds serve to mitigate or accentuate feelings of loss (Bonanno et al., 2004; Field & Friedrichs, 2004; Neimeyer et al., 2006; Root & Exline, 2014). It is likely that they can do both. Reminders of the deceased can evoke feelings of comfort and safety associated with the attachment relationship, but such reminders can also reactivate sorrow about the loss. Indeed, although one or the other function might predominate in certain individuals or at certain stages of grief, particular contexts or events may evoke a mixed or ambivalent reaction, namely concurrent feelings of both comfort and sorrow.

Granted the evidence that children can spontaneously think and talk about absent attachment figures from an early age, we might expect that, like adults, children will also show continuing bonds to a deceased attachment figure (Silverman & Nickman, 1996). Indeed, two developmental reports based on a relatively large sample (N = 125) of children aged 6–17 years enrolled in the Boston-based Child Bereavement Study, echo the findings with adults (Silverman et al., 1992; Silverman & Worden, 1992). When interviewed 4 months after losing a parent, 90% said that they were still thinking about their dead parent several times a week, 81% thought that the dead parent was somehow watching them, 77% kept something personal belonging to their parent, and finally more than half (57%) reported speaking to him or her. Despite the frequency with which children acknowledged these continuing ties to the dead parent, very few (only 3%) stated that they were unable to believe that the parent had actually died.

In a more fine-grained analysis of 24 children from the Child Bereavement Study, Normand et al. (1996) emphasize that for the majority of children (70%), the form of their continuing bond changed over the 2-year study period. For example, some children shifted from a connection based on the preservation of memories to a more interactive communication whereas other children shifted from interactive communication toward a conscious emulation of the goals and behaviors of the deceased parent. Synthesizing the findings from the Child Bereavement Study with those from a parallel study of Israeli children, Silverman and Nickman (1996) conclude that their data on children’s continuing bonds with a deceased parent challenge: “the traditional clinical practice of encouraging the bereaved to disengage from the deceased” (p. 86).

Reviewing a range of clinical reports, Packman et al. (2006) argue that the concept of continuing bonds can also be applied to children who have lost a sibling. Foster et al. (2011) reinforce this claim based on an interview study with bereaved siblings, ranging from 8 to 17 years of age. Siblings were invited to talk about whether and how they stayed “in touch” with their sibling who had died. Like all their parents, almost all of the bereaved siblings (92%) reported the use of various deliberate reminders, such as visual representations, keeping personal belongings, communicating with the deceased sibling, visiting locations associated with the deceased sibling and thinking about him or her. Moreover, consistent with the reports of their parents, these widespread expressions of continuing bonds appeared to be deliberately sought out rather than involuntary. Reports of non-purposeful or involuntary reminders were much less frequent. They were mentioned by only 8% of the bereaved siblings. Finally, bereaved siblings, like their parents, were more likely to mention the comforting as opposed to the discomforting effects of continuing bonds.

Overall, the findings from children and adults indicate that a deceased attachment figure, whether it is a spouse, a child, a parent or a sibling, “lives on” in the thoughts and memories of the bereaved. Indeed, by turning their thoughts to the deceased or retaining their belongings, the bereaved make deliberate efforts to ensure such continuity. Moreover, in many cases—both among children and adults—that continuity is not confined to memories of the past. It can also involve a sense that the deceased remains present, watches over the bereaved, and thereby provides a source of comfort or reassurance.

Can we reconcile the existence of such continuing bonds with the traditional, cognitive-developmental approach to children’s conception of death? That approach has emphasized children’s gradual consolidation of a biological “theory” of death implying the termination of all mental and bodily processes. Indeed, by the age of 9 or 10, most children grasp that death is inevitable, irreversible, and is the fate of all living creatures (Kenyon, 2001). A major contributor to that biological understanding of death is children’s growing appreciation of the body as a set of interconnected organs whose functioning is essential for living processes (Slaughter & Lyons, 2003).

It is conceivable that among bereaved children, this biological account of death as an endpoint or terminus is denied or overridden, especially among those who deliberately maintain continuing bonds with a deceased loved one. However, as noted earlier, among children enrolled in the Boston Bereavement Study, very few stated that they were unable to believe that the deceased parent was actually dead. A more plausible interpretation is that children—and indeed adults—endorse two different accounts of death: a biological account in which death is accepted as the end of all living processes and an afterlife account in which death is not seen as a terminus but as a departure. In the context of this latter account, the dead are presumed to have some type of continued existence, albeit elsewhere and in some altered form. Accordingly, we may ask how far children endorse each of these two accounts, so that a belief in some form of afterlife coexists with children’s construal of death as a terminal biological event.

Harris and Giménez (2005) presented Spanish 7- and 11-year-olds with two narratives about the death of a grandparent. One narrative ended with a biological framing: a doctor explained to a bereaved grandchild that the hospital staff had not been able to cure their sick and elderly grandparent. The other narrative ended with an afterlife framing: a priest explained to a bereaved grandchild that their grandparent was now with God. Following each narrative, children were asked a series of questions about the dead grandparent, notably whether various psychological capacities (e.g., seeing, remembering) and various bodily capacities (e.g., the eyes, the brain) continued to function or not and to justify their answers.

Two key findings emerged. First, children were more likely to say that psychological capacities continue to function after death than bodily capacities. Second, children produced such continuity judgments and justifications more often in the context of the afterlife framing as compared with the biological framing. Subsequent research has shown that this dual stance—in which a biological and an afterlife account of death and its sequelae coexist alongside each other—is not confined to children growing up in Catholic Spain. It has been observed among children and adults from diverse cultural settings, notably the United States (Lane et al., 2016; Watson-Jones et al., 2017), Madagascar (Astuti & Harris, 2008), and the island of Tanna, Vanuatu (Watson-Jones et al., 2017). Moreover, it emerges quite early in development. In the course of an extended interview about death, Rosengren et al. (2014) found that more than a third of US children aged 3–6 years displayed signs of coexistence thinking even in the absence of any explicitly religious framing by the interviewer.

Arguably, such beliefs in a continued existence are part of a cultural narrative—an account that children are led to endorse given the assumptions and practices of their community (Harris, 2012). Indeed, children are likely to learn about death and the afterlife through discussion with parents (Harris, 2012; Menendez et al., 2020) and from observing and participating in family and community rituals marking the departure of the deceased from this world (Astuti, 2011) or their temporary return (Gutiérrez et al., 2015, 2020). However, granted children’s ability—from an early age, as just discussed—to keep loved ones in mind despite their absence, it is also possible that children are receptive to such assumptions and practices because, in key respects, they are consistent with the way that children think about those who are absent, whether through departure or death.

In a striking passage, Marcel Proust (1992) remarks on the parallels between our thoughts about those who have journeyed elsewhere and those who are dead:

People do not die for us immediately but remain bathed in a sort of aura of life which bears no relation to true immortality but through which they continue to occupy our thoughts in the same way as when they were alive. It is as though they were traveling abroad. It is a very pagan survival. (In Search of Lost Time, Volume 5)

Proust remarks that such thoughts offer only a “pagan” afterlife—rather than “true immortality.” But from a less doctrinal perspective, it is feasible that the tendency to think about the dead as if they were elsewhere serves as an early-emerging conceptual building-block for the acquisition and consolidation of afterlife beliefs.

Scrutiny of the euphemistic metaphors that are used in everyday language to talk about death lends support to this conjecture. In a variety of languages, death is conceptualized not as an endpoint but as a departure or transition. Consider the way that death is talked about in English: “She has passed (away),” “He is gone,” “She is in a better place,” “His family said ‘goodbye,’” “He has left this world,” “He is no longer with us,” “Our dear departed,” “She’s been taken from us.” All these phrases treat death as an exodus or leave-taking, not as a terminus or cessation. At the same time, echoing Proust’s remark, any presumption of continuity in these metaphors is not explicitly couched in religious terms—it is a pagan or pre-religious form of continued existence. Nor is this pattern confined to English. Studies of Mandarin, Polish, Serbian, Spanish, Turkish, EkeGusii (a Bantu language of Western Kenya), Paiwan, and Seediq (two Formosan languages) confirm that the metaphor of death as a departure is widespread (Harris, 2018).

Reviewing the argument so far, Attachment theory has focused primarily on negative reactions to separation. However, infants often cope well with separation from one caregiver provided another familiar caregiver is available. Indeed, from 24 months and sometimes earlier, they can keep an absent caregiver in mind—spontaneously talking about him or her. Arguably, this early-emerging ability to keep absent persons in mind impacts the way that children grieve. Like adults, they can maintain continuing bonds to the deceased. Even though children increasingly conceive of death as a biological terminus, they also judge that certain capacities of the deceased, particularly psychological capacities are retained after death. Such beliefs may be rooted in a “pagan” or pre-religious conception of death as a form of life that continues elsewhere, in the wake of a departure from this world. Indeed, the metaphor of death as a departure recurs across the world’s languages.

This proposal synthesizes diverse findings, but there is an important lacuna in the pattern of evidence. In recent cognitive-developmental studies, children have been asked to think about death in relatively impersonal terms. For example, they are presented with a narrative about an unknown or fictional grandparent and asked to judge whether various bodily and psychological capacities of the grandparent will continue after death or not. Moreover, children are typically asked about the continuation of capacities that have no obvious connection to them personally. For example, they are asked whether the dead grandparent can still see something, can still feel happy and sad, can still think, and so forth. Thus, children have not been asked about whether a deceased person to whom they have an attachment continues to have thoughts and feelings vis-à-vis them as individuals.

These considerations raise the possibility that young children who are questioned about an actual bereavement might answer differently from children who are questioned about a hypothetical bereavement. More specifically, if children were questioned about the death of someone to whom they had an attachment, they might be especially prone to conceive of the deceased as continuing to exist, albeit elsewhere—especially if the questions posited the children themselves as the targets of the potential thoughts and feelings of the deceased. Thus, children might readily endorse the possibility that the dead attachment figure continues to think about them, watches over them, or cares about them.

Conclusion

Separation from an attachment figure can be viewed as a challenge to existing affective ties and in the context of Attachment theory, despite early dissent, it has generally been studied as such. The ability of young children to maintain ties to absent caregivers has been studied less often. Yet when earlier and more recent findings are drawn together, they underline the ability of young children to cope with separation from an attachment figure. Particularly, in the presence of another familiar caregiver, children do not show the classic pattern of protest, despair, and detachment. Indeed, children often think and talk about a caregiver who is absent. Arguably, such thoughts and conversations serve to maintain ties and provide reassurance. Indeed, among bereaved children, there are clear indications of such continuing bonds, comparable to those found among bereaved adults. Consistent with the persistence of such bonds, death is often conceptualized not as the end of life but as a departure from this life.