2.1. What is grandparental investment?

Resources invested in one's offspring are referred to as parental investment (see Geary Reference Geary2000). Parental investment was originally defined as “any investment by the parent in an individual offspring that increases the offspring's chance of surviving (and hence reproductive success) at the cost of the parent's ability to invest in other offspring” (Trivers Reference Trivers and Campbell1972, p. 139). Trivers' key point was that the currency of parental investment is its cost to the parent's ability to invest in other offspring. Therefore, the definition of parental investment was highly restrictive. However, a broader definition has evolved that incorporates:

We interpret parental investment and, by extension, grandparental investment using this broader frame (see also Dawkins Reference Dawkins2006; Geary Reference Geary2000; Hertwig et al. Reference Hertwig, Davis and Sulloway2002). Proxy outcome measures of parental investment and – to the extent that grandparents influence these outcomes – grandparental investment that have been used include child mortality, birth weight, breast-feeding, infanticide, homicide, abuse, attachment/bonding between parents and infants, inheritance, interbirth interval, and educational investment (Hagen et al. Reference Hagen, Barrett and Price2006). The pathway through which grandparental investment affects a child's development can be direct (e.g., direct support, advice, or as a role model) and indirect via a third party (most often the grandchildren's parents).

2.2. Why grandparental investment?

Cooperative breeding species such as humans can increase their inclusive fitness directly by reproducing themselves or indirectly by helping their relatives, who share a proportion of their genotype by descent, to transmit their genes into future generations (Hamilton Reference Hamilton1964). Grandparents can boost their inclusive fitness by devoting resources to their kin, helping to ensure the conception, birth, survival, growth, development, and reproduction of their relatives. The opportunity to increase indirect fitness is afforded by the long human lifespan in general, and in particular by the extensive postreproductive lifespan of human females that is unique among cooperative breeding animals (Hrdy Reference Hrdy, Voland, Chasiotis and Schiefenhövel2005b). Although various forms of reproductive termination occur across a range of species (Cohen Reference Cohen2004; Packer et al. Reference Packer, Tatar and Collins1998; Paul Reference Paul, Voland, Chasiotis and Schiefenhövel2005), humans are the only primate species in which reproductive termination is a distinct and universal trait (Caro et al. Reference Caro, Sellen, Parish, Frank, Brown, Voland and Borgerhoff Mulder1995; Pavelka & Fedigan Reference Pavelka and Fedigan1991). The survival of an organism beyond reproduction represents a biological anomaly: “[T]here should be little or no postreproductive period in the normal lifecycle of any species” (Williams Reference Williams1957, p. 407). Attempting to explain this anomaly, Williams hypothesized that it would be adaptive to cease reproduction before death if the benefits to the woman's survival – that is, removal of the risks associated with dying during childbirth, and the increased parenting effort that she could devote to her living children and grandchildren – outweighed the benefits to reproductive fitness brought to the woman by further reproduction.

Couched in terms of “Hamilton's Rule” (Hamilton Reference Hamilton1964), postreproductive helping behavior will evolve when an individual can help a related individual, who has a probability r (i.e., coefficient of relatedness) of sharing this same gene by common descent, to survive and reproduce. This holds as long as the benefit (B, i.e., benefit to the relative receiving help), multiplied by the probability of sharing that gene (r), outweighs the cost (C) to the helper, in terms of his own survival and forgone reproductive opportunities. To wit, postreproductive helping behavior should evolve when rB>C. It is likely that this condition held for human ancestors because (a) grandparents are closely related to their grandchildren (grandparents, on average, share 25% of their genes with each of their grandchildren, over and above the large proportion of genes that all humans share in any case; Dawkins Reference Dawkins2006), (b) helping comes with a relatively low opportunity cost, in terms of the grandparent's own reproductive success (especially in the case of postreproductive grandparents), and (c) there is a potentially large benefit to grandchildren in terms of survival (and subsequent reproduction) during the high-risk times of infancy and childhood.Footnote ¹

Instigated by Williams' (Reference Williams1957) suggestion of the potential adaptive value of a postreproductive period, two related lines of reasoning have emerged, both emphasizing the vast investments required by the long childhood development period: the good-mother hypothesis and the grandmother hypothesis (reviews by Jamison et al. Reference Jamison, Cornell, Jamison and Nakazato2002; Peccei Reference Peccei2001a; Voland et al. Reference Voland, Chasiotis, Schiefenhövel, Voland, Chasiotis and Schiefenhövel2005).Footnote ² The good-mother hypothesis holds that burdened with the long dependence of human children, a postreproductive period shields the mother from survival risks (i.e., pregnancy and giving birth) and the child from health risks (e.g., Down syndrome), and enables the mother to free up the resources to keep her children, particularly the youngest and most helpless, alive. This investment contributes to the mother's direct reproductive success by ensuring that her existing children survive. Thus, one would expect to see the presence of a mother and her increased longevity improving the survival chances and health state of her children, and we do.Footnote ³

The grandmother hypothesis suggests that a postreproductive grandmother can foster her inclusive fitness by supporting her reproductive daughter and her grandchildren, thus increasing her daughter's fertility and improving her grandchildren's chances of surviving (e.g., Hrdy Reference Hrdy1999). Moving the provisioning of children from exclusively the mother (i.e., breast-feeding) to allomothers, as the grandmother hypothesis proposes, has immediate reproductive consequences that could translate into inclusive fitness advantages for the allomothers (grandmothers): For example, ceasing to breast-feed terminates the mother's lactational amenorrhea (Ellison Reference Ellison1990), thus potentially reducing the interval to the daughter's next pregnancy. Comparative data confirm that humans wean their children at a younger age and have shorter inter-birth intervals than do orangutans, gorillas, and chimpanzees (Hawkes et al. Reference Hawkes, O'Connell, Blurton Jones, Alvarez and Charnov1998; Kramer Reference Kramer2005a; Lancaster et al. Reference Lancaster, Kaplan, Hill and Hurtado2000). The notion of the helpful grandmother, however, should not be conceptualized as selection to blindly maximize the number of offspring, because any increase in number of descendents is often accompanied by a decrease in investment per offspring and potentially survival (e.g., Coall et al. Reference Coall, Meier, Hertwig, Wänke and Höpflinger2009; Gibson & Mace Reference Gibson and Mace2005; Walker et al. Reference Walker, Gurven, Burger and Hamilton2008). Thus, we would expect to see the presence of a grandmother and her increased longevity improving the fertility of her daughter, but not at the expense of the survival and health of her grandchildren.

Although the good-mother hypothesis and the grandmother hypothesis, which are not mutually exclusive, predict a selective advantage associated with a postreproductive lifespan, attempts at modeling this advantage in contemporary, largely natural fertility, human populations have failed to support these hypotheses (e.g., Hill & Hurtado Reference Hill and Hurtado1991; Reference Hill and Hurtado1996; Rogers Reference Rogers1993; see also Austad Reference Austad1994). The benefits of early reproductive cessation in women – diverting resources from reproduction to maternal care and avoiding the costs of continuing to reproduce – do not appear to outweigh the benefits of having additional children. However, consideration of additional costs, such as the decline in fertility with older age, increased perinatal mortality risk associated with giving birth at an older age, and increased competition with increased family size, as well as the variety of additional benefits – earlier reproduction in daughters, improved growth and survival of grandchildren – suggests that the benefits of switching from reproduction to parenting could indeed outweigh the costs (Hrdy Reference Hrdy, Hewlett and Lamb2005a; Mace Reference Mace2000; Turke Reference Turke1997). Moreover, a recent analysis of data from two Gambian villages (1950–1975) showed that menopause may have a fitness advantage when the combined influence of both the increased maternal mortality associated with giving birth at an older age and the increased childhood survival associated with having a maternal grandmother present are taken into account (Shanley et al. Reference Shanley, Sear, Mace and Kirkwood2007; see also Sear et al. Reference Sear, Mace and McGregor2000; Shanley & Kirkwood Reference Shanley and Kirkwood2001). Thus, the joint positive influence of a “good mother” and a “helpful grandmother” on survival during childhood may have created a selective advantage for postreproductive survival in women.

2.3. Why postreproductive lifespan may not be adaptive

In contrast to the aforementioned views, there are evolutionary theories proposing that the early cessation of reproduction, which implies the availability of postreproductive grandmothers, is an exaptation. It is not or was not originally adaptive (for reviews, see Peccei Reference Peccei2001a; Voland et al. Reference Voland, Chasiotis, Schiefenhövel, Voland, Chasiotis and Schiefenhövel2005). On this view, menopause occurs because a woman has lived long enough to deplete the supply of primary oocytes with which she was born (Ellison Reference Ellison2001; but see Johnson et al. Reference Johnson, Canning, Kaneko, Pru and Tilly2004; Skaznik-Wikiel et al. Reference Skaznik-Wikiel, Tilly, Lee, Niikura, Kaneko-Tarui, Johnson and Tilly2007). Several theories of senescence propose that small fitness benefits early in an individual's life will be selected for even if they exact a cost to fitness later in life. The reduced strength of natural selection at older ages permits these negative effects to remain unchallenged (Kirkwood Reference Kirkwood1977; Medawar Reference Medawar1952; Williams Reference Williams1957). Therefore, rather than being advantageous itself, menopause may be a by-product of the adaptive benefit of producing, protecting, and storing a complete stock of primary oocytes at birth (Pavelka & Fedigan Reference Pavelka and Fedigan1991). Consistent with this interpretation is the similarity of human and chimpanzee ovarian follicle depletion rates (Jones et al. Reference Jones, Walker, Anderson, Lacreuse, Robson and Hawkes2007; for gorillas see Atsalis & Margulis Reference Atsalis and Margulis2008), which suggests that menopause may be a phylogenetic legacy and not unique to humans.

Even if this is the case, and menopause is a non-adaptive epiphenomenon that occurs because of humans' longevity, its presence creates the stage for helpful grandmothers. Bogin and Smith (Reference Bogin and Smith1996) proposed a biocultural model in which menopause is non-adaptive but helpful grandmothers still play a central role. Once women started living past their reproductive ceiling, the only way that they could improve their inclusive fitness was by caring for their children and grandchildren. Whether human longevity and specifically the postreproductive longevity associated with menopause is adaptive or an epiphenomenon, grandparents are still in the right place, at the right time.

2.4. Does grandparental investment enhance fitness? Evidence from natural fertility societies

The influence of helpful grandmothers has been most intensively studied in historical and contemporary natural fertility societies with little or no access to modern medical care, including contraception. In these societies, grandparental investment matters for mothers' reproductive success. Sear and Mace's (Reference Sear and Mace2008) review of 45 studies investigating the presence of kin (father, grandparents, older siblings) supports the beneficial influence of postreproductive relatives: The presence of a maternal or paternal grandmother was associated with an increase in her grandchildren's probability of surviving in 69% (9 of 13 studies) and in 53% (9 of 17 studies) of cases, respectively. For illustration, in their time series analysis of the Oromo agro-pastoralists of southern Ethiopia, Gibson and Mace (Reference Gibson and Mace2005) found that the presence of a hard-working maternal grandmother, who helped the mother with heavy domestic tasks (e.g., grinding maize), increased the probability of a grandchild surviving to three years of age by 25%. This effect is comparable to that achieved by installing a new water supply (Gibson & Mace Reference Gibson and Mace2006).

Omnipresent grandparents, however, are not inevitably advantageous. Sear and Mace (Reference Sear and Mace2008) found evidence that the presence of grandparents has been associated with detrimental consequences for child survival. In 83% (10 of 12 studies) of cases, the presence of maternal grandfathers had no effect, and in 75% the paternal grandfather had no effect or even a negative one on survival. There is currently little evidence supporting the adaptive significance of grandfathering (see Lahdenperä et al. Reference Lahdenperä, Russell and Lummaa2007). Negative effects have also been reported for paternal grandmothers who may, in contrast to maternal grandmothers, increase their daughter-in-law's fertility at the expense of grandchild survival (Leonetti et al. Reference Leonetti, Nath, Hemam, Neill, Voland, Chasiotis and Schiefenhövel2005; Reference Leonetti, Nath and Hemam2007; Mace & Sear Reference Mace, Sear, Voland, Chasiotis and Schiefenhövel2005). For example, in a historical German population (1720–1874), the presence of a paternal grandmother increased a grandchild's risk of stillbirth by 35% (Voland & Beise Reference Voland, Beise, Voland, Chasiotis and Schiefenhövel2005) and infant mortality up to one month of age by 85% (Voland & Beise Reference Voland and Beise2002). The majority of these studies have been conducted in patrilineal societies; recent evidence from a matrilineal society suggests that even maternal grandmothers may have a negative effect under some circumstances (Sear Reference Sear2008; but see Leonetti et al. Reference Leonetti, Nath, Hemam, Neill, Voland, Chasiotis and Schiefenhövel2005). The beneficial influence of grandparents cannot be taken as given. But there is evidence that the presence of maternal grandmothers in particular boosts the survival chances of their grandchildren.

2.5. The issue of confounding effects

The grandparental investment literature consists predominantly of correlational studies showing associations, or lack thereof, between grandparental presence and grandchild survival, for example. Of course, these studies do not show that this relationship is causal (Borgerhoff Mulder Reference Borgerhoff Mulder2007; Sear & Mace Reference Sear and Mace2008). Some argue that shared genes or environment effects may equally account for the observed associations. Grandparents who live under good socioeconomic conditions or who have a fortuitous genetic complement are more likely to live long and healthy lives, and it would make sense that their grandchildren are more likely to survive, are healthier, or have better cognitive abilities (e.g., Modin & Fritzell Reference Modin and Fritzell2009). Although this possibility cannot be ruled out, several pieces of evidence speak against it. First, if the effect is due to shared genes or environment, one would expect that all grandparents who share the same proportion of genes by common descent or share their residence with a grandchild to have similar effects on grandchild survival. Clearly, this is not the case, as we discuss in the next section. Second, the detailed ethnographic accounts that accompany many correlational studies not only corroborate associations between kin presence and survival, growth, and development, but also highlight potential behaviors that promote these beneficial effects (e.g., Gibson & Mace Reference Gibson and Mace2005; Leonetti et al. Reference Leonetti, Nath and Hemam2007). Third, many studies allow adjustment for a wide range of confounding variables (e.g., grandparental health, age, socioeconomic status, and residential distance) that at least partially control for shared environment effects (e.g., Borgerhoff Mulder Reference Borgerhoff Mulder2007; Coall et al. Reference Coall, Meier, Hertwig, Wänke and Höpflinger2009; Pollet et al. Reference Pollet, Nettle and Nelissen2006). Taken together, this evidence suggests that it is unlikely that these effects can be explained away by shared gene or shared environment effects.

3.1. Genetic relatedness

Hamilton's rule implies that the amount of resources transferred to an individual should depend on the relatedness of the provider (r, the coefficient of relatedness). The closer the genetic relationship between two individuals, the more likely helping behaviors will evolve and the smaller the benefit needs to be for this altruism to be maintained (Hamilton Reference Hamilton1964). There has been a long history of research suggesting that parents allocate their resources to children according to their genetic relatedness (e.g., Anderson Reference Anderson2005; Daly & Wilson Reference Daly and Wilson1980; but see Hofferth & Anderson Reference Hofferth and Anderson2003). Evidence that grandparents favor genetically related grandchildren also exists. The investment that grandparents make in a family reduces as the proportion of stepchildren in the family increases (Eggebeen Reference Eggebeen1992). Relatedly, relationships with step-grandparents are not considered to be as close as those with biological grandparents (Aldous Reference Aldous1995). In lesbian-mother families, grandchildren have more contact with the biological grandparents compared with non-biological grandparents (Patterson et al. Reference Patterson, Hurt and Mason1998). Although the size of the variance accounted for in these studies and the necessity of a focus on genetic relatedness has been questioned (Rose & Rose Reference Rose and Rose2001), the patterns appear robust (see Anderson Reference Anderson2005).

3.2. Paternity certainty

Paternity uncertainty is the (perceived) risk that a male is not the biological father of his children. In contrast to males, because of internal fertilization mammalian mothers know categorically who their children are (Trivers Reference Trivers and Campbell1972). In the few mammalian species in which males provide parental investment, they run the risk that their investment is being wasted on another male's offspring (Alexander Reference Alexander1974; Clutton-Brock Reference Clutton-Brock1991; Geary Reference Geary2000; Trivers Reference Trivers and Campbell1972). Both theoretical and empirical studies suggest that lower levels of paternity certainty are associated with reduced male parental care (see Anderson et al. Reference Anderson, Kaplan and Lancaster2007).

This logic can be extended to grandparents and their inclination to invest resources: Because the maternal grandmother is certain of her relationship with her daughter and her daughter's relationship with her grandchildren, it is predicted that maternal grandmothers will invest the most. Paternal grandfathers, on the other hand, are predicted to invest the least because there are two points of uncertainty between themselves and their grandchildren: They are not 100% certain of their relationship with their son nor of their son's relationship with their grandchildren. The maternal grandfather and paternal grandmother are indistinguishable between these two extremes and will both invest an intermediate amount, because they both have one point at which their relationship certainty with their grandchildren could be severed (Smith Reference Smith, Crawford, Smith and Krebs1987; Reference Smith and MacDonald1988).

Consistent with these predictions, across human societies maternal kin invest much more than paternal kin during the prenatal, delivery, and postnatal periods (Huber et al. Reference Huber, Linhartova, Cope and Lacy2004; Meehan Reference Meehan2005), and paternity uncertainty seems to be associated with reduced intergenerational investment from the father (Anderson et al. Reference Anderson, Kaplan and Lancaster2007; Kurland Reference Kurland, Chagnon and Irons1979) and the father's relatives (Gaulin & Schlegel Reference Gaulin and Schlegel1980; Hartung Reference Hartung1985). Yet, some caution should be taken in interpreting the effect of paternity uncertainty. Paternity certainty varies considerably across human societies (Anderson Reference Anderson2006). Moreover, throughout recent human history paternity certainty has increased substantially in many cultures (Gaulin Reference Gaulin1980). Although the rebuttal to this statement is usually that paternity certainty was lower in our ancestral environment, and it is to this that humans are tuned, the findings that investment varies by paternity certainty within communities suggest that humans respond to local conditions. Other authors state that focusing on paternity certainty alone is inadequate (Euler & Weitzel Reference Euler and Weitzel1996; Pashos & McBurney Reference Pashos and McBurney2008) and that in patrilateral and patrilocal societies it is unable to account for the higher levels of investment by paternal relatives (Pashos Reference Pashos2000; but see Gibson & Mace Reference Gibson and Mace2005).

3.3. Sex-specific reproductive strategies

Several authors have addressed the limits of paternity certainty by incorporating sex-specific reproductive strategies into their models of grandparental investment (Euler & Michalski Reference Euler, Michalski, Salmon and Shackelford2007; Euler & Weitzel Reference Euler and Weitzel1996; Huber & Breedlove Reference Huber and Breedlove2007). Because of the large energetic investment in the egg, as compared with the sperm, females are physiologically obligated to provide higher levels of parental investment than males (Bateman Reference Bateman1948). In mammals, internal gestation and the production of milk for offspring further increase this asymmetry (Clutton-Brock Reference Clutton-Brock1989). Whereas females may focus on parental investment, producing a few high-quality offspring, males (because of their low levels of parental investment) can increase their reproductive success by mating with additional females. Therefore, investment put into female kin is more likely to be transformed into parental care, whereas resources invested in a son may be used for mating or parenting effort.

In light of this, individuals are assumed to be more inclined to invest in female relatives, whether providing practical help in the form of childcare or more social and psychological support (Euler & Weitzel Reference Euler and Weitzel1996; Euler et al. Reference Euler, Hoier and Rohde2001). Extending this logic to grandparents, maternal grandparents who invest in their daughters and their children are expected to invest more than paternal grandparents who are investing in their sons and their sons' children (Euler & Weitzel Reference Euler and Weitzel1996). In fact, this female kin bias in grandparental investment has been found across disciplines and societies (e.g., Gibson & Mace Reference Gibson and Mace2005; Huber et al. Reference Huber, Linhartova, Cope and Lacy2004; Rossi & Rossi Reference Rossi and Rossi1990).

When combined with the notion of paternity certainty, sex-specific reproductive strategies further differentiate between the maternal grandfather and paternal grandmother. Because the maternal grandfather invests in his daughter's children, he invests more than the paternal grandmother who invests in her son's children (Euler & Weitzel Reference Euler and Weitzel1996). Evidence supports the significantly higher investment by maternal grandfathers compared with paternal grandmothers in the frequency of face-to-face interactions and emotional closeness (Hoffman Reference Hoffman1980). Thus, the combination of paternity uncertainty and sex-specific reproductive strategies predicts the often-found pattern of grandparental investment where the maternal grandmother invests the most, followed by the maternal grandfather, the paternal grandmother, and the paternal grandfather (e.g., Bishop et al. [2009] and their summary of the data; but for different interpretations, see Huber & Breedlove Reference Huber and Breedlove2007; Pollet et al. Reference Pollet, Nettle and Nelissen2006).

3.4. Reproductive value

An individual's age-specific future reproductive potential, the reproductive value, is another likely factor that influences how many resources he or she receives. If a parent or grandparent has the option of investing in his or her child who has just negotiated puberty and is beginning her reproductive career, or in that child's sibling who is equally related (r=.5) but is approaching the end of her reproductive career, the child's higher reproductive value may strongly influence this decision. Relatedly, one would expect grandparents to transfer resources to grandchildren as a function of their reproductive value, with grandparents favoring those grandchildren who have a higher probability of reproducing in the future (Leek & Smith Reference Leek, Smith and Smith1991; Smith Reference Smith and Smith1991; Smith et al. Reference Smith, Kish and Crawford1987).

Reproductive value may help us to understand, in part, why grandparents go to such enormous lengths and often self-sacrifice to support their teenage daughters who become parents. Individuals who have recently gone through puberty and are ready to begin their reproductive careers are at their peak reproductive value, which will gradually decline with age (Fisher Reference Fisher1930; Hamilton Reference Hamilton1966; Williams Reference Williams1957). If the aim is to maximize one's number of descendants in the future, preferentially investing in these high reproductive value young adults may be wise. Analyses conducted by Hughes (1988, as cited in Barrett et al. Reference Barrett, Dunbar and Lycett2002, pp. 53–55) suggest that it is these young adults with high reproductive value who have the largest influence on any related individual's inclusive fitness. This influence is greater than that of any other segment of the community. Thus, the high cost that some grandparents incur in helping their children through teenage parenthood may be outweighed by the inclusive fitness benefits, both in contemporary and in ancestral environments.

3.5. Sex-biased grandparental investment

Under certain conditions, biasing investment preferentially towards sons or daughters may increase the probability that an individual will have grandchildren in the future (Trivers & Willard Reference Trivers and Willard1973). In populations where members of one sex have higher potential reproductive success, grandparents may increase their inclusive fitness by preferentially investing in grandsons over granddaughters, or vice versa (see Euler & Weitzel Reference Euler and Weitzel1996; Jamison et al. Reference Jamison, Cornell, Jamison and Nakazato2002; Smith Reference Smith and MacDonald1988). In theory, this may be possible to explore; however, studies examining this Trivers-Willard hypothesis in humans across just two generations have shown mixed results, and examining it over three generations will be even more challenging. To our knowledge, there is currently no convincing evidence of sex-biased grandparental investment in humans.

Table 1 summarizes our review of the evolutionary factors that can be expected to moderate grandparents' inclination to invest their resources. We now turn to two factors that are likely to affect the availability of grandparental resources.

Table 1. Factors associated with the inclination to invest grandparental resources (factors are derived using an evolutionary framework)

3.6. The availability of resources: Family size and birth order

Actual investments presuppose the availability of resources, which, in turn, depends on numerous variables, such as the socioeconomic status of the grandparents and their state of health. Two factors – family size and birth order – are often incorporated into evolutionary analyses of parental and grandparental investments (e.g., Hagen et al. Reference Hagen, Barrett and Price2006; Kaplan Reference Kaplan1994; Laham et al. Reference Laham, Gonsalkorale and von Hippel2005; Lawson & Mace Reference Lawson and Mace2009; Leonetti et al. Reference Leonetti, Nath, Hemam, Neill, Voland, Chasiotis and Schiefenhövel2005), and both affect the amount of resources to be distributed.

Although larger families can recruit older siblings to provide resources for a family (Kramer Reference Kramer2002; Reference Kramer2005a), increased family size typically dilutes the resources available for each child (Blake Reference Blake, Lancaster, Altmann, Rossi and Sherrod1987; Hertwig et al. Reference Hertwig, Davis and Sulloway2002; Marks Reference Marks2006) and grandchild (Laham et al. Reference Laham, Gonsalkorale and von Hippel2005; Leonetti et al. Reference Leonetti, Nath, Hemam, Neill, Voland, Chasiotis and Schiefenhövel2005; Uhlenberg & Hammill Reference Uhlenberg and Hammill1998). In a study of 787 Australian university students, Laham et al. (Reference Laham, Gonsalkorale and von Hippel2005) observed that the emotional closeness that a grandchild reported to his/her maternal grandfather or paternal grandmother depended on the availability of other kin (see also Smith Reference Smith and MacDonald1988). Moreover, the general finding that maternal grandfathers provide more investment to grandchildren than do paternal grandmothers (Eisenberg Reference Eisenberg1988; Euler & Weitzel Reference Euler and Weitzel1996; Hoffman Reference Hoffman1980; Pashos Reference Pashos2000) only held when the paternal grandmothers had other children in whom to invest.

Just as increased family size dilutes the resources available for each child and grandchild, so too does a higher birth order (Coall et al. Reference Coall, Meier, Hertwig, Wänke and Höpflinger2009; Van Bavel Reference Van Bavel2006). Furthermore, there is good evidence that birth order has an additional nonlinear influence on the distribution pattern of parental and grandparental investment. Even if parents aim to invest equally in all of their offspring, an unavoidable unequal distribution of resources results due to the effects of birth order (Hertwig et al. Reference Hertwig, Davis and Sulloway2002). The reason is that only the first and last born children have periods of investment that are not jeopardized by competition from siblings, thus undercutting middle-borns in a way that may have long-term consequences for family cohesion and relationships (Salmon & Daly Reference Salmon and Daly1998; see review of evidence in Hertwig et al. Reference Hertwig, Davis and Sulloway2002). By analogy, grandparents who invest equally in all of their grandchildren will also undercut middle-born grandchildren (but not so for bequests; see Bernheim & Severinov Reference Bernheim and Severinov2000).

Last but not least, as grandparents are making the investment, grandparent's family size may matter as well. In a study of 580 grandparent–grandchild relationships, Coall et al. (Reference Coall, Meier, Hertwig, Wänke and Höpflinger2009) found that grandparent's family size (i.e., number of children and grandchildren) explained approximately three times more of the variance in grandparental investment (shared contacts, occasions to meet, discussions, activities, interests, and important roles) than grandchild's family size (i.e., number of siblings and birth order).

4.1. The economic view: Altruism versus mutual reciprocity

In the economic literature, grandparental investment is typically treated as part of the same type of transfers along the intergenerational chain that characterizes parental transfers to children (although there are three-generation variants of exchange, see e.g., Cigno Reference Cigno1993). Parental investment, in turn, has been an important subject of economic analysis at least since Becker (Reference Becker1974; Reference Becker1991) and Barro's (Reference Barro1974) influential work on why parents care about the well-being of their progeny. Since then, however, as Arrondel and Masson (Reference Arrondel, Masson, Kolm and Ythier2006) noted, there has been an “impressive blossoming of miscellaneous models and motivations introduced in the literature” (p. 980), possibly due to the fact that the diverse manifestations of intergenerational transfers – for instance, post-mortem or inter-vivos (transfers among living members of the family), financial or time transfers, upward or downward transfers – are assumed to have different determinants and to serve different goals. Consequently, there is no single overarching model of parental, let alone grandparental, investment, although most models rest on the utility maximization and rational choice framework. Reviewing these manifold accounts of intergenerational transfers is beyond the scope of this article; moreover, such reviews already exist (e.g., Arrondel & Masson Reference Arrondel, Masson, Kolm and Ythier2006; Cigno Reference Cigno, Kolm and Ythier2006; Cremer & Pestieau Reference Cremer, Pestieau, Kolm and Ythier2006; Laferrère & Wolff Reference Laferrère, Wolff, Kolm and Ythier2006). Instead, we focus – at the expense of other motives (e.g., a “warm glow” associated with the very act of giving; Andreoni Reference Andreoni1990) – on the two competing motives that a large proportion of economic models of intergenerational transfers have emphasized to explain inter-vivos intergenerational transfers: altruism and self-interested exchange.

Why do parents desiring to maximize their self-interest make enormous investments in their children? According to the Barro–Becker view, the solution to this puzzle is that a parent's welfare is influenced by the welfare of future generations. Formally, the parent's utility function is augmented by the child's likely lifetime utility (e.g., Barro Reference Barro1974; Becker & Tomes Reference Becker and Tomes1976; Reference Becker and Tomes1979), thus explaining why parents transfer resources to their children. Parents try to maximize total child quality as defined by the sum of their children's adult wealth. The model posits that parents allocate the resources as a function of child quality (e.g., the child's own skills and abilities), and later use wealth transfers to equalize outcomes across children (redistributive neutrality). Successive generations are linked by recursive altruistic preferences, that is, individuals have an altruistic concern only for their children, who in turn care altruistically for their children. Parents thus expect their children to adopt similarly altruistic behavior towards their children, and so on; thereby all generations of a family are linked together by a chain of intergenerational transfers.

According to the second motive, parents' transfers are less “pure,” but part of a bargaining between parents and children (see Laferrère & Wolff Reference Laferrère, Wolff, Kolm and Ythier2006). Transfers are a form of investment, like an investment into a portfolio, and the context of the investment is one in which parents anticipate that when they become old they will need help. Parents thus invest now (e.g., education expenses, gifts, loans) and in the future (promise of inheritance) to make their child behave altruistically toward them in the future – an intertemporal exchange between self-interested parents and children.

Both altruistic and exchange models face a number of empirical challenges, and according to Arrondel and Masson (Reference Arrondel, Masson, Kolm and Ythier2006), several surveys of the evidence have emphasized both models' “poor empirical performance” (p. 975). For the altruistic model, the challenges range from the empirical failure of the prediction that the amount of transfer should decrease with a child's income if the property of redistributive neutrality holds, to the fact that altruistic parents should transfer their wealth mainly in the form of gifts when children with limited liquidity need them most, rather than much later through bequests (Arrondel & Masson Reference Arrondel, Masson, Kolm and Ythier2006). For models of exchange, the crucial question is whether there exists a specific quid pro quo (intertemporal) exchange. If so, one would, for instance, expect that the levels of aid and attention that children give to their parents is positively correlated with their gift and inheritance expectations – a prediction for which there is, at best, mixed evidence (Arrondel & Masson Reference Arrondel, Masson, Kolm and Ythier2006).

Few economic models of the altruism and exchange variety and beyond have explicitly considered grandparental investment (e.g., Wolff Reference Wolff2000), or treated it differently from a replication of the parent–child relationship. Exchange concerns may make the transfer motive most suitable for application to the grandparent–parent and grandparent–grandchild relationships, respectively. But again, there is little empirical evidence for a quid pro quo exchange. Using data from the German Aging Survey, Hoff (Reference Hoff2007), for example, found that extensive financial transfers from grandparents to grandchildren existed, as did a small amount of instrumental support from grandchildren to grandparents; however, in only 10 out of 371 (2.7%) cases was any exchange of resources evident. Only for the 80–85-year-old grandparents did the proportion of grandchildren providing instrumental support to grandparents exceed the proportion receiving financial transfers from them. The relative value of these time transfers is hard to evaluate (Arrondel & Masson Reference Arrondel, Masson, Kolm and Ythier2006) and may be crucial to the minority of grandparents that receives them; however, it is clearly outweighed by the preceding decades of their investment.Footnote ⁵

4.2. The sociological view: Values, norms, and institutions

In sociology, the influential modernization paradigm has stressed the emergence of the nuclear family and the parallel weakening of bonds beyond the nuclear family. Probably owing to these predictions, the topic of grandparental investment has received scant attention from sociologists (Kohli Reference Kohli and Silverstein2004). This, however, changed in the closing years of the twentieth century, when demographic dynamics and economic pressure on state-funded old-age pension systems put the topic of intergenerational transfers and intergenerational solidarity (Monserud Reference Monserud2008; Wood & Liossis Reference Wood and Liossis2007) on sociologists' scientific agenda. Since then, their inquiries have mostly turned to structural factors, social institutions, and cultural values that shape the process of intergenerational transfer. Structural factors relate to macro-level properties, such as female participation in the labor force, or income and wealth distributions. Social institutions relevant for transfer concern, for instance, laws that regulate testamentary freedom (e.g., in Germany the surviving spouse and children have a legal right to half of the estate, whereas in the United States there are virtually no constraints on the estate holder), and how wealth is taxed at death. Cultural values – underlying norms of reciprocity, family obligations, and responsibility – have been suggested as shaping intergenerational transfers (e.g., Finch & Mason Reference Finch and Mason1993). In a representative sample of the German population (the German Aging Survey), Kohli and Künemund (Reference Kohli, Künemund, Bengtson and Lowenstein2003), for example, observed that retirees (>55 years) who agree to an unconditional solidarity norm (“If my family members need help, I will always be there”) are much more likely to provide monetary transfers (to their kin during the 12 months prior to the survey) than those who espouse a strategic norm of direct reciprocity (“If I help my family members, I can expect help from them in return”). Closely related to investigations of such norms are studies of the different roles that grandparents can adopt (Kivett Reference Kivett1993), and the meaning of (Werner et al. Reference Werner, Lowenstein and Katz1998) and satisfaction with (Reitzes & Mutran Reference Reitzes and Mutran2004) these roles.

As of now, the sociological inquiries have yielded a rich repertoire of empirical findings. On the one hand, it seems fair to say that this wealth of findings has not been born out of a strong theoretical framework, nor has it culminated in one, notwithstanding repeated calls from within sociology, and its crossover with psychology, for a theoretical synthesis (Smith & Drew Reference Smith, Drew and Bornstein2002; Reference Smith, Drew, Hohhughi and Long2004; Szinovacz Reference Szinovacz and Szinovacz1998b). On the other hand, sociologists' emphasis on variables and theoretical constructs (e.g., values and norms) that evolutionary researchers tend to neglect has revealed a number of key robust properties of intergenerational transfer and solidarity. Among these are the following: First, intergenerational support is not only a function of needs and the availability of resources, but also of the donor's (e.g., grandparent's) norms and values system (e.g., unconditional vs. conditional solidarity). Second, the development of public welfare systems – most profoundly implemented in European societies – has not crowded out support from beyond the nuclear family: There is a net flow of material resources from retirees to their adult children and grandchildren (Albertini et al. Reference Albertini, Kohli and Vogel2007). Third, for monetary resources, welfare state provision appears to enable rather than crowd out transfers (Attias-Donfut et al. Reference Attias-Donfut, Ogg and Wolff2005); furthermore, transfers in European countries to some extent depend on countries' welfare regimes (e.g., Albertini et al. Reference Albertini, Kohli and Vogel2007). Fourth, comparative analyses, for instance, between East and West Germany, have demonstrated that transfers from the elderly to their children occur even under unfavorable economic conditions (Kohli Reference Kohli and Silverstein2004).

Lacking an encompassing theoretical framework of grandparental investment, sociologists have made steps toward such a framework. One such step is the recent rational grandparent model (Friedman et al. Reference Friedman, Hechter and Kreager2008). It depicts grandparents as rational actors who use their relationship with their grandchildren as a conduit to a reciprocal relationship with the grandchildren's parents. That is, grandparents attempt to reduce the agonizing uncertainty regarding their quality of life in old age by investing differentially in the children of that child most likely, in the grandparent's perception, to reciprocate in the future. The mechanism assumed is that acts of investment in grandchildren activate children's norms of reciprocity. Clearly, this model shares aspects with other exchange theories (e.g., the indirect reciprocities model; Arrondel & Masson Reference Arrondel, Masson, Kolm and Ythier2006), but it also makes a number of unique predictions, some of which, however, are blatantly at odds with evolutionary theory, others with currently available evidence. For example, the model predicts, contrary to the kin altruism account of grandparental investment, that grandparents are indifferent as to whether their grandchildren are related to them or not (i.e., adopted). The model also assumes that to overcome the potential diffusion of responsibility among their children, grandparents must make investments “demonstrably and publicly differential” (Friedman et al. Reference Friedman, Hechter and Kreager2008, p. 42; these authors' emphasis). Such favoritism, however, is in conflict with both the parents' espoused value of equal treatment of children in egalitarian societies (Zervas & Sherman Reference Zervas and Sherman1994) and the “equal division puzzle” (Bernheim & Severinov Reference Bernheim and Severinov2000); that is, the consistent observation that today's bequests are to a large extent distributed equally across all children within a family.

4.3. An alternative Darwinian view: Evolutionary continuity and the empathy mechanism

Recent years have seen sustained challenges to the standard categories of altruistic motivations in evolutionary theory and economics. According to those, people desire to enhance the welfare of others at a net welfare loss to themselves because they are related to those others (kin altruism); or because they expect future material benefits in repeated interactions (reciprocal altruism); or because they are concerned about building a reputation (reputation-based cooperation). Two bodies of evidence have challenged the sufficiency of these categories. The first body of evidence, gathered by experimental economists and psychologists, suggests that a substantial portion of people are strongly motivated by “other-regarding preferences” and are guided by concerns for fairness that cannot be easily reduced to the “weak reciprocity” that is motivated by long-term self-interest in repeated interactions.

Perhaps the most famous social game challenging the weak reciprocity view is the Ultimatum Game (Güth et al. Reference Güth, Schmittberger and Schwarze1982). In its simplest form, the Ultimatum Game involves a single round in which one person (the proposer) proposes to split a fixed pie (say $100). The proposed split represents an ultimatum to the other person, the responder, who must accept or reject it. If the proposed distribution is accepted, it will be implemented. A rejection, however, results in a default payment – typically zero – for each player. Because of the extreme asymmetry in available actions, the equilibrium for this game involves very asymmetric payoffs: If the responder is self-regarding, she will accept any positive payoff in excess of the default value, even an offer of $1, because it is better than nothing. Knowing this, a self-regarding proposer will offer just that, $1. This prediction hinges on the assumption that the responder is rational and acts in a self-regarding manner, and that the proposer is convinced that the responder is rational and will act in a self-regarding manner. Conflicting with this prediction, a robust result in the Ultimatum Game, across hundreds of experiments, is that the vast majority of the offers to the responders are between 40 and 50% of the available pie (Fehr & Schmidt Reference Fehr, Schmidt, Kolm and Ythier2006). Moreover, those proposals that offer the responder less than 20% of the pie are rejected with a substantial probability.

The most influential interpretation of this behavior – which conflicts with the assumption made routinely by most economists (see Fehr & Schmidt Reference Fehr, Schmidt, Kolm and Ythier2006) that material self-interest is the sole motivation of all people – has been in terms of altruistic punishment (Fehr & Gächter Reference Fehr and Gächter2002). Altruistic punishment means that individuals punish unfair and uncooperative behavior, notwithstanding the fact that the punishment is costly for them and yields no material gain. In the Ultimatum Game, the proposer is assumed to anticipate the risk of altruistic punishment – both responder and proposer will receive nothing if the responder rejects the allocation – and thus offers a more equitable distribution than the equilibrium solution suggests. The threat of altruistic punishment of defectors has been suggested as a key condition of why humans, unlike other animals, are able to frequently cooperate with genetically unrelated strangers (i.e., no kin altruism), often in large groups, with people they will never meet again (i.e., no reciprocal altruism), and when reputation gains are small or absent (no reputation-based cooperation). Although altruistic punishment may be selectively disadvantageous at the individual level, it may be selectively advantageous at the group level (Wilson & Wilson Reference Wilson and Wilson2007), and thus could be the glue that keeps human societies together (e.g., Fehr & Gächter). Moreover, neuro-scientific evidence indicates that, first, negative emotions towards defectors and, second, hedonic responses in reward-related brain areas when people cooperate or punish others for violations of widely accepted social norms (e.g., de Quervain et al. Reference de Quervain, Fischbacher, Treyer, Schellhammer, Schnyder, Buck and Fehr2004; Sanfey et al. Reference Sanfey, Rilling, Aronson, Nystrom and Cohen2003) are the proximate mechanisms behind altruistic punishment.

Ethologists, primatologists, and psychologists have gathered the second body of evidence challenging the traditional categories of altruism. It suggests that precursory elements of key aspects of human moral behavior and other-regarding preferences, such as empathy, sympathy, consolation, perspective taking, theory of mind, and a sense of fairness, also exist in our closest primate relatives. De Waal (Reference de Waal2006), in particular, has attacked what he calls the “veneer theory,” according to which human morality is merely a fragile cultural layer, scarcely concealing the brutish natural tendencies that we have in common with other species. In contrast, he has argued that morality has evolved gradually in our forebears, especially in the great apes. The possibility of other-regarding preferences in nonhuman primates has triggered numerous fascinating studies of primates, and a heated debate. Some of these studies clearly support the notion of an evolutionary continuity between humans and primates with regard to crucial aspects of altruism (e.g., Brosnan & de Waal Reference Brosnan and de Waal2003; Brosnan et al. Reference Brosnan, Schiff and de Waal2005; Burkart et al. Reference Burkart, Fehr, Efferson and van Schaik2007; de Waal et al. Reference de Waal2008; Warneken et al. Reference Warneken, Hare, Melis, Hanus and Tomasello2007), whereas others cast doubt on the existence of profound other-regarding preferences among primates in general and specifically in contexts involving food acquisition and distribution (e.g., Jensen et al. Reference Jensen, Hare, Call and Tomasello2006; Silk et al. Reference Silk, Brosnan, Vonk, Henrich, Povinelli, Richardson, Lambeth, Mascaro and Schapiro2005; Vonk et al. Reference Vonk, Brosnan, Silk, Henrich, Richardson, Lambeth, Schapiro and Povinelli2007).

Possibly the single most important aspect of both of these roiling debates over the origins and the nature of human altruism for grandparental investment is the evolution of empathy. In several publications, de Waal (Reference de Waal2008; Preston & de Waal Reference Preston and de Waal2002) has proposed empathy to be a candidate mechanism that underlies directed altruism, which is altruistic behavior in response to another's pain, need, or distress. On this view, evolution may have selected empathy in animals as the main proximate mechanism to generate directed altruism. The mechanism, however, is not in opposition to the ultimate causes of behavior, but empathy-based altruistic acts are allocated in accordance with kin and reciprocal altruism theory (de Waal Reference de Waal2008, p. 279).

The empathy-mechanism thesis highlights the distinction between ultimate and proximate causes of behavior (Mayr Reference Mayr1961). Proximate mechanisms have received scant attention, with the kin altruism explanations of grandparental investment focusing instead on the investments' ultimate causes (reproductive success; section 2). According to de Waal's (Reference de Waal2008) argument, the ultimate benefits of altruistic acts could be quite time-delayed, and thus unable to provide the motivational force necessary to implement acts of directed altruism. Empathy in terms of a rapid emotional connectedness mechanism, however, could bridge the gap between altruistic acts now and ultimate reproductive success benefits in the future. The core of the empathic capacity lies in the access that an organism gains to the subjective state of another organism through the subject's own neural and body representation. The mechanism works so well because it gives individuals an emotional stake in the welfare of others.

How did the empathy mechanism evolve? De Waal (Reference de Waal2008) suggests that it “likely started in the context of parental care long before our species evolved” and “once the empathic capacity existed it could be applied outside the rearing context and play a role in the wider network of social relationships” (p. 282). Via the empathy mechanism, de Waal puts altruism back into altruism. Analogously, via the empathy mechanism one could also put altruism back into grandparental altruism. That is, the proximate cause for why grandparents care for the welfare of their children and grandchildren is not that they bet on return benefits of altruistic behavior, but may lie in their capacity for empathy, or in Hrdy's (Reference Hrdy2009) term, in humans' hypersociality. Empathy alerts them to their kin's needs, helps them to identify emotionally with them, and provides them with psychological benefits (e.g., satisfaction) from giving support.

5.1. Shared lifespan

The recent increase in human life expectancy has expanded the shared lifespan between grandparents and grandchildren, thus creating theoretically an ideal situation for grandparents to impact their grandchildren's welfare. Using cohort-specific survival rates and age-at-first-birth data from Switzerland, Höpflinger et al. (Reference Höpflinger, Hummel and Hugentobler2006) estimated that of the grandchildren who reached 20 years of age in 1900, 27% would have had one or more living grandparents. This figure had risen to 92% in 2000. At 35 years of age, only 1% of grandchildren would have had at least one surviving grandparent in 1900; in the year 2000, even at 35 years of age, 39% of grandchildren would have had one or more surviving grandparents (for U.S. data, see Uhlenberg Reference Uhlenberg1996).

How will this trend continue in the future? Using mortality, fertility, and average age at first birth data for England and Wales, Murphy and Grundy (Reference Murphy and Grundy2003) estimated the proportion of people born between 1911 and 2050 with living mothers and women with living children. From the grandparent's point of view, the future looks bright. A combination of the decline in age at first childbirth to a low in the 1970s and steadily declining mortality rates means that 60% of 70-year-olds currently have two or more living children and by extension, potentially grandchildren. It is estimated that this proportion will remain above 50% until at least 2030. From the grandchildren's point of view, we may be about to leave a golden age of shared lifespan. A combination of an increasing age at first childbirth since the 1970s and a less marked decline in mortality rates means that it is more likely a mother will be dead for a given age of her children (and grandchildren). Therefore, the proportion of adults with a living mother, and thus the proportion of grandchildren with a living grandmother, has started to decline from those individuals born around 1970. In terms of human history, grandparents and grandchildren may currently be at the zenith of their shared lifespan.

More generally, even though the average age at first childbirth is rising (Billari et al. Reference Billari, Kohler, Andersson and Lundström2007), and hence delaying the onset of grandparenthood, in industrialized societies the typical grandparent is young (relative to life expectancy) and healthy (Manton et al. Reference Manton, Gu and Lowrimore2008). For example, Schwartz and Waldrop reported in Reference Schwartz and Waldrop1992 that nearly 50% of grandparents in the United States were aged less than 60 years, 33% less than 55 years, and only 20% older than 70 years. Some have argued that healthier grandparents are more likely to be pursuing their own, perhaps costly, interests outside their family (e.g., traveling). However, in industrialized societies, healthier and younger grandparents have been found to invest more in their grandchildren (Höpflinger et al. Reference Höpflinger, Hummel and Hugentobler2006; see also Euler & Weitzel Reference Euler and Weitzel1996), and to increase the chance of grandchildren surviving in historical populations (Lahdenperä et al. Reference Lahdenperä, Lummaa, Helle, Tremblay and Russell2004).

5.2. Low fertility rates

According to Kohler and colleagues, “The majority of the world's population is living in countries with near-replacement or below-replacement fertility” (Kohler et al. Reference Kohler, Billari and Ortega2002, p. 641). In Europe, for instance, Turkey is the only country among 46 countries studied where generation replacement is guaranteed (with a fertility level higher than 2.1 children in 2002; Council of Europe 2005). At the turn of the millennium, 14 countries in Europe had a total fertility rate below 1.3, which implies halving of the stable population size every 45 years (Kohler et al. Reference Kohler, Billari and Ortega2002). As a consequence, the probability of becoming a grandparent is falling, and those people who do become grandparents are likely to have fewer biological grandchildren, relative to the past.

5.3. Increased individual wealth and investment per child

The influence that grandparents, like parents, have on the development of grandchildren is largely dependent on resource availability (see Borgerhoff Mulder Reference Borgerhoff Mulder2007; Gibson & Mace Reference Gibson and Mace2005; Hadley Reference Hadley2004; Lawson & Mace Reference Lawson and Mace2009). In industrialized societies, ensuring that descendants' skill base and wealth makes them competitive, means that the cost of raising children has increased exponentially (see Borgerhoff Mulder Reference Borgerhoff Mulder1998; Kaplan Reference Kaplan1996; Mace Reference Mace1998). Thus, even though low childhood mortality and fertility rates may translate into fewer, wealthier descendants, the need for grandparental investment may not necessarily diminish. Rather, the increased cost of children may further exacerbate the demands on grandparents. Consistent with this interpretation, higher socioeconomic status grandparents invest more in family members (Cao Reference Cao2006) and show higher levels of affectual solidarity with their grandchildren (Wood & Liossis Reference Wood and Liossis2007). However, the aspect of socioeconomic status being investigated (e.g., income or education) and the form of investment being made (e.g., time or money) may strongly influence the association between grandparental wealth and investment.

In industrialized societies, higher socioeconomic status groups tend to delay reproduction. Even people who want to have children often consciously delay reproduction to accumulate resources (Hammarberg & Clarke Reference Hammarberg and Clarke2005). Increasingly, people are turning to reproductive technologies to assist conception later in life (Oakley et al. Reference Oakley, Doyle and Maconochie2008). The fact, however, that most industrialized countries cannot match demand for fertility treatments (Hoorens et al. Reference Hoorens, Gallo, Cave and Grant2007), and the increased risk of poorer outcomes for the mother and child associated with such technologies (Fisher et al. Reference Fisher, Hammarberg and Baker2005; Kalra, & Molinaro Reference Kalra and Molinaro2008), means that the probability of becoming a grandparent continues to fall, and that those people who become grandparents may need to provide especially high levels of investment.

To conclude, grandchildren and grandparents have never had and may never again have more shared lifetime than today. Moreover, intergenerational transfers remain a robust property of grandparent–grandchild relationships in industrialized societies, although they are dependent on a range of factors such as grandparents' socioeconomic status, their role within the family, grandparents' occupational status, and the presence of elderly parents to care for in addition to children and grandchildren (e.g., Attias-Donfut & Segalen Reference Attias-Donfut and Segalen2002; Cooney & An Reference Cooney and An2006). One likely consequence of the substantial increase in shared lifespan and the lower birthrate is that the time, money, and affection that grandparents invest is spread across fewer grandchildren, increasing the potential investment per grandchild. Paradoxically, however, their investments are likely to be less beneficial than ever – at least when measured in terms of fitness consequences. Does that mean that grandparental investment has become a vain endeavor?

6.1. Low-risk family contexts: Grandparental investment and the physical and mental well-being of children and grandchildren

Could the beneficial influence of grandparents express itself in industrialized societies in terms of successful development on various psychological dimensions? In search of an answer, we conducted a systematic review of the grandparental investment literature. Empirical studies examining the influence of grandparents on their grandchildren have been conducted in sociology, psychology, medicine, biology, economics, and education. To identify these studies, we searched seven databases: PsychINFO (1970–2007), Medline (1970–2007), ERIC (1970–2007), Sociological Abstracts (1970–2007), EconLit (1970–2007), Biosis Preview (1980–2007), and ISI Web of Science (1970–2007), limiting the search to English-language journal articles. The same 11 searches were conducted in each database. The key words “grandparent” and “grandchild” were searched in the title, abstract, and keywords fields in each database, in combination with terms focusing on a range of measures reflecting childhood development: “development,” “verbal ability,” “verbal scores,” “SAT scores,” “academic performance,” “school performance,” “grade point average,” “cognitive performance,” “cognitive ability,” “adjustment,” and “behavioral development.” This search was completed in December 2007.

All searches were cross-matched for duplication, yielding 196 publications. We read the title and abstract of each study to identify those that could be relevant. If there was any indication that grandparental investment (in terms of time, money, care, contact, etc.) in industrialized societies was empirically studied, we read the complete article. We then culled this group, keeping only those articles that (a) provided direct measures of grandparental investment (rather than indiscriminate extended kinship investment), (b) provided direct measures of child outcome (e.g., SAT scores, school performance, grade point average, behavioral development), and (c) investigated a low-risk family context in industrialized societies. Because of the low-risk criterion, we excluded studies involving three-generation households, households where grandparents were the primary caregivers, and households where the influence of grandparents was examined following parental divorce – circumstances that typically represent higher risk family situations, in which grandparents and their families are often disadvantaged themselves (Brandon et al. Reference Brandon, Heyworth and Griffen2007; Lavers & Sonuga-Barke Reference Lavers and Sonuga-Barke1997). When one of the articles cited a study that appeared relevant, we looked it up and included it if the study met the described criteria; similarly, we looked up all studies that cited the article (using Web of Science).

This produced only 13 publications, and, of these, merely four publications met all of our criteria: Tinsley and Parke (Reference Tinsley and Parke1987), Falbo (Reference Falbo1991), Scholl Perry (Reference Scholl Perry1996), and Fergusson et al. (Reference Fergusson, Maughan and Golding2008).Footnote ⁶ Three features of the literature may account for why our search identified so few studies. The first is that very few studies provide direct measures of grandparental investment, but focus instead on outcomes, such as educational attainment as proxies for grandparental investment. The second feature is that many studies focus on co-residence, which occurs more frequently with non-intact families (high-risk context), but is generally low in countries such as the United States (Hill Reference Hill2006). Finally, many studies investigate the impact of help from the broad category “other adults,” rather than specifically grandparents (e.g., Runyan et al. Reference Runyan, Hunter, Socolar, Amaya-Jackson, English, Landsverk, Dubowitz, Browne, Bangdiwala and Mathew1998; Surkan et al. Reference Surkan, Ryan, Carvalho Vieira, Berkman and Peterson2007). We now turn in more detail to the few relevant studies that we retrieved.

6.1.1. Infant physical and mental development

Tinsley and Parke (Reference Tinsley and Parke1987) investigated measures of physical and mental development as a function of the frequency of grandparent–grandchild contacts in a sample of 30 seven-month-old Caucasian infants and their families in the United States. All of the infants were healthy and none of the grandparents were daily caregivers to their grandchildren, lived in the same household, considered themselves retired, or lived more than 50 miles (80 km) away from their grandchildren. Each parent and grandparent was observed playing with the infant for five minutes, and adult–infant interaction and adult global behaviors were rated. To measure grandparental investment, parents completed a social support questionnaire and grandparents completed a questionnaire measuring their contact (frequency) with their grandchildren. Finally, the Bayley Scales of Infant Development, administered on a separate home visit, measured the infant's mental and physical development.

Table 2 shows the statistically significant relationships between grandparent–infant patterns of interaction and infant development scores, and reports the effect size measure etaFootnote ² (η²). First, grandchildren whose grandparents engaged in more stimulating and engaging behavior had higher Bayley scores across both the physical and mental domains (only significant effects were reported). All significant effects were positive and of a large size. Second, infants who had high levels of contact with their grandparents had higher Bayley Mental Development Index scores compared with infants with low contact (η²=0.148 and 0.146, respectively). There was no influence, however, on the Bayley physical development scales.

Table 2. Relationships between grandparent–infant interaction patterns (median split) and infant development scores (adapted from Tinsley & Parke Reference Tinsley and Parke1987) ^a

^a Unfortunately, only those relationships found to be statistically significant were presented in the original text and are reproduced here. It is likely that a broad range of unreported small and medium effects also exist.

^bThe Bayley Scales of Infant Development measured infant mental and physical development across cognitive, motor, and behavioral domains. This produced four scores: (a) a raw score on the Bayley Mental Scale, (b) an adjusted Bayley Mental Development Index, (c) a raw score on the Bayley Physical Scale, and (d) an adjusted Bayley Physical Development Index.

^cThe original text reported the F statistics, therefore the effect size calculated was eta² (η²; Cohen Reference Cohen and Wolman1965, p. 105). Eta² can be interpreted as the proportion of variance in the dependent variable that is attributable to each effect (η²; small [0.01], medium [0.06], and large [0.14] effects; Cohen Reference Cohen1988):

To the best of our knowledge, Tinsley and Parke (Reference Tinsley and Parke1987) were the first to show that more stimulating and interactive play and more contact with grandmothers and grandfathers (but without differentiating between lineages) have positive influences on infant mental development. However, from their study, it is not possible to distinguish between direct influences of grandparenting and indirect influences via parenting (see Bridges et al. Reference Bridges, Roe, Dunn and O'Connor2007), and thus to distinguish between predictions from the good-mother and grandmother hypotheses. In sum, these results document a suggestive, not necessarily causal, association between grandparents' behaviors and infant mental development that deserves further investigation.

7.1. Teenage pregnancy

For good reason, the majority of teenage pregnancy studies focus on grandmother–mother–grandchild co-residence and its consequences. Co-residence represents a high-risk situation. Generally, grandparental co-residence is more common when the teenage mother is younger (Black & Nitz Reference Black and Nitz1996), and when the grandparent is younger and has multiple responsibilities, including work and family commitments (Fuller-Thomson Reference Fuller-Thomson2005; Moore & Brooks-Gunn Reference Moore, Brooks-Gunn and Bornstein2002). Moreover, grandparental co-residence does not necessarily ensure that children and grandchildren receive higher levels of care (Lee et al. Reference Lee, Ensminger and LaVeist2005).

The most common source of support for teenage mothers is their own mothers, the soon-to-be – if not already – grandmothers (Tolson & Wilson Reference Tolson and Wilson1990). Support from grandparents, particularly grandmothers, can have a beneficial effect on young teenage mothers and their children's adjustment and development (e.g., Apfel & Seitz Reference Apfel and Seitz1991; Pope et al. Reference Pope, Whiteside, Brooks-Gunn, Kelleher, Rickert, Bradley and Casey1993; Stevens Reference Stevens1988). On the other hand, evidence also shows a negative influence of grandparental co-residence on teenage mothers and their children (e.g., Chase-Lansdale et al. Reference Chase-Lansdale, Brooks-Gunn and Zamsky1994; Schölmerich et al. Reference Schölmerich, Leyendecker, Citlak, Miller, Harwood, Voland, Chasiotis and Schiefenhövel2005). These negative effects seem to arise especially where the grandparent–grandchild co-residence (a) occurs under conditions of poor family cohesion (Kalil et al. Reference Kalil, Spencer, Spieker and Gilchrist1998), (b) is over an extended period of time (Black & Nitz Reference Black and Nitz1996), (c) occurs when the mother is older (Black et al. Reference Black, Papas, Hussey, Hunter, Dubowitz, Kotch, English and Schneider2002), or (d) occurs when there are high levels of grandparental care that may be considered intrusive or may limit the development of autonomy (Spieker & Bensley Reference Spieker and Bensley1994; Tomlin Reference Tomlin and Szinovacz1998). Taken together, these findings suggest that it is most likely that grandmothers have a beneficial effect when they help rather than completely take over the mother's role or take no part in the grandchild's life (Tinsley & Parke Reference Tinsley, Parke and Lewis1984).

7.2. Maternal depression

By 2020, depression is predicted to be second only to ischemic heart disease as the worldwide leading cause of years of life lost from premature death and or disability (Murray & Lopez Reference Murray and Lopez1997). From the grandchild's point of view, adversities early in life – for instance, the disruption of kinship networks, and the concomitant loss of social support and material security – contribute to an increased risk of developing depression (Korkeila et al. Reference Korkeila, Korkeila, Vahtera, Kivimaki, Kivela, Sillanmaki and Koskenvuo2005; Strassman & Dunbar Reference Strassman, Dunbar and Stearns1999). Maternal depression represents such a disruption of the family network. Individuals whose mother is depressed are at a higher risk of developing depression themselves (e.g., Weissman et al. Reference Weissman, Wickramaratne, Nomura, Warner, Pilowsky and Verdeli2006). Can grandparents in general and a high-quality relationship with one of their grandparents in particular attenuate this risk?

Conducting (in January 2008) the same literature search structure as detailed previously, but using the additional key word “depression,” produced 63 studies. We found merely one study that examined the degree to which cohesion in grandchild–grandparent relationships buffered grandchildren against depressive symptoms. Silverstein and Ruiz (Reference Silverstein and Ruiz2006) analyzed data from 2,280 grandchildren and their mothers who participated in two waves of a nationally representative U.S. sample. Parents (at t₁) and children (at t₂, when they were 18–34 years old) reported the frequency with which they had experienced depressive symptoms in the past week. Cohesion in the grandchild–grandparent relationship was measured in terms of children's responses to questions regarding emotional closeness, frequency of contact, and ability to confide (when they were 10–23 years old). Children were divided into three groups, with weak, moderate, and strong cohesion. After controlling for numerous variables, such as parental income and marital history, Silverstein and Ruiz found that maternal depression (at t₁) was “transmitted to grandchildren with weak and moderately strong ties to grandparents [adjusted r ²=0.017 and 0.015, respectively], but not to those with the strongest ties [adjusted r ²=0.001]” (Silverstein & Ruiz Reference Silverstein and Ruiz2006, p. 608).

This longitudinal study suggests that strong ties with grandparents can help to attenuate the risk of intergenerational transfer of maternal depression. This possibility raises the question of how robust this buffering effect is across other disruptions of the family network due to a parent suffering from psychiatric or psychological disorders (e.g., addiction), parental separation, or death. A recent study by Attar-Schwartz et al. (Reference Attar-Schwartz, Tan, Buchanan, Flouri and Griggs2009), which came to our attention after our literature search had been concluded, showed that a buffering effect could be particularly valuable when the structure of the nuclear family is changed. In a representative sample of 1,515 adolescents from England and Wales, the authors examined the association between degree of grandparental involvement (from the “closest grandparent”) and adolescents' behavioral and emotional adjustment as a function of three family structures: two-parent biological families, lone-parent families, and families with one step-parent. A higher degree of grandparental involvement, which did not differ across family structures, was strongly associated with reduced adjustment problems among adolescents from lone-parent and step-parent families. Parallel beneficial effects, although mostly in the same direction, were not significant for two-parent biological families. Taken together, these observations suggest that the support of emotionally involved grandparents can be an important protective resource for their grandchildren.

8.1. Grandparents as public health targets: Future research

Equipped with expertise and the motivation to share it, grandmothers in non-Western and indigenous societies have been targets of public health promotion to great effect: increasing birth weight in Australia (d'Espaignet et al. Reference d'Espaignet, Measey, Carnegie and Mackerras2003), improving breast-feeding practices in Cambodia (Crookston et al. Reference Crookston, Dearden, Chan, Chan and Stoker2007), or improving nutritional knowledge in young mothers (Aubel Reference Aubel2005). Grandparents also function as health educators in industrialized societies (Watson et al. Reference Watson, Rudolph and Lyons2005), and it has been proposed that interventions target the resources grandparents bring to families (see Denham & Smith Reference Denham and Smith1989). Our empirical review suggests that the socio-emotional support grandparents provide is key to understanding how grandparental investment may influence grandchildren in industrialized societies. Evolutionary theory suggests maternal grandparents are most inclined to help, especially the maternal grandmother, and demographic data suggest that she is most likely available. Hence, one policy approach informed from an evolutionary framework is to target phenomena where the mother's family is likely to have an influence and that are influenced by socio-emotional support. Next, we illustrate two such phenomena.

8.2. Grandparental investment: A one-way street?

In our aging societies, it is increasingly important also to consider the benefits that grandchildren can provide for grandparents (beyond increased fitness), especially grandparental health, rather than focusing exclusively on the reversed relationship. What could such benefits be? One obvious benefit that follows immediately from the view that grandparental investment is an intertemporal exchange between self-interested grandparents, parents, and grandchildren is that grandparents who have cared for their grandchildren may be more likely to receive support from them or their parents in times of need (Friedman et al. Reference Friedman, Hechter and Kreager2008; Stack Reference Stack1975). Such help matters. Esbensen et al. (Reference Esbensen, Østerlind and Hallberg2007) showed that direct support from grandchildren could improve a grandparent's health during times of severe physical illness. Relatedly, in a study of 2,200 older Japanese people, Okabayashi et al. (Reference Okabayashi, Liang, Krause, Akiyama and Sugisawa2004) found that, in those individuals who did not have a spouse (n=677), social support from children and grandchildren was associated with more positive well-being, fewer depressive symptoms, and reduced cognitive impairment.

Another short-term benefit of grandparental investment is that grandchildren provide company. Individuals who live in the company of more kin generally have improved health and a higher probability of surviving crisis situations (Sugiyama Reference Sugiyama2004). Consistent with such findings, in their study of 442 grandparents from the Longitudinal Study of Generations, Drew and Silverstein (Reference Drew and Silverstein2007) observed that grandparents who lost contact with at least one of their grandchildren exhibited a steadily increasing incidence of depressive symptoms between 55 and 80 years of age. In contrast, in grandparents who experienced no loss of contact with their grandchildren, depressive symptomatology remained stable. This effect was robust to the influence of gender, education, marital status, number of children, and linear change in health.

Although these studies observed ways in which grandchildren can benefit grandparents, they did not condition these effects on grandparents' past investment behavior. The two studies of which we are aware that examine the future effects of grandparental good deeds are by Fujiwara and Lee (Reference Fujiwara and Lee2008) and Hughes et al. (Reference Hughes, Waite, LaPierre and Luo2007). Fujiwara and Lee found that grandfathers and fathers (not grandmothers and mothers) who provided moderate amounts of informal or financial support had a lower risk of developing major depression three years later. This association was robust to the influence of additional factors associated with the risk of major depression, including major depression at baseline. Using data from the Health and Retirement Study, Hughes et al. found that grandmothers who babysat for their grandchildren showed higher levels of self-rated health and were more likely to exercise two years later compared with grandmothers who provided no care.

To conclude, the available evidence suggests that altruistic acts towards others (e.g., grandchildren) can have beneficial consequences for the altruist's (e.g., grandparent's) own physical and mental health (McClellan et al. Reference McClellan, Stanwyck and Anson1993), including a reduced risk of morbidity and mortality (Brown et al. Reference Brown, Consedine and Magai2005; Brown et al. Reference Brown, Nesse, Vinokur and Smith2003). However, as before, let us not lose sight of the boundary conditions of grandparental altruism. When the demands of helping become too much, the impact on health can be negative (Goh Reference Goh2009; Lee et al. Reference Lee, Colditz, Berkman and Kawachi2003; Post Reference Post2005). That is, just as grandparents who lose contact with their grandchildren experience deterioration in their physical and emotional health (Drew & Silverstein Reference Drew and Silverstein2007; Drew & Smith Reference Drew and Smith2002), so too do those who assume primary caregiver responsibilities (Heywood Reference Heywood1999; Ross & Aday Reference Ross and Aday2006).

These findings suggest that moderation, as with many things in life, in the amount of care grandparents give is likely to optimize grandparental well-being. Figure 1 stylizes this hypothesized nonlinear relationship between grandchild care and grandparental well-being. Myriad factors, such as the availability of resources, gender, and life transitions (e.g., retirement), will affect the nonlinear relationship depicted in Figure 1. Perhaps one of the most important moderating factors is the degree to which grandparents are overstrained by the simultaneous demands of caring for both their ailing parents and grandchildren (Cooney & An Reference Cooney and An2006; Grundy & Henretta Reference Grundy and Henretta2006). Interestingly, the same nonlinear relationship may be true for the grandchild's well-being: Very low and very high levels of grandparental involvement can have a negative impact on the development of grandchildren (Lavers & Sonuga-Barke Reference Lavers and Sonuga-Barke1997).

Note. ^a Well-being encompasses various positive emotions such as satisfaction and contentment, and positive activities, such as spending time in company.

Figure 1. A schematic representation of the hypothesized nonlinear relationship between the care grandparents take of their grandchildren and the grandparents' well-being.

8.3. Contemporary grandparental investments and fertility: How and why are they related?

Let us return to one key evolutionary account of grandparental altruism. Could grandparents in industrialized societies boost their own fitness by investing resources in their children's and grandchildren's survival and reproduction? We do not know of any study into this possibility, and this lack is not surprising given the demographic changes in industrialized societies (sect. 5). There are, however, separate lines of research documenting fertility trends and levels of grandparental childcare across countries. Connecting these two lines reveals an interesting picture for Europe. Using data from the 2004 Survey of Health, Ageing and Retirement in Europe (SHARE), Hank and Buber (Reference Hank and Buber2009) have reported the percentage of grandparents who provided any care (over the last 12 months), and who provided regular care (i.e., almost weekly or more often over the last 12 months) across ten European countries. The trends for both kinds of care are reversed. The prevalence of any care is highest among Danish, Dutch, French, and Swedish grandparents and lowest among Italian and Spanish grandparents. In terms of regular care, however, Greek, Italian, and Spanish grandparents score highest, whereas Danish, French, and Swedish grandparents contribute least. When one correlates the likelihood of regular care with total fertility rates in the ten European countries studied by Hank and Buber, one finds a strongly negative correlation (Spearman rank correlation=–.88, p=0.008; we calculated this correlation using estimated fertility rates for 2009; Central Intelligence Agency 2009). That is, surprisingly, high levels of investment by grandparents are associated with low fertility rates.

Why is that? Among the key variables contributing to this relationship are likely to be women's participation rate in the labor market and the availability of institutional care (Billari & Kohler Reference Billari and Kohler2004). In Europe, countries that value gender equality are likely to provide state-subsidized day care and other institutional support for working women, thereby reconciling the conflict between motherhood and work. In contrast, countries such as Greece, Italy, and Spain have relatively little state-financed day care, and women are more likely to stay at home after they become mothers. Indeed, Italy and Spain have markedly lower labor force participation of women than, for example, Denmark, France, and Sweden. Because of the lower female participation rate, there is, as Hank and Buber (Reference Hank and Buber2009) suggest, less demand for grandparents to help out in Italy and Spain, relative to, say, Denmark. If, however, a Mediterranean mother decides to have children and continue to work, she has to rely on grandparents' support on a regular basis (for U.S. data, see Presser Reference Presser1989). Consequently, parents of a working mother might have been key to their daughters' decision to have children. For these grandparents, grandparental care and reproductive success might indeed be closely related. More generally, this initial and speculative analysis illustrates that studying the interplay of evolutionary (e.g., fertility), economic (e.g., women's participation rate in the labor market), and institutional variables (e.g., availability of institutional care) promises to yield new insights into the impact of grandparental investment in industrialized societies.