OXT and vasopressin (AVP)

OXT and AVP are two evolutionarily closely related peptides that derived from a single peptide more than 400 million years ago (Acher, Reference Acher1980). Both are synthesized in the supraoptic and paraventricular nucleus of the hypothalamus, from where they affect the central nervous system following different pathways. Peripheral targets are reached by release in the bloodstream via the pituitary, and OXT and AVP are projected to diverse regions in the brain via axonal transport. Local extracellular diffusion and transport by the cerebrospinal fluid are functionally relevant pathways (Landgraf & Neumann, Reference Landgraf and Neumann2004; Veening, de Jong, & Barendregt, Reference Veening, de Jong and Barendregt2010). In mammalian species, OXT is of critical importance for reproductive function, such as the onset of uterine contractions during labor and for milk letdown during suckling. It exerts its effects by acting on a single receptor (OXTR). AVP is known as an antidiuretic hormone and acts to retain water and increases vasoconstriction. It acts on a few receptors of which the AVP receptor 1a (AVPR1a) has shown most relevance to mammalian social behavior. Studies in several mammalian species have shown that OXT release during parturition is critical in the formation of the mother–infant bond (Neumann, Reference Neumann2009), and that OXT and AVP stimulate mutual recognition and maternal bonding after parturition (Curley & Keverne, Reference Curley and Keverne2005). AVP is also important in the regulation of maternal and paternal behavior after parturition, and both OXT and AVP have been implicated in maternal aggression (Bosch & Neumann, Reference Bosch and Neumann2012; Wang, Ferris, & De Vries, Reference Wang, Ferris and De Vries1994). Besides parental care behavior, receptor distributions of OXT and AVP have shown to be critical for monogamous pair bonding in voles, with higher densities of OXTR in the ventral striatum in female monogamous voles and higher densities of AVPR1a in the lateral septum in monogamous males (Insel & Young, Reference Insel and Young2001). These findings suggest a role for these neuropeptides in adult pair bonding and also show the sexually dimorphic effect these neuropeptides have on mammalian social behavior. Besides these effects of OXT and AVP on social–emotional behavior, both peptides are implicated in the endocrine responses to stress. Acute stressors in both sexes lead to pulses of OXT and AVP release, where in the short term OXT might increase arousal and activation of the sympathetic nervous system (Carter, Reference Carter2014), whereas AVP increases goal-directed behavior and anxiety (Engelmann, Landgraf, & Wotjak, Reference Engelmann, Landgraf and Wotjak2004), both facilitating adaptive responses to threat and challenge. Afterward OXT can have downregulating effects on endocrine stress responses by reducing anxiety and reinstating homeostasis (Carter, Reference Carter2014).

Estradiol

Estradiol acts on the estrogen receptor (ER), and its primary function is to act in concert with progesterone (PROG) to facilitate pregnancy. In several mammalian species, estradiol is also critical for the formation of maternal bonding after parturition by priming the maternal brain for synthesis of OXT and OXT receptors (Nelson, Reference Nelson2005). In primates, estradiol levels during pregnancy have been shown to predict maternal caregiving behavior after parturition, although the relation of estradiol levels after pregnancy with maternal caregiving are less clear (Saltzman & Maestripieri, Reference Saltzman and Maestripieri2011). Although the role of estrogens in paternal behavior are unknown, preliminary data from primates have further shown increased estrogen levels in fathers exposed to infant odors (Ziegler, Peterson, Sosa, & Barbard, Reference Ziegler, Peterson, Sosa and Barnard2011).

Testosterone/hypothalamus–pituitary–gonadal (HPG) axis

Testosterone acts on the androgen receptor (AR). It is the end product of the HPG axis and is critical for mammalian reproductive, sexual, and aggressive behavior. Testosterone is easily converted into estradiol and is functionally relevant in both males and females (Bos, Panksepp, et al., Reference Bos, Panksepp, Bluthe and Honk2012). With regard to mammalian caregiving behavior, studies on maternal behavior are scarce, but testosterone levels are generally negatively associated with paternal behavior. In primates, high caring fathers generally have lower levels of testosterone and show decreases in testosterone levels when interacting with their infants (Storey & Ziegler, Reference Storey and Ziegler2016). Nonetheless, in rodents testosterone has been related to maternal behavior (González-Mariscal, McNitt, & Lukefahr, Reference González-Mariscal, McNitt and Lukefahr2007) and indirectly might contribute to paternal behavior by conversion to estradiol (Trainor & Marler, Reference Trainor and Marler2002).

Cortisol (CORT)/hypothalamus–pituitary–adrenal (HPA) axis

CORT acts on the glucocorticoid receptor (GR) and mineralocorticoid receptor (MR) and is the end product of the HPA axis. CORT serves a broad function to recruit neural and peripheral resources to deal with threat and to restore homeostasis in the aftermath of stress (Nelson, Reference Nelson2005). In the central nervous system, CORT can facilitate shifts in brain networks to adaptively deal with stressful situations and to restore emotional reactivity (Hermans, Henckens, Joëls, & Fernández, Reference Hermans, Henckens, Joëls and Fernández2014). During pregnancy CORT levels rise in most mammalian species, including primates (Nelson, Reference Nelson2005; Saltzman & Maestripieri, Reference Saltzman and Maestripieri2011). Correlational studies in primates indicate that CORT is positively related to maternal and paternal responsiveness toward infants and can facilitate arousal-dependent initiation of caregiving behavior, especially in unexperienced parents (Saltzman & Maestripieri, Reference Saltzman and Maestripieri2011; Storey & Ziegler, Reference Storey and Ziegler2016). Studies in animals have further shown that action of corticotropin-releasing hormone (a component of the HPA axis preceding CORT) during parturition is critical for maternal responsiveness after birth (Broad, Keverne, & Kendrick, Reference Broad, Keverne and Kendrick1995), likely acting in concert with peptides such as AVP, OXT, and opioids (Olza-Fernández, Gabriel, Gil-Sanchez, Garcia-Segura, & Arevalo, Reference Olza-Fernández, Gabriel, Gil-Sanchez, Garcia-Segura and Arevalo2014). Thus, the relation between CORT and parenting behavior might be brought forth by the stress-induced upregulation of the HPA axis during parturition, which facilitates maternal recognition and bonding directly postpartum. An indication that such a role of CORT remains relevant later in life comes from a study that shows that in monogamous mammals, exposure to stress and CORT injections both induce partner preference (DeVries, DeVries, Taymans, & Carter, Reference DeVries, DeVries, Taymans and Carter1996).

Prolactin (PRL)

The physiological functions of PRL are diverse, but in mammalian species it is most known for its role in lactation. The function of this hormone might however extend to stress reduction, social learning, and maternal protection (Neumann, Reference Neumann2009). In rodents, PRL has convincingly been demonstrated to induce maternal behavior (Nelson, Reference Nelson2005), but its role in primate maternal behavior is unknown and has not received much attention (Saltzman & Maestripieri, Reference Saltzman and Maestripieri2011). PRL has been studied in relation to primate paternal behavior, and these studies show that PRL is most likely important to instigate caring behavior by motivating fathers for caregiving behavior (Storey, Noseworthy, Delahunty, Halfyard, & McKay, Reference Storey, Noseworthy, Delahunty, Halfyard and McKay2011), similar to the role of CORT described above.

Opioids

Opioids act on several receptor types to inhibit pain processing and are an important component in the neural reward circuitry by mediating affective responses to pleasurable stimuli (Nelson, Reference Nelson2005). During parturition, endogenous opioids may alleviate pain and together with OXT stimulate the onset of maternal behavior (Keverne & Kendrick, Reference Keverne and Kendrick1994). After parturition, opioids regulate attachment behaviors and mother–infant bonding by reducing separation anxiety and increasing motivation for social interaction (Nelson & Panksepp, Reference Nelson and Panksepp1998). Opioids also modulate primate maternal behavior, as variability in genes coding for the opioid receptor and pharmacological manipulations of the opioids system have demonstrated strong effects on maternal behavior (Curley, Reference Curley2011; Saltzman & Maestripieri, Reference Saltzman and Maestripieri2011).

PROG

In humans, during pregnancy PROG levels gradually increase, and this increase is important for the maintenance of the pregnancy. Although the role of PROG during pregnancy in relation to maternal behavior in most mammalian species is not entirely clear, in rats it is proposed to sensitize the brain for infant stimuli and to control the timing of this responsivity (Bridges, Reference Bridges2015). However, PROG levels during pregnancy follow a different profile in rats and humans and, thus, might serve a different function. In other mammals, such as in sheep, priming of OXT receptors by PROG (together with estradiol as described above) is essential for mother–infant bonding (Keverne & Kendrick, Reference Keverne and Kendrick1994). Data from primates has shown involvement of PROG in maternal caregiving, but no consistent pattern has emerged so far. Both positive and negative relations with maternal behavior have been found for PROG, depending on the species involved and the methodology used to measure endocrine levels (Saltzman & Maestripieri, Reference Saltzman and Maestripieri2011).

Endocrine interactions

The hormones and neuropeptides described in this review orchestrate complex social–emotional behavioral repertoires, which rely on neural networks that, although conserved throughout the mammalian lineage, have evolved over the course of millions of years. Hormonal factors can adaptively change the activation patterns of these neural networks, often in coordination with peripheral release, to deal with environmental cues. As such, variation in hormonal control allows for dynamic variation in response patterns between and within species (Goodson & Kabelik, Reference Goodson and Kabelik2009). Considering the coordinated fashion in which hormonal factors are involved in behavioral regulation, a single hormonal factor being responsible for a specific behavioral outcome is likely to be an exception. Although the role of the above hormones on behavior are generally described in isolation, it is important to stress that none of the reported endocrine factors function in isolation. Diverse interactive effects have been observed between steroid hormones, neuropeptides, and several neurotransmitter systems. For example, there is bidirectional control between the HPA and HPG axes, which might explain some of the sex-specific responses to stress (Viau, Reference Viau2002). As mentioned above, testosterone is easily converted into estradiol via aromatase in both males and females. Both estradiol and testosterone affect dopaminergic and serotonergic activity (de Souza Silva, Mattern, Topic, Buddenberg, & Huston, Reference de Souza Silva, Mattern, Topic, Buddenberg and Huston2009; McEwen & Alves, Reference McEwen and Alves1999) and have stimulating effects on the expression of OXT and AVP in several brain regions (de Vries, Reference de Vries2008). Estradiol further interacts with PROG and PRL function, as well as with the noradrenergic system (McEwen & Alves, Reference McEwen and Alves1999). In addition, several of the effects of OXT are established by interacting with the dopamine and opioid system (Rilling & Young, Reference Rilling and Young2014), and when released in high quantities, OXT and AVP can activate each other's receptors (Carter, Reference Carter2014). Because these examples are only a fraction of all possible interactions between the endocrine factors described in this review, it is important to realize that although the hormones and neuropeptides in the remainder of this review are discussed in relative isolation, this does not do justice to the complexity of the exact neuroendocrine mechanisms that mediate social–emotional behavior. Instead of representing approximations of the exact neuroendocrine mechanisms involved in human caregiving and its intergenerational transmission, the models proposed in this review are to be seen as heuristic models, guiding further research, based on which new hypotheses can be formulated.

OXT and AVP

Of all of the hormonal factors, which effects on parenting are summarized in Figure 1, OXT has certainly received the most attention due to both the ease of administration (Guastella & MacLeod, Reference Guastella and MacLeod2012) and the strong link with parental and social bonding as evidenced by animal research as described above. Based on the considerable number of studies that employed OXT administration procedures in humans, it has convincingly been demonstrated that OXT increases emotion recognition in others (Shahrestani, Kemp, & Guastella, Reference Shahrestani, Kemp and Guastella2013), trust toward ingroup members (van IJzendoorn & Bakermans-Kranenburg, Reference van IJzendoorn and Bakermans-Kranenburg2012), and cognitive empathic behavior (Bos, Panksepp, et al., Reference Bos, Panksepp, Bluthe and Honk2012). OXT administration has further been shown to increase neural empathic responses to infant crying (Riem, Bakermans-Kranenburg, et al., Reference Riem, Bakermans-Kranenburg, Pieper, Tops, Boksem, Vermeiren and Rombouts2011) and functional connectivity of the amygdala with the orbitofrontal cortex (OFC; an important region critical for reward evaluation) in response to infant laughter (Riem, van IJzendoorn, et al., Reference van IJzendoorn, Huffmeijer, Alink, Bakermans-Kranenburg and Tops2011). This increased connectivity of the amygdala with frontal brain regions also indicates facilitation of emotion regulation (Kim et al., Reference Kim, Loucks, Palmer, Brown, Solomon, Marchante and Whalen2011). In addition, OXT responses in mothers after interaction with their child predicted neural reward responses in the ventral striatum toward their children's faces (Strathearn, Fonagy, Amico, & Montague, Reference Strathearn, Fonagy, Amico and Montague2009), indicating increased sensitivity for the rewarding properties of infant stimuli. Administration of OXT in fathers tended to result in less hostility (Naber, van IJzendoorn, Deschamps, van Engeland, & Bakermans-Kranenburg, Reference Naber, van IJzendoorn, Deschamps, van Engeland and Bakermans-Kranenburg2010) and increased touch and social reciprocity toward their children during interactive play (Weisman, Zagoory-Sharon, & Feldman, Reference Weisman, Zagoory-Sharon and Feldman2012). Outside the context of parent–infant interaction, OXT administration also leads to a reduction of physiological stress responses during social interaction (Heinrichs, Baumgartner, Kirschbaum, & Ehlert, Reference Heinrichs, Baumgartner, Kirschbaum and Ehlert2003) and to increased social bonding of adult partners during conflict (Ditzen et al., Reference Ditzen, Schaer, Gabriel, Bodenmann, Ehlert and Heinrichs2009). These effects might rely on the neurobiological mechanisms by which OXT exerts its soothing effects during mother–infant interaction after birth (Carter, Reference Carter1998; Panksepp, Reference Panksepp2009). For example, breastfeeding, which leads to upregulated OXT levels in both mother and infant, also results in reduced stress responses in mothers (Heinrichs, Neumann, & Ehlert, Reference Heinrichs, Neumann and Ehlert2002).

Figure 1. Model of the hormones and peptides relevant for human caregiving. The behavioral effects of the described hormones and neuropeptides are indexed depending on the relevant context, either in safe environments to elicit tending and caring behavior or in more unsafe or challenging environments supporting protecting and exploring behavior. Arrows next to the descriptions of the behavior are used to indicate whether the hormonal factors are expected to increase or decrease this type of behavior. *Behavioral effects observed in mammals, but not convincingly demonstrated in humans.

The OXT literature has also demonstrated the context dependency of the effects of OXT (Bakermans-Kranenburg & van IJzendoorn, Reference Bakermans-Kranenburg and van IJzendoorn2013; Bartz, Zaki, et al., Reference Bartz, Zaki, Bolger and Ochsner2011). A meta-analysis shows that the effects of OXT on trust are selective for ingroup members (van IJzendoorn & Bakermans-Kranenburg, Reference van IJzendoorn and Bakermans-Kranenburg2012), and OXT has also shown to increase protective decision making selectively for ingroup members (De Dreu et al., Reference De Dreu, Greer, Handgraaf, Shalvi, van Kleef, Baas and Feith2010). OXT has even been demonstrated to have antagonistic effects toward outgroup members (De Dreu, Greer, Vsn Kleef, Shalvi, & Handgraaf, Reference De Dreu, Greer, van Kleef, Shalvi and Handgraaf2011), effects that can all be interpreted as serving a protective function, whereby the prosocial effects of OXT are reserved for closely related others. Such a function of OXT bears similarities with increased maternal aggression toward intruders after OXT administration observed in rodents (Bosch, Meddle, Beiderbeck, Douglas, & Neumann, Reference Bosch, Meddle, Beiderbeck, Douglas and Neumann2005).

A neuropeptide related to OXT of which animal studies have shown considerable involvement in parenting behavior is AVP (e.g., Bosch & Neumann, Reference Bosch and Neumann2008; see Figure 1). Although studies on the role of AVP in human parenting are scarce, a recent study found a positive relation between AVP levels and parent–infant joint attention, as well as stimulatory touch (Apter-Levi, Zagoory-Sharon, & Feldman, Reference Apter-Levi, Zagoory-Sharon and Feldman2014). In addition, administration of AVP in expectant fathers increased their attention for infant stimuli compared to control fathers (Cohen-Bendahan, Beijers, van Doornen, & de Weerth, Reference Cohen-Bendahan, Beijers, van Doornen and de Weerth2015). Other studies on social behavior have shown that AVP administration increases sensitivity to social stress (Ebstein et al., Reference Ebstein, Israel, Lerer, Uzefovsky, Shalev, Gritsenko and Yirmiya2009) and aggressive facial responses to others (Thompson, George, Walton, Orr, & Benson, Reference Thompson, George, Walton, Orr and Benson2006), but also increased social cooperation (Brunnlieb et al., Reference Brunnlieb, Nave, Camerer, Schosser, Vogt, Münte and Heldmann2016; Rilling et al., Reference Rilling, DeMarco, Hackett, Chen, Gautam, Stair and Patel2014) and empathic concern for others (Tabak et al., Reference Tabak, Meyer, Castle, Dutcher, Irwin, Han and Eisenberger2015). Some of these effects are contrary to those of OXT, indicating that both peptides could have very different functions in relation to parenting behavior, relating to the strong sex differences that are observed after administration (Rilling et al., Reference Rilling, DeMarco, Hackett, Chen, Gautam, Stair and Patel2014; Thompson et al., Reference Thompson, George, Walton, Orr and Benson2006). Based on an overview of the animal and human literature, it is proposed that AVP and OXT are both important, yet serve different functions in challenging situations and in dealing with threat (see Figure 1). Especially OXT is likely to be more important in the reinstatement of homeostasis after stress and in coping with chronic stress by directing behavior more toward social engagement (Bos, Panksepp, et al., Reference Bos, Panksepp, Bluthe and Honk2012; Carter, Reference Carter2014; Neumann, Reference Neumann2009). The distinct roles these neuropeptides have in stress responding might translate to the proposed different roles they have in parental caregiving (Apter-Levi et al., Reference Apter-Levi, Zagoory-Sharon and Feldman2014). In this light it is important to consider the sex-specific effects that OXT and AVP can have on human social–emotional behavior. Although data on humans is scarce, research in other mammalian species show that sexes vary in the expression of OXT and AVP and in the location and density of their respective receptor distributions (Dumais & Veenema, Reference Dumais and Veenema2016). Such sex differences might contribute to the differences between maternal and paternal behavior and sex-specific dealing with stressful situations. However, due to paucity of data in humans, together with the often selective inclusion of only males or females in human clinical studies (Bos, Panksepp, et al., Reference Bos, Panksepp, Bluthe and Honk2012), such interpretations have to be considered with care.

Testosterone and estradiol

In contrast to the mostly beneficial effects of OXT on the behavioral repertoire of sensitive caregiving, the functions of testosterone are generally antagonistic to such parenting behavior. For example, testosterone reduces cognitive empathic behavior (van Honk et al., Reference van Honk, Schutter, Bos, Kruijt, Lentjes and Baron-Cohen2011), interpersonal trust (Bos, Terburg, & van Honk, Reference Bos, Terburg and van Honk2010), and automatic empathic facial responses toward others (Hermans, van Wingen, Bos, Putnam, & van Honk, Reference Hermans, van Wingen, Bos, Putman and van Honk2009). In addition, testosterone reduces neural connectivity of the amygdala with the prefrontal cortex (PFC; Bos, Hermans, Ramsey, & van Honk, Reference Bos, Hermans, Ramsey and van Honk2012; van Wingen, Mattern, Verkes, Buitelaar, & Fernandez, Reference van Wingen, Mattern, Verkes, Buitelaar and Fernandez2010), symptomatic for reduced emotion regulation (Kim et al., Reference Kim, Loucks, Palmer, Brown, Solomon, Marchante and Whalen2011). In fathers, basal testosterone levels are inversely related to the amount of caregiving they provide (Mascaro, Hackett, & Rilling, Reference Mascaro, Hackett and Rilling2013), and testosterone responses to erotic behavior in nonfathers are inversely related to interest in babies (Zilioli et al., Reference Zilioli, Ponzi, Henry, Kubicki, Nickels, Wilson and Maestripieri2016). Furthermore, longitudinal data has shown that parenthood results in reduced basal testosterone levels in both mothers and fathers, which in the light of the effects of testosterone indicates adaptive downregulation of testosterone function (Gettler, McDade, Feranil, & Kuzawa, Reference Gettler, McDade, Feranil and Kuzawa2011; Kuzawa, Gettler, Huang, & McDade, Reference Kuzawa, Gettler, Huang and McDade2010).

However, in some contexts, similar to the role of AVP in challenging situations, testosterone also has effects on social–emotional behavior that might be beneficial for caregiving behavior. Testosterone increases social vigilance (Hermans, Ramsey, & van Honk, Reference Hermans, Ramsey and van Honk2008), sensitivity to emotional facial expressions (Bos, Hermans, et al., Reference Bos, Hermans, Ramsey and van Honk2012; Bos, van Honk, Ramsey, Stein, & Hermans, Reference Bos, van Honk, Ramsey, Stein and Hermans2013), and sensitivity for reward (Hermans et al., Reference Hermans, Bos, Ossewaarde, Ramsey, Fernandez and van Honk2010). In fathers, basal testosterone levels correlate positively with neural activation in reward regions when observing their own children (Kuo, Carp, Light, & Grewen, Reference Kuo, Carp, Light and Grewen2012), and in nonmaternal women higher basal testosterone levels predict increased reward values for cuteness of infant faces (Hahn, DeBruine, Fisher, & Jones, Reference Hahn, DeBruine, Fisher and Jones2015). Administration of testosterone also increases neural activation in response to crying infants in the thalamocingulate circuit (Bos, Hermans, Montoya, Ramsey, & van Honk, Reference Bos, Hermans, Montoya, Ramsey and van Honk2010), a key circuit for parental responsiveness (MacLean, Reference MacLean1985). As such, testosterone might facilitate caregiving by serving a protective function when facing challenge or threat, but also in successfully dealing with infant distress. Tentative support for this notion comes from studies showing increases in testosterone levels in fathers listening to infant crying (Fleming, Corter, Stallings, & Steiner, Reference Fleming, Corter, Stallings and Steiner2002), whereas no such increase in testosterone is observed when listening to crying was combined with active nurturing behavior (van Anders, Tolman, & Volling, Reference van Anders, Tolman and Volling2012). Perhaps testosterone directly increases in response to crying, in order to motivate the parent for active caring behavior, followed by a fast decrease when care is provided. This view is not in conflict with the longitudinally observed decreases in basal testosterone levels, but merely shows that low baseline in combination with fast reactivity of the HPG axis could be optimal for caregiving behavior. Increased parental behavior was related to stronger declines in testosterone levels in fathers after the strange situation procedure (Kuo et al., Reference Kuo, Saini, Thomason, Schultheiss, Gonzalez and Volling2016). In addition, while baseline testosterone levels had no effect, stronger diurnal variability in testosterone levels in fathers was related to more sensitivity toward the child during play behavior (Endendijk et al., Reference Endendijk, Hallers-Haalboom, Groeneveld, van Berkel, van der Pol, Bakermans-Kranenburg and Mesman2016). Mothers however showed a different pattern, with higher baseline testosterone levels predicting more sensitivity. However, diurnal variability was inversely related to sensitivity toward the child, indicating a sex-specific role of testosterone in caregiving behavior. Perhaps similar to the context-dependent functions of OXT described above, the role of testosterone might be critical for “protective” parenting behavior, which is an indispensable behavioral repertoire in addition to more “tending” behavior, in order to guarantee reproductive success. As such, the role of testosterone might possibly be more important in paternal compared to maternal protective behavior, as maternal aggression is mediated partly by OXT. However, because testosterone administration studies have mostly been performed in females, and most baseline studies in males, more comparative studies between males and females such as Endendijk et al. (Reference Endendijk, Hallers-Haalboom, Groeneveld, van Berkel, van der Pol, Bakermans-Kranenburg and Mesman2016) should be performed to give more insight.

Another potential role of testosterone in parenting is indirectly, by conversion of testosterone to estradiol (see Endocrine interactions section). In humans, estradiol is known for its role in the development of women's secondary sexual characteristics and facilitation of ovulation and implantation during the menstrual cycle. It is surprising that only a few studies have been conducted investigating the role of estradiol in relation to human parenting. One study shows that mothers with higher levels of estradiol report higher postpartum feelings of attachment (Fleming, Ruble, Krieger, & Wong, Reference Fleming, Ruble, Krieger and Wong1997), whereas more indirect evidence for a role of estradiol in human parenting comes from a study in which variation of the ER polymorphism correlated with neural responses toward one's own child (Lahey et al., Reference Lahey, Michalska, Liu, Chen, Hipwell, Chronis-Tuscano and Decety2012). Research on the role of estradiol in other human social–emotional behavior is equally scarce, although a recent study in which estradiol was administered in males shows increased vicarious responses to pain observed in others (Olsson, Kopsida, Sorjonen, & Savic, Reference Olsson, Kopsida, Sorjonen and Savic2016), pointing at increased affective empathy after estradiol. In addition, there is evidence for a role of estradiol in emotional memory and fear learning in females (Zeidan et al., Reference Zeidan, Igoe, Linnman, Vitalo, Levine, Klibanski and Milad2011), which might relate to the role of estradiol in maternal memory for infant cues in other mammals (Banerjee & Liu, Reference Banerjee and Liu2013). Overall, this scarcity of studies on the relation between estradiol and human parenting, or other forms of social–emotional behavior, is surprising considering the number of women, including mothers, who take estradiol in the form of oral contraceptives on a daily basis. Comparative studies between naturally cycling women and those using oral contraceptives have shown differences in partner preference and selection (Little, Burriss, Petrie, Jones, & Roberts, Reference Little, Burriss, Petrie, Jones and Roberts2013), in emotional memory and stress responses (Nielsen, Segal, Worden, Kim, & Cahill, Reference Nielsen, Segal, Worden, Yim and Cahill2013), and in mood and neural responses to emotional stimuli (Gingnell et al., Reference Gingnell, Engman, Frick, Moby, Wikström, Fredrikson and Sundström-Poromaa2013). These effects might be due to the altered levels of estradiol in the oral contraception group, and therefore, more research on the role of estradiol in human (parenting) behavior is of utmost importance.

CORT

In the context of parenting, and similar to testosterone, CORT also increases in response to infant crying and predicts alertness to such crying (Fleming et al., Reference Fleming, Corter, Stallings and Steiner2002). Furthermore, CORT administration upregulates hippocampal responses to infant cry sounds (Bos, Montoya, Terburg, & van Honk, Reference Bos, Montoya, Terburg and van Honk2014). The hippocampus is also a critical region for social recognition (Hitti & Siegelbaum, Reference Hitti and Siegelbaum2014), and activation of the hippocampus has previously been related to levels of maternal bonding (Musser, Kaiser-Laurent, & Ablow, Reference Musser, Kaiser-Laurent and Ablow2012). In line with this, mothers with higher CORT levels are better able to distinguish infant cry sounds and respond more affectionately to these sounds (Stallings, Fleming, Corter, Worthman, & Steiner, Reference Stallings, Fleming, Corter, Worthman and Steiner2001). Similar effects are observed for different modalities, as higher CORT levels in mothers also predict increased sensitivity and recognition of olfactory infant cues (Fleming, Steiner, & Corter, Reference Fleming, Steiner and Corter1997). CORT might however also act on more deliberate and reflective behavior, as mothers with higher CORT also hold more positive attitudes toward their own children (Fleming, Ruble, et al., Reference Fleming, Ruble, Krieger and Wong1997). Thus, in line with studies in other primates, CORT in humans might similarly facilitate arousal-dependent initiation of caregiving behavior. Higher levels of CORT in males predicted stronger feelings of closeness toward a previously unknown interaction partner after social stress (Berger, Heinrichs, von Dawans, Way, & Chen, Reference Berger, Heinrichs, von Dawans, Way and Chen2016). Furthermore, a recent study showed that administration of a MR agonist results in more self-reported empathic concern for others (Wingenfeld et al., Reference Wingenfeld, Kuehl, Janke, Hinkelmann, Dziobek, Fleischer and Roepke2014), similar to effects of AVP and OXT. In contrast to these positive relations between CORT and human caregiving behavior, opposite relations have also been reported, which might be due to persistent high levels of CORT in the context of continued stress or depression, which can lead to suppression of maternal motivation (Saltzman & Maestripieri, Reference Saltzman and Maestripieri2011).

Other candidates: Opioids, PRL, and PROG

Despite the theoretical background on the role of these hormonal factors in shaping parental behavior, experimental studies in humans are scarce. The studies on human social behavior that have been performed so far show that opioids reduce social pain after rejection and facilitate sensitivity for social reward (Chelnokova et al., Reference Chelnokova, Laeng, Eikemo, Riegels, Løseth, Maurud and Leknes2014; Hsu et al., Reference Hsu, Sanford, Meyers, Love, Hazlett, Wang and Korycinski2013). Administration of naltrexone, an opioid antagonist, decreased feelings of social connection both in the laboratory and in daily life, which is a strong indicator of involvement of the opioid system in human social bonding (Inagaki, Ray, Irwin, Way, & Esenberger, Reference Inagaki, Ray, Irwin, Way and Eisenberger2016). PRL has been studied in relation to parenting, and in males, increased PRL levels have been observed in anticipation of fatherhood, as well as in fathers compared to nonfathers (Storey & Ziegler, Reference Storey and Ziegler2016). Other data from fathers shows that levels increase when listening to infant crying, and that higher levels predicted more alert and positive responses to the cry sounds (Fleming et al., Reference Fleming, Corter, Stallings and Steiner2002). In addition, baseline levels of PRL predicted exploratory play behavior in fathers when interacting with their 6-month-old infants (Gordon, Zagoory-Sharon, Leckman, & Feldman, Reference Gordon, Zagoory-Sharon, Leckman and Feldman2010). Overall, while these studies show that higher PRL levels predict increased paternal behavior, levels generally decline during father–infant interaction, which might indicate that PRL, comparable to the role of testosterone, is most important in initiation of paternal behavior rather than the actual caregiving behavior itself (Storey & Ziegler, Reference Storey and Ziegler2016). Nonetheless, these findings are promising and can hopefully spawn more research on the role of PRL in human parenting, although research in this direction is challenging due to the limited availability of methods to pharmacologically alter PRL levels in humans. Synthetic PRL, for example, is currently not available for administration purposes in humans. In contrast, PROG can be pharmacologically administered in humans, but effects of administration on human social behavior have not yet been investigated. Nonetheless, a few studies measured baseline PROG levels in response to social manipulations, and these studies suggest a role for PROG in facilitating affiliation during or after stress. More specifically, PROG increases after the observation of a movie clip showing social rejection, and correlated with the participants’ motivation to affiliate (Wirth & Schultheiss, Reference Wirth and Schultheiss2006). Another study measured increased PROG levels after social interaction with an unfamiliar other, which predicted altruistic responses to this person a week later (Brown et al., Reference Brown, Fredrickson, Wirth, Poulin, Meier, Heaphy, Cohen and Schultheiss2009). Finally, a recent study showed that after social rejection, PROG increased when given the opportunity to affiliate with others, specifically in individuals sensitive for social rejection, whereas socially anxious subjects showed a decrease in PROG after social rejection (Maner, Miller, Schmidt, & Eckel, Reference Maner, Miller, Schmidt and Eckel2010). These few studies suggest that, similar to opioids and PRL, future research on PROG hold great promise for increasing our understanding of the neurobiology underlying human social bonding and caregiving.

Genetic variation of endocrine receptors

The primary ontogenetic factor that can explain variability in endocrine sensitivity is the variation in genes coding for the receptors of hormones and neuropeptides. Most research has been done on two single nucleotide polymorphisms (SNPs) of the OXTR gene, rs53576 and rs2254298, and these studies have generally attempted to investigate the direct effect of individual variation in SNPs on measures of social behavior. This approach has resulted in some promising findings. Individuals homozygous for the G allele of OXTR rs53576 polymorphism expressed more neural empathic responses to pain in others (Luo et al., Reference Luo, Li, Ma, Zhang, Rao and Han2015), showed stronger differentiation of neural responses to infant emotions in mothers (Peltola et al., Reference Peltola, Yrttiaho, Puura, Proverbio, Mononen, Lehtimäki and Leppänen2014), and showed higher levels of sensitive parenting (Bakermans-Kranenburg & van IJzendoorn, Reference Bakermans-Kranenburg and van IJzendoorn2008). Children homozygous for the G allele show stronger structural abnormalities in the ventral striatum after increasing early life trauma experiences (Dannlowski et al., Reference Dannlowski, Kugel, Grotegerd, Redlich, Opel, Dohm and Baune2015), pointing at increased vulnerability to early life stress in this group. In adults higher self-reported childhood attachment predicted alterations in neural structure and activation, in regions related to emotional processing, selectively in individuals homozygous for the G allele (Schneider-Hassloff et al., Reference Schneider-Hassloff, Straube, Jansen, Nuscheler, Wemken, Witt and Kircher2016). However, several inconsistent findings have been reported, such as reduced levels of positive parenting and accompanying neural empathic responses toward child stimuli for participants homozygous for the G allele (Michalska et al., Reference Michalska, Decety, Liu, Chen, Martz, Jacob and Lahey2014). In addition, a recent meta-analyses failed to find evidence for significant effects of the rs53576 and rs2254298 SNPs on social behavior (Bakermans-Kranenburg & van IJzendoorn, Reference Bakermans-Kranenburg and van IJzendoorn2014).

A difficulty in the approach of directly predicting behavior from variation in SNPs is that the underlying molecular function of these SNPs is currently unknown, which makes it difficult to compare the effects of these studies to endocrine administration and baseline studies. For example, the above-mentioned beneficial effects of a homozygous G allele on empathic responses (Luo et al., Reference Luo, Li, Ma, Zhang, Rao and Han2015) seem in contrast to a study in which administration of OXT results in reduced empathic neural responses to pain in others (Bos Montoya, Hermans, Keysers, & van Honk, Reference Bos, Montoya, Hermans, Keysers and van Honk2015). Thus, critical to further our understanding of how variation in OXTR SNPs contributes to variation in caregiving behavior is understanding the functional significance of these SNPs on the neurobiological level. A first step is a study in humans that shows that rs2254298 GG carriers have lower levels of peripheral OXT, which subsequently predict higher depression and risk of child pathology (Apter-Levy, Feldman, Vakart, Ebstein, & Feldman, Reference Apter-Levy, Feldman, Vakart, Ebstein and Feldman2013). A promising approach that can give critical insight in how SNPs alter neuroendocrine sensitivity is the combination of hormone administration paradigms with genetic analyses. Employing this approach, Marsh et al. (Reference Marsh, Henry, Pine, Gorodetsky, Goldman and Blair2012) show that rs53576 GG allele carriers are more sensitive to infant faces after OXT administration, pointing at increased sensitivity for OXT. Another recent study following a similar approach, yet focusing on several OXTR SNPs, shows that the rs53576 A allele was associated with increased emotion recognition performance under OXT versus placebo (Chen et al., Reference Chen, Kumsta, Dvorak, Domes, Yim, Ebstein and Heinrichs2015), which together with the study by Marsh et al. (Reference Marsh, Henry, Pine, Gorodetsky, Goldman and Blair2012) is in line with the studies described above by Luo et al. (Reference Luo, Li, Ma, Zhang, Rao and Han2015) and Bakermans-Kranenburg and van IJzendoorn (Reference Bakermans-Kranenburg and van IJzendoorn2008). In addition, a less well-studied SNP (rs401015) modulated activation of the amygdala to adult faces after OXT administration (Montag, Sauer, Reuter, & Kirsch, Reference Montag, Sauer, Reuter and Kirsch2013). Studies such as these are promising attempts to bridge the gap in the causal chain between genetic variation and variation in caregiving behavior.

A genetic polymorphism in which biochemical function is better understood is variation in the number of repeats of CAG trinucleotide in the first exon of the AR. The number of CAG repeats is inversely related to sensitivity of the AR, on which testosterone exerts most of its genomic effects (Chamberlain, Driver, & Miesfeld, Reference Chamberlain, Driver and Miesfeld1994). Consistent with this inverse relation is the finding that men with shorter CAG repeat lengths exhibited greater upper body strength, scored higher on self-report measures of dominance (Simmons & Roney, Reference Simmons and Roney2011; but see Hurd, Vaillancort, & Dinsdale, Reference Hurd, Vaillancourt and Dinsdale2011), and had increased activity of the amygdala to facial expressions of negative affect (Manuck et al., Reference Manuck, Marsland, Flory, Gorka, Ferrell and Hariri2010). These are all measures typically related to testosterone function (Bos, Panksepp, et al., Reference Bos, Panksepp, Bluthe and Honk2012). In addition, shorter CAG repeat lengths also predicted larger testosterone responses when males interacted with young women (Roney, Simmons, & Lukaszewski, Reference Roney, Simmons and Lukaszewski2010) and less empathic neural responses to infant crying in fathers (Mascaro, Hackett, Gouzoules, Lori, & Rilling, Reference Mascaro, Hackett, Gouzoules, Lori and Rilling2014). Furthermore, baseline salivary levels of testosterone have shown to correlate positively with CAG repeat lengths (Manuck et al., Reference Manuck, Marsland, Flory, Gorka, Ferrell and Hariri2010). This might indicate a compensation mechanism where higher sensitivity downregulates physiological levels of testosterone, which illustrates the difficulties of studying endocrine parameters in isolation.

This notion becomes even clearer when considering the diverse pathways by which testosterone can influence brain function. One of these pathways is the conversion of testosterone to estradiol by aromatase. This pathway is of great relevance, because aromatase can act quickly and estradiol has diverse effects on social behavior and the brain, both genomically and nongenomically. Because variation in SNPs for the ER predicted mother's neural responses to their own children, and indirectly predicted observed negative parenting, this pathway might account for some of the effects ascribed to testosterone. With regard to fast effects of aromatase, this gives infinite temporal flexibility (Balthazart, Taziaux, Holloway, Ball, & Cornil, Reference Balthazart, Taziaux, Holloway, Ball and Cornil2009; McEwen & Alves, Reference McEwen and Alves1999). Considering that aromatase is densely located in regions of the human brain such as the amygdala (Biegon et al., Reference Biegon, Kim, Alexoff, Jayne, Carter, Hubbard and Fowler2010), some of the effects seen after testosterone administration might be caused by local conversion to estradiol. A further step in scrutinizing this pathway is to find the genes that can explain variation in the functionality of this cascade. Such a gene is the recently identified transcription factor retinoic acid receptor-related orphan receptor alpha (RORA), which regulates expression of aromatase (Sarachana, Xu, Wu, & Hu, Reference Sarachana, Xu, Wu and Hu2011). In addition, the expression of RORA itself is controlled by steroid hormones, as it is downregulated by the androgen dihydrotestosterone and upregulated by estradiol, making RORA an additional interesting target to study individual variation in relation to caregiving or other steroid regulated social behavior.

Another candidate is the gene coding for the mu-opioid receptor M1 (OPRM1), which predicts variation in neural and dispositional sensitivity for social rejection in humans (Way, Taylor, & Eisenberger, Reference Way, Taylor and Eisenberger2009), as well as self-reported fearful attachment (Troisi et al., Reference Troisi, Frazzetto, Carola, Di Lorenzo, Coviello, Siracusano and Gross2012). Animal studies have shown a strong link between OPRM1 and mother–infant attachment (Curley, Reference Curley2011). Finally, variation in the genes coding for AVP receptor 1a (AVPR1a) and MR are of interest because the function of these receptors have both been related to empathy. Although it warrants replication, a single study reported that variation in the AVPR1a genes correlated with self-reported trait cognitive empathy (Uzefovsky et al., Reference Uzefovsky, Shalev, Israel, Edelman, Raz, Mankuta and Ebstein2015). The relevance of the MR is demonstrated by a study showing that administration of an MR agonist increased self-reported empathic concern (Wingenfeld et al., Reference Wingenfeld, Kuehl, Janke, Hinkelmann, Dziobek, Fleischer and Roepke2014).

Despite the importance of genetic variation as a first etiological factor for endocrine sensitivity, the vast complexity of the intermediate steps from genes to phenotypic outcomes limits the predictive values of the “candidate genes approach” (Chabris et al., Reference Chabris, Hebert, Benjamin, Beauchamp, Cesarini, van der Loos and Laibson2012). Regarding this complexity, a highly interesting and relatively new research area is the field of epigenetics, which looks at how genetic expression can be changed under the influence of environmental factors, such as early childhood or even prenatal experiences, without changing the genetic code itself (Zhang & Meaney, Reference Zhang and Meaney2010). Although the field of human epigenetics is young, the few studies that have been done in humans on epigenetic expression of endocrine receptors are promising. For example, increased methylation of the OXTR gene, which generally causes reduced expression of the receptor, resulted in stronger neural responses to negative emotional expressions and reduced connectivity of the PFC with the amygdala, which could suggest less efficient emotion regulation (Puglia, Lillard, Morris, & Connelly, Reference Puglia, Lillard, Morris and Connelly2015). This finding is in line with a study showing increased OXTR methylation in children that develop callous–unemotional traits (Cecil et al., Reference Cecil, Lysenko, Jaffee, Pingault, Smith, Relton and Barker2014) and in individuals with autism (Gregory et al., Reference Gregory, Connelly, Towers, Johnson, Biscocho, Markunas and Ellis2009). However, a first study that investigated depression found decreased methylation of the OXTR gene compared to healthy controls, and further showed that this effect depended on both the exact location of the methylation and on the polymorphism of the OXTR gene (Reiner et al., Reference Reiner, van IJzendoorn, Bakermans-Kranenburg, Bleich, Beutel and Frieling2015). This preliminary finding forewarns that the relation between epigenetic expression and phenotypic outcome is complex. In addition, another recent finding demonstrates the temporal dynamics of such methylation. Subjects exposed to social stress showed increased OXTR methylation directly after stress exposure, which normalized again in a follow-up the year after (Unternaehrer et al., Reference Unternaehrer, Luers, Mill, Dempster, Meyer, Staehli and Meinlschmidt2012).

Research on the epigenetic expression of the CORT receptors has revealed strong sensitivity of the GR and the FK506 binding protein 5 (FKBP5) gene methylation to early life stress in humans (Klengel et al., Reference Klengel, Mehta, Anacker, Rex-Haffner, Pruessner, Pariante and Bradley2013; Kundakovic & Champagne, Reference Kundakovic and Champagne2015). FKBP5 is an important regulator of the GR complex, and childhood trauma affects transcription of FKBP5, which results in long-term alteration of the stress system and hippocampal structure (Klengel et al., Reference Klengel, Mehta, Anacker, Rex-Haffner, Pruessner, Pariante and Bradley2013). Because epigenetic expression of the GR and FKBP5 polymorphisms have both been related to mood and anxiety disorders (e.g., Klengel et al., Reference Klengel, Mehta, Anacker, Rex-Haffner, Pruessner, Pariante and Bradley2013; Vukojevic et al., Reference Vukojevic, Kolassa, Fastenrath, Gschwind, Spalek, Milnik and Demougin2014), studies on epigenetic expression of endocrine receptors can contribute to increased understanding of the etiology of stress-related psychopathology. In addition, these studies also pave the way for an explanatory mechanism for the environmental effects of endocrine sensitivity and subsequent variation in caregiving behavior. Already starting before conception, the environment alters the epigenetic expression of individuals (Gapp et al., Reference Gapp, Jawaid, Sarkies, Bohacek, Pelczar, Prados and Mansuy2014). In the remainder of this review, the focus will be on the sensitive periods from fetal development to early childhood.

Fetal programming

The time spent in the womb is the first sensitive period after conception that is important in the regulation of endocrine sensitivity. During this period, testosterone levels have strong organizational effects on the brain. A line of studies has investigated the effect of fetal testosterone as measured from amniotic fluid in pregnant women on the development of children after birth. These studies show that fetal testosterone predicts the children's brain structure in a sexually dimorphic way (Lombardo, Ashwin, Auyeyng, Chakrabarti, Taylor, et al., Reference Lombardo, Ashwin, Auyeung, Chakrabarti, Lai, Taylor and Baron-Cohen2012). That is, in areas involved in cognitive empathy, such as the temporoparietal junction and the inferior frontal gyrus, gray matter volume was predicted by fetal testosterone consistent with observed sex differences in these children. Fetal testosterone further predicted sex-typical play behavior (Auyeung et al., Reference Auyeung, Baron-Cohen, Ashwin, Knickmeyer, Taylor, Hackett and Hines2009) and neural responses to socially rewarding stimuli (Lombardo, Ashwin, Auyeung, Chakrabarti, Lai, et al., Reference Lombardo, Ashwin, Auyeung, Chakrabarti, Lai, Taylor and Baron-Cohen2012), similar to the effects of testosterone described above. Finally, in a large longitudinal data sample, elevated levels of testosterone and PROG were found in amniotic fluid in individuals diagnosed with autism later in life (Baron-Cohen et al., Reference Baron-Cohen, Auyeung, Nørgaard-Pedersen, Hougaard, Abdallah, Melgaard and Lombardo2015). A different line of studies investigated effects in women exposed to elevated levels of androgens due to the rare genetic disease congenital adrenal hyperplasia. These women show masculinized behavior and interests, including reduced interest in infants (Leveroni & Berenbaum, Reference Leveroni and Berenbaum1998) and motherhood (Ehrhardt & Meyer-Bahlburg, Reference Ehrhardt and Meyer-Bahlburg1981). These findings show that fetal testosterone has organizational effects on the infants’ brain, which can lead to sex-typical differences in caregiving behavior later in life. In addition, fetal testosterone also seems to alter sensitivity to testosterone surges later in life. A study by van Honk et al. (Reference van Honk, Schutter, Bos, Kruijt, Lentjes and Baron-Cohen2011) showed that administration of testosterone reduced cognitive empathic behavior in women, and that the effect was dependent on those women with a low ratio of the second and fourth digit of the right hand (2D/4D ratio), which is a marker of fetal exposure to testosterone (Breedlove, Reference Breedlove2010). Thus, increased exposure to fetal testosterone can increase later life sensitivity to testosterone, resulting in subsequent alterations in caregiving behavior. Preliminary data shows that a low 2D/4D and low maternal warmth predicted self-reported anxious attachment later in life (Blanchard, Lyons, & Nelson, Reference Blanchard, Lyons and Nelson2014), whereas self-reported attachment is related to more positive parenting behavior (Jones, Cassidy, & Shaver, Reference Jones, Cassidy and Shaver2015).

Furthermore, a recent finding suggests that the effects of prenatal priming by testosterone interact with sensitivity for OXT (Weisman, Pelphrey, et al., Reference Weisman, Pelphrey, Leckman, Feldman, Lu, Chong and Ebstein2015). The study shows that cognitive empathy was predicted by the 2D/4D ratio selectively in carriers of the GG allele of the OXTR gene rs53576, whereas in carriers of the A allele, no such effect for this proxy of fetal testosterone was found. Whether exposure to OXT and estradiol in the womb during pregnancy can affect later life sensitivity to these endocrine factors has not yet been investigated, but a few studies have demonstrated that OXT levels in mothers during pregnancy predict maternal behavior postpartum, such as increased maternal bonding (Feldman, Weller, Zagoory-Sharon, & Levine, Reference Feldman, Weller, Zagoory-Sharon and Levine2007) and also a decreased risk of postpartum depression (Skrundz, Bolten, Nast, Hellhammer, & Meinlschmidt, Reference Skrundz, Bolten, Nast, Hellhammer and Meinlschmidt2011). These studies should be interpreted with caution, because the measurement procedure for OXT in these studies have been proven unreliable (McCullough, Churchland, & Mendez, Reference McCullough, Churchland and Mendez2013); therefore, replications with more sensitive OXT measures are needed. Nonetheless, because both OXT and testosterone can easily cross the placenta, the above findings give ground to speculate that testosterone and OXT prenatally affect both the child's and the mother's neuroendocrine system underlying social bonding behaviors. Although the relation between prenatal estradiol and caregiving behavior has not yet been studied, a recent study does show that epigenetic markers for postpartum depression are related to changes in estradiol during pregnancy (Osborne et al., Reference Osborne, Clive, Kimmel, Gispen, Guintivano, Brown and Braier2015). This indicates that epigenetic variation of these markers might contribute in the etiology of postpartum depression by mediating endocrine sensitivity.

The programming effects of prenatal stress on the HPA axis in children and adults are relatively well studied (Bowers & Yehuda, Reference Bowers and Yehuda2016). Extensive animal literature shows downregulating effects of prenatal stress on epigenetic MR and GR expression, which is most pronounced at the level of the hippocampus (Zhang, Labonte, Wen, Turecki, & Meaney, Reference Zhang, Labonte, Wen, Turecki and Meaney2013). This literature also shows increased HPA-stress responses, most likely caused by reduced downregulating effects of the hippocampus on the HPA axis (Lupien, McEwen, Gunnar, & Heim, Reference Lupien, McEwen, Gunnar and Heim2009). Studies in humans suggest similar increased HPA axis activation in infants after prenatal stress, and preliminary data indicate that maternal anxiety and CORT levels affect epigenetic expression of the infants’ GR, pointing at comparable mechanisms in humans to those described in rodents (Hompes et al., Reference Hompes, Izzi, Gellens, Morreels, Fieuws, Pexsters and Freson2013). In addition, CORT levels measured from amniotic fluid in mothers predicted low infant birth weight, which subsequently predicted infant distress at 3 months (Baibazarova et al., Reference Baibazarova, van de Beek, Cohen-Kettenis, Buitelaar, Shelton and van Goozen2013). Finally, animal work shows that some of the effects of prenatal stress can be countered by postnatal infusion of OXT (Lee, Brady, Shapiro, Dorsa, & Koenig, Reference Lee, Brady, Shapiro, Dorsa and Koenig2007), indicating a possible therapeutic role of OXT administration for altered function of the HPA axis after prenatal stress. Taken together, the above findings show that prenatal exposure to testosterone and CORT results in long-term endocrine alterations, which could change the developmental trajectory in which child–parent interactions shape the child's caregiving behavior later in life.

Parturition and early postpartum

Perhaps the least studied period in human development during which endocrine factors can mark human development is the timeframe of parturition and the early postpartum period. That this period is understudied is surprising considering the strong cultural and historical variation in caregiving behaviors associated with this period. For example, medical advances in Western societies have resulted in the widely applied practices of epidural anesthesia, synthetic OXT injections to facilitate contractions, caesarean section (CS), immediate mother–infant separation upon birth, bottle feeding, and reduced body contact due to separate sleeping. Some of these routines became common practice as recent as in the last few decades. Whereas the 1970 CS rate in the United States was only 5% (Sewell, Reference Sewell1993), today over 30% of deliveries in the United States are performed by CS, with numbers up to 48% for other Western countries (Olza-Fernández et al., Reference Olza-Fernández, Gabriel, Gil-Sanchez, Garcia-Segura and Arevalo2014). This strong rise in CS is concerning, because CS has been related to adverse health outcomes in children (Blustein & Liu, Reference Blustein and Liu2015). Preliminary data also suggest a link between CS and autism, as well as mood and anxiety disorders (Olza-Fernández et al., Reference Olza-Fernández, Gabriel, Gil-Sanchez, Garcia-Segura and Arevalo2014). Although such epidemiological data suffers from methodological difficulties, a critical difference between CS and normal delivery is the absence of a hormonal cascade resulting in high levels of OXT, AVP, CORT, and catecholamines such a noradrenaline in both mother and infant during CS. This hormonal cascade during natural birth has been brought forth as a critical period in the etiology of autism (Ben-Ari, Reference Ben-Ari2015). Studies in mammals have further shown that administration of OXT and AVP in infants directly after birth has long-lasting effects on social–emotional behavior, including parenting behavior (Hammock, Reference Hammock2015; Miller & Caldwell, Reference Miller and Caldwell2015), and that in some species OXT is essential for the formation of the mother–infant bond (Kendrick, Reference Kendrick2000). In humans, OXT levels in the infant directly after birth predicted infant social engagement at 3 and 6 months of age (Clark et al., Reference Clark, John, Pasca, Hyde, Hornbeak, Abramova and Penn2013). Postnatal CORT levels in mothers are related to maternal sensitivity and affection (Fleming, Ruble, et al., Reference Fleming, Ruble, Krieger and Wong1997; Stallings et al., Reference Stallings, Fleming, Corter, Worthman and Steiner2001) and to better olfactory recognition of their infants (Fleming, Steiner, et al., Reference Fleming, Steiner and Corter1997). Infants delivered by CS, but exposed to contractions before CS, show better olfactory recognition in the first days after birth compared to infants not exposed to contraction (Varendi, Porter, & Winberg, Reference Varendi, Porter and Winberg2002), suggesting that the endocrine cascade during childbearing facilitates social interaction and olfactory learning in both mother and infant. The absence of this hormonal cascade could also account for the reduced neural responses in mothers to their infants’ crying after CS compared to normal delivery (Swain et al., Reference Swain, Tasgin, Mayes, Feldman, Constable and Leckman2008).

Besides CS, another recent practice with possible longer lasting endocrine effects is the common use of administration of synthetic OXT (pitocin) to facilitate delivery. Use of pitocin has been related to an increased risk for autism (Weisman, Agerbo, et al., Reference Weisman, Agerbo, Carter, Harris, Uldbjerg, Henriksen and Dalsgaard2015) and has been implicated in problems with initiation of breastfeeding, with subsequent effects on OXT function (Odent, Reference Odent2013; Olza-Fernández et al., Reference Olza-Fernández, Gabriel, Gil-Sanchez, Garcia-Segura and Arevalo2014). Although conclusive studies are not available, similar problems with breastfeeding have been related to the use of epidural anesthesia (French, Cong, & Chung, Reference French, Cong and Chung2016). Thus, pitocin administration, CS (but also reduced body contact between mother and infant), and bottle-feeding compared to breastfeeding strongly reduce maternal and infant exposure to OXT after birth, which can interfere with the synchronous neurobiological mechanisms in mother and infant and can thereby hamper the establishment of mother–infant attachment (Feldman, Reference Feldman2015). For example, longer duration of breastfeeding is positively associated with sensitive responsiveness and attachment security (Tharner et al., Reference Tharner, Luijk, Raat, IJzendoorn, Bakermans-Kranenburg, Moll and Tiemeier2012), effects that might be related to upregulation of OXT function. In addition, mother–infant separation in the first 2 hr after birth compared to skin to skin contact and suckling resulted in reduced maternal sensitivity, infant self-regulation, and mother–infant interaction 1 year later (Bystrova et al., Reference Bystrova, Ivanova, Edhborg, Matthiesen, Ransjö-Arvidson, Mukhamedrakhimov and Widström2009). Findings such as these suggest a considerable impact of the early postpartum period on development and stresses the need for more research on this sensitive period.

Early life environment

One of the strongest environmental factors during early development that affect neuroendocrine sensitivity is the experience of stressful situations. Although stress can be caused by various sources, most detrimental is stress caused, or accompanied by, inadequate caregiving behavior. Because the primary caregivers are considered a pivotal source of social support and stress buffering (Hostinar & Gunnar, Reference Hostinar and Gunnar2013), parenting practices such as physical and emotional abuse and neglect not only induce stress but also deprive the child of a valuable protective factor against stress: the safety and security provided by caregivers. Although it is difficult to experimentally study in humans, the effects of such harsh caregiving practices on the neuroendocrine outcomes of child development have been substantially studied, mostly due to the strong effects this behavior has on social–emotional behavior and the risk it brings for psychopathology (e.g., Andersen & Teicher, Reference Andersen and Teicher2009; Cicchetti & Toth, Reference Cicchetti and Toth2009; Heim & Nemeroff, Reference Heim and Nemeroff2001).

Puberty, adolescence, and young adulthood

Of final importance is the period between early childhood and adulthood, with early adulthood being the phase when most individuals make the transition to parenthood. With respect to endocrine variability, during puberty and adolescence hormonal factors have strong organizational effects on the human brain (Crone & Dahl, Reference Crone and Dahl2012; Peper, Rol, Crone, & van Hook, Reference Peper, Pol, Crone and van Honk2011). Work in other animals has brought forth adolescence as an additional sensitive period in which behavioral profiles are shaped, because environmental circumstances during this period can alter neuroendocrine functions and later life social–emotional behavior, such as adult social bonding (Sachser, Kaiser, & Hennessy, Reference Sachser, Kaiser and Hennessy2013). In humans, where adolescence is considered critical for the development of cognitive control, emotion regulation, and social competence, this developmental phase is characterized by changes in endocrine function, including vast increases in levels of steroid hormones such as estradiol and testosterone (Crone & Dahl, Reference Crone and Dahl2012). Neuroimaging studies have attempted to shed light on the functional consequences of these endocrine changes for neural development and for social–emotional behavior. Data of these studies show that testosterone levels in both adolescent boys and girls are positively related to neural responses in reward processing (de Macks et al., Reference de Macks, Moor, Overgaauw, Güroğlu, Dahl and Crone2011) and that adolescents’ testosterone levels are also positively associated with increased risk taking (Peper, Koolschijn, & Crone, Reference Peper, Koolschijn and Crone2013). In boys this association between testosterone and risk taking was potentiated by smaller OFC volumes, which might explain impaired emotion regulation. Testosterone levels in both adolescent boys and girls were also related to lower quality of the structural connections between subcortical regions and the frontal and temporal cortices (Peper, de Reus, van den Heuvel, & Schutter, Reference Peper, de Reus, van den Heuvel and Schutter2015). The quality of these connections was further inversely related to increased expressive forms of aggression, indicating the relevance of increases of steroid hormones, and especially testosterone, for social–emotional behavior in this developmental phase. In addition, in young adults, quality of the structural connectivity between the frontal cortex and the striatum predicted lower levels of impulse control and higher levels of testosterone (Peper, Mandl, et al., Reference Peper, Mandl, Braams, de Water, Heijboer, Koolschijn and Crone2013). Taken together, these studies indicate that higher levels of especially testosterone during adolescence and early adulthood might predispose to more reward-seeking behavioral profiles characterized by less impulse control and higher levels of aggression, which could have negative effects on caregiving behavior. However, the direct relation between endocrine variation in adolescence and early adulthood and later life caregiving is currently unexplored.

Research on animals has also demonstrated that social environmental factors interact with endocrine function during adolescence. For example, the number of competitive interactions in male rodents leads to changes in reproduction patterns under the influence of altered testosterone levels and function of the HPA axis (Sachser, Reference Sachser, Carter, Ahnert, Grossmann, Hrdy, Lamb, Porges and Sachser2005; Sachser et al., Reference Sachser, Kaiser and Hennessy2013). In humans, a comparable mechanism has been proposed, in that environmental factors that indicate instability interact with the endocrine stress system to accelerate the onset of puberty in girls (Ellis, Reference Ellis2004). As such, in a stressful environment, evolutionary pressure might stimulate higher fecundity at the cost of less investment in offspring (Trivers, Reference Trivers1972). Early life stress in humans has been shown to result in more adultlike hormonal patterns earlier in life compared to controls, and this effect was specific for girls (Ruttle, Shirtcliff, Armstrong, Klein, & Essex, Reference Ruttle, Shirtcliff, Armstrong, Klein and Essex2015). Effects of early life stress on the adolescent endocrine system were further demonstrated in a study that found flattened CORT levels, indicative of reduced sensitivity to CORT, in girls who were exposed to inconsistent discipline by their mother (Martin, Kim, & Fisher, Reference Martin, Kim and Fisher2016). Thus, during adolescence, characterized by the organizational effects of steroid hormones on the neuroendocrine systems involved in caregiving behavior, environmental factors can alter the developmental trajectories, which makes this developmental phase a fruitful direction for further research.