The Neurobiology of Safety and Threat Learning in Infancy

Abstract

What an animal needs to learn to survive is altered dramatically as they change from dependence on the parent for protection to independence and reliance on self-defense. This transition occurs in most altricial animals, but our understanding of the behavioral neurobiology has mostly relied on the infant rat. The transformation from dependence to independence occurs over three weeks in pups and is accompanied by complex changes in responses to both natural and learned threats and the supporting neural circuitry. Overall, in early life, the threat system is quiescent and learning is biased towards acquiring attachment related behaviors to support attachment to the caregiver and proximity seeking. Caregiver-associated cues learned in infancy have the ability to provide a sense of safety throughout lifetime. This attachment/safety system is activated by learning involving presumably pleasurable stimuli (food, warmth) but also painful stimuli (tailpinch, moderate shock). At about the midway point to independence, pups begin to have access to the adult-like amygdala-dependent threat system and amygdala-dependent responses to natural dangers such as predator odors. However, pups have the ability to switch between the infant and adult-like system, which is controlled by maternal presence and modification of stress hormones. Specifically, if the pup is alone, it will learn fear but if with the mother it will learn attachment (10–15 days of age). As pups begin to approach weaning, pups lose access to the attachment system and rely only on the amygdala-dependent threat system. However, pups learning system is complex and exhibits flexibility that enables the mother to override the control of the attachment circuit, since newborn pups may acquire threat responses from the mother expressing fear in their presence. Together, these data suggest that the development of pups’ threat learning system is not only dependent upon maturation of the amygdala, but it is also exquisitely controlled by the environment. Most notably the mother can switch pup learning between attachment to threat learning in a moment’s notice. This enables the mother to navigate pup’s learning about the world and what is threatening and what is safe.

Graphical Abstract

Introduction

As we assess the development of learning it is critical to understand that the young brain is not an immature version of the adult brain. Rather it is specifically constructed to ensure learning is age-specific to accommodate the changing ecological niche typically associated with a temporally constrained learning period referred to as a sensitive period. A child learns about the world from the caregiver, including the fundamentally important ability to navigate the complex world by safety and threat signals. However, what is safe and what is threatening changes during development, suggesting the supporting learning circuitry must also change. For example, altricial infants such as humans and rodents might perceive separation from the caregiver as a danger signal and proximity to the caregiver as safety. This would require a learning system where learning the characteristics of the caregiver and expressing approach responses to the caregiver would be beneficial, while removal of caregiver cues might be defined as a threat. With maturation, and the ability of the infant to crawl away from the caregiver for brief time periods might require adding a more sophisticated learning network and the emergence of an ability to identify proximal dangers but evoke a behavioral response of return to the nest and engagement of the parent’s defense system. This would require retention of the caregiver as a safety signal. Finally, as the preparation for independence progresses, one might expect a system where dependence on the parent for protection would wane combined with the emergence of behavioral expression that ensures a flexible self-defense system to navigate a complex environment. Indeed, emerging evidence indicates that the learning neural network of the young brain is not an immature version of the adult brain: it morphs to ensure learning is age-specific to accommodate the changing ecological niche as age-specific demands change during the transition to independence.

This major transition from complete dependence on the caregiver to complete independence occurs in most altricial species, including humans, nonhuman primates and rodents. This transition occurs in three short weeks in rat pups and is accompanied by complex changes in responses to both natural and learned threats, and safety, with corresponding changes in the supporting neural circuitry. Development continues after weaning and considerable rapid learning must occur to further prepare for the emergence of adulthood and reproduction in a little more than one month later. These transition and periods of rapid learning all occur as the brain continues to develop and be shaped by environmental experiences.

One key concept for understanding infant learning is that learning in early life is biased towards attachment learning. While attachment has been considered innate, evidence now indicates a high degree of learning that activates a biologically predisposed attachment circuit used to initiate and maintain this bond. The importance of learning within attachment in humans is indicated by the rapid attachment between young babies and nonbiological parents of either gender. This is further supported by situations where atypical caregiver behaviors (i.e. neglect, abuse by the caregiver) within biological and nonbiological parent-infant dyads emerge, yet attachment occurs in the offspring, suggesting a learning system with unique features that support learning bond formation even with trauma. Here we focus on threat and safety learning within an age-specific behavioral neurobiology from the infant’s perspective, which must begin with the biological demands of infancy – to attach and remain attached to the caregiver until independent self-defense emerges.

Historical perspective: Early life safety learning within attachment and later emerging threat learning

An attachment neural circuitry for humans was first proposed in Bowlby’s Attachment Theory, which relied heavily on integration of research on early learning about the caregiver in avian imprinting (Ainsworth, 1969; Bowlby, 1977, 1978, 1984). In imprinting, hatchling birds have a temporally limited access to learning about the caregiver (a sensitive period for imprinting). During the sensitive period for imprinting, the chicks will learn to follow a moving object encountered in its immediate environment of the nest, which is typically the caregiver. However, the caregiver can be replaced by a human, box, or a stuffed bird and imprinting still occurs, which indicates the importance of learning and the wide range of stimuli that can be identified and learned as the caregiver. Imprinting also provides an illustration of how this early learning produces unique learned responses, which includes proximity seeking. The unique learning and unique expression of proximity seeking ends as the sensitive period for imprinting ends (Bolhuis & Honey, 1998). Overall, this work suggested that learning to attach to the caregiver is learned, but once learned, it engaged a biologically predetermined learned action of following the caregiver. Considerable behavioral neurobiological research has documented this imprinting learning circuit (Bateson, 2015; Solomonia & McCabe, 2015; Bolhius et al., 2015; Bolhuis & Honey, 1998; Gottlieb, 1965; Hoffman & DePaulo, 1977). Importantly, what happens to the chick during this imprinting process does not follow the traditional learning rules: providing these chicks with electric shock during imprinting still results in imprinting supporting a strong attachment (Rajecki et al., 1978; Kovach & Hess, 1963). As will be reviewed below, this phenomenon is a characteristic of attachment learning seen in other species, including rodents, nonhuman primates and abused children (see: Fig. 1 for a rodent model of a dominance of attachment learning and an emergence of threat learning). This paradoxical feature of attachment learning was noted by Bowlby, where poor quality of care, including an abusive caregiver, still supports attachment formation (Walden & Ogan, 1988).

Fig. 1. Transition from the dominance of attachment learning to adult-like threat learning.

Until around postnatal day (PN) 10 the threat system is quiescent and learning is biased towards learning attachment (sensitive period). Around PN 10, adult-like fear learning emerges. However, during a transient period between PN10 and PN 15 referred to as the Transitional Sensitive Period, pups can switch between attachment learning and threat learning based on maternal presence: if the mother is present odor-pain produces attachment learning and if pups are alone, amygdala-dependent threat is learned. At PN16 and older, pups no longer have access to the attachment learning system, even if the mother is present.

Neurobiology of attachment learning: odor learning and safety

Infants must learn to identify and remember the caregiver by learning about the caregiver’s sights, voice, touch and odor. Previous work provides some insight into the mechanisms that support this learning in humans. For example, in humans and rodents, the mother’s natural odors and food odors experienced in utero are preferred after birth (Mennella 2014; Cooke & Fildes, 2011; Abate et al., 2008; Mennella et al., 2001; Varendi et al., 1996; Pedersen & Blass, 1982), although lack of subsequent exposure to these odors decrease this preference, and novel postnatal odors (natural, perfume) of the mother and other nonrelated caregivers are quickly learned (Al Ain et al., 2016; Schaal, 2015; Mennella et al., 2011; Schaal et al., 2009; Mennella & Beauchamp, 1999). Indeed, in newly born human infants a few minutes of simply pairing a novel odor with tactile stimulation on the first day of life causes that odor to be preferred but also elicits head turns towards that odor and mouthing (Sullivan et al., 1991). The mother’s odor and other characteristics of the mother (face, touch, and voice) can take on qualities that help regulate the infant’s behavior and physiology (Hostinar et al., 2014). The caregiver also acquires the ability to signal safety for the infant, or serve as a safe haven, which is well-documented by human and animal studies demonstrating that under threat infants rapidly return to the caregiver (Ainsworth, 1969; Raineki et al., 2010d; Sanchez et al., 2015), whose presence, in turn, decreases the infant’s stress hormone levels (Hostinar et al., 2014; Gunnar et al., 2015) and the amygdala activity (Gee, et al. 2014).

A similar phenomenon exists in infant rodents, which permits the exploration of neural mechanisms for this learning and provides insights for these processes in humans. In infant rat pups, pairing a novel odor with a reward quickly produces a learned preference for that odor, but this odor also becomes a characteristic of the mother and takes on the ability to regulate the infant’s physiology and behavior similarly to the natural maternal odor (Hostinar et al., 2014; Gunnar et al., 2015; Sanchez et al., 2015; Sullivan and Perry, 2015; Hennessy et al., 2009; Hennessy et al., 2015}. This learning, while similar to classical conditioning in many ways, seems to have some unique features, including the brain areas involved and a unique dependence upon the neurotransmitter norepinephrine (NE) (Sullivan et al., 1991, 1992, 2010; Yuan et al., 2000; Bordner and Spear, 2006). In an incredibly simplistic circuit, maternal stimulation of pups, including a gentle touch or stepping on pups, or an experimentally applied squirt of milk, a tailpinch or an electric shock, all have the common feature of activating the locus coeruleus (LC) (Moriceau et al. 2009; Nakamura et al. 1987; Sullivan et al., 2000b). The LC is a major source of NE within the brain and the sole source of NE to the olfactory bulb (see: Fig. 2).

Fig. 2. Rapid Attachment Learning in Early Infancy.

Newborn rat pups rapidly learn a new maternal odor. This learning is mostly activated by presumably pleasurable stimuli, such as maternal licking (1) or gentle tactile stimulation with a paintbrush (2) but also by painful stimuli, such as tailpinch (3) or moderate electric shock (4). Attachment learning in pups is supported by a simplistic neural circuit that requires the pairing of the novel (maternal) odor with increased locus coeruleus (LC) norepinephrine (NE) release into the olfactory bulb (OB). Attachment learning-induced synaptic plasticity has been identified in the OB and the piriform cortex (PIR). The functional emergence of recurrent collaterals in the LC around postnatal day (PND) 10 allows stimulation of α₂ presynaptic receptors, which downregulates the LC noradrenergic release and results in insufficient NE levels in the OB to support the synaptic plasticity necessary for learning about maternal odor.

Importantly, the young infant LC is different from the older infant and adult LC: stimulation of the infant LC releases large amounts of NE, and the LC fails to habituate to repeated stimulation and shows extensive electrical coupling to synchronize LC activity (Christie, 1997; Christie & Jelinek, 1993; Christie, Williams, & North, 1989; Nakamura et al., 1987; Williams & Marshall, 1987; Winzer-Serhan et al., 1996). This unique characteristic of the infant’s LC results in copious amounts of NE release that continue each time the pup is repeatedly stimulated by the mother. The olfactory bulb target of the LC, which is critical for learning the maternal odor, also shows responses to NE that facilitate neural plasticity (Pandipati & Schoppa, 2012; Wilson & Leon, 1988; Yuan et al., 2000, 2002; Okutani et al., 1999). The LC appears to take on adult-like features at 10 days old, just as the sensitive period for attachment ends (Nakamura et al., 1987; Winzer-Serhan et al., 1996). This olfactory bulb NE has been shown to be critical in memory formation during a sensitive period in rat pups {Pandipati & Schoppa, 2012; Yuan et al., 2014; Sullivan et al., 2000b); Todrank et al., 2011; Sullivan et al., 1992). Although we are unsure if the mechanism supporting this learning in human infants is the same as in rodents, there is some suggestion that NE’s role in attachment is a phylogenetically conserved system. NE has been shown to play a critical role in social bond formation throughout development in numerous species (Nelson & Panksepp, 1998; Numan & Young, 2016; Kaba & Huang, 2005; Corona & Levy, 2015; Calamandrei et al., 1992), as well as in modulation of learning throughout the lifespan (McGaugh, 2015; Maity et al., 2015; Shea et al., 2008; Gold, 2015). While children have high levels of NE during the first two years of life, the role of NE or any other hormone or neurotransmitter in early life attachment in humans is yet to be demonstrated, as it has been established by the rodent literature (Slotkin & Langercrantz, 1986).

Another feature of young pup learning that is unique is that the odor memory is limited to a few key structures within the brain: primarily the olfactory bulb and the piriform (olfactory) cortex (Modaresi et al., 2016; Raineki et al., 2009; Roth & Sullivan, 2006; Sullivan et al., 1990; Yuan et al., 2002). These regions are also important in the adult brain and play a role in assigning the hedonic value to a learned odor (Haberly, 2001; Schwob & Price, 1984; Swanson & Petrovich, 1998; Wilson & Stevenson, 2003). These brain sites appear to play a similar role in infancy (Moriceau & Sullivan, 2006b; Moriceau et al., 2006; Roth & Sullivan, 2005b). It is important to note that both natural maternal odor and a learned artificial maternal odor generate the same responses from the olfactory bulb (Raineki et al., 2010d; Roth & Sullivan, 2005b).

Sensitive period neurobiology of pain related/abusive attachment: Why attachment learning occurs with pain

During the sensitive period for attachment, even painful interactions with the mother or pain delivered by the experimenter still support odor preference learning {Roth et al., 2005; Raineki et al., 2015; Raineki et al., 2010b; Sullivan et al., 2000). The mechanism underlying attachment learning with pain is relatively simple: the amygdala, the brain region supporting fear learning (Maren & Fanselow, 1996; Phelps & LeDoux, 2005, Debiec & LeDoux, 2009; Janak & Tye, 2015) is not mature enough to reinforce this learning until pups are 10 days old. Thus, when pups experience odor-shock or odor-abusive mother pairing, the locus coeruleus NE attachment learning system is activated but it does not engage the amygdala. Specifically, pups feel pain (Barr, 1995; Collier & Bolles, 1980; Emerich et al., 1985; Stehouwer & Campbell, 1978), and this information arrives in the amygdala, yet the amygdala fails to show learning induced plasticity until pups reach 10 days of age (Barr et al., 2009; Moriceau et al., 2006; Sullivan et al., 2000). Abuse-related attachment formation is phylogenetically conserved and occurs in a number of species, including chicks that form attachments after being shocked during imprinting (Hess, 1962; Rajecki et al., 1978; Salzen, 1970), dogs (Stanley, 1962) and monkeys raised with a wire surrogate that inflict pain (Harlow & Harlow, 1965). More recently, this abuse-related attachment has been modeled in nonhuman primates and again shows that infants learn and retain strong preference for the abusive caregiver (Maestripieri et al., 1999; O’Connor & Cameron, 2006; Sanchez et al. 2001; Suomi, 2003). This broad species representation of abuse-related attachment was noted by Bowlby in his Attachment Theory (Bowlby, 1977, 1984; Harlow & Harlow, 1965; Harmon et al. 1984). Findings from human studies in adults suggest that abuser-related bonding depends on a number of psychosocial factors including a history of abuse in childhood, type and duration of abuse, gender, relationship with the perpetrator, social isolation, as well as accompanying cognitive patterns and distortions (Karantzas et al., 2016; Canton-Cortes et al., 2015; McCloskey, 2013; Nicholas, 2013; Julich, 2005), and thus likely involves a number of neural mechanisms outside of the amygdala threat processing circuitry. While it is unclear whether humans’ and other species’ failure to learn aversions to their abusive caregiver is also caused by lack of amygdala functioning, amygdala maturation appears delayed in many species, including humans (Gee et al., 2013; Goksan et al., 2015).

Pups during the sensitive period (<PN10) also lack responses to natural stimuli that elicit defense responses in adults, such as predator odors {Moriceau et al., 2004; Takahashi et al., 1991; Takahashi & Rubin, 1993; Wiedenmayer et al., 2003). Yet, it is important to note that adult-like threat responses are delayed in many species until they are ecologically relevant when the animal emerges from the nest or borrow (Pongrácz & Altbäcker, 2000; Putman et al., 2015). However, this does not mean that the very young animal has no defensive responses, rather it means that the adult system has not emerged. Indeed, responses that increase survival within a threatening context do exist in very young animals across species, but they typically involve age-specific stimulus-responses uniquely adapted to the early life niche of the nest or borrow. For example, some birds will be quieted (become still) by vibrations similar to that which would occur if a predator landed near the nest and rat pups will decrease activity in response to intrusion of an unfamiliar adult. These quieting responses typically involve neural circuitry distinct from adult defensive behaviors.

Maternal Control: Turning on and off the sensitive period for attachment

While the sensitive period for attachment was initially thought to end at PN10, additional research indicated that the mother is capable of rapidly turning on and off pups sensitive period. Indeed, pups’ attachment sensitive period was showing considerable flexibility, and it’s remarkable control by maternal presence and the mother’s emotional state provided a rapid method for further defining what pups learn. This is counter to our traditional view of sensitive period as strict temporally defined developmental windows of learning. While the environment is capable of changing the duration of the sensitive period, (Raineki et al., 2010b; Callaghan et al., 2013), there was no previous demonstration of any event or social partner rapidly switching on and off any sensitive period during development. Here we review two distinct situations where the mother can control switching on/off the sensitive period for attachment: one involves the mother turning on the sensitive period for attachment in older pups and suppressing threat learning (illustrated in Fig. 3) and the other involves the mother rapidly turning on the threat learning in very young sensitive period pups and suppressing the attachment system (illustrated in Fig. 5).

Fig. 3. Transitional Sensitive Period for Attachment Learning (PN 10–15).

As pups mature (starting around PN 10) and begin exploring the area surrounding the nest, threat learning emerges. Away from the mother, when exposed to an innately noxious event, such as mild electric shock, pup’s threat processing system is activated (1). A response to threat involves the amygdala (AMY) and the hypothalamus-pituitary-adrenal gland (HPA) axis activation resulting in the release of the stress hormone corticosterone (CORT), which modulates synaptic plasticity and supports fear/aversion learning in the AMY. However, until around PN 15, maternal presence suppresses threat-induced HPA axis activation by blocking NE action in the paraventricular nucleus of the hypothalamus (PVN) and prevents threat/aversion learning enabling attachment/preference learning (2). Although, maternal stress buffering effects continue, from PN 16 on, maternal presence loses the ability to completely abolish amygdala-dependent fear learning.

Fig. 5. Social Transmission of Fear from Mother to Infant.

A frightened mother activates in pups a number of areas involved in processing threat and stress, including the amygdala (AMY) and hypothalamus (HYP), and leads to the rise of CORT levels (1). Pups with a mother expressing fear to the previously conditioned odor, acquire fear responses to this odor, a fear transmission that involves alarm chemosignaling processing areas (including Grueneberg ganglion, GG) (2).

Here we present an example of the mother turning on the sensitive period for attachment and suppresses threat learning in older pups (see Fig. 3). Specifically, our experiments indicated that the mother was opening and closing this extended (transitional) sensitive period by manipulating pup’s stress hormone, corticosterone (CORT) levels (Moriceau & Sullivan, 2006; Shionoya et al., 2007). The clue to the role of CORT in controlling sensitive period termination came from a diverse literature. The first clue was that increasing systemic CORT levels in very young pups induced precocious expression of fear to predator odor, which is typically delayed until 10 days old (Takahashi & Rubin 1993). The second clue was provided by the literature documenting that pups younger than 10 days old have a stress hyporesponsive period (SHRP), which means that the hypothalamic-pituitary-adrenal (HPA) axis does not respond to most stressors (including shock, tailpinch) (Grino et al., 1994; Henning, 1978; Levine, 1962; Rosenfeld et al. 1992; Walker et al., 1986; Dallman, 2000; Stanton & Levine, 1985). The final clue came from the social buffering literature. Social buffering is a phenomenon where a significant social figure blocks activation of the HPA axis, a process that occurs across species and development. As pups leave the developmental period of the SHRP, shock and other painful stimuli increase CORT, social buffering of stress and fear emerges (Levine et al., 1985; Ditzen & Heinrichs, 2014; Gee et al., 2014; Hennessy et al., 2009; Hennessy et al., 2015; Hostinar & Gunnar, 2013; Kikusui et al., 2006; Moriceau & Sullivan, 2006; Nachmias et al., 1996; Sanchez, 2006; Sanchez et al., 2015; Shionoya et al., 2007; Stanton & Levine, 1990; Takahashi et al., 2013). One of the most powerful effects of social buffering is the mother’s social buffering of offspring (Gee et al., 2014; Hennessy et al., 2009; Hostinar et al., 2014; Kikusui et al., 2006; Levine, 2001; McCormack et al., 2009; Sanchez, 2006; Sanchez et al., 2015; Stanton & Levine, 1990; Suchecki et al., 1995; Sullivan & Perry, 2015; van Oers et al., 1998). In social buffering, a significant social figure typically attenuates stress hormone release throughout development, although in early life the mother can completely block release of CORT (Shionoya et al., 2007; Suchecki et al., 1993; Stanton et al., 1987; Moriceau et al., 2006) (Fig. 3). Animal models suggest the mother blocks the hypothalamic-pituitary-adrenal (HPA) axis at the level of the hypothalamic paraventricular nucleus (PVN) NE (Moriceau et al., 2006; Shionoya et al., 2007). Specifically, previous work from our group showed that maternal presence during odor-shock conditioning on PND 12 attenuated both infant’s PVN neural activity and PVN NE levels, and intra-PVN infusions of the α₁-adrenergic receptor antagonist prazosin blocked fear conditioning with maternal absence, while intra-PVN α₁-adrenergic receptor agonist phenylephrine permitted fear conditioning with maternal presence (Shionoya et al., 2007).

CORT control of the sensitive period learning produces profound effects on pup learning about fear but can also turn off the attachment learning circuit. Specifically, increasing CORT during the sensitive period, either naturally via mother’s milk (stressed mother), early separation from the mother or by rearing pups with an abusive caregiver, or pharmacologically (systemic injections or by intra-amygdala microinfusions), can prematurely end sensitive period learning and enable pups to learn amygdala-dependent fear. This enables the amygdala to undergo learning-related plasticity and allows pups to learn fear and avoidance (Moriceau et al., 2004; Moriceau et al., 2009; Moriceau & Sullivan, 2004, 2006; Callaghan & Richardson, 2011, 2012; Cowan et al., 2013). This indicates the young amygdala receives the shock and odor information and is capable of supporting fear learning, but it requires CORT to engage the mechanisms of learning plasticity (Thompson et al., 2008).

Conversely, a reduction of either systemic CORT (injection or pharmacological intervention at the level of the hypothalamus) or blockade of CORT limited to the amygdala disrupts learned fear/avoidance but also reinstates the sensitive period learning to induce a learned preference (Sullivan & Holman, 2000; Moriceau et al., 2006; Moriceau & Sullivan, 2006; Shionoya et al., 2007). The ability to turn on and off fear learning mirrors the ability of CORT to turn on and off fear expression to predator odor in infant rats, a threat behavior that also emerges at PN10 (Wiedenmayer et al., 2003; Moriceau et al., 2004; Takahashi & Rubin, 1993; Takahashi et al., 1991). In terms of fear and safety learning, between PN10 and PN15, the “transitional sensitive period”, maternal presence can act as a switch to turn off the fear circuitry and turn on the attachment circuitry during odor-shock conditioning simply by blocking CORT (Moriceau & Sullivan, 2006). The ability of CORT to switch between pup’s fear and attachment safety learning ends at PN15 (Upton & Sullivan, 2010).

Together, these data indicate that learning mechanisms in early infancy favor attachment/preference learning and threat/aversive learning in newborns remains quiescent (Fig. 1). This suppression of threat system, notable during the transitional sensitive period is controlled by a caregiver whose presence, at default, buffers infant’s stress and threat responses and prevents fear learning (Fig. 3).

Development of associative threat learning

As pups transition towards weaning, their trips outside the nest become longer in distance and duration. During these first days of out-of-the-nest explorations, pups are exposed to expeditiously increasing number and variety of environmental cues which poses for them a demand to process their new experiences, including encounters with novel environmental threats. Rapid neural development and the emergence of new sensory, motor and cognitive functions advance pups’ ability to learn. The development of associative learning depends on the emergence and maturation of sensory systems, output sites directly controlling threat responses, and the connections of both with the threat system, including the main player in threat processing, the amygdala (Deal et al., 2016; Debiec & LeDoux, 2009; Hunt & Campbell, 1997; Rudy, 1993). Previous studies in rodents established that associative learning emerges in a sensory-specific sequence and that an emergence of a sensory modality precedes the occurrence of learning which involves this modality (Richardson et al., 2000; Hunt & Campbell, 1997; Moye & Rudy; 1985; Hyson & Rudy, 1984; Alberts, 1984). From birth on, pups have functional olfactory and tactile modalities which during the first postnatal days support mainly attachment learning (Fig. 2), and, starting on PN 10, also support threat learning (Fig. 3). By the end of the second postnatal week, pups are able to reflexively respond to acoustic stimuli (Hyson & Rudy, 1984). Auditory associative learning begins to emerge at PN 14 (Hyson & Rudy, 1984) and auditory fear conditioning becomes adult-like by PN 17–18 (Deal et al., 2016; Landers & Sullivan, 2012; Hunt et al., 1994; Rudy, 1993) (Fig. 4). At PN 15, pups are capable of detecting visual stimuli, yet visual fear conditioning starts emerging as early as on PN 16–17 (Hunt & Richardson, 2010; Moye & Rudy, 1985) (Fig. 4).

Fig. 4. Sensory Development and the Emergence of Threat Conditioning.

Olfactory system is functional at birth and olfactory fear conditioning emerges at PN 10. At PN 12 auditory system becomes functional and auditory fear conditioning emerges between PN 14–17. At PN 15, pups can effectively process visual stimuli and visual fear conditioning emerges at PN 16–17. Functional contextual fear conditioning emerges during the periweaning period (PN 18–24).

As weaning approaches at PN 21, pups are able to discriminate and learn a variety of stimuli from distinct modalities: olfactory, tactile, gustatory, auditory and visual (Deal et al., 2016; Hunt et al., 1994; Moye & Rudy, 1985; Rudy & Hyson, 1984; Hyson & Rudy, 1984; Vogt & Rudy, 1984). During the periweaning period, the development of the hippocampus allows the processing and encoding of more complex configurations of stimuli which make the learning context (Debiec et al., 2013; Pearce & Bouton, 2001; Fanselow, 2000), and by PN 24 contextual threat conditioning emerges (Raineki et al., 2010; Rudy, 1993) (Fig. 4). Although, contextual fear may occur even before weaning at PN 18, this learning is short-lived, a phenomenon known as infantile amnesia (Rudy & Morledge, 1994). Recent studies in mice suggest that infantile amnesia may be at least in part due to high levels of neurogenesis in the young pup’s hippocampus (Akers et al., 2012, 2014).

The developmental trajectory of associative threat learning outlined above is sensitive and may be modified by environmental factors. Previous work in rats has shown that early-life atrocities, such as maternal deprivation may accelerate the emergence of threat learning (Cowan et al, 2013). Indeed, a recent human brain imaging study suggests that early maternal deprivation accelerates the development of the threat processing system, involving the amygdala (Gee et al., 2013).

Social transfer of fear from the mother to infant

It is also well documented the caregiver can facilitate fear learning in the offspring learning through a process known as Social referencing (de Rosnay et al., 2006; Murray et al., 2008; Aktar et al., 2013, 2014), which we have recently demonstrated in rodents (Debiec & Sullivan, 2014). In social referencing, a child uses caregiver’s emotional signaling to appraise and respond to uncertain situations, such as when approached by a stranger (Walden & Ogan, 1988; Frith, 2008). Social referencing studies show that infants are especially sensitive to caregiver’s fear and anxiety (de Rosnay et al., 2006; Murray et al., 2008; Aktar et al., 2013, 2014). Consistent with these reports, we found in preweaning rat pups that an exposure to a frightened mother increases activity of a number of brain sites involved in processing threat, stress and pain, including the amygdala and hypothalamic nuclei (Debiec & Sullivan, 2014; Chang & Debiec, 2016). This activation of the threat system in the infant brain was accompanied by the elevation of plasma CORT levels (Debiec & Sullivan, 2014). Thus, although the caregiver’s presence at default buffers stress and fear, the presence of the frightened caregiver induces fear in pups. We also found that newborn pups exposed to their mother expressing fear to the previously conditioned olfactory cue acquire threat responses to this cue from their mother (Debiec & Sullivan, 2014, Chang & Debiec, 2016) (Fig. 5).

This social transmission of fear was present during the sensitive period (as early as on the day of birth) (Chang & Debiec, 2016) and required functional amygdala and intact HPA axis response (Debiec & Sullivan, 2014). Pharmacological inactivation on the pups’s amygdala or blocking of CORT synthesis prevented acquisition of socially transmitted maternal fear (Debiec & Sullivan, 2014). The early social transfer of fear was mediated by the olfactory modality and engaged areas involved in alarm chemosignaling, such as Grueneberg ganglion (GG) (Debiec & Sullivan, 2014; Brechbuhl 2008), although later in life other sensory modalities are also likely to be involved (Kim et al., 2010; Jeon et al., 2010). Maternal fear signaling may thus override natural mechanisms preventing aversive learning in infancy and serves as an unconditioned stimulus reinforcing fear learning (Debiec et al., 2010). This social fear learning occurs throughout development and has been well-documented in children and in adults (Aktar et al., 2013, 2014; Murray et al., 2008; Askew & Field, 2007; Olsson et al., 2007).

Concluding remarks

Learning occurs throughout life but what we learn and how we learn it changes to fit each developmental epoch, typically using transient “sensitive periods” that filter what is learned by enhancing some aspects of learning and inhibiting others. These defined periods of altered learning provide streamlined learning that ensures survival. The traditional view of sensitive period learning has been that it gradually or abruptly provides access and then restriction to the specialized learning circuitry. However, it is now clear that these specialized periods of learning can be turned on and off naturalistically in rat pups by the mother’s presence and emotional affect. This can be mimicked through a direct manipulation of the stress response by pharmacologically altering stress hormone CORT levels or indirectly, through changing maternal emotional states. Understanding how the caregiver’s presence alters the neurobiology of infant’s safety and threat learning may provide insight into mechanisms of adaptive and maladaptive fears, such as in anxiety disorders.

Highlights.

• In newborn rat pups attachment learning is dominant and threat learning is quiescent.

• Around postnatal day 10 classical threat conditioning emerges.

• Maternal presence may prevent threat conditioning in pups until postnatal day 10.

• From birth pups can acquire threat responses from the mother through social learning.