Defining Immediate Effects of Sensitive Periods on Infant Neurobehavioral Function

Abstract

During a sensitive period associated with attachment, the infant brain has unique circuitry that enables the specialized adaptive behaviors required for survival in infancy. This infant brain is not an immature version of the adult brain. Within the attachment relationship, the infant remains close (proximity seeking) to the caregiver for nurturing and survival needs, but the caregiver also provides the immature infant with the physiological regulation interaction needed before self-regulation matures. Here we provide examples from the human and animal literature that illustrate some of these regulatory functions during sensitive periods, recent advances demonstrating the supporting transient neural mechanisms, and how these systems go awry in the absence of species-expected caregiving.

Introduction

Sensitive periods are typically considered epochs during early life when specific experiences have a unique impact on both immediate and long-term neurobehavioral outcomes [1]. Here we focus on how unique neural mechanisms during the sensitive period go beyond long-term programming to define immediate infant behaviors. We know that specific and distinct neural mechanisms are critical for initiating, maintaining and terminating sensitive periods, and that the brain uses many different mechanisms to support myriad physiologically distinct sensitive periods [2–4][5–7]. Based on decades of eloquent research by developmental psychologists and clinicians, we know sensitive periods are also present in humans associated with attachment [8–10]. Careful and elegant studies have shown that our experiences, especially during infancy and adolescence, impact development, forming a basis for an interest in sensitive periods. Less well understood are the mechanisms initiating, maintaining and terminating these sensitive periods, although it is clear that they are supported by myriad specialized brain functions in early life. These specialized brain functions serve a dual purpose, programming of later life cognition and emotion but also impacting the child’s immediate behavioral capabilities and functions [11]. Here we focus on the latter and ask, how is the infant brain different during the sensitive period to support immediate changes in behavior? We focus on attachment and integrate the human literature with rat research employing invasive procedures impossible in humans that enable us to define causal mechanisms.

Altricial species, such as humans and rats, form bonds to promote activities critical for individual and species survival, such as defense when threatened, procuring food, and reproduction. Across altricial species, this bond extends to the mother-infant bond formed at birth, where the mother stays with the infant to nurture and feed the young and the infant remains with the mother showing prosocial behaviors that enhance the bond [12–15]. This bond formation has received considerable attention in both the animal and human literature, at times converging on theoretical perspectives, but mostly diverging. We suggest this is a time of convergence, where bi-directional exchange across research domains has resulted in our advancements in our ability in improve the lives of children.

Circuits for the Sensitive Period for Attachment: Not Immature Version of Adult Learning

We begin with an overview of attachment in the infant rodent, an animal model that has provided a rich literature on mechanisms and circuits supporting attachment in the developing brain. After the mother rat gives birth, she gathers her pups into the nest, grooms them and goes into a nursing posture, at which point pups quickly nipple attach. For decades we thought of this as an automatic process, dependent upon a maternal pheromone and fixed action patterns, but we now know this process requires pups to first learn the maternal odor in order to perform both approach behaviors and nipple attachment [16,17]. The attachment learning circuit that enables a novel odor to become a maternal odor is relatively simple: any odor paired with a maternal behavior (i.e. licking, being stepped on) produces a new maternal odor. We can readily mimic this learning by experimentally delivering maternal inputs to the pup: licking can be mimicked by tactile stimulation with a small brush, maternal thermoregulation mimicked by a warm environment, and occasionally stepping on pups simulated with tailpinch or a 0.5mA electric shock, all of which engage the attachment learning network and produce a new maternal odor [18]. It is important for pups to continue to learn the maternal odor because it is diet-dependent and changes as the mother eats new foods [19,20].

The neural circuitry supporting this rapid learning of the maternal odor has been well-studied in the typically-developing pup. Maternal behavior or experimental simulating maternal behavior activates the noradrenergic locus coeruleus (LC) to release copious levels of norepinephrine (NE) into the olfactory bulb [21]. The joint action of the odor and NE within the olfactory bulb produces prolonged excitatory activity of mitral cells (primary output neurons of the bulb), which together with small interneurons called granule cells, induces plasticity [21–26]. The infant LC has unique function and responsivity, in that it responds to a very broad range of stimuli and it has a very prolonged response that results in large levels of NE at the target site to induce plasticity. This attachment learning terminate at postnatal day (PN) 10, due in part to maturational changes in the LC [27–29]. Similarly to avian imprinting, this illustrates a unique learning system in the infant brain, which is not simply an immature version of an adult-like learning [30,31]. Of course, it is difficult to test whether a similar NE-dependent attachment system is present in children. Although there are very high levels of NE during the first years of life, this is true for myriad neurotransmitters and hormones [2,32]. However, the importance of NE in attachment is preserved, in concert with other neurotransmitters and hormones, across the lifespan in other species, including rats, mice, and sheep [33–36] .

Unique Threat Response during Sensitive Period for Attachment: Approach Mother

Presenting a threat to activate the fear system has been a powerful tool to understand human attachment [8] and uncover deficits in attachment [37,38]. When encountering a threat, the defenseless child goes to the caretaker for protection [39], and failure to approach the caregiver or approach-avoidance behavior is associated with attachment deficits [40]. It has also been a useful tool in infant rats to uncover characteristics of the sensitive periods for attachment. Specifically, we use fear (threat) conditioning with or without the mother present to explore attachment sensitive period mechanisms and neurobiology. Fear conditioning pairs a neural stimulus with a painful stimulus, which results in the neutral stimulus becoming feared [41,42]. In rat pups (<PN10), pairing a novel odor with 0.5mA shock or tailpinch does not support amygdala-dependent fear learning. In fact, pups learn to prefer the odor [43], and more surprisingly, the odor takes on qualities of the maternal odor as indicated by its ability to promote social interactions with the mother and nipple attachment [20]. The neural substrate for this early life unusual learning is the olfactory bulb- LC attachment system, rather than the adult-like amygdala-dependent circuit for learning threat. Although the new maternal odor engages a fear response, pups approach the odor because the age-appropriate response to fear is to use the mother as a safe haven [44–46].

Transitional Sensitive Period: Fear Learning and Maternal Regulation of the Amygdala

Our initial assessment of the neurobiology and development of fear learning involved removing pups from the litter and conditioning pups alone to ensure we had a controlled environment. This body of work demonstrated that amygdala-dependent fear learning emerges at PN10 [43,47]. However, in these models we had neglected to test pups in their most salient ecological context – with the mother. Emerging animal data was suggesting this was important, as maternal presence was shown to block threat-induced stress hormone release [48–51] and innate fear expression in pups was dependent upon the stress hormone corticosterone [52–54]. Repeating our experiments in the presence of a anesthetized mother (to prevent behavioral interference) blocked pups’ ability to learn amygdala-dependent fear. Notably, we could mimic the effect of maternal presence in pups conditioned alone by pharmacologically blocking pups’ shock-induced stress hormone release [55–57]. This demonstrated a causal role for maternal suppression of stress hormones in blockade of fear learning. We went on to characterize a Transitional Sensitive Period for a pup from PN10–15, where amygdala–dependent fear learning was suppressed if the mother was present through a simple mechanism of reducing amygdala stress hormones [34, 51][34], via systemic blockade of the hypothalamic-pituitary-adrenal axis (HPA) [34,58]. After PN15, maternal presence and corticosterone blockade fail to block or buffer fear learning.

This phenomenon has recently been shown to occur in children (Figure 1). Specifically, caregiver presence during an emotional task suppresses amygdala reactivity in children younger than ten years old, although this not observed in adolescence [59–62]. And similarly to rat pups, maternal presence was shown to block behavioral expression of fear learning in young children: pairing an aversive stimulus with a neutral shape when alone produced an avoidance of the shape in a choice test, but if the mother was present during the conditioning, the children later chose to approach the shape paired with aversion [63]. This effect is likely due to the ability of mothers to block their child’s threat-induced cortisol increase at young ages [64] . Thus, it appears that maternal presence blocks fear learning via suppression of the amygdala across species and suggests a specific mechanism for infants approaching the mothers as a safe haven when threatened.

Figure 1. Maternal presence switches avoidance of threatening cues to approach during the Transitional Sensitive Period in development.

In both rodents [43] and human children [63], maternal presence during the presentation of fear learning (novel stimulus paired with a threat) causes young to subsequently approach the previous novel stimulus. Rodent research shows failure of the amygdala to be engaged in fear learning during the sensitive period is causally related to the infant’s failure to learn fear [103]. This is in sharp contrast to post-sensitive period fear learning, where fear conditioning with or without parental presence supports amygdala-dependent fear learning [47].

Mechanisms for maternal blockade of infant fear learning

As our understanding of the profound impact of the mother on pups’ neurobehavioral functioning emerged, we began to search for possible mechanisms for her intervention (Figure 2). Again using a rodent model, we focused on the paraventricular nucleus of the hypothalamus, a brain area known to modulate hypothalamic-pituitary-adrenal (HPA) system [65]. PVN function was assessed by measuring PVN activation, quantifying PVN NE levels (major neurotransmitter transmitting stress information in PVN) with microdialysis, and blocking or activating PVN NE receptors to show the causal link for maternal blockade of pups’ fear learning. This set of studies showed that maternal presence blocks activation of the HPA axis to prevent stress hormone release [34] preventing support of plasticity since the infant amygdala is uniquely dependent upon CORT to enable plasticity mechanisms [56]. Together, these data suggest that pups are able to mount a stress response and respond/learn about threat while alone, but capitalize on defense from the caregiver while in her presence.

Figure 2. Unique neural circuitry supports maternal regulation of infant threat learning during the Transitional Sensitive Period.

Using a threat conditioning paradigm of odor-shock presentations has enabled us to uncover a developmentally unique learning system in pups that typically supports attachment learning. Data indicate that during the sensitive period for attachment learning (PN<9), pairing an odor with a shock activates the attachment learning neural circuit involving increased NE to produce approach responses to the odor [43,104]. The odor also takes on qualities of the maternal odor to support nipple attachment and enhance prosocial behaviors to the mother. In pups older than PN9, this fear conditioning paradigm accesses the amygdala to support fear/threat learning if the pup is alone. However, if the mother is present, she socially buffers (attenuates) the pup’s stress response, and pups revert to sensitive period learning and learn an odor preference. This mother-controlled switch between fear and attachment learning is mediated through the mother’s ability to control pups’ CORT [104]. A more adult-like fear learning system, which cannot be switched on/off by CORT, develops by PN15 [34,96,105–107]. Upstream of this circuit, the VTA releases dopamine (DA) in the BLA to promote fear learning when pups are alone, which can be suppressed by maternal presence[57].

The next set of studies was designed to address how maternal presence modulates amygdala plasticity. One of our first experiments was to use microdialysis to sample amygdala neurotransmitter activity during fear conditioning with or without the mother present. Measuring catecholamines, dopamine (DA) emerged as having a robust response to threat and maternal presence during conditioning. Next, using microinfusions of DA receptor agonists and antagonists into the amygdala, we uncovered that DA has an important causal role in the infant threat response and its suppression, although it worked in conjunction with increases in systemic stress hormones [53]. Further work verified that the source of BLA DA, the ventral tegmental area (VTA) was being modulated by the mother during fear conditioning, and this change in DA regulated whether the activation or suppression of molecular and synaptic processes required for learning [57].

The Mother is Modulating the Infant Brain Even When Infant is Not Stressed

Next, we integrated these data into the broader theoretical framework of maternal regulation of the infant. We were guided by the human developmental literature on regulation [66–68], as well as Myron Hofer’s concept of Hidden Regulation [69], in which sensory stimuli hidden within the complex maternal behavior and mother-infant interactions has the power to alter physiological function (heart rate, respiration, stress and growth hormones). We asked, was the mother impacting the amygdala function when pups were not threatened? We focused on the amygdala because of its known role in processing cue valence through amygdala DA [53,70–72], but also because we had shown that the mother was modulating amygdala DA levels at its VTA source [57].

In these studies, pups were implanted with an electrode in either their amygdala or cortex and placed back in the nest where we conducted untethered local field potential (LFP) recordings during periods of time spent alone and with the mother performing typical maternal behaviors, such as nursing and grooming [73]. LFP provide a readout of the rhythmic neural activity of groups of cells, with a range of frequencies from low (delta associated with sleep) to high (gamma typically associated with neural processing), all of which serve distinct physiological functions and are important for plasticity and network communication within and between brain regions [74]. In infancy, these rhythmic frequencies are also known to guide brain development [75]. The literature had already highlighted the ability of social stimuli to modify neural oscillations throughout human development [76,77], although the attachment figure (biological or adoptive caregiver) is documented to be particularly effective [67,78]. Since this had not been assessed mechanistically using an animal model during mother-infant interactions, we questioned whether infant brain activity was influenced directly by interactions with the mother within the natural nest environment. We showed that maternal absence from the nest increased high frequency (gamma) oscillatory power, and replacing the mother reduced gamma power and increased slow-wave activity. Perhaps most striking was the immediate impact of maternal simulation of pups: when the mother groomed pups or gave pups milk, it produced a rapid, transient surge in high frequency oscillations. Notably, the mother’s ability to alter pups’ cortical oscillations waned with pup maturation. Finally, capitalizing on animal research and understanding the importance of NE in attachment, we injected a NE blocker, which reduced maternal regulation of infant cortical activity. Taken together, these data demonstrate maternal regulation of the infant brain occurring in real-time.

It must be noted that there is no experiment that can be done within a naturalistic environment to show the effects discussed here are due to anything other than the mother. For this reason, we have done studies mimicking specific features of the mother to test causation[79,80]. For example, maternal behavior is not critical but maternal presence/maternal odor during repeated stress is responsible for some of the enduring effects of adversity. Other studies are ongoing in the lab to mimic rough handling outside the nest to assess outcomes of this specific feature of adversity-rearing.

Compromised Sensitive Period Attachment Neurobiology by Adverse-Rearing

Early life adversity compromises the mother’s ability to regulate the child, suggesting a subtle but immediate impact of adversity. To integrate our animal model of regulation with dysfunctional regulation in humans, we used a model of Scarcity-Adversity Rearing with low resources (mother has insufficient bedding for nest building (for review, see [81,82]). In this model, the mother still nurtures pups but frequently handles pups roughly during repetitive nest-building. Importantly, pups still shows robust attachment to the mother rats that treated them roughly, replicating abusive attachment observed in many species [14,83,84]. However, type of rearing appears to degrade the value of maternal signals to pups: maternal odor produces attenuated approach and attenuated neural responses throughout the brain [84]. This devaluation of the maternal odor appears to have wide-spread impact on pups. For example, following Adversity-Rearing, maternal presence fails to block fear learning in PN12–14 pups. Activation of the ventral tegmental area (VTA) is not buffered by maternal cues and these cues fail to block amygdala plasticity [56,57] (Figure 3A).

Figure 3. Early adversity impacts developing circuits supporting fear and attachment.

A) Maternal presence during fear learning engages distinct connectivity patterns in the amygdala-VTA network across development and after early adversity. B) Adversity-rearing blunts the impact of nurturing maternal care (eg. grooming) on pup cortical oscillations, with minimal effect on rough handling responses. Data reproduced from [57] and [85].

Recent work identifies specific features of maternal presence and behavior that compromise maternal buffering of the pup threat circuitry. Again recording LFP from the infant neocortex, we observed that during adversity-rearing, the mother fails to regulate pup cortical oscillations [85] (Figure 3B). Unexpectedly, the most robust decrements in maternal regulation were in the cortical response to nurturing behaviors, such as milk and grooming, while LFP responses to being roughly handled (which occasionally occurs in control rearing) did not differ between groups. These effects were stress-hormone dependent, as blocking pup stress hormones during adversity-rearing restored maternal regulation of oscillations, as well as pup attachment behaviors. The role of stress hormones was also demonstrated in a parallel series of studies which isolated the effects of stress, maternal presence, and adverse maternal behavior [79]. Whereas the mother typically regulates acute stress responses in the infant, repeated stress in the presence of the mother produced attachment deficits and amygdala dysfunction. Notably, repeated stress alone (no mother) was not sufficient to mimic the effects of adversity rearing. Overall, these studies help identify promising biomarkers of later-life psychopathology following adversity, as well as generate testable hypotheses in children.

These data suggest that adversity compromises the experience-dependent plasticity afforded by the sensitive period such that the maternal input fail to regulate the pup brain. In the above referenced examples of cortical oscillations and DA signaling in the BLA, adversity impairs the mother’s ability to regulate the pup during the typical sensitive window for this regulation. However, the timing of the window itself can be shifted by early adversity. For example, amygdala plasticity to promote threat learning can be induced prior to PN10 in adversity-reared pups [86,87]. Furthermore, plasticity mechanisms, such as elaboration of perineural nets, can be shifted later or earlier by early adversity to terminate sensitive windows for developmentally-specific amygdala function ([88,89]. Clinical research and Developmental Psychology research in children shows similar effects of early care quality on the timing of processes including regional activation and dynamic functional connectivity between threat processing brain regions [90–92].

Shifting these windows through early stress manipulations has long-term outcomes for social, cognitive and emotional processing across species[84,93,94]. Although the behavioral expression of these brain changes may be considered aberrant or even pathological, an alternative interpretation that confers adaptive behavioral flexibility cannot be discounted[95]. For example, on some level, early caregiving adversity in the pup spares/preserves the pup attachment system but modulates the strength of the attachment [96]. This occurs at the cost of decreased behavioral flexibility in response to trauma and less social interaction with peers. However, the latter may be adaptive, as novel social partners generally represent a greater risk than the mother [97]. Furthermore, changes in brain function associated with adversity may confer some adaptive advantage in terms of promoting resilience to future stressors [98–101]. However, these interpretations depend greatly on the degree to which early stressors fall within the spectrum of species-expected or species-typical range of experience [102].

Conclusion

A rich literature has documented sensitive periods across altricial species, including humans, with animal work identifying many neural substrates. Here we have focused on fear learning in the developing infant, which is regulated by the caregiver during a circumscribed period in early life. Whereas typical maternal regulation of infant threat responses can occur on the levels of behavior, network oscillations, brain regions, neurotransmitters, and proteins, early adversity compromises maternal regulation across all of these measures. Ongoing translational research efforts will be critical for identifying age-appropriate therapies and interventions following adversity.

Research Highlights.

• 1)Infant attachment learning involves a neural circuit distinct from adults
• 2)Defenseless infants approach the caregiver for protection when threatened
• 3)Caregiver presence suppresses the adult-like amygdala-dependent threat system
• 4)Caregiver modulates pups neural oscillations, even in the absence of a threat
• 5)Adversity-rearing compromises pups’ neural circuitry for attachment behaviors