Double Helix: Reciprocity between juvenile play and brain development

Abstract

This review summarizes what is presently known about the function, sexual differentiation, and neural circuitry of juvenile rough-and-tumble play. Juvenile rough-and-tumble play is a unique motivated behavior that is widespread throughout the mammalian order and usually occurs more often in males. Immediate early gene studies indicate that cortical and subcortical circuits, many of which are sensitive to sex steroid hormones, mediate juvenile play. Sex differences in rough-and-tumble play are controlled in part by neonatal exposure to androgens or their estrogenic metabolites. Studies indicate that testicular androgens during play are also necessary to stimulate male-like levels of play initiation. The resemblance of rough-and-tumble play to aggression and sexual behavior has led some to question whether male-typical adult behavior is contingent upon the experience of play. Attempts to control the amount of play through social isolation show that social experience during adolescence is critical for male-typical adult behaviors to be expressed. This well-established finding, together with evidence that play induces neural plasticity, supports the hypothesis that juvenile play contributes to male-typical brain development that ultimately enables the expression of adult social and reproductive behavior.

1. Introduction

Adolescence is fraught with drama in that the risk of psychiatric conditions like affective disorders and schizophrenia, drug abuse, and addiction increases dramatically (Wagner and Anthony, 2007). Adolescents seek greater novelty and are more prone to take risks (Spear, 2000). Puberty, which usually coincides with adolescence, culminates in a sexually mature individual with re-oriented and broadly expanded social interests and proclivities (Forbes and Dahl, 2010). The simultaneous changes in the motivational, emotional, and hormonal state of adolescents, as well as their susceptibility to mental illnesses, means that the circuits and systems that control these behaviors must undergo considerable plasticity during this time. Because it is so fraught with risk, both for physical injury and mental dysfunction, it is essential to understand the neurodevelopmental processes that occur during adolescence.

Perhaps the most dramatic upheaval during adolescence occurs in the social realm. Before adolescence, boys and girls prefer same-sex play partners: boys play together in large groups, whereas girls generally prefer one-on-one and less competitive relationships (Maccoby, 1998). During adolescence, however, sexual attraction becomes a powerful force driving social behavior such that children must learn to navigate through the treacherous seas of the other sex. At the same time, inter-sex competitiveness can increase, particularly among boys as they jockey for dominance (Fry, 2005). How does this reorientation of social priorities occur? Studies in laboratory rodents have shown that the development of male-typical adult behaviors depends in part on social experience during adolescence, particularly the experience of play.

Rough-and-tumble social play is widespread throughout the mammalian order, but object-centered play has also been observed in animals diverse as birds, fish, and lizards (Heinrich and Smolker, 1988, Burghardt, 2010). Rough-and-tumble play often, but by no means always, occurs more frequently in males (Ellis et al., 2008). Yet play has no well-established immediate benefit for juveniles and is almost by definition a purposeless behavior. The sexually dimorphic frequency of rough-and-tumble play in many species suggests that sexual selection has led to its evolution. This notion is supported by the costs of play: animals at play expend considerable amounts of energy, are at higher risk of predation, and incur more accidents (Byers, 1998). Studies show that rough-and-tumble play has long-term benefits for males that are realized when they are confronted with the challenges of sexual maturity: establishing dominance and having sex.

What then is play? The aim of this paper is to explore this question from a developmental neuroscientific perspective. The title of this paper, Double Helix, was chosen to highlight the reciprocal relationship between sex-typical behavior and sex-typical development. Perinatal exposure to gonadal steroids induces pervasive change in the connectivity and neurochemistry of the brain, increasing the motivation to play, i.e., to approach another conspecific and engage in a wrestling match until someone wins and someone loses. Playful experiences induce changes in brain structure and function, as evidenced by play-induced immediate early gene expression and the results of social deprivation experiments. Thus, over developmental time, juvenile play-driven neural plasticity shapes circuits in a sex-specific manner that in turn enable sex-typical adult behaviors to be expressed. To paraphrase Kuo (1999) adult socio-sexual behaviors are the functional product of the dynamic relationship between the playful male and its hormonal and social environment. In an analogous fashion, the DNA molecule requires each strand of nucleotides to complement the other in order to perform its function in biology.

We will describe hypotheses about the adaptive value of play, experimental work that demonstrates the importance of juvenile play to brain and behavioral development, as well as what is known about how play is sexually differentiated. We should mention that the possible functions of play are explored in The Playful Brain (Pellis and Pellis, 2009). Significant comparative work on juvenile play has been conducted, some of which is collected in two superb volumes (Fagen, 1981, Bekoff and Byers, 1998). However, the vast majority of studies investigating the neurobiological substrates of play motivation have been conducted in rodents. Thus, this review is restricted mainly to studies of the order Rodentia.

1.1. The function of play

The superficial resemblance of rough-and-tumble play to actual fighting, as well as the sex difference favoring males in rough-and-tumble play frequency would suggest that play is practice for adult male behavior, such as aggression or mating (Fagen, 1981, Takahashi and Lore, 1983, Nunes et al., 2004). If play were practice for aggression, one would expect young rats to execute the actual movements seen during fights, and to perform the most difficult moves of aggression most frequently (Pellis and Pellis, 1998). Yet neither of these expectations is empirically supported. For example, when adult rats fight, they attempt to bite the flanks and back of their opponent, whereas the nape of the neck is targeted during play and it is vigorously investigated, not bitten. Defensive postures are also distinctly different between juvenile play and adult aggression (Smith et al., 1998). Spotted hyenas (Crocuta crocuta) display intense, sometimes fatal, inter-sibling aggression only minutes after parturition, followed several weeks later by rough-and-tumble play (Drea and Hawk, 1996, Drea et al., 1997). Thus, the usual sequence of behaviors is reversed, and if play were necessary practice for aggression, it could not occur in this species.

The targets of Djungarian hamster playful attacks are often the same as those targeted during pre-copulatory behavior (Pellis and Pellis, 1998), and mounts are occasionally observed during rough-and-tumble play among rats. But the critique of the play-as-practice for aggression theory applies here as well: if play were practice for mating behavior, the most difficult move, intromission, would surely be executed more frequently than mounts. Yet intromission is rarely observed among playful rodents. Thus, the practical effects of play on adult behavior are likely to be indirect.

A clue to the purpose of play is its competitiveness, rough-and-tumble play's quiet essential quintessential characteristic. Rats at play pounce at one another, often aiming toward the opponent's nape, and struggle to gain advantage over their playmate, concluding bouts with a pin, a dominance posture, or when the victor vigorously buries his muzzle into the nape of his opponent's neck (a montage of a typical juvenile play bout can be found in Fig. 3). In spite of being pinned, the defeated rat will nonetheless get up and immediately pounce toward the individual that had dominated him a moment before, and the struggle for dominance will begin anew. The inherently social quality of play suggests that it helps animals to understand the intentions of others, enabling flexible social behavior throughout life (Bekoff and Allen, 1998). Although learning to “read minds” is undoubtedly important for both sexes in a social species like the rat, their polygynous reproductive strategy requires intense male–male competition (Trivers, 1985). Semi-naturalistic observations of Rattus norvegicus affirm that males compete vigorously with each other to gain access to an estrus female, and are more aggressive generally (Calhoun, 1962). The competitiveness of play may therefore allow adult males to anticipate other males’ intentions, a key skill in a social ecology that entails intense intra-sexual competition.

Fig. 3.

(A–D) Typical play bout sequence, beginning with a pounce, and concluding with a pin. (E) Gonadectomized rats initiated fewer pounces than those with their testes. Cooke, B., Woolley, C.S, 2009. Developmental Neurobiology, 69, 141–152. This material is reproduced with permission from Wiley & Sons.

Another, not mutually exclusive possibility, is that play is a means of self-assessment. Each play bout has an unambiguous binary outcome that provides immediate feedback about the success or failure of one's actions, enabling the animal to modify its ongoing behavior, as well as to gauge its future ability to compete with its conspecifics for food, resources, and mates (Thompson, 1998). Best to learn to resign from contests that one may lose than to repeatedly engage in battles in which defeat is the only outcome.

The internal environment is also a source of feedback from play. As motor commands are generated, an efferent copy of the command is sent to sensory areas, where it is used to suppress the sensory processing elicited by the self-generated action. For example, the auditory evoked potential is smaller for one's own speech than when one hears another's, despite being matched in amplitude (Ford et al., 2001). The cerebellum is hypothesized to generate an error signal if there is a discrepancy between the actual and the predicted feedback (Leube et al., 2003, Houk et al., 2007). Thus, the seeming purposelessness of juvenile play may actually be the activity-dependent sculpting of cerebellar circuits so as to maximize the precision of error detection during behavior. In line with this possibility is the finding that the most significant phase of synaptic remodeling in the rodent cerebellum occurs during adolescence (Byers, 1998), and that play behavior in juvenile squirrels predicts improved motor skills (Nunes et al., 2004). Furthermore, as described below, play deprivation experiments show that play may indeed serve as an experience-dependent mechanism that is necessary for normal motivational, social, and emotional development.

In the only study of its kind, Fagen and Fagen (2004) related the survival of brown bear cubs (Ursus arctos) to the rate of play, while controlling for food availability, the cub's condition at birth, and the quality of maternal care. Independently of these factors, bears that played more were more likely to survive into their second year. The authors could not identify the mechanism that linked play to short-term survival, although they speculated that play relieves stress and improves resistance to future stressors. This hypothesis seems reasonable in light of findings that playful hamsters have higher corticosterone levels (Cheng et al., 2008), and that rats deprived of play show abnormally high levels of corticosterone when confronted with an aggressive conspecific (van den Berg et al., 1999a). But it is of course also possible that there is no causal relationship between juvenile play and immediate survival at all. It may simply be that healthier bear cubs were able to play more.

Speculation about the ultimate function of play has ranged from it being directly beneficial, e.g., as practice for adult behavior, to being indirectly beneficial. The play-as-practice theory seems unlikely, given that playful animals do not practice the most difficult behaviors they need to execute as adults. The indirect benefits of rough-and-tumble play could include the refinement of sensori-motor circuits through sensory feedback and error detection, improved judgment of others’ intentions, better awareness of one's ability to compete in a social environment, and/or activity-dependent refinement of the stress response system. Another possibility is that rough-and-tumble play may facilitate the sex-specific “wiring up” of neural networks that mediate adult sex-typical social behavior so as to enable hormone-driven adult behaviors to occur. As described below, we hypothesize that juvenile play exerts a trophic effect on the endogenous opioid system that allows adult social behavior to be more rewarding.

1.2. Effects of play on the brain

Given interest in the ultimate function of play, what are the effects of play on the brain? Experiments that address this question can be categorized into those that employ short- or long-term social isolation to control play behavior. Regardless of their duration, however, social deprivation experiments are fraught with the potential for misinterpretation. As noted originally by Lehrman (1953), an isolated animal continues to have experiences. These experiences can fundamentally alter the trajectory of development, such that an isolated animal cannot be thought of as one from whom the effect of social experience has been “subtracted”. With that caveat in mind, social isolation during the epoch of maximal rough-and-tumble play devastates adult socio-sexual behavior, suggesting that adolescent social experience is necessary for proper adult behavioral development.

1.2.1. Neurobehavioral effects of short-term deprivation

When juvenile rats are socially isolated for several hours, they engage in play more frequently upon reunion than those left in standard conditions, where play can be sporadic. A typical short-term isolation experiment is one in which both groups of rats are allowed to play, are isolated for 12–24 h, one group is reunited while the others remain separate, and both groups are perfused following the play period.

When studied in this way, rats that played had widespread c-fos mRNA expression in areas involved in sensori-motor processing such as the parietal cortex, superior and inferior colliculi, dorsa land ventral striati, pontine nuclei, and periaqueductal gray (Gordon et al., 2002). c-fos is an immediate early gene (IEG) transcribed when neurons fire at high rates, making it a cellular-level marker of neural activity. The proteins produced as a result of IEG expression are transcription factors that control the expression of other genes that ultimately underlie the synaptic, dendritic, and network-level plasticity that constitutes developmental change in behavior. The implication of c-fos expression, then, is not only that those particular brain regions participated in the expression of rough-and-tumble play, but that plasticity most likely occurred within those networks as well.

In a study of Syrian hamsters, play increased fos protein in the medial prefrontal cortex (mPFC), particularly the pre- and infralimbic areas (Cheng et al., 2008). These areas send strong descending projections to the medial amygdala (MeA), and posterior bed nucleus of the stria terminalis (BNST; McDonald et al., 1996, McDonald et al., 1999), which also showed high levels of play-induced fos, as did the lateral septum and the vasopressinergic neurons of the hypothalamus. The MeA, BNST, and lateral septum participate in a distributed neural network that collectively regulates social behavior, including the processing of olfactory cues, neuroendocrine regulation, and the control of species-specific acts, such as pouncing, defensive postures, and vocalizations (Newman, 1999). Bell et al. (2010) reported that the mPFC of juvenile female rats housed with an adult, as opposed to with peers, had longer and more branched dendrites. Interestingly, the opposite effect was observed in the orbitofrontal cortex: females housed with three peers had greater dendritic branching in the orbitofrontal cortex than those housed with an adult or those with a single peer. This may be the first report of juvenile social experience affecting the female brain, exposing a gaping hole in the empirical literature. The underlying mechanism of these particular effects is, however, unknown. The lack of peer-related social experience may have prevented dendritic pruning or, conversely, housing with an adult may have induced dendritic growth. The behavioral and functional effects of this plasticity, if any, are of course also unknown.

In the human, the mPFC inhibits aggression (Bufkin and Luttrell, 2005), monitors conflict between the tendencies to approach and avoid other people (Hall et al., 2010), and is engaged in tasks when self-monitoring is necessary (Kiefer et al., 1998, Egner et al., 2008). Rodent studies indicate that the mPFC also inhibits the stress response system via its connectivity with the MeA, the central nucleus of the amygdala, and the BNST (Figueiredo et al., 2003, Herman et al., 2005). The presence of fos and dendritic remodeling in the mPFC means that play-initiated synaptic plasticity may ultimately improve its ability to modulate the social behavior network that encompasses the medial extended amygdala, lateral septum, and hypothalamus.

The clearest evidence yet that rough-and-tumble play promotes brain development comes from two studies that link play to growth factor expression. Playful rats had increased brain-derived neurotrophic factor (BDNF) mRNA in the amygdala and prefrontal cortex (Gordon et al., 2003), and, in a separate study, insulin-like growth factor 1 (IGF1)-related genes and protein in the frontal and posterior cortices (Burgdorf et al., 2010). Furthermore, injection of IGF1 into the cerebral ventricles increased rough-and-tumble play, including hedonic 50 kHz vocalizations, suggesting that juvenile play and IGF1 signaling is mutually reinforcing (Burgdorf et al., 2010). As the name implies, IGF1 is associated with neuronal growth and remodeling, and protection from cellular insult (Torres-Aleman, 2010). While very intriguing, these reports are nonetheless only circumstantial evidence that play induces plasticity that in turn is necessary for adult behaviors to emerge. Much stronger support for this hypothesis would be obtained if the prevention of c-fos, BDNF, or IGF1 expression following play blocked the development of male-typical adult behavior.

Dopamine is widely known as being critical for incentive motivation. Given that juvenile play is a highly motivated behavior, dopamine antagonists consistently reduce juvenile play, as one would expect (Niesink and Van Ree, 1989, Siviy and Panksepp, 2011). Interestingly, dopamine and IGF1 both belong to a growing group of ligands capable of activating estrogen receptors in a steroid hormone-independent manner (Mani et al., 2009, Marin et al., 2009). Given the sexually dimorphic nature of play, increased dopamine and IGF1 during juvenile play could potentially mediate sex-specific changes through activation of estrogen receptors, which have a masculinizing effect (see below).

An important piece of evidence that rough-and-tumble play does in fact promote brain development and adult behavior comes from work investigating the effects of brain lesions. Male rats were subjected to lesions of the medial preoptic area (mPOA) at P28. Some rats were housed in social groups and allowed to play whereas others were housed alone. As adults, the rats were tested for their sexual behavior. Only those rats housed with their peers showed any recovery. Social isolates and those separated from their peers with a perforated barrier rats had severe deficits in mounting, intromissions, and ejaculations (Twiggs et al., 1978, Leedy et al., 1980). These reports are all the more remarkable considering how critical the mPOA is to male sexual behavior, as lesions there permanently abolish male copulation in adult males (Heimer and Larsson, 1967). No studies to date have identified how the lesioned brain is rewired as a result of juvenile social experience, but that is clearly the implication of this work.

“Happiness is never better exhibited than by young animals, such as puppies, kittens, lambs, etc., when playing together, like our own children.” (Darwin, 1872). As Darwin recognized, play is universally rewarding. Play is fun. Rats will learn discrimination tasks in order to play, and come to prefer locations where they have played (Humphreys and Einon, 1981, Calcagnetti and Schecter, 1992). The pleasurable aspect of play requires the endogenous opioid system, as shown by Panksepp and others (Panksepp et al., 1980, Trezza et al., 2010). The selective μ opioid receptor (MOR) antagonist naltrexone reduces play, while morphine increases it (Vanderschuren et al., 1995a, Vanderschuren et al., 1995b). Juvenile play induces endogenous opioid release in the periaqueductal gray, as indicated by reduced MOR binding in autoradiography experiments (Panksepp and Bishop, 1981). The notion that play induces opioid release was supported when Vanderschuren et al. (1995c) quantified MOR binding in male rats that played after 24 h social isolation vs. those that did not. Play itself reduced MOR binding (and thus, increased opioid release) in the nucleus accumbens, suggesting that opioid release controls to some extent the incentive value of play. Social experience, regardless of prior housing condition, caused a small but significant reduction in MOR binding in the paraventricular nucleus of the hypothalamus, which could reflect an activation of the stress response system due to the novelty of the situation. MOR binding was increased in the mPFC and parafascicular area after seven days of isolation, suggesting that longer-term isolation may reduce opioid release in those areas (Vanderschuren et al., 1995c).

Although the effect of play-induced opioid release on brain development is unknown, a deficit in juvenile opioid signaling can impair adult behavior. Van den Berg et al. (1999b) reported that morphine treatment during juvenile social isolation ameliorated isolation's deleterious effects on adult social behavior, suggesting that play-induced opioid release participates in the development of the social brain. We know that endogenous opioids play an important role in brain development, increasing neural and glial proliferation (Stiene-Martin et al., 2001, Zagon et al., 2002). Endogenous opioids directly upregulate BDNF mRNA in the frontal cortex of the adult (Torregrossa et al., 2006, Zhang et al., 2006), and can serve as growth factors themselves through the opioid growth factor receptor (Zagon et al., 2002). Interestingly, low doses of naltrexone reportedly improve sociability in autistic children temporarily, increasing rough-and-tumble play, eye contact, and less self-injurious behavior (Leboyer et al., 1992, Lensing et al., 1992). Given the importance of opioids in brain development and the abnormally high levels of β endorphin in the autistic brain (Tordjman et al., 2009), early and prolonged treatment with naltrexone could potentially lead to lasting improvement in autistic symptoms.

Taken together, these reports suggest that play-induced opioid release regulates the ontogeny of the opioid system itself as well as brain circuits modulated by opioids, such as the mesolimbic incentive motivational system, the hypothalamus, and the amygdala. Maturation of the endogenous opioid system, in conjunction with rising levels of circulating sex hormones, could ultimately allow adult forms of social behavior like courtship, mating, and parental behavior to be pleasurable as well, thus enabling the adolescent re-orientation of social behavior. If true, this would support the notion of a mutually dependent interplay over time between the experience of rough-and-tumble play and sex-typical brain development, like the strands of a double helix.

1.2.2. Neurobehavioral effects of long-term social isolation

Long-term social isolation (days to weeks) has long been known to induce significant sociosexual behavior deficits in males (Gerall et al., 1967, Einon et al., 1978, Larsson, 1978, Hol et al., 1999). For example, isolation between postnatal days 28 and 35 prevents isolates from displaying submissive postures in the presence of a dominant conspecific. At the same time, adominant rat evoked significantly greater corticosterone levels in the isolates than in controls (van den Berg et al., 1999a). Thus, little more than a week of complete social isolation was sufficient to permanently alter a male rat's social behavior and stress response system. It is easy to imagine how poorly such a rat would fare in ecologically realistic circumstances. In another first-of-its kind study, Hermes et al. (2010) examined the social behavior and synaptic protein content of females that had been socially isolated from weaning through P70. As seen in males, isolation reduced social behavior and exploratory behavior, indexed by the number of physical contacts with conspecifics and the number and duration of entries to the center zone of the open field. Prefrontal cortex levels of synapsin, GluR1 and NR1 protein levels were all reduced in the isolates. This, and the Bell study cited earlier highlight the significance of social behavior during adolescence for normal development in both sexes, and underscores the importance of including both sexes in studies of non-sexual social behavior.

The MeA is necessary for rats to respond appropriately to dominant conspecifics (Luiten et al., 1985), and also modulates the hypothalamic–pituitary–adrenal axis (Herman et al., 1996, Herman et al., 2005). With this in mind, we asked whether post-weaning social isolation could lesion the MeA. We compared the sexual motivation, sexual performance, and the brains of rats that had been isolated from P21 to P60 with socially housed controls (Cooke et al., 2000). As expected, the isolates were sexually impaired, displaying one sixth as many non-contact penile erections in the presence of an inaccessible estrus female, and executing far fewer mounts, intromissions, and ejaculations. After the tests, the rats were perfused and their brains were sectioned and stained with thionin. We then used stereology to estimate the regional volume of the four subnuclei in the MeA, the size of the neurons within them, as well as the volume and soma size of the sexually dimorphic nucleus of the preoptic area (SDN-POA).

Social isolation selectively reduced regional volume and soma size in the sexually dimorphic posterodorsal subnucleus (MePD) and the SDN-POA. Other MeA subnuclei were unaffected by the isolation. Seminal vesicle weight was slightly and non-significantly lighter in the isolates, suggesting that isolation may have reduced androgens, leading to atrophy of the hormone-sensitive MePD and SDN-POA. Another non-mutually exclusive possibility is that MePD and SDN-POA development requires male-typical social experience to occur, independent of circulating androgen levels. To test this idea, it would be necessary to mimic in isolates and controls the dramatic peri-pubertal flux in circulating androgen levels, which may be impossible given the potential effect of social experience on androgen metabolism in what are supposed to be hormonally “clamped” males (Raouf et al., 2000). Contemplation of this hypothetical shows just how closely intertwined are the effects of social experience on hormones and the effects of hormones on social experience.

To determine whether the behavioral effects of isolation are due to play deprivation per se, and if the physical inactivity imposed by isolation contributes to later deficits, we then compared rats reared between P28 and P50 in one of four conditions: with 2 peers, with a non-playful adult male, isolated with a running wheel, or isolated with a locked running wheel (Arnold and Cooke, 2010). Rats were tested in the open field, given standard sexual behavior tests, and perfused to analyze MeA dendritic morphology.

The isolates required many more intromissions to achieve an ejaculation than either of the socially housed rats, which was reflected in their lower intromission ratio (Fig. 1A). Social experience without play was therefore sufficient to rescue sexual performance from the isolation-induced deficit. This impairment may have been in part due to more fearfulness, as the isolates spent less time in the central region of the open field test in spite of having identical locomotor activity. A functional running wheel affected only one aspect of sexual behavior: rats with a locked wheel made twice as many unsuccessful mounts as the socially housed rats (Fig. 1B), suggesting that physical activity during adolescence improves coordination in mating. Isolates had slightly lighter seminal vesicles and slightly heavier adrenal glands, but neither difference was statistically significant. Not surprisingly, given their social deprivation, isolates had significantly lighter brains, particularly those housed with a locked running wheel. Classic studies show that social and/or physical deprivation leads to lower brain weight due to the atrophy of cortical pyramidal cells (Rosenzweig et al., 1972).

Fig. 1.

Sexual performance in isolates and social controls. (A) Socially housed rats intromission ratio (intromissions/intromissions + mounts) was greater than in isolates (t-test, ***p < 0.001), indicating better performance. (B) Isolates housed with a locked running wheel attempted significantly more mounts than socially housed animals (t-test, *p < 0.05).

After the tests, the rats were perfused and their brains prepared for diolistic labeling of neurons in the MeA. A single investigator traced the dendritic arbors of well-labeled neurons. Sufficient numbers of neurons were obtained only within the anterodorsal subnucleus of the MeA, the MeAD (Fig. 2A and B). Social isolates had dendritic lengths that were on average 200 μm longer than controls, due to more dendritic branches. While these findings are only based on a few cells, they suggest that social isolation can either prevent dendritic remodeling or that social experience causes dendritic remodeling.

Fig. 2.

(A) Tracings of diolistically labeled MeAD neurons. (B) Isolated rats MeAD neurons had longer dendrites due to more branches, particularly at 60 μm, 260 μm and 370 μm from the soma.

1.3. The sexual differentiation of play

1.3.1. Organizational–activational effects

Males generally engage in more rough-and-tumble play than females. Higher frequencies of rough-and-tumble in play occur in male rodents, felids, canids, non-human primates, and children (Meaney and Stewart, 1981a, Thor and Holloway, 1986, Caro, 1988, Lovejoy and Wallen, 1988, Ward and Stehm, 1991, Maccoby, 1998, Nunes et al., 1999, Ellis et al., 2008). Many neonatal hormone ablation and hormone treatment experiments have shown that the sex difference in play is controlled by exposure to gonadal hormones during perinatal life (Beatty et al., 1981, Meaney and Stewart, 1981b, Tönjes et al., 1987), consistent with a so-called “organizational” effect of gonadal hormones. Congenital adrenal hyperplasia exposes the fetus to abnormally high levels of adrenal androgens. Girls with this condition show male-typical rates of rough-and-tumble play (Ehrhardt and Meyer-Bahlburg, 1981, Hines, 2003, Auyeung et al., 2009). The female spotted hyena is exposed to high levels of gestational androgens and is more aggressive and dominant as an adult than males, and female hyenas engage in more rough-and-tumble play than males (Pedersen et al., 1990).

In rodents, testosterone is aromatized in the brain into estrogen, where it acts on estrogen receptors (ERs) that are intensely expressed within nuclei in the midbrain, hypothalamus, BNST, MeA, and hippocampus. Androgens also act directly via the androgen receptor (AR), which is expressed in largely the same areas as the ER. Activation of either receptor during perinatal life appears to be sufficient to masculinize rough-and-tumble play. For example, neonatal treatment with 100 μg estradiol benzoate masculinized juvenile play in rats (Auger and Olesen, 2009), whereas pharmaceutical blockade of ARs prevented juvenile play masculinization (Casto et al., 2003, Hotchkiss et al., 2003). Male rats rendered insensitive to androgens from a mutation in the AR gene (Tfm) initiated play as frequently as wild-type control males, but they continued to use defensive postures characteristic of juveniles as adults (Field et al., 2006). Collectively, these results indicate that the rodent AR and ER target overlapping sets of genes whose products ultimately result in the expression of male-typical play. The ER is apparently unnecessary to masculinize human play behavior, however. Androgen-insensitive human males carrying the Tfm mutation display feminine patterns of juvenile play (Hines, 2004).

To determine whether pre-pubertal testicular androgens contribute to rough-and-tumble play, we gonadectomized male rats at P22 and counted play initiation pounces until P50 (Cooke and Woolley, 2008). As expected, the number of pounces in the sham control rats followed an inverted U-shaped curve, peaking in frequency around P30. Gonadectomized males played significantly less than controls, however, particularly at P30 (Fig. 3). We then measured miniature synaptic current frequency and dendritic spine density in the MeA of these animals and found significant reductions, indicating a loss of functional excitatory synapses in the gonadectomized males. This finding indicates that the low but rising testicular androgen levels in the male rat at this time increases play motivation, and may do so via trophic effects on MeA synapses. The prepubertal adrenal gland produces considerable amounts of two other androgens, 5α-androstane-3α-17β-diol and dehydroepiandrosterone (Moger, 1977, Pignatelli et al., 2006). A future study should compare the play of gonadectomized with adrenalectomized rats to assess each gland's relative contribution to play motivation.

Gonadal hormones obviously affect behavior, but the social environment is also known to affect hormones. For example, male Syrian hamsters defeated in an agonistic encounter have lower androgens (Huhman, 2006). Although gonadal hormone levels have not been compared in playful vs. non-playful rodents. Lupo di Prisco et al. (1978) reported that aromatase activity was greater in the brains of juvenile male rats housed in unisexual groups than those housed in heterosexual or isolated conditions. Because the amount of rough-and-tumble play was undoubtedly greater in the unisexual housing condition than any other, this finding suggests that aromatase obeys the “challenge hypothesis” in juvenile rats, rising during play behavior so as to increase the amount of available estrogen, perhaps to enable play-induced neural plasticity.

Why do males typically play more than females? Males’ greater aggressiveness is probably essential to this sex difference in motivation. Sex differences in aggression make males more likely to try to dominate their peers, and this is reflected in the competitiveness of rough-and-tumble play as indexed by the number of playful contacts. As one might expect, overlapping androgen-sensitive neural circuits regulate juvenile play and adult aggression, with the MeA, the anterior and ventromedial nuclei of the hypothalamus, and the periaqueductal gray being implicated in both behaviors (Potegal et al., 1996, Siegel et al., 1999, Gregg and Siegel, 2001, Cheng et al., 2008). At the same time, play must be more rewarding to males. The higher overall expression levels of MORs and enhanced β endorphin release in males (Cicero et al., 2002, Krzanowska et al., 2002, Zubieta et al., 2002, Cataldo et al., 2005, Craft, 2008) could enable play-related stimuli to be more rewarding. Males are less concerned for their physical welfare during aggressive conflicts, reflecting another potentially important sex difference that being juvenile males’ higher pain threshold (Butkevich and Vershinina, 2001). A higher pain threshold would make rough-and-tumble play less aversive, and simultaneously permit the enjoyment of the rough-and-tumble play experience.

A third factor that underlies sex differences in play is not play-specific but rather is sex-specific. When rats are confronted with a conspecific that attempts to steal food, males and females display subtly different patterns of evasion. Both sexes rotate away from the attacker, but males remain closer to the attacker than females (Pellis, 2002). These patterns of evasion carry over to play, meaning that another system, presumably at the musculoskeletal level, controls this sex difference.

1.3.2. Neural circuits that control sex-typical play

The MeA, and the sex steroid-sensitive, sexually dimorphic circuit it is part of, is a prime candidate to control sex differences in play. Known to regulate many adult sociosexual behaviors (Newman, 1999, Simerly, 2002), the MeA has also been implicated in the sexual differentiation of juvenile play. Lesions that encompass the MeA reduce play solicitations to female-like levels (Meaney et al., 1981a) and intracerebral cannula packed with testosterone and implanted into the neonatal MeA masculinize play (Meaney et al., 1981b). Given the latter finding, it is not surprising that the prepubertal MeA is sexually dimorphic. We have shown that the regional volume of the MeA is ∼10% greater in P28 males than in age-matched females (the volumetric sex difference increases dramatically during puberty; Cooke et al. (1999). The volumetric sex difference does not appear to be the result of more neurons. Instead, male MeA neurons have longer dendrites and consequently more glutamatergic synapses per neuron (Cooke and Woolley, 2005, Cooke et al., 2007b). The behavioral relevance of this sex difference, if any, is unknown. One possibility, however, is that more excitatory synapses per neuron increases the representational capacity of the MeA, allowing male rats to identify a greater number of conspecifics, which may confer advantages in its highly competitive behavioral ecology.

Methyl-CpG-binding protein 2 (mecp2) is a repressor of gene transcription that is mutated in Rett syndrome and other sex-linked progressive neurodevelopmental disorders. Kurian et al. (2007) reported that mecp2 expression in the amygdala and ventromedial hypothalamus was greater in postnatal day 1 males than females. Testing whether suppression of this gene could affect sex-typical behavior, they found that suppression in the neonatal amygdala reduced juvenile play levels in males but not in females (Kurian et al., 2008). The same group also found that nuclear receptor corepressor (NCoR), another gene coregulator, controlled juvenile play, but this time in the opposite direction: Suppressing its function in the amygdala with small interfering RNA led to an increase in rough-and-tumble play in males (Jessen et al., 2010). Because the amygdala is too small at P1 to target specific nuclei within it, the authors could not determine where the suppression of these coregulators had their effects, nor could they identify where sex differences in their expression were found. However, the MeA is the only amygdaloid nucleus known to be sexually dimorphic in terms of regional volume, cell size, and neurotransmitter content (Cooke, 2006). Thus, suppressing mecp2 and NCoR most likely acted there to affect the frequency of juvenile play.

In a nucleus usually thought to have more brain matter (i.e., neuropeptides, dendrites, axons, glia, synapses, etc.) in males than females, it was surprising to learn that treatment of neonatal female rats with the cannabinoid receptor agonist WIN 55,212-2 reduced astrocyte proliferation to male-like levels in the MeA and masculinized play behavior (Krebs-Kraft et al., 2010). Given the myriad responsibilities of astrocytes in normal brain function, including glutamate scavenging, synaptogenesis, and synaptic transmission (Ullian et al., 2001, Halassa and Haydon, 2010), consideration of the causal connection between reduced astrocyte proliferation and increased male-like behavior, if there is one, is intriguing. Endocannabinoids acutely affect juvenile play too. For example, treating rats with an anandamide transporter inhibitor, which increases the endocannabinoid content only at active synapses, facilitates play (Trezza and Vanderschuren, 2009). Endocannabinoids also interact in complex ways with the endogenous opioid system to influence play motivation and reward. These phenomena are discussed in an excellent recent review (Trezza et al., 2010).

The last decade has seen a revitalization of interest in sex differences, spurred in part by the prevalence of sex-linked neurodevelopmental disorders such as autism and Rett syndrome, and sex-linked psychiatric conditions such as depression and anxiety. As underlined by the last two findings, we now know that the mechanisms by which the brain becomes sexually dimorphic are vastly more complex than was believed during the late twentieth century when this field was young (Arnold and Breedlove, 1985, McCarthy and Arnold, 2011). From our perspective, there are two equally interesting questions that should be asked about the cellular–molecular mechanisms underlying sex differences: (1) How do gonadal steroids change the brain in such a way as to increase the frequency of rough-and-tumble play, and (2) Does male-typical play make the brain more male-like? The MeA and the sexually dimorphic network with which it is connected seem like a good place to begin asking these questions.

2. Conclusion

The transformation of a child into a sexually mature adult entails a profound re-orientation of social priorities, away from same-sex friendships and toward romantic relationships, usually with the other sex. During puberty, gonadal hormones induce significant morphological and neurochemical plasticity in the MeA of males (Romeo and Sisk, 2001, Cooke et al., 2007a, Cooke, 2010) that is associated with changes in the salience of social stimuli. Certain social cues become more likely to arouse feelings of limerance toward attractive conspecifics, whereas those from a rival become more likely to arouse anger. While gonadal hormones during puberty are important, they are not sufficient to explain these changes in behavior. The resemblance of juvenile play to some aspects of sex-specific adult behavior has led many to speculate that without play, sex-typical adult behavior could not occur. While play is clearly not practice for specific adult behaviors, play may function to educate the young animal about its own sensori-motor and social capabilities, learning its place in the hierarchy before the consequences of social defeat become too dear. Although social deprivation experiments should be interpreted cautiously, studies with males indicate that isolation during the period of peak juvenile play makes rats more fearful, devastates some aspects of adult socio-sexual behavior, and de-masculinizes two sexually dimorphic brain areas. However, as we observed, the presence of an adult conspecific mitigated the effect of peer deprivation on two sexual behaviors. Thus, the function of juvenile social experience generally and play in particular could be to sculpt and refine neural circuits that ultimately control adult social behavior. And because social experience not surprisingly affects the brain and the behavior of females (Bell et al., 2010, Hermes et al., 2010), it is critical to include them in future studies to uncover commonalities and differences between the sexes that may be instructive in terms of understanding the ontogeny of sex-typical behaviors and mental illnesses. That said, although there is no evidence that juvenile social isolation can impair the expression of mating behavior in female rats, social deprivation of female rhesus macaques results in social and sexual deficits (Deutsch and Larsson, 1974, Kempes et al., 2008). Thus, it may simply be that socio-sexual deficits in female rats have never been examined.

Since the work of Stewart and Meaney 30 years ago, we have known that rough-and-tumble play is masculinized, in part, by the action of gonadal hormones during perinatal life. Furthermore, we know that neonatal testosterone specifically in the MeA masculinizes play frequency, while MeA lesions de-masculinize it. These findings indicate that the MeA is a key site for sex differences in play. What perinatal gonadal hormones do in the MeA to increase play is unknown, although recent papers have demonstrated unexpected and intriguing links between juvenile play, the neonatal endocannabinoid system, and glial proliferation, and between juvenile play and neonatal nuclear coregulator expression in the amygdala. Activational effects of gonadal hormones are usually thought to occur during puberty and into adulthood, but we have found that prepubertal gonadectomy diminishes juvenile play motivation, and may do so via a loss of synapses in the MeA.

Whether or not rough-and-tumble play promotes brain and behavioral masculinization remains an open question. We do not have evidence that clearly demonstrates the necessity of play-induced plasticity to the expression of socio-sexual behavior. It is certain, however, that play experience increases the expression of genes that are known to increase plasticity. Endogenous opioids regulate the expression of rough-and-tumble play while simultaneously increasing BDNF in the prefrontal cortex. The expression of these genes in the MeA, BNST, lateral septum, medial prefrontal cortex, and periaqueductal gray, among many other areas, suggest that juvenile play remodels synapses in the ‘social behavior network’ that ultimately enables adult animals to live together so that they may struggle for dominance, mate, and raise their offspring.