Broadly seen in most mammalian species from rodents to humans, social play (also known as rough-and-tumble play or play-fighting) has fascinated neuroscientists and evolutionary biologists alike for decades. Though animals spend upwards of 20% of their time participating in play during adolescence (Pellis and Pellis, 2009), this well-conserved behavior serves no obvious purpose, yet appears fundamental for appropriate development. Various studies have found that juvenile social isolation in rats leads to impairments in social behavior (Hol et al., 1999; Van den Berg et al., 1999; Von Frijtag et al., 2002), cognition (Einon et al., 1978; Baarendse et al., 2013; Yusufishaq and Rosenkranz, 2013), and sexual behavior (Gerall et al., 1967; Cooke et al., 2000), increases anxiety- and depression-like behavior (Parker and Morinan, 1986; Wright et al., 1991; Arakawa, 2003; Leussis and Andersen, 2008; Lukkes et al., 2009; Cuesta et al., 2020), and impacts susceptibility to addiction-related behaviors (Whitaker et al., 2013; Baarendse et al., 2014).

While these studies provide insight on the impacts of juvenile social isolation writ large, there are three important caveats that preclude the ability to apply them to assess the importance of social play specifically. First, isolation prevents all social interaction—both play and non-play—so it is unclear in most of these studies whether and to what extent any identified impairments are due to the lack of play vs. the lack of general social interaction, as well as the additional stress that total isolation induces (Begni et al., 2020). Second, in many of these studies, animals were isolated through adulthood, when testing was conducted; as such, it is also unclear whether and to what extent there may be an effect of acute isolation on the observed phenotypes. Finally, and most notably, the majority of these studies were conducted in only one sex—typically, male subjects—thereby preventing the ability to assess whether there are sex differences in the identified impairments.

This final caveat is of critical importance given there are robust sex differences in social play that are seemingly as well-conserved as play itself. Across nearly all species that play, from rodents to humans, male juveniles play more frequently and intensely than females (see VanRyzin et al., 2020a for review). Additionally, there are known sex differences in the qualitative characteristics of social play interactions (Pellis and Pellis, 1990, 1997). Studies investigating the neural underpinnings of this behavioral sex difference have identified various nodes within the larger circuitry of social behavior, including the medial amygdala (VanRyzin et al., 2019) and lateral septum (Bredewold et al., 2014), which exert sex-specific influences on play following sexual differentiation early in life as part of typical brain development. Indeed, deficits in play are core symptoms of neurodevelopmental disorders like autism spectrum disorder, attention-deficit/hyperactivity disorder, and early onset schizophrenia (Alessandri, 1992; Jones et al., 1994; Møller and Husby, 2000; Jordan, 2003; Helgeland and Torgersen, 2005), many of which also exhibit robust sex differences in diagnosis and symptomology (Aleman et al., 2003; Ramtekkar et al., 2010; Arnett et al., 2015; Halladay et al., 2015; Giordano et al., 2021; Prosperi et al., 2021). The robust and conserved nature of this sex difference, then, speaks to its importance and begs the question: does play serve a different purpose in males compared to females?

Here, we investigate the impact of juvenile play experience on later-life behaviors and how this may differ as a consequence of sex. Juvenile rats of both sexes were deprived of play via one of two different methods or were housed under normal conditions, i.e., controls. For the first play deprivation method, we created perforated Plexiglass cage dividers (“play barriers”) that could be placed into standard home cages to physically separate the two juvenile rats inside (Figure 1A). Improving upon the previous methodology, this manipulation prevents play while still allowing for other forms of social interaction in the visual, auditory, olfactory, and tactile domain. Interaction with a conspecific across a similar physical barrier has been shown to be socially rewarding (Peartree et al., 2012) and to reduce anxiety-like behavior as compared to full isolation, an additional benefit (Klapper-Goldstein et al., 2020). For the second play deprivation method, we eliminated all social interactions (play and non-play) by socially isolating animals via single-housing, as done by others. Animals were placed in these or control housing conditions as juveniles, then re-housed in standard group housing around puberty, after which we assessed the impact on various adult behaviors. Supporting our hypothesis, we found that social play experience impacts later-life endpoints in a sex-specific manner. Play deprivation preferentially impacted behaviors within the socio-sexual domain, decreasing sexual behavior, increasing sociability, and increasing aggressiveness in adulthood in males but not females, providing valuable insight into sex differences in the function of this fundamental adolescent behavior.

To ensure our manipulation did not induce overt behavioral pathology on its own, we began by assessing the acute effects of play deprivation. During the juvenile age (at either P26 or P21 for the open field and playfulness tests, respectively), same-sex sibling pairs of rats of both sexes were placed into control (CTRL) or barrier (BARR) cages or were single-housed in isolation cages (ISO). One week later, we assessed anxiety-like behavior via the open field test. In a separate cohort of BARR and ISO animals, we also assessed playfulness. We found no effect of sex or housing condition on center time in the open field (Figure 1B). As seen in previous studies of juvenile isolation (Panksepp and Beatty, 1980; Ikemoto and Panksepp, 1992), we found that the ability and motivation to participate in play was not only intact in BARR and ISO animals but was increased relative to our historical data for control-housed animals (Figure 1C; Argue and McCarthy, 2015). Interestingly, when we assessed the BARR and ISO play data, we detected a significant main effect of both sex (p = 0.036) and treatment (p = 0.026), with males of both housing conditions and ISO animals of both sexes playing significantly more than females and BARR animals, respectively. However, this effect of the condition seems to be driven more strongly in females than in males. Although the sex × condition interaction did not reach statistical significance (p = 0.13), the mean difference in play levels between female BARR and ISO animals (12.7 vs. 29.0 play events) is much larger than that of male BARR and ISO animals (28.6 vs. 31.9 play events). This suggests that the increased social interaction allowed by the BARR housing condition is sufficient to reduce play motivation in females but not males, an unexpected finding speaking to the differential role that play may serve in females compared to males. Overall, these initial studies indicate that the modified housing used to prevent play does not in itself increase anxiety-like behavior or impair the ability to play if given the opportunity and provides intriguing evidence in support of the idea that play may serve different purposes across the sexes.

For adult behavioral experiments, rats were weaned on P21 into same-sex sibling pairs. In the short duration barrier experiments (SDB), rats were then placed into barrier or isolation cages starting on P26 or remained in control housing conditions (Figure 1D). Animals remained in these conditions until P40, a 2-week period encompassing the peak of the window in which social play is maximally expressed (Panksepp, 1981), after which they were returned to standard group housing for the remainder of the experiments. To determine whether there were any sex differences in resiliency to play deprivation, we repeated our behavioral assays in the long duration barrier experiments (LDB). In this study, animals were placed in control, barrier, or isolation conditions for as long as could be tolerated based on size: starting at weaning (P21) and lasting until P45, an age when rats of both sexes have typically entered puberty, at which point playfulness dramatically decreases (Meaney and Stewart, 1981; Panksepp, 1981). Given the relatively short period of time in which playfulness is seen in rats, the additional 10 days of play deprivation in the LDB study (an additional 5 days before and 5 days after the SDB deprivation period) represents an appreciable expansion in play deprivation time as compared to the SDB study, while maintaining specificity to the time period in which play is the predominant social behavior.

Here, we report persistent alterations in the expression of various adult behaviors following juvenile play deprivation in rats. We demonstrate that social play experience is critical for the development of appropriate socio-sexual behavior in males, as males prevented from playing as juveniles exhibited altered sexual behavior, sociability, and aggressive behavior. These effects are complex and pervasive, seen weeks to months after all animals were returned to group housing, and impact reproductively critical behaviors. In females, however, play is more dispensable. While rats of both sexes were hyperactive shortly after play deprivation and displayed impairments in a test of empathy motivation when acutely prevented from playing, we observed no effects in females on any other test in the diverse battery we conducted.

Given the importance for reproductive fitness, the most striking of our findings is the lasting effect of play deprivation on male sexual behavior. Males prevented from playing for as little as 2 weeks (in the SDB study) showed a significant reduction in mount rate across two tests of sexual behavior. This was seen in both BARR and ISO males, indicating that the lack of play, not the lack of social interaction writ large, was the critical driver of this impairment. Moreover, this impairment did not seem to represent a simple delay in the learning process, but rather a persistent reduction in the rate of copulatory behavior; while we present the data here as an average, the deficit in BARR and ISO males was evident across both successive tests. Furthermore, males prevented from playing for as little as 3.5 weeks (in the LDB study) also exhibited deficits in ejaculation. Around 40% of BARR and ISO males did not ejaculate in either test of copulatory behavior, a significant deviation from what was seen in CTRL males (12.5%) and from the expected distribution of non-ejaculators (~10%) based on decades of research on male rat sexual behavior (Dewsbury, 1972; Pattij et al., 2005; Portillo et al., 2006). Remarkably, these impairments were close to the magnitude seen in other studies in which male copulatory behavior is altered by more substantial pharmacological or genetic manipulations, despite the fact that our manipulation merely prevented juveniles from playing for a short window of time. For example, the decrease in the number of mounts we observed was of similar magnitude to that seen in males treated with a cyclooxygenase-2 inhibitor for 2 weeks during the perinatal sensitive period (Amateau and McCarthy, 2004), and the reduction in the percentage of animals which successfully ejaculated we observed was similar to that seen in males in which both the androgen receptor and estrogen receptor alpha were genetically deleted (Trouillet et al., 2022).

We also found that play deprivation induces hypersociability in adulthood. In the SDB study, BARR males exhibited social preference that was significantly higher than that seen in all other groups. This hypersocial phenotype was seen across multiple SDB cohorts, and the effect size was substantial (Cohen’s d ranging between 1.9 and 2.9 for all pairwise comparisons between Male BARR and the other groups for percent of time near the social box; Figure 6A). However, there was no effect on social preference in the LDB animals. This differential effect in the SDB compared to the LDB study may be due to an as-yet-unidentified sensitive period for social development around the time of weaning. Work from Hol et al. (1999) demonstrated that social isolation from P22–28, but not P29–35, decreases sociability in adulthood in male rats. Isolation across the full window, from P22–35, also decreased sociability. This finding was mirrored by a study in females which found that isolation from P19–70 also decreases sociability (Hermes et al., 2011). In contrast, isolation starting later in adolescence, from P28–70, was shown to produce hypersociability in both sexes, although this study was conducted in mice (Rivera-Irizarry et al., 2020). Looking at the intersection of these studies, it is possible that two distinct sensitive periods exist, perhaps governed by the maturation of distinct sets of social circuitry: one window from ~P19–28, in which the lack of play experience causes hyposociability, and another starting at ~P35, in which the lack of play experience causes hypersociability. In this view, it is possible we observed no effect on sociability in the LDB study as the play deprivation spanned across those windows (from P21–45), canceling out any effects. It is also possible that the additional stress caused by full isolation may alter the direction and magnitude of these effects, given ours were seen only in BARR, but not ISO, males. Future studies should more closely investigate the possibility of these distinct sensitive periods and whether they may differ based on sex, given the hypersocial phenotype we observed was seen only in males.

One of many theories regarding the function of social play is the “play-as-practice” theory, which argues that the individual behaviors constituting social play (e.g., pounces, pins, and boxing behaviors in rats) resemble adult agonistic behaviors for a reason: because play serves as practice for later-life aggressive encounters (Fagen, 1981; Smith, 1997). This theory has received criticism for various reasons, primarily that there are several key differences in the microstructure of juvenile play-fighting and adult aggressive behaviors, thus hindering play’s applicability as a means of physical “practice” (Pellis and Pellis, 1987). We here provide further evidence against the play-as-practice theory, as we demonstrate that juvenile play deprivation increases aggressive behavior in adult males, not decreases as a straightforward interpretation of the theory would predict. Nonetheless, a more complex interpretation of the theory may be better justified. While it seems evident that play does not serve as simple practice for the motor skills needed for later-life agonistic behavior, perhaps what is gained from play is experience in its reciprocal nature. This reciprocity—necessitating monitoring one’s own actions and those of a partner, recognizing social signals indicating the partner’s responses and intentions, and adjusting one’s actions appropriately to maintain the back-and-forth of the play bout—is a fundamental feature of social play (Pellis and Pellis, 2017), so play may serve to provide experience in these social skills, allowing for the maturation of brain circuitry to the same effect. Animals deprived of play may lack this experience, thereby hindering their ability to appropriately recognize and/or respond to social signals to de-escalate an agonistic encounter, resulting in increased aggression as we observed. Indeed, recent work in Drosophila identified a transcriptional regulator believed to govern this same effect, suppressing aggression in a social experience-dependent manner (Ishii et al., 2022), and long-term post-weaning social isolation has been shown to induce qualitative differences in the attack patterns of male rats (Toth et al., 2011).

We also conducted a test of prosocial motivation in animals acutely deprived of play starting a week before and continuing throughout the 12 days of testing. BARR animals of both sexes performed worse on this task, taking significantly longer to free their distressed cagemate from a confinement box on the first 6 days of testing. This impairment was resolved by the latter half of testing, with the performance of play-deprived animals improving to that seen in controls on days 7–12 of the paradigm. This finding may serve as another example of the concept that play aids in developing the ability to appropriately sense and respond to social signals. In this view, BARR animals may perform worse on the empathy test due to a deficit in their perception of and/or responsiveness to the social signals generated by their distressed cagemate. This impairment may have improved over time due to the increased social experience afforded as part of the test (while animals were allowed only 10 s to interact with their cagemate following release from the confinement box, this social experience may have accumulated over time), or because animals had sufficient experience with the paradigm by the latter half of testing such that it allowed BARR animals to “catch up” to controls. We also observed a sex difference in this assay: males generally performed worse than females, as they exhibited a significantly higher latency to release their cagemate on average in the first half of testing. This too may speak to the heightened importance of play experience in males compared to females, given even at baseline males are less successful on this socially-driven task.

Importantly, we saw no effects of play deprivation on any other assessed behavior, with the exception of a small but significant effect on locomotion shortly after animals were returned to standard group housing. Our findings in which we observed play deprivation had no effect—on anxiety-like behavior, object memory, sex preference, and social recognition in both sexes and on sex behavior and social preference in females—are just as informative as our findings in which we did observe an effect, in that they show what play experience does not seem to be important for. The effects of play deprivation (and therefore, the importance of play itself) seem to preferentially center around socio-sexual behaviors in adulthood, a concept further supported by our factor analyses, which in an independent and unbiased manner identified factors in both the SDB and LDB studies that represent socio-sexual behavior and significantly differed in control and play-deprived males.

This specificity to socio-sexual behavior makes sense, given play is itself a social behavior and involves the activity of many regions in the social circuitry, such as the medial amygdala, lateral septum, and hypothalamus (summarized in Siviy, 2016 and VanRyzin et al., 2020a). These regions have distinct roles in socio-sexual behavior in adulthood as well, often driven by distinct, separable cell type-specific and projection-specific mechanisms (Unger et al., 2015; Wong et al., 2016; Ishii et al., 2017; Kohl et al., 2018). The maturation of these cell subtypes and/or projections, then, may be modulated in an experience-dependent (and, potentially, sex-dependent) manner based on playfulness. Like the importance of visual experience for the development of ocular dominance in V1 neurons (Wiesel and Hubel, 1963) or somatosensory experience for dendritic spine dynamics in the barrel cortex (Lendvai et al., 2000), play experience may be fundamental to the developmental firing and wiring of individual cells and projections that establishes appropriate adult circuit organization. However, the fact that the impairments we observed following play deprivation were generally specific to males begs the question: in this view, what, if anything, governs social circuit maturation in females? Does social experience (play or non-play) have any role? In our initial studies, we found that the additional non-play social experience provided by the BARR vs. the ISO housing condition is sufficient to decrease play motivation in females but not males (Figure 1C), suggesting that play itself is the critical feature driving the motivation to play in males. Perhaps in females, the significance of play is more so the social experience in general, and less so the experience of play itself. Even then, it appears little social experience during the juvenile window is necessary for appropriate social development in females, given we observed few impairments following even full social isolation. Future studies should investigate whether the window for experience-dependent social circuit organization is shifted in females compared to males, or if, perhaps, female social circuit maturation is not experience-dependent at all.

While our BARR manipulation improves upon previous methodology in that it prevents play while allowing for other forms of social interaction, there are nonetheless limitations. BARR animals are also deprived of other tactile non-play social interactions such as social grooming, and there may be other variables such as the difference in physical space allowed by the BARR vs. CTRL and ISO conditions that may affect development. Nevertheless, the convergence of evidence from our study and others using different manipulations (e.g., reducing play by rearing juveniles with an adult or a peer from a less playful strain) points to play experience being the critical driver of the impairments observed (Schneider et al., 2016; Burleson et al., 2016; Pellis et al., 2017; Stark and Pellis, 2020, 2021). Additionally, while females appeared largely resilient to play deprivation across the myriad of behavioral assays we conducted, this may not be true for all traits, or there may be latent differences in our assessed traits not detectable by the paradigms we used to evaluate them. Regardless, this female resilience is notable, supporting our hypothesis that the purpose of play differs across the sexes, at least in the rat.

Embedded within our studies is also a deep assessment of sex differences and similarities across a wide variety of behavioral tests in rats. When possible (i.e., in assays that are not only applicable to animals of one sex), males and females were included in all experiments, all of which were adequately powered to detect sex differences if they were present. We statistically tested sex as a variable, calling a finding a “sex difference” if and only if we detected a significant main effect of sex or a significant interaction. This was as opposed to independently assessing males and females and noting a sex difference if we observed an effect in one sex but not the other, a common misconception of the appropriate way to assess sex differences (Garcia-Sifuentes and Maney, 2021). As the research community works to better address sex as a biological variable (SABV), especially following the implementation of the United States National Institutes of Health’s 2016 policy requiring consideration of SABV in all funded research, we hope this work is useful to behavioral neuroscientists interested in the differences and similarities in male and female rats across various behavioral domains.

Finally, our results are especially relevant in the wake of the COVID-19 pandemic. We began these studies shortly before shutdowns began in early 2020, impacting access to play for millions of children worldwide, and as such our experiments became even more relevant. While our studies were conducted in rats, social play is remarkably similar across the many mammalian species in which it is expressed and has strong face validity in humans. Our results would argue that boys may be more vulnerable to a reduction in access to play than girls. This potential sex difference in the influence of the pandemic on child health and brain development is something that should be monitored as we collectively move forward.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The animal study was reviewed and approved by Institutional Animal Care and Use Committee, University of Maryland School of Medicine.

AM and MM designed research and wrote the manuscript. AM, JV, and RF performed research. AM and JV analyzed data. All authors contributed to the article and approved the submitted version.