We assessed sex differences in monkeys’ toy preferences, as was done previously [9, 10], but using a choice test paradigm with socially isolated animals, a novel experimental approach in this scientific domain. Rhesus monkeys were tested with toys that we assigned to different gendered categories (i.e., “masculine”, “feminine”, “neutral”, “ambiguous”) based on the previous study in rhesus macaques [10]. Overall (considering all trials), male and female monkeys had similar likelihoods of interacting with different types of toys: both sexes preferred neutral and masculine toys, more than ambiguous and feminine toys. Both males and females also had a greater preference (as operationalized by likelihood to interact) for toys made of hard material as compared to plush toys, and for non-zoomorphic over zoomorphic toys.

When they interacted with the toys, males interacted more with the doll than with most toys and interacted more with the doll than the females did. This was the only between-sex difference that emerged in our study. When they interacted with the toys, both males and females interacted more with neutral than with masculine toys (hard vehicles) but while this was the only preference expressed by males related to gendered categories, females showed more preferences and interacted the least with feminine toys. Finally, females interacted more with toys made of hard materials than with plush toy while males did not show preferences for materials. Multiple aspects of our results contrast with the two previous studies in monkeys (vervets: [9], rhesus: [10]). In particular, our results contrast with the previous study in rhesus monkeys [10] that reported male monkeys had preferences for masculine toys and female monkeys had no preference for either masculine or feminine toys.

Unlike the previous studies testing toy preferences in monkeys, we used a choice test in which we presented two toys simultaneously to each subject monkey. Such methods are recommended if the goal is to measure preferences [43] because it makes animals pick between options (rather than quantifying “preference” based on time spent or frequency of interaction with a single object, and then comparing those metrics across objects). Forced choice paradigms like this have been used to test human toy preferences, revealing stronger typical sex differences than other methods [23]. This paradigm is very different from that used by Alexander and Hines [9] who presented only one toy at a time and compared the time vervets spent manipulating each object. However, [10] presented two toys simultaneously as we did on our study. We discuss hereafter multiple possible explanations for why our results do not match the results found in their study and more broadly, do not match the typical pattern of sex differences in toy preferences in humans [11, 23].

Our study evaluated preferences for a larger variety of toys and toys categories than did Hassett et al. [10] and used a slightly different analysis plan to estimate preferences through both the likelihood to interact with the toys and the number of interactions. To make sure discrepancies in the analysis plan and choice of toys did not drive the differences between the studies, we conducted an additional analysis on the average number of interactions for individuals who interacted at least 5 times with toys during the trials presenting only masculine and feminine toys (respectively wheeled and plush toys) mirroring the analysis approach from Hassett et al. [10]. This analysis did not reveal the pattern identified in Hassett et al. [10] (Fig. 7) suggesting that the analysis strategy itself is not the cause of across study differences.

Fig. 7 Mean (± SEM) number of interactions with masculine and feminine toys from male and female rhesus monkeys tested in group (A adapted from [10] or isolated, B the present study). We modified our analysis to match with the study by Hasset et al. [10]: we limited our study to trials offering choices between wheeled (masculine) and plush (feminine) toys only. All our animals generated more than 5 interactions with the toys overall (criteria of data inclusion in the previous study [10]), and were consequently all included in the analysis. Bars that share a letter do not differ significantly from each other. Bars that do not share a letter differ from each other and denote significant difference Full size image

The experimental design of our study followed the toy classification made by Hassett et al. [10], as this is the only study that properly tested preferences of toys in nonhuman primates. We consequently classified wheeled toys as male toys and plush toys as female toys. The classification of the different toys to masculine and feminine categories is not standardized among authors [23]. While vehicles are generally considered as male toys, classification for plush toys is more controversial (e.g., [42]) and past studies were equally likely to classify them as female toys or as neutral toys [23], except plush dolls, typically considered female toys [11, 23]. A wrong a priori classification of toys is at risk of misleading conclusions, but we argue here that our results would not be fundamentally different would we have decided to consider plush toys as neutral toys. Indeed, males were not more likely than females to interact with vehicles and when they did interact with vehicles, they did not interact more than females did. Females were not more likely to interact with the doll than males did, and when they did interact with the doll, they interacted less than males did. Therefore, our results would still contrast with the typical pattern of sex differences in toy preferences in humans would we have decided to classify plush toys, other than the doll, as neutral toys.

Another methodological difference between studies was that in the Hassett et al. [10] study, animals could freely interact with the toys while the toys could not be moved from their initial locations in our study. Toys in our study could be physically manipulated but remained in the small mesh compartment. This could have reduced the ability of animals to generate certain exploratory behaviors and potentially affected our measures of preferences through the number of interactions with the toys. For instance, the propensity of males (both human infants and non-human primates) for vehicle-like toys has been posited as relying on their preference for object that afford propulsive movement [9, 10, 50]. Female monkeys have been proposed to show preferences for objects that resemble features of monkey infants [51] due to their greater interest in younger infants [52]. In the wild, young chimpanzee females tended to carry sticks in a way that suggests a rudimentary doll play more often than males did [53]. In this view, sex differences may not have emerged in our study because while monkeys could touch and manipulate the toys, they could not propel them through space or cradle them. This idea is called into question by the fact that studies in humans have demonstrated that the emergence of sex differences in toys does not require the motor activity associated with the toy [31]. Further, differences in the ability to propel objects does not explain why the most potent sex difference that we identified was that males manipulated the doll more than females.

For a variety of reasons, we also believe that it is unlikely that the limited access to the toys affected likelihood of interacting at all with the toys—which was our measure of preferences. Whether or not the monkeys interacted at all with the toys should be contingent more on the visual attractiveness of the different toys than their ability to propel them (which would have a greater impact on longer lasting manipulation behaviors). This index revealed no preference of males for vehicles and no preference of females for plush toys (or for the doll, with a more restrictive categorization) in our monkeys. This result is inconsistent with results found in human infants showing visual preference for sex-congruent toys as early as 3–8 months old [31]. While infants at this age are posited to not seek conformism with external referents, they are nevertheless exposed during infancy to gender congruent toys which might have an experience-dependent effect on their visual preferences [31, 54]. Therefore, it is possible that the visual preference for sex-congruent toys in humans relies more on such socially induced familiarity with specific toys than on biologically rooted sex differences in visual preferences for toys characteristics. Supporting this hypothesis, when 1 year old children are tested for visual preference for shape or texture rather than for specific toys, they do not show sex differences [55]. Similarly, we did not find sex differences in the likelihood to interact with the toys based on toys characteristics, consistently with previous work in nonhuman primates [51].

We propose that the emergence of human-typical sex differences in nonhuman primate’s toy preferences depend on the social context in which interactions with toys is observed. The human sex difference pattern [11], previously observed in nonhuman primates (rhesus: Hassett et al. [10],vervets: Alexander and Hines [9]) emerged when group housed monkeys were tested in social groups. In contrast, our group-reared, pair-housed monkeys were tested alone and did not generate this pattern of behavior. This could have impacted preferences by influencing who can access the toys (e.g., a particular type of toy could be monopolized by some members of the group) or because individuals from one sex are less likely to compete for it. Dominance status, for instance, is likely to influence access to the different toys. In the previous study of rhesus, higher ranking monkeys interacted more with toys and the females with no toy preferences were the lowest ranking [10]. Similarly, higher ranking vervets, regardless of sex, tended to interact more with masculine and less with feminine toys than lower ranking individuals [9]. Based on how we tested animals, the group dynamics that impacted the results of the previous studies were absent in our study.

Access to toys is not the only reason that group testing might influence behavioral outcomes. Another possibility is that the group social context is necessary for the emergence of human-typical sex differences in toy preferences because males and females do not differ in their individual preferences for the toys but differ in their preferences for the social activities that are or could be facilitated by the toys. Toy preferences have been proposed to parallel these social activity differences in humans [56] but also in nonhuman primates. According to this hypothesis, male monkeys prefer objects allowing for propulsion due to their generally more social active playstyle [57,58,59]. Therefore, preferences are not observed if the social activity cannot be expressed.

Finally, the social environment of our monkeys could have impacted our results. The monkeys in the previous studies lived and were tested in large social groups, which can create conditions in which the social behavior of males and females is fairly different. The dynamics of rich social groups are said to maintain such social differences while sex differences are reduced when animals are housed in mixed sex dyads [60]. Therefore, the social conditions in which the animals in the present study were housed may have contributed to our effects.

On a final note, our observations highlight another novel result: the only between-sex difference we observed is that, when males interacted with the doll, they interacted more with it than females did. One argument is that a lack of interest of the females for the doll could be a result of their reproductive history since our females had infants in the past which could have led to a decreased interest in play-mothering activities [53]. However, females and males were equally likely to interact with the doll, suggesting a similar visual interest. Additionally, most of the monkeys included in the two previous studies were old enough to be parous and authors did not report any effect of age or parity on toy preferences [9, 10]. The higher number of interactions of males with the doll is a new result that needs to be replicated by further studies in nonhuman primates assessing toy preferences in different social conditions.

Perspective and significance

Our study found that males and females rhesus monkeys tested alone did not show the same pattern of toy preferences as young male and female children or as monkeys in previous experiments. This result calls into question the existence of a strong biological basis supporting sex differences in toy/object preferences in humans. The existence of a strong biological basis to sex differences in toy preferences in humans is mainly supported by visual preferences for gender congruent toys at a very young age [31] and preference for masculine toys in young girls with CAH [29]. However, these latter findings can also be largely driven by social factors. Young boys and girls can develop differences in visual preferences at a very young age due to divergences in the environments that parents design for their sons and daughters [55]. Similarly, young girls with CAH preferences for boy toys could be induced by their preference for playstyles preferred by males and for male playmates, which could also socially induce their preference for masculine toys. Further investigation is required to better understand how the past and current social environment influence the expression of sex differences in primates, including humans.