Puberty, as a turning point of adolescent period, is accompanied by prominent physiological (e.g., sex hormone changes and menstruation), physical (e.g., secondary sex characters and height) and psychosocial (e.g., gender role stereotyping) changes (Patton and Viner, 2007; Blakemore et al., 2010; Marceau et al., 2011; Guyer et al., 2014). Each of them has been considered a potential risk factor for adaptation stress and affective disorder during adolescence (Hankin and Abramson, 2001; Ge et al., 2003; Spear, 2009; Duke et al., 2014). In fact, epidemiological studies have consistently shown a higher prevalence of emotional disorders in adolescents relative to children, such as depression, panic disorder, social anxiety, obssessive-compulsive disorder and so on (Hankin, 2006; Patton and Viner, 2007), and this prevalence is higher in girls than in boys (Kessler et al., 1993; Nolen-Hoeksema and Girgus, 1994; Hankin and Abramson, 2001).

However, the mechanisms underlying this phenomenon have not been sufficiently studied. A number of observational studies have suggested a couple of psychosocial factors that contributed to increasing incidence of emotional disorders in adolescence, such as gender role intensification (Aubé et al., 2000), body image concern (Hankin and Abramson, 2001), parent-offspring conflicts (Laursen et al., 1998), interpersonal stress (O’Shea et al., 2014) and academic stress (Quach et al., 2015). While these studies have found that these factors are associated with affective disturbances, and these conclusions were only drawn from the observation designs and psychometric data, such as ICD-10, SCL-90, etc. (Bülow et al., 2002; Vicente et al., 2012; Wesselhoeft et al., 2015). Though observational design is important in showing links between potential contributors and mental health, the lack of experiment manipulation made it hard to answer by causation how puberty and sex interact to influence adolescents’ susceptibility to affective disturbances. Previous studies suggest that the emotional susceptibility was closely related with affective disturbances (Hofer et al., 2006; Yuan et al., 2009), and this susceptibility is embodied by one’s brain sensitivity to emotional stimuli (Carretié et al., 2004; Hofer et al., 2006; Williams and Gordon, 2007; Yuan et al., 2009). In this regard, it is necessary to design an experiment exploring the interactive effects of sex and puberty on brains’ susceptibility to emotional stimuli. However, currently few studies have directly investigated this issue, despite existence of several relevant studies.

For instance, using startling EMG activity for eyeblink as an index of defensive motivation, Quevedo et al. (2009) showed that the mid-/late pubertal adolescents showed enhanced startle reflex amplitudes compared to pre-/early pubertal adolescents, irrespective of picture valence. This suggests that pubertal development is linked with enhanced defensive motivation, which predicts greater levels of fear and anxiety (Pine et al., 1998). However, the lack of behavioral and brain activity measures made it unable to depict a comprehensive profile of how emotional sensitivity varies across puberty and sex. In addition, Fujisawa and Shinohara (2011) investigated sex differences in the recognition of emotional prosody in late childhood and adolescence, and the authors observed that girls were more sensitive to sad and happy prosodies than boys in adolescence but not in childhood. However, this study used an affect recognition task that measured one’s recognition of facial expressions, leaving one’s natural emotional reaction to evocative stimuli undetermined. Moreover, using functional magnetic resonance imaging, Hardee et al. (2017) observed reduced amygdala and precentral gyrus activation for negative vs. neutral words with age increase (ranging from 8.5 to 17.6 years) in males but not in females, which was coupled with age-related increase in internalizing symptomatology experiences for females but not for males. More relevant to the current study, using a cross-sectional design and the time-frequency analysis of EEGs induced by negative pictures, a prior study in our lab has showed that pubertal transition was associated with enhanced gamma oscillations for negative pictures in girls but not in boys, in line with the epidemiological reports of increased prevalence of affective disorders in girls than in boys after entry into puberty (Yuan et al., 2014). Since gamma oscillatory activities were proved as an index of the emotional arousal effect (Bastiaansen and Hagoort, 2006; Balconi and Lucchiari, 2008; Balconi et al., 2009), this result suggests that pubertal development has enhanced girls’ emotional arousal level for negative pictures compared with boys. However, this work did not include positive stimulation, leaving it unknown how positive emotional sensitivity varies across puberty and sex.

Thus, the present study explores how puberty and sex interact to influence the humans’ susceptibility to emotionally negative and positive stimuli, using both behavioral (e.g., experiment-induced mood) and event-related potentials measures. To better assess emotional susceptibility in the behavioral level, we used a block-wise design wherein only emotional stimuli of a specific valence (positive or negative) were presented in a given block. Then, self-ratings of mood were collected before and after either experimental block, to examine how either experimental procedure alters one’s mood and how this modulation varies with puberty and sex. On the other hand, in neurophysiological levels, we varied the valence strength of emotional stimuli in either category since emotion-related individual differences are manifested by both response magnitude differences and different threshold of emotion induction (Yuan et al., 2009, 2012; Lou et al., 2016; Lu et al., 2016). In order to increase ecological validity of emotional induction, we used an implicit task wherein emotional stimulus was presented infrequently and unpredictably, and emotional assessment of the stimulus was not overtly requested (Yuan et al., 2009; Lou et al., 2016).

On the other hand, emotion-related gender differences were mainly manifested by three aspects in prior studies, from attention allocation (Campanella et al., 2004; Yuan et al., 2009); cognitive evaluation (Rhudy and Williams, 2005; Maffei et al., 2015) to late emotional arousal/experiences (Maffei et al., 2015). For example, females, instead of males, probably allocated more attentional resources to the Mildly Negative (MN) stimuli, a processing step that has been shown to be represented by centrally-peaking N2 or P3a in brain potentials during an oddball task (Delplanque et al., 2005; Yuan et al., 2009). In particular, P3a activity was proved as involuntary attention orientating to biologically important, salient stimuli (Friedman et al., 2001; Delplanque et al., 2005; Yuan et al., 2008). In addition, it has been shown that women tend to give a more negative evaluation to ambiguous emotional pictures compared with men (Krohne and Hock, 2008), and this evaluative processing is often illustrated by P3b activity when ERP measures are used (Ito et al., 1998; Delplanque et al., 2005). Also, it has been shown that the same threat cues, like negative film clips, elicited enhanced emotional arousal in women than in men (Rhudy and Williams, 2005; Maffei et al., 2015), and the emotional arousal has been established to covary with the amplitudes of Late Positive Potentials (LPPs) in ERP (Moser et al., 2006; Foti and Hajcak, 2008; Krompinger et al., 2008). Based on these considerations, the current study hypothesized that sex and puberty differences in susceptibility to emotional stimuli would be reflected by attention allocation (N2/P3a), evaluative processing (P3b) and later emotional arousal/experience (LPP) components.

As prior studies have shown that more advanced pubertal status is associated with increased incidence of various affective disorders and this modulation is more pronounced in girls (Hankin and Abramson, 2001; Hankin, 2006), we predict that mid-/late relative to pre/early pubertal subjects would show more pronounced emotion effects for negative and less pronounced emotion effects for positive stimuli in self-rated mood and in attentive (P3a), evaluative (P3b) or experiential (LPP) processing stages, and these pubertal effects would be more robust in girls than in boys. Additionally, accumulating evidences have shown that positive and negative affects are functionally independent in assessing one’s susceptibility to affective disturbances (Larsen and Ketelaar, 1991; Crawford and Henry, 2004; Ding et al., 2015). For instance, the reduced positive affect is a more sensitive predictor for depression compared to the increased negative affect (Dyck et al., 1994; Jolly et al., 1994; Crawford and Henry, 2004), while anxiety is linked with the increase of negative affect and seldom changes of positive affects (Dyck et al., 1994; Jolly et al., 1994). Hence, the current study examined the sex and pubertal effects in sensitivity to positive and negative stimuli, separately, to depict a unique profile for positive and negative emotion varying as a function of sex and pubertal development.

Prior studies have extensively investigated the profiles of brain development (both gray and white matter; Paus et al., 1999; Somerville et al., 2010) during adolescence and how physiological maturation may interact with psychosocial factors in the occurrence of adolescent psychopathology (Paus et al., 2008). Nevertheless, how pubertal development may interact with sex in modulating brains’ susceptibility to emotional stimuli and its neurophysiological correlates have yet to be systematically studied. The examination of this issue is helpful to the understanding of the sex-related prevalence of affective disturbances from adolescence. To this aim, the current study used ERP technique and manipulated the emotion intensity of positive and negative stimuli in an implicit emotional task. The results show, from behavioral and brain potential levels, that pubertal development is associated with reduced reaction to pleasant stimuli across sexes. Moreover, puberty enhanced attention bias for negative stimuli in females but not in males, though females exhibited enhanced cognitive and experiential sensitivity to negative stimuli than males, regardless of puberty.

First, irrespective of pubertal status, girls rather than boys showed significantly more pronounced P3b amplitudes for MN compared to neutral stimuli, while the emotion effect for HN stimuli in LPP amplitudes was more pronounced in girls than in boys. These results suggest that girls weigh aversive, threatening contents more heavily and elicit higher experiential arousal to these stimuli than boys, irrespective of puberty. Though there was no female preponderance in the overall prevalence of affective disorders during childhood (Nolen-Hoeksema and Girgus, 1994; Wichstrøm, 1999; Wesselhoeft et al., 2015), epidemiological studies have consistently shown more incidence of phobia-related and anxiety disorders in girls compared to boys during prepubertal, 6–9 year old childhood (Almqvist et al., 1999; Kroes et al., 2001), which are characterized by heightened responses to uncertain or actual threats (Pflugshaupt et al., 2005; Grupe and Nitschke, 2013). These evidences are consistent with our findings that girls exhibited higher P3b and LPP amplitudes for aversive stimuli than boys, irrespective of pubertal status.

Second, the results showed that pre-/early but not mid-/late adolescents exhibited enhanced LPP amplitudes for pleasant relative to neutral stimuli, irrespective of sex. The lack of pleasure and interest in response to hedonic stimuli has been proven an essential feature of depressive disorder (Dyck et al., 1994; Crawford and Henry, 2004; Sherdell et al., 2012). Consistently, there is abundant evidence showing that pubertal transition leads to a higher prevalence of depression and related behavioral disorders, irrespective of sex (Hankin, 2006; Patton and Viner, 2007). For instance, it was reported that anhedonia and psychomotor retardation symptoms of depression tend to increase and become more prevalent with the transition from childhood into adolescence (Hankin, 2006). Prospective longitudinal studies show that average levels of depressive mood and symptoms rise substantially from childhood to middle adolescence across sexes (Ge et al., 1994, 2003), due to increasing stressful life events such as academic stress, parent-offspring and other interpersonal conflicts during adolescence (Galambos and Almeida, 1992; Laursen et al., 1998; LaRue and Herrman, 2008; Quach et al., 2015). It is worth noting that the current study also observed decreased pleasant mood ratings during mid-/late relative to pre-/early adolescents, irrespective of block category and sex. All these data suggest that pubertal transition is linked with decreased pleasant emotion sensitivity, a key element that predisposes an adolescent to a depressive state (Sherdell et al., 2012).

Third, we observed higher P3a amplitudes in girls than in boys in mid-/late adolescents, and higher P3a amplitudes for mid-/late relative to pre-/early girls, during the negative block. This result is consistent with the previous finding that pubertal transition is associated with enhanced gamma oscillations for negative pictures in girls but not in boys (Yuan et al., 2014). Also, these results suggest that pubertal development brings girls a greater attention bias for negative stimuli compared with boys. This provides an explanation for the epidemiological reports of girls’ increased incidence of affective disturbances relative to boys that begins from early adolescence (Kessler et al., 1993; Nolen-Hoeksema and Girgus, 1994; Ge et al., 2001; Marcotte et al., 2002; Hyde et al., 2008). Prior studies suggested a couple of biological or psychosocial factors that mediate girls’ vulnerability to affective disorders during adolescence. For instance, the start of menstrual circle, which symbolizes pubertal transition of girls, leads girls to more fluxes in reproductive hormones than boys, and this flux increases emotional disturbances (Altemus, 2006; Ziomkiewicz et al., 2012). In addition, pubertal development is associated with a couple of psychosocial vulnerabilities that are more prominent in girls, such as gender role differentiation that stereotypes girls to be feminine (e.g., compliant and passive); increased interpersonal dependance, body image concerns and emotion-focused coping (Wichstrøm, 1999; Hyde et al., 2008). For example, recent studies show that the wellbeing of pubertal girls relies more on peer acceptance compared to that of prepubertal girls, while this pattern of puberty effect is not significant in boys (Guyer et al., 2014). These factors may contribute to our finding of different patterns of sex differences in attention bias for negative stimuli across pubertal stages.

It is worth noting that boys showed no significant puberty effect in P3a amplitudes, and pre-/early adolescents exhibited no sex differences in this component. This is consistent with prior reports of no sex differences in the occurrence of affective disorders in pre-pubertal childhood (Nolen-Hoeksema and Girgus, 1994; Wichstrøm, 1999; Marcotte et al., 2002). It has been indicated that pubertal development of boys is associated with increased social requirement of trait masculinity: to be less emotion-focused, more assertive, confident and action-oriented with physical resemblance to adult males (Wichstrøm, 1999; Hyde et al., 2008). On the other hand, there are evidences showing that emotional-expressive suppression is able to downregulate the emotional impacts of aversive events in young men but not in women, while both sexes benefit from the reappraisal strategy (Gross and John, 2003; Cai et al., 2016). This suggests that boys may benefit from more flexible choices of regulation strategies according to contextual requirements. Moreover, there is evidence that the increasing testosterone level during boys’ puberty is linked with more developed top-down control of prefrontal cortices (Stanton et al., 2009), which helps to regulate subcortical emotional inputs (Lieberman et al., 2007). Thus, though the puberty of boys is also linked with psychosocial stressors (McCabe and Ricciardelli, 2001; Hankin, 2006), the increased testosterone, masculinity intensification, and the flexibility of coping strategies may constitute compensatory factors for these risks. This also provides an explanation for why the incidence of affective disorders in adolescent boys is not as prevalent as that in adolescent girls. However, we need to acknowledge the limitation that our findings of puberty effect is based on cross-sectional comparison instead of being based on the longitudinal data, which may otherwise depict a fine-grained profile of emotional sensitivity varying as a function of puberty directly. Nevertheless, considering that emotion-related individual difference measures (e.g., anxiety, depression) were controlled across the four samples prior to the experiment, the conclusions based on the cross-sectional methods should be considered reliable.

In summary, in addition to our observation that females had enhanced cognitive and experiential sensitivity to negative stimuli than males, irrespective of puberty; the present study observed that puberty increased attentional bias for negative stimuli in girls but not in boys, and puberty reduced brains’ experiential sensitivity to pleasant stimuli across sexes. These patterns of pubertal developmental changes contribute to our understanding of the electrophysiological bases underlying the greater prevalence of affective disturbances in girls vs. boys during adolescence.

JYa and JYu designed the study. SZ conducted the experiment. JYa, SZ and JYu performed data analysis and prepared the manuscript. YLo assisted in data analysis and article revision. QL assisted in the experimental operation and data analysis. YLi and SX assisted in experimental operation. JYu supervised the whole research and wrote the article.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.