Oxytocin, sexually dimorphic features of the social brain, and autism


Hidenori Yamasue, MD, PhD, Department of Neuropsychiatry, Graduate School of Medicine, University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8655, Japan. Email: yamasue-tky@umin.ac.jp


The common features of autism spectrum disorder, a highly heritable representative pervasive developmental disorder with significant heterogeneity and multiple-genetic factors, are severe dysfunction in social reciprocity, abnormalities in social brain regions, and disproportionately low probability in the female gender. Concomitantly, certain domains of mental function, such as emotional memory and social reciprocity, show a significant sex difference. In addition, recent neuroimaging studies have shown significant sexual dimorphisms in neuroanatomical correlates of social cognition. Recently, some sexually dimorphic factors, including oxytocin, vasopressin, and genes linked with the x-chromosome, have received attention because of their possible contribution to mental development especially in the social cognitive domain. Taking this evidence together, it is hypothesized that a sexually dimorphic factor associated with social reciprocity could affect characteristics of autism spectrum disorder including dysfunction in social reciprocity, abnormalities in social brain regions, and disproportionately low probability in female gender. This review article overviews sexual dimorphisms in clinical features of autism spectrum disorder, in normal social cognition, and in social brain function and structure. The association of oxytocin with sexual dimorphisms, social reciprocity, neural correlates of social cognition, and the pathogenesis of autism spectrum disorder were further summarized. Recent studies have suggested that oxytocin plays a role in social attachment in experimental animals, in enhancing social interactive ability in human adults, and in the pathogenesis of autism spectrum disorder. Thus, the ongoing accumulated evidence suggests that oxytocin deserves to be examined as a candidate that causes the sexually dimorphic aspect of human social reciprocity, social brain development and the pathogenesis of autism spectrum disorder.

AUTISM IS A pervasive developmental disorder (PDD) characterized by severe social and interpersonal dysfunction, abnormalities in so-called social brain regions, and disproportionately low probability in female subjects.1 This review paper begins by overviewing the sexually dimorphic features that have been identified with respect to the clinical characteristics of autism, social cognition and social brain regions, and oxytocin (OT). The paper then summarizes the relationship between OT and social cognition, and further reviews recent studies suggesting its implication in the pathophysiology of social dysfunction in autism. Finally, the possibility of OT as a candidate molecule to uncover the sexually dimorphic aspects of normal social interaction and the social interactive deficits in autism is discussed.


Fifteen years ago the prevalence of autism spectrum disorder (ASD) or PDD was reported at 30 per 10 000, but a recent study put it at approximately 60 per 10 000.2 This change in the reported incidence seems to be at least partially due to a broadening of the concept and diagnostic criteria for autism as well as increased awareness and improved detection at all ages and all levels of intellectual ability.3,4 The sex ratio of prevalence is, however, a stable male preponderance, although it varies depending on ASD subtype. While male subjects with autistic disorder are three–fourfold as likely as female subjects with autistic disorder, the figure is fivefold in PDD not otherwise specified (PDD-NOS) and ninefold in Asperger disorder.5 Moreover, the sex ratio is modified by IQ level. There were proportionally more female subjects with autism in the lower IQ range, particularly below an IQ of 35; in contrast, at higher IQ levels male subjects with autism were more frequent than female subjects even if the general sex ratio in autism is considered.6,7 In the PDD-NOS group, however, variability of sex ratio in proportion to IQ is not reported.7 Other measures of severity of autism, such as the age of recognition, ICD-10 symptom score, and adaptive behavior, were not different between male and female subjects in either autism or PDD-NOS groups when IQ was entered as a covariate.7 Moreover, the male preponderance weakened for ASD with epilepsy such that epilepsy was more common in female than in male subjects with ASD.8 The finding might reflect, however, the greater severity of associated mental retardation in female subject with ASD because the ASD with epilepsy group included individuals with more severe mental retardation than the ASD without epilepsy group. Thus, further research incorporating control of participant IQ level is needed to clarify the sex difference in ASD with epilepsy.

In neuropsychological and neurobiological indices, such as profiles of executive function and brain morphological and functional cortical substrate, few studies have compared those indices between male and female subjects with ASD. Most of the studies excluded female subjects with ASD or included very few in accordance with the sex ratio in prevalence of ASD. A recent magnetic resonance imaging (MRI) study, however, using voxel-based morphometry, recruited female subjects only9 and reported reduced gray matter density in fronto-temporal cortices and limbic regions of female subjects with ASD in similar locations to those reported in men with ASD.

Sex difference in autistic traits was investigated in ASD subjects without mental retardation using the Autism spectrum quotient, the Empathy quotient and the Systemizing Quotient.10–18 These were self-report questionnaires to assess social cognition and empathizing and systematizing ability in high-functioning subjects with ASD. All three quotients scores did not show any differences between male and female subjects with ASD.14,19


While some clinical characteristics of ASD showed significant sex differences, social interactive ability and social cognition are also sexually dimorphic in healthy human individuals. In healthy adults, women tend to show strong cooperativeness across nations and cultures,10,13,20 although previous studies contained some inconsistency.21 Baron-Cohen and his colleagues have found sex differences in the several psychometric measures developed by their research group.10–18 Based on these findings, they suggested that the male brain may be defined psychometrically by poor performance in terms of empathizing or friendship and significantly better systemizing ability, while the female brain is defined as the opposite cognitive profile.

Regarding neural correlates of social cognition, significant sexual dimorphisms have also been suggested. Recent functional imaging studies have reported activation of posterior superior temporal gyrus, posterior inferior frontal gyrus, anterior medial prefrontal cortex, anterior insula, and fusiform gyrus as neural correlates of human social reciprocity and related factors such as empathy, understanding other's emotion, and interpersonal interaction.22–27 Furthermore, a few studies have suggested sex differences in these activations.28–31 For example, Azim et al. found sex differences in brain activation elicited by humor.30 While the same brain regions (including temporal–occipital junction and temporal pole; structures implicated in semantic knowledge and juxtaposition, and the inferior frontal gyrus; likely to be involved in language processing) were activated in both sexes, female subjects had greater activation in the left inferior frontal cortex and caudate than male subjects during comprehension of humor. Platek et al. also founds sex differences in the neural correlates of perception for child facial resemblance.29 They showed using functional MRI (f-MRI) that female subjects had greater activation in the fusiform gyrus in response to child faces than did male subjects when resemblance was not modeled. In contrast, male subjects had greater cortical activity than female subjects in response to children's faces that resembled their own face. These results were related to theories of sexual selection. Furthermore, Singer et al. examined sex differences in the modulation effect of perceived fairness on brain empathic responses induced by viewing another person's pain.31 Response was measured on f-MRI and perceived fairness was determined by whether the person experiencing pain had previously played an economic game fairly or unfairly. Both sexes exhibited empathy-related activation in pain-related brain areas (fronto-insular and anterior cingulate cortices) towards the fair person, while these empathy-related responses were significantly reduced in male subjects when observing an unfair person receiving pain. Based on the results, it was suggested that empathic responses are shaped by valuation of other's social behavior, such that people empathize with fair opponents while favoring the physical punishment of unfair opponents.

In addition, sexual dimorphisms in neuroanatomical substrates of social cognition might extend to the brain structural level. Human altruistic cooperativeness, one of the most important components of our highly organized society is, along with a greatly enlarged brain relative to body size, a spectacular outlier in the animal world. The social brain hypothesis suggests that human brain expansion reflects an increased necessity for information processing to create social reciprocity and cooperation in our complex society. The author's research group reported that young women (n = 66) showed greater Cooperativeness as well as larger relative global and regional gray matter volumes than the matched men (n = 89), particularly in the social brain regions including bilateral posterior inferior frontal and left anterior medial prefrontal cortices.32 Moreover, in female subjects, higher cooperativeness was tightly coupled with larger relative total gray matter volume and more specifically with regional gray matter volumes in most of the regions, indicted larger-in-female sex dimorphism. Correlations between global, and most regional gray matter volumes, and Cooperativeness were significantly specific to female subjects (Fig. 1). Our results suggest that sexually dimorphic factors may affect the neurodevelopment of these ‘social brain’ regions, leading to higher cooperativeness in female subjects.

Figure 1.

Sexually-dimorphic correlation between whole relative gray matter volume and Cooperativeness.32 Scatterplots depicting correlations between total gray matter volume / intracranial volume (ICV) ratio and individual variability in cooperativeness scored by Temperament and Character Inventory in 89 healthy adult males and 66 healthy adult females. The sex difference of these correlations was statistically significant (Fisher's r to z transformation; z = 2.37, p = 0.018).

Our morphological findings may also have an implication for the pathophysiology of ASD, characterized by severe dysfunction in social reciprocity, abnormalities in social brain, and disproportionately low probability in female subjects.1 Baron-Cohen proposed the extreme male brain theory of autism,12 in which the male brain may be defined psychometrically by poor performance in terms of empathizing or friendship and significantly better systemizing ability, while the female brain is defined as the opposite cognitive profile.10–18 Using these definitions, autism can be considered as an extreme of the normal male profile. Recently, the hypothesis further suggests that specific aspects of autistic neuroanatomy may also be extremes of typical male neuroanatomy.18 Our MRI results are consistent with this hypothesis. Previous studies using structural MRI demonstrated smaller anterior cingulate,33,34 superior temporal gyrus, prefrontal cortex,35–40 thalamus,41 posterior inferior frontal gyrus,42 and enlarged amygdala, cerebellum in subjects with ASD (Fig. 2).42,43 Our study32 found sex dimorphism in brain anatomy at a similar location and in the same direction as these previous studies of individuals with ASD. Thus, the study32 may add supportive evidence for Baron-Cohen's extreme male brain theory of autism at the level of brain structure.

Figure 2.

Neuroanatomy in monozygotic twins with Asperger's disorder discordant for comorbid depression.39 (Top row) Plots of adjusted voxel-based morphometry responses at each brain region. (a–d) Gray matter voxels with reduced density in the twins with Asperger's disorder as compared with normal controls (n = 2 vs n = 82) rendered onto orthogonal slices of the normal template magnetic resonance imaging. Significantly abnormal regions: (a) left superior temporal gyrus including superior temporal sulcus, (b) left fusiform gyrus, (c) right prefrontal cortex, and (d) right amygdala. (e) Statistical parametric maps in the axial projection showing gray matter voxels with reduced density in the twins (n = 2; upper map) and each twin (lower maps) as compared with normal controls (n = 82). A significant reduction in the right amygdala (d) found in twin A with comorbid major depression was absent in twin B.


OT is a sexually dimorphic neuropeptide, consisting of six amino acid rings with three amino acid tails. OT is synthesized primarily in the central nervous system, although smaller amounts of this peptide are also produced in other tissues. In addition to various reproductive effects, OT affects emotional and social behavior in both male and female animals.44 OT seems to be associated with attachment, affiliative and social behaviors in experimental animals. For example, Takayanagi et al. generated mice with a null mutation in the OT receptor gene and compared their behavior with that of OT-deficient mice.45 They showed that the OT receptor plays a critical role in regulating several aspects of social behavior, including maternal behavior, ultrasonic vocalization to social isolation, and aggressive behaviors. Recently, Jin et al. showed that adult CD38 knockout female and male mice show marked deficits in maternal nurturing and social behavior, respectively, with higher locomotor activity.46 CD38 knockout mice also had strongly decreased plasma OT level, and replacement of OT improved social behavior and maternal care in CD38 knockout mice. They suggested that CD38 plays a key role in neuropeptide release, thereby critically regulating maternal and social behaviors, and is an element in neurodevelopmental disorders. The research group further showed that the number of ultrasonic vocalizations was lower in CD38 knockout mice than in wild-type mice.47


Rapidly accumulating evidence has recently demonstrated that OT could shape the development of social behavior and cognition not only in experimental animals but also in humans. Zak et al. examined 134 young human adults, 50% female, to explore the association between the intentionality of trust assessed by a trust game and blood OT level.48 OT levels in subjects who received a monetary transfer that reflected an intention of trust were higher than in those who received an unintentional monetary transfer of the same amount. In addition, higher OT levels are associated with trustworthy behavior. The results suggest that perceptions of intentions of trust affect levels of OT in adult human individuals. A previous study has shown that neuropeptides cross the blood–brain barrier within 30 min after intranasal administration.49 Recent studies have utilized intranasal OT administration to test the effect of OT on social cognition in humans. Previous studies have shown that OT nasal administration enhances trust48 and the benefits of social support during social stress tasks.50 For example, Kosfeld et al. showed that intranasal administration of OT causes a substantial increase in trust during the trust game among 178 young men, thereby greatly increasing the benefits from social interactions.51

There is further evidence in humans that OT enhances the perception of faces. OT increases gaze to the eye region of human faces,52 and improves the identification of emotion from the eyes of others.53 In addition, two recent studies have reported a significant enhancement of memory in face recognition induced by OT administration, although the influence of facial expression on the enhancement was inconsistent.54,55 Guastella et al. reported that the enhancement of identity recognition memory was observed for previously shown happy faces but not for neutral and angry faces,54 while Savaskan et al. observed enhancement of identity recognition memory for neutral and angry faces but not for happy faces.55 These two studies also differ in the timing of pharmacological administration; Guastella et al. administered OT before the encoding phase,54 whereas Savaskan et al. did this after it.55 Thus, although Guastella et al. argued that OT improves the encoding of positive social memory,54 Savaskan et al. suggested that OT enhances the consolidation and retrieval of negative social memory.55 Most previous studies examined the OT effect in male subjects only, while Savaskan et al. examined the OT effect in female subjects as well,55 but the authors reported no significant interaction between OT effects and gender on the memory enhancement. A consistent finding across the recent studies overviewed here, however, is that OT nasal administration improves human behaviors associated with the processing of social information.

Although the neural background of these OT effects has not yet been fully clarified, a few f-MRI studies have reported changes in brain activation associated with the OT effect on human emotional and social functions.56–58 Kirsch et al. found that human amygdala function is modulated by OT.56 They used f-MRI in region-of-interest amygdala activation by fear-inducing visual stimuli, for example fearful faces and scenes, in 15 healthy male subjects after double-blind cross-over intranasal application of placebo or OT. Compared with placebo, OT potently reduced activation of amygdala and reduced functional connectivity of the amygdala to brainstem regions implicated in autonomic and behavioral manifestations of fear. Domes et al. also reported that OT administration was associated with reduced amygdala activation in response to emotional faces stimuli regardless of emotional content or valence of facial perception in 13 healthy men. They also showed reduced activation in several frontal and temporal brain regions.57 Baumgartner et al. examined the neural correlates of the behavioral difference during a trust game with and without OT administration in 49 healthy young male subjects. They demonstrated that the behavioral difference induced by OT is associated with a specific reduction in brain activation of amygdala, the midbrain regions, and the dorsal striatum.58 Petrovic et al. further examined the effect of OT on affective responses to faces associated with fear as a function of their social relevance.59 They used presentation of face stimuli, some that had previously been fear conditioned by pairing with shocks and some that had not. They then assessed the effect of OT on the conditioning-induced change in affective ratings of faces. Affective ratings were more negative for those faces that had been fear conditioned, but this difference was abolished by treatment with OT; an effect associated with a decrease in activity of the amygdala and anterior cingulate cortex. In addition, this modulation was stronger for faces with direct gaze, relative to averted gaze, consistent with a relative amygdala and fusiform face area specificity for socially relevant cues. They suggested that OT modulates the expression of evaluative conditioning for socially relevant faces via influences on amygdala and fusiform face area, an effect that may explain its prosocial effects. Although it seems that these previous studies have detected reduced brain activations associated with decreased cautiousness or negative emotions, the studies have not reported enhanced brain activation associated with enhanced performance in social interaction.


Previous studies have repeatedly demonstrated both functional and structural brain abnormalities in individuals with ASD compared with individuals with typical development. In this section, previous functional neuroimaging studies, especially those indicating abnormalities in so-called social brain regions such as amygdala, fusiform gyrus, medial prefrontal cortices, superior temporal sulcus, and posterior inferior frontal cortices, are briefly overviewed here.

Several previous studies have reported neural substrates of face and gaze processing deficits in individuals with ASD,60–67 although potential confounds have recently been noted.68 Schultz et al. reported that the subjects with ASD showed less activation in the fusiform gyrus and greater activation in the right inferior temporal gyrus during face discrimination, the same activation pattern as seen in the controls during object discrimination.60 Several studies also showed that subjects with ASD fail to activate the fusiform face area during a face perception task.60,62 Pierce et al. further examined the effect of facial familiarity on the activation of the fusiform face area in individuals with ASD.63,64 They showed that viewing a familiar face, such as that of their mother, elicited a significant activation in the fusiform face area of children with ASD64 as well as adults with ASD.63 Dalton et al. examined the effect of gaze fixation on the aberrant activation in fusiform face area of individuals with ASD or in unaffected siblings of subjects with ASD.66,67 In the autism group, activation in the fusiform face area was strongly and positively correlated with time spent fixating the eyes, suggesting that diminished gaze fixation may account for the previously reported hypoactivation of fusiform face area in response to faces in ASD.67 Furthermore, the unaffected siblings of subjects with ASD had a similar pattern to that of subjects with ASD.69

Previous studies have suggested that the medial prefrontal and cingulate cortices seem to be implicated in mentalizing and theory of mind: understanding other's intentions and emotions.69 Several studies have reported that these cortices contribute to the deficits in social interaction observed in subjects with ASD.70,71 For example, a recent study indicated that subjects with high-functioning ASD had lesser activation in cingulate cortices while playing an interactive trust-game with a human partner than did controls.72 The diminished cingulate activation was associated with clinically assessed symptom severity.

Among the temporal cortices engaging in processing multimodal perception such as auditory and visual information, previous studies have suggested that the superior temporal sulcus processes the perception of socially relevant behaviors of others, such as voice73,74 and biological motion.75–80 Individuals with ASD had dysfunctions in this brain region. Castelli et al. reported that healthy adults showed increased activation in a previously reported mentalizing network (medial prefrontal cortex, temporal poles, and superior temporal sulcus at the temporoparietal junction) while viewing animations of triangles that elicited mentalizing compared with randomly moving shapes.70 During the same f-MRI task, the adults with ASD had less activation than the normal groups in all these regions. Gervais et al. indicated that subjects with ASD failed to activate superior temporal sulcus voice selective regions in response to vocal sounds, whereas they showed a normal activation pattern in response to non-vocal sounds.81 It was suggested that cortical processing of socially relevant auditory information is abnormal in subjects with ASD. Pelphrey et al. demonstrated a difference in the response of the superior temporal sulcus underlying eye gaze processing in subjects with ASD, suggesting that a lack of modulation of this brain region by gaze shifts that convey different intentions contributes to the eye gaze processing deficits associated with ASD.65

Because superior temporal sulcus, inferior parietal and inferior frontal cortices are implicated in mirroring others' actions, these brain regions have recently been thought to form a human mirror neuron system.82 Several studies have focused on the mirror neuron system as a possible neural substrate of deficits in social reciprocity of subjects with ASD.83,84 For example, Dapretto et al. showed that children with autism had less than normal activity in the inferior frontal gyrus (pars opercularis) while imitating and observing emotional expressions.84 Because the reduced inferior frontal activation was inversely related to symptom severity in the social behavioral domain, it was suggested that a dysfunctional mirror neuron system may underlie the social deficits observed in ASD. Wang et al. have further examined children with ASD using f-MRI to study neural correlates of irony comprehension. They found aberrant activation in the inferior frontal gyrus.85 Furthermore, reduced activity in the medial prefrontal cortex and right superior temporal gyrus was observed in children with ASD relative to typically developing children during the comprehension of irony. Importantly, a significant group × condition interaction in the medial prefrontal cortex showed that activity was modulated by explicit instructions to attend to facial expression and tone of voice only in the subjects with ASD. Finally, medial prefrontal cortex activity was inversely related to symptom severity in children with ASD such that children with greater social impairment had less activity in this region.71


As overviewed in the previous section, OT, a sexually dimorphic neuropeptide, is associated with normal social interaction, while recent studies have also suggested that OT may play a role in the pathogenesis of ASD. For example, it has been hypothesized that administration of OT during labor can generate excess OT in the fetal brain. Such excesses might lead to downregulation of OT receptors and, subsequently, to imbalance of the OT system and unavailability of OT for further signal transduction cascades. Several previous studies have examined the rate of labor induction, mainly using pitocin (a synthetic analog of OT), in children with ASD. One study reported that ASD had significantly higher rates of pitocin induction compared to normal children,86 while other studies have reported that there were no differences in pitocin induction rates, failing to support an association between exogenous exposure to OT and neurodevelopmental abnormalities.87,88

Another line of study has indicated that children with ASD tend to be characterized by lower levels of plasma OT. The first study that examined peripheral OT level in subjects with ASD analyzed midday samples from prepubertal children.89 It was found that the ASD group had significantly lower plasma OT levels than the normal group. Plasma OT increased with age in the normal children but not in those with ASD. Elevated OT was associated with higher scores on social and developmental measures in the normal children, but was associated with lower scores in children with ASD. Moreover, the same research group further found that children with ASD had alterations in OT peptide forms.90 In contrast to these previous studies, Jansen et al. showed that high functioning adults with ASD had increased plasma OT levels, and that OT levels were not correlated with impairment in social interaction or communication, nor with stereotyped behavior.91 One of the explanations for the discrepancy between these findings may be that the former studies included ASD children of various levels of intellectual function while the latter examined high-functioning adults with ASD.

While several studies have reported the effect of OT administration on interpersonal information processing in healthy adults,50–55 a limited number of studies has reported the effect in subjects with ASD.92,93 In addition, although intranasal OT administration is more likely to pass the blood–brain barrier, previous OT challenge in subjects with ASD involved i.v. infusion of OT. Previous studies using infusion of OT have reported the reduction of repetitive behaviors, and increased retention of social cognition in patients with ASD.92,93 One study examined the impact of OT on repetitive behaviors in adults with ASD via randomized double-blind i.v. infusion of OT and placebo challenges. The primary outcome measure was an instrument rating six repetitive behaviors: need to know, repeating, ordering, need to tell/ask, self-injury, and touching. Individuals with ASD had a significant reduction in repetitive behaviors following OT infusion in comparison to placebo infusion.92 A more recent study from the same research group explored the effect of i.v. OT administration on the retention of social information in adults with ASD. Comprehension of affective speech (happy, indifferent, angry, and sad) in neutral content sentences was tested. All subjects showed improvements in affective speech comprehension from before to after infusion; but whereas those who received placebo first tended to revert to baseline after a delay, those who received OT first retained the ability to accurately assign emotional significance to speech intonation on the speech comprehension task.93 These findings have provided support for the use of OT in the treatment of ASD. Future OT challenge researchers, however, should use OT cautiously, especially in infants and/or female subjects with ASD, because novel unknown side-effects, although not occurring in experiments in men, could emerge in the case of infants and/or female subjects.

As overviewed here, previous studies have found suggestive evidence linking OT and ASD, although the underlying biological and genetic mechanisms of this connection have not yet been uncovered. OT exerts its effects in the brain through OT ligands and receptors and the corresponding genes. Linkage and association studies of ASD have provided ongoing evidence regarding the potential role of OT in the etiology of ASD. Specifically, a combined analysis of Autism Genetic Resource Exchange (AGRE) and a sample of Finnish families of probands with ASD implicated region 3p24-26, harboring the OT receptor gene, as a susceptibility region for ASD.93 Another whole-genome scan also pointed to this region as the second best locus.95 Association studies with Chinese Han and US Caucasian families of probands with ASD implicated the OT receptor gene as a candidate gene for ASD.96,97 A third association study, in an Israeli population, showed a significant association between OT receptor gene and ASD, and this association was also observed with IQ and the Vineland adaptive behavior scales.98 Furthermore, another more recent study conducted with US Caucasian families became the fourth to report the association between OT receptor gene and ASD.99 These recent findings have confirmed the hypotheses that the OT receptor genes are involved in the development of affiliative behaviors in the manifestation of ASD. A pharmacogenetic approach, made feasible by these investigations of OT administration, offers a worthwhile strategy for the treatment of core social skill deficits associated with ASD.


The recently accumulated evidence overviewed here indicates that OT is associated with the disruption of social communication at various levels including the genetic, brain structural, brain functional, and behavioral levels in humans as well as experimental animals. It might be further suggested that the OT molecule could be an important key to understanding the sexually dimorphic features of social cognition100 and ASD,101 and further could be a candidate agent to treat the social dysfunction of subjects with ASD. Future studies linking the effect of OT on social behavior with brain functional, structural and genetic background are expected to uncover the pathophysiology of ASD and aid the development of treatment strategies.102,103 Toward this goal, future studies are also expected to identify biological markers that could predict individual differences in the efficacy of OT on social behavior.