Priming the Pathway: Combining Oxytocin and Behavioral Intervention to Improve Outcomes in Autism Spectrum Disorder

Autism spectrum disorder (ASD) is a neurodevelopmental disorder characterized by social-communication deficits and the presence of restricted interests and/or repetitive behaviors. There are currently no psychopharmacologi-cal agents approved to treat core symptoms of ASD. As such, behavioral interventions are the most effective method for improving symptoms. In the current chapter, we propose that administering the neuropeptide oxytocin in conjunction with evidence-based behavioral interventions may lead to improved outcomes in social-communication for children with ASD. From a mechanistic perspective, we hypothesize that oxytocin may “prime” social reward circuitry in the brain, thereby allowing behavioral interventions designed to increase social motivation/ initiation to be more effective. Extant literature related to theories of ASD, oxytocin administration in children with ASD, and behavioral intervention outcomes are reviewed, and considerations for individual characteristics (e.g., genetics, oxytocin availability, age, behavioral profile, etc.) that may affect efficacy are discussed.


Introduction
It is estimated that 1 in 59 children in the United States has a diagnosis of Autism Spectrum Disorder (ASD) [1], which is characterized by life-long social communication deficits and the presence of restricted interests and/or repetitive behaviors [2]. Though all individuals with ASD meet common diagnostic criteria, they display a wide spectrum of behavioral manifestations, thus producing a wide array of heterogeneous phenotypes. As such, a multitude of behavioral interventions exist to address common symptoms of ASD, many of which have been tailored to address the heterogeneity of the disorder. Behavioral interventions are currently one of the most effective methods for improving social-communication skills, as no medications have been approved to "treat" social communication behaviors in ASD.
In the current chapter, we propose that administering the neuropeptide oxytocin in conjunction with evidence-based behavioral interventions may lead to improved outcomes in social-communication for children with ASD. Specifically, we propose

Social motivation hypothesis of ASD
The social motivation hypothesis posits that early impairments in social attention (due to social stimuli being less rewarding for individuals with ASD compared to their neurotypical peers) set a series of negative developmental consequences in motion. Initial impairments in social attention--which often manifest as decreased orienting to one's own name, diminished eye contact, and decreased social initiations in early life--lead to fewer opportunities for social learning, which in turn lead to deficits in social communication, social skills, and social-cognitive development [3,4]. The proposed neural mechanisms underlying the social motivation hypothesis are the reward centers in the brain including the amygdala, ventral striatum, and orbito-frontal cortex [4]. Importantly, social motivation is thought to involve both oxytocin and the dopaminergic reward pathways [5,6]. Given the hypothesized role of oxytocin in social motivation, it is not surprising that oxytocin has been considered as a potential mechanism for understanding social deficits in ASD--and that oxytocin administration has been considered as a potential therapeutic agent.

Oxytocin overview
Oxytocin is a neuropeptide produced in the hypothalamus [7]. Colloquially referred to as the "love hormone," oxytocin has been implicated in the formation of pair bonds in a variety of species, including rats and prairie voles [8-10]. As such, oxytocin has been extensively studied and utilized for its effects on social cognition and prosocial behaviors [11][12][13]. It has been hypothesized that interactions between oxytocin and the dopaminergic reward system support social affiliative behaviors and social bonds [14]. Further evidence of the connection between oxytocin and the dopamine reward system comes from overlap in locations of receptor binding sites and neuronal fibers [15]. Furthermore, oxytocin cells in the hypothalamus have been found to express dopamine receptors [16]. Importantly for neurodevelopmental disorders such as ASD, oxytocin is developmentally regulated and its receptors are malleable, particularly in response to parent-child interactions [17,18].
In contrast to other conditions in which both behavioral and medical treatments are supported by empirical evidence, no medication is currently approved to improve core symptoms of ASD. Although two pharmaceutical medications, risperidone and aripiprazole, have been approved by the Federal Drug Administration for use in ASD, they do not address core symptoms. Rather, they have been approved to treat ancillary symptoms of ASD, such as aggression and irritability [19]. Due to the lack of pharmacological interventions to address the core symptoms of ASD, there

Behavioral findings
To our knowledge, eight studies have been conducted to measure behavioral results of oxytocin administration for children with ASD [22,[24][25][26][27][28][29][30][31][32][33]. Though the current chapter focuses on children, see the meta-analysis by Ooi and colleagues [32] for a review of randomized control trials of oxytocin administration in both children and adults. See Table 1.
In 2010, Guastella and colleagues [25] completed a double-blind placebocontrolled study with 16 adolescent boys with ASD (ages 12-19). 45 minutes after oxytocin (or placebo) administration, participants completed the Reading the Mind in the Eyes Task, or RMET [33]. Compared to placebo, oxytocin improved performance on the REMET for 60% of participants. When the authors split REMET items into "easy" and "hard" categories, the effect of oxytocin was particularly significant for "easy" items. This was the first investigation of the results of oxytocin administration to children/adolescents, and suggested that oxytocin improves emotion recognition in a group of males with ASD.
Tachibana and colleagues [26] measured behavioral changes associated with an open label (participants knew they were receiving oxytocin), single arm (no placebo condition) long-term oxytocin administration in 8 boys with ASD (ages 10-14). Participants received intranasal oxytocin twice a day for approximately 7 months. Oxytocin dosage was increased in a stepwise fashion every 2 months (from an initial dose of 8 IU to 24 IU). Outcome measures were scores on the Autism Diagnostic Observation Schedule-Generic (ADOS-G; [34]), Aberrant Behavior Checklist (ABC; [35]), and Child Behavior Checklist (CBCL; [36]). ADOS-G scores for items comprising the "communication," and "social interaction" sub-scales and the sum of both sub-scales were significantly improved after oxytocin administration. No change in ADOS-G scores related to play/imagination of restrictive/ repetitive behaviors (RRBs) were observed. No significant changes were observed for the ABC or CBCL after oxytocin administration though some scores improved at a trend level. The authors note that oxytocin administration appeared to specifically affect social communication and interaction, but did not affect RRBs or play. However, the study was open label, single arm. Therefore, caution is needed when interpreting these findings. Anagnostou and colleagues [27] administered intranasal oxytocin to 15 male and female adolescents (ages 10-17) twice per day for 12 weeks in a single arm, open label design. Participants completed a follow-up session 12 weeks after discontinuation of oxytocin administration. Though the authors measured change on a variety of behavioral measures related to social function, social cognition, anxiety, and repetitive behaviors, the aim of the study was to measure the safety of varying dose levels of oxytocin in this population. Safety results suggested that .4 IU/kg twice per day for 12 weeks produced no serious or severe adverse events. Using the Clinical Global Improvement-Social (CGI), almost half of the sample were classified as "responders" at 12 weeks, and most responders at 12 weeks maintained improvements 12 weeks after the end of the study (e.g. 24 weeks from the beginning of the study). At 12 weeks, significant improvements were observed for the The authors note that not all children demonstrated behavioral improvements in response to oxytocin administration, and that more information is needed about individual differences in response to oxytocin. Nevertheless, a broad array of measures was, on average, shown to increase after oxytocin administration.
Dadds and colleagues [28] completed a 5-day double-blind, placebo controlled, randomized controlled trial with 38 boys with ASD (ages 7-16). This study is unique as it combined oxytocin administration with consecutive parent-child interaction training sessions over a period of 4 days. The authors did follow-up testing 3 months after the completion of the 5-day intervention. The parent-child interaction training consisted of emotion recognition training using the Mindreading (MR) program [45], and short video clips demonstrating successful client interactions (e.g. eye contact, body language, responsiveness). Outcomes included parent-child interaction tasks (free play, emotion talk, and an "I love you" task in which a parent expresses positive emotions and the child's response is recorded), and parentquestionnaire measures (Social Skills Rating Scale, or SSRS, [46]; SRS, [37]), and an emotion recognition task (UNSW Facial Emotion Task, [28]). Analysis of the parent-child interaction tasks, parent rating scales, and emotion recognition tasks indicated behavioral improvements over time in both the placebo and oxytocin groups, but no significant benefit of oxytocin versus placebo. We note that although this study combined oxytocin and behavioral intervention, the intervention was not one previously validated for improving core symptoms of ASD, and the study timeline was short (5 days).
Guastella and colleagues [29] completed a double-blind placebo-controlled trial with 50 male participants with ASD (ages 12-18). Participants received either placebo or oxytocin nasal spray twice a day for 8 weeks. Participants were assessed again 3 months after completion of the study. Primary outcome measures were the SRS [37] and clinician ratings on the Clinical Global Impressions--Improvement scale (CGI-Improvement . Additionally, authors measured plasma oxytocin concentrations both before and after participation in the trial, as well as expression of genes related to oxytocin receptors (Oxytocin Receptor Gene, OXTR, and Vasopressin Receptor 1a; V1AR). Results suggested that children who received oxytocin demonstrated greater improvements on the SRS compared to those receiving placebo. Additionally, pre-treatment plasma oxytocin negatively predicted the magnitude of improvement on the SRS. That is, individuals who had the lowest concentrations of oxytocin pre-treatment demonstrated the largest improvements. The authors hypothesized conflicting findings of previous oxytocin trials may be attributable to variability in pre-treatment concentrations of oxytocin. That is, previous trials did not measure pre-treatment oxytocin concentrations which may have reduced their ability to accurately measure improvement. When the authors did not include pre-treatment oxytocin, concentration findings were non-significant. In fact, inclusion of these pre-treatment "biomarkers" improved the statistical models by 43%, which underscores the importance of measuring individual variability in pre-post measurement designs.
Strathearn and colleagues [31] completed a double-blind, placebo-controlled crossover study with 16 children with ASD and 16 neurotypical children (ages 8-19). Participants completed an eye tracking paradigm on two occasions: once after receiving oxytocin and the other after receiving a placebo nasal spray. The authors measured participants' eye gaze while viewing images that varied on how organized or structured they were. The authors' primary focus was whether or not oxytocin affected eye gaze patterns when viewing stimuli that varied on levels of systemization. Results demonstrated that after oxytocin administration, children and adolescents with ASD displayed a decreased preference for highly systemized stimuli, whereas neurotypical children and adolescents displayed increased preference for systemized images. Overall, when participants received placebo, children and adolescents with ASD displayed more of a preference for systemized stimuli compared to their neurotypical peers. After both groups received oxytocin this difference was no longer significant (e.g. participants with ASD looked more similar to their neurotypical peers, and neurotypical children and teens looked more similar to their peers with ASD). The authors concluded that oxytocin administration may have differential effects on individuals with and without ASD, but that oxytocin appears to decrease preference for highly organized (e.g. systemized) stimuli in ASD.
Overall, studies of the behavioral effects of oxytocin in children in ASD are mixed, with some investigations observing significant behavioral changes [22,[25][26][27]30, 31], whereas others find no evidence for change as a function of oxytocin versus placebo administration [28,29]. A likely explanation for these disparate findings is outlined by Parker and colleagues [22]. The authors noted that measuring participant's plasma oxytocin levels provided a critical "biomarker" for predicting a given individual's response to oxytocin administration. Most studies do not measure oxytocin concentration prior to administration and therefore may miss individual differences which significantly impact outcomes. Parker and colleagues [22] noted that without the addition of pre-administration plasma oxytocin concentration into their predictive models, their results would have indicated no difference between oxytocin and placebo. When pre-administration oxytocin levels were included, their models were improved by 43%. This underscores why it is critical to consider biological differences in individuals prior to administration of oxytocin in order to more accurately understand variability within samples of children with ASD. When such individual differences are considered, behavioral effects of oxytocin may be more detectable, and/or may help us understand which children are most likely to benefit from oxytocin.

Neuroscience findings
To our knowledge, only 3 studies have investigated the effects of oxytocin on brain activity in children with ASD [54-56]; see [57] for a review. See Table 2.
Gordon and colleagues [54] used functional magnetic resonance imaging (fMRI) to measure brain activity after oxytocin versus placebo in 17 children and adolescents with ASD (ages 8-16.5) using a randomized, double-blind, crossover design. Fortyfive minutes after administration of either oxytocin or placebo, participants completed an fMRI emotion recognition task based on the RMET [33]. In a control fMRI condition, participants were asked to label the category of automobile presented in pictures. Findings suggested enhanced brain activity after oxytocin administration in brain areas hypothesized to be critical to reward (dorsal and ventral striatum), social attention and cognition (posterior cingulate, inferior parietal lobule, posterior superior temporal sulcus), and reasoning about mental states (medial prefrontal cortex). In addition, several brain regions were less active during nonsocial judgements after oxytocin administration. Interestingly, the effect of oxytocin on brain activity in response to social judgements differed as a function of symptom severity. Children with ASD with less severe symptoms (measured with the SRS) exhibited more 'typical' brain activity after oxytocin administration compared to children with ASD with more severe symptoms. Taken together, the authors hypothesized that oxytocin may increase the reward value and/or salience of social stimuli for children with ASD while simultaneously decreasing the salience of nonsocial information.
Gordon and colleagues [55] employed a randomized, double-blind, placebo-controlled crossover design with 20 children with ASD (ages 8-16.5). Participants completed an fMRI scan twice: once receiving oxytocin and the other after receiving placebo nasal spray. The authors measured participants' brain activity in response to two social perception tasks: a biological motion task and an affective voices task. The authors' primary aim was to measure whether oxytocin increased brain activity in areas related to the reward system and the connections between reward and social brain regions  (e.g. mesocorticolimbic pathways and communication between this pathway and socially-relevant brain regions) compared to placebo. Results were consistent with the author's hypotheses: oxytocin enhanced neural responses to biological versus random motion in the posterior superior temporal sulcus (pSTS), and in the amygdala and hippocampus in response to angry versus happy voices. Across both fMRI tasks, oxytocin increased neural connectivity both within reward regions and between the reward pathway and regions associated with social perception. The authors concluded that oxytocin administration appeared to enhance salience/reward of social stimuli (as measured in the biological motion task), but also increased the salience of negative social stimuli (as measured by the affective voices task).
Greene and colleagues [56] conducted a placebo-controlled, double-blind study with 28 children and adolescents with ASD (ages 10-17). Participants completed an fMRI incentive delay task on two occasions: once after receiving oxytocin, and once after receiving a placebo nasal spray. The fMRI task involved both social and non-social reward conditions so the authors could compare the effects of oxytocin versus placebo on brain activity related to both social and nonsocial rewards. Results demonstrated that compared to placebo, oxytocin increased brain activity in the caudate nucleus, left anterior cingulate cortex, frontal pole, insular cortex, and orbito-frontal cortex in the nonsocial reward versus social reward condition. Additionally, the authors found a positive relationship between symptom severity (measured using the SRS; [37]) and activation in the frontal pole and anterior cingulate during nonsocial reward anticipation, and between symptom severity and activation in the precentral gyrus and left caudate during nonsocial reward processing. Interestingly, these findings do not support the hypothesis that oxytocin selectively enhances reward-related brain activity to social stimuli, but rather may be associated with increased reward anticipation and processing for nonsocial stimuli. The authors hypothesize that reward-related effects of oxytocin may be sensitive to task-specific features, and noted that their findings strengthen the body of evidence that oxytocin acts on the brain's reward system.
Taken together, these findings provide further evidence that intranasal oxytocin administration appears to act on the neural reward system. However, evidence is mixed regarding whether oxytocin specifically acts on the social reward system [54, 55] or the reward system more broadly [56]. It is likely that differences in neuroscience tasks, as well as differences in regions of interest, and participant sample characteristics may explain disparate findings. Importantly for our theoretical framework, intranasal oxytocin administration appears to impact the reward system in children with ASD. This provides an empirical basis for our hypothesis that combining a pharmacological agent with behavioral interventions that act on similar brain regions/networks may improve outcomes related to social initiation and communication.

Behavioral Interventions in ASD
Given the heterogeneous behavioral manifestations of ASD, it is not surprising that multiple behavioral interventions have been developed to improve social communication and decrease challenging behaviors [58-60]. Multiple systematic reviews have been published on the success of behavioral interventions for children with ASD (e.g. [61-64]), and an extensive review is beyond the scope of the current chapter. Of primary relevance to the current chapter are behavioral interventions designed to increase the salience, relevance, and reward value of social interactions in order to improve social-communicative symptoms. This section will briefly review two of these behavioral interventions, specifically the Early Start Denver Model (ESDM) and Pivotal Response Training (PRT) as both of these interventions are based on principles of Applied Behavioral Analysis (ABA) but emphasize naturalistic teaching strategies to promote generalization.

Early start denver model (ESDM)
ESDM is an empirically validated, manualized intensive early intervention program designed for children between the ages of 1-4 [60, 71]. ESDM uses teaching strategies including: interpersonal exchange and positive affect, engagement with real-world activities and materials, adult responsivity and sensitivity, and focus on both verbal and nonverbal communication. These strategies are grounded in ABA principles including operant conditioning, shaping, and chaining. Importantly, ESDM is not conceptualized as a behavioral intervention that must occur in a table-top or structured situation, nor is it a "one size fits all" approach. Each child's program is individualized, and parent's roles are emphasized. Parents are taught ESDM strategies and encouraged to utilize them during daily activities (e.g. feeding, bath time, play). In a 2010 randomized controlled trial, Dawson and colleagues compared the efficacy of ESDM versus treatment as usual (TAU) over a 2-year period [60]. Results found significant improvements in cognitive abilities in the ESDM group compared to TAU. These group differences appeared driven by improvements in both expressive and receptive language in the ESDM group. Additionally, significant group differences were observed in adaptive behaviors. Whereas children in the ESDM group remained steady in their adaptive behaviors across time, children in the TAU group exhibited lower scores across time when compared to their neurotypical peers. This difference appeared driven by increasing gaps between the TAU group and their neurotypical peers in a variety of adaptive skills, whereas children in the ESDM group displayed improvement in their communication abilities. A recent meta-analysis corroborates the efficacy of ESDM in improving cognition and language for children with ASD [72].

Pivotal response training (PRT)
Pivotal response training (PRT), sometimes referred to as Pivotal Response Teaching or Pivotal Response Therapy, is a naturalistic behavior intervention based on principles of ABA. The underlying assumption of PRT is that children's challenges can be improved with behavioral and environmental manipulations including reinforcement, contingencies, consequences, and extinction [73]. The term "pivotal" is important as it refers to pivotal behaviors that, when targeted, can lead to improvements in other areas of behavior not specifically targeted. The behaviors/function areas most commonly targeted in PRT are: motivation, initiation, responding to multiple cues, and self-management [74,75]. To increase motivation, teaching strategies include: following the child's lead, offering choices, providing clear opportunities for response, varying tasks, including both maintenance and acquisition tasks, contingent and natural reinforcement, and reinforcing all attempts at target skills [73,74]. Similar to ESDM, PRT emphasizes the importance of implementing the intervention in the child's natural environment and the involvement of parents and other caregivers in the intervention [73]. Results of a systematic review indicate that PRT is largely effective at increasing self-initiation and improving language, communication, and play skills in children with ASD [76]. A 2016 meta-analysis concluded that PRT is effective at teaching behaviors to children with ASD [77].

Neuroscience findings after behavioral intervention in ASD
Despite a wealth of research examining the behavioral utility of empiricallybased interventions for children with ASD, there is a relative paucity of literature utilizing neuroscience as an outcome measure or predictor of behavioral intervention response (see [78] for a review). To our knowledge, seven studies have been published using neuroscience as either an outcome measure or predictor of response for an empirically supported behavioral intervention designed to help with core symptoms of ASD. See Table 3.
The first published paper using neuroscience as an outcome measure was Dawson and colleagues in 2012 [79]. 29 children with ASD (ages 48-77 months) participated in the study and were randomly assigned to either receive two years of ESDM intervention or two years of treatment as usual, TAU. Compared to toddlers in the community intervention, those who received ESDM demonstrated faster neural signatures of attention (the Nc ERP component) when viewing faces versus objects. Interestingly, when brain activity patterns of the two groups of toddlers with ASD were compared to brain activity of neurotypical (TD) toddlers, the ESDM and TD groups exhibited increased cortical activation when viewing faces versus objects, whereas toddlers in the TAU group evidenced more cortical activation when viewing objects vs. faces. The authors concluded that participation in ESDM appeared to lead to "normalization" of attention to faces. However, the authors did not include measures of brain activity prior to intervention, so they were unable to directly infer whether participation in ESDM significantly changed brain activity from pre to post intervention.
Voos and colleagues [80] used fMRI to measure brain activity in two five-yearolds with ASD before and after four months of PRT. The fMRI paradigm involved watching point-light displays that were either attached to an adult who performed actions (e.g. biological motion) or scrambled light displays created with random selections of light points (scrambled motion). When measuring brain activity from brain areas implicated in ASD from previous research [81], the authors found that one participant had increased brain activity after intervention in the left fusiform gyrus and left dorsal prefrontal cortex, and the other participant had greater activation in the left ventrolateral prefrontal cortex, right posterior superior temporal sulcus, and fusiform gyrus. All of these brain regions have been previously implicated as relevant to processing social stimuli and biological motion. Despite the small sample size, the results suggest that PRT can increase brain activity in important regions associated with social stimuli.
Ventola and colleagues [82] also measured brain activity before and after participation in PRT. 10 children with ASD (aged 4-7) completed an fMRI biological motion task both before and after 16 weeks of PRT. 5 neurotypical children were tested twice (once at "baseline" and once 16 weeks later) but they did not receive PRT. Based on activity in the parietal temporal sulcus (pSTS) at baseline compared to neurotypical participants, children with ASD were separated into two groups: hypo-and hyper-active. Children with hypo-activation were hypothesized to have decreased social motivation (evidenced by hypoactivity in the pSTS compared to neurotypical children), and children with hyper-activation were hypothesized to be hypersensitive to stimuli (evidenced by hyperactivity in the pSTS compared to neurotypical children). After PRT, children in the hypo- group had increased brain activity in the pSTS and ventral striatum when viewing biological motion, and decreased ventral striatum activity in response to scrambled motion. Children in the hyperactive group evidenced decreased brain activity in the pSTS, amygdala, thalamus, and hippocampus when viewing biological motion. The authors' hypothesized that participation in PRT differentially affected brain activity of children depending on their baseline characteristics (e.g. children who displayed too little brain activity in response to social stimuli evidenced increased brain activity after intervention, whereas children who displayed too much brain activity in response to social stimuli evidenced decreased activity after intervention).
Van Hecke and colleagues [83] used electrophysiology (EEG) to measure brain activity before and after participation in a randomized control trial of the Program for the Education and Enrichment of Relational Skills (PEERS; [84, 85]) versus a waitlist control (WLC) condition. PEERS is an empirically validated, manualized program which focuses on making and keeping friends for adolescents with ASD [86].  35 adolescents (13-14 years old) with ASD were randomly assigned to receive PEERS, 31 were randomly assigned to WLC, and 30 neurotypical (TD) adolescents were assessed prior to the intervention for comparison purposes. The authors measured brain activity patterns during "resting state" during which adolescents were instructed to focus on a fixation point for three minutes. Of particular interest was patterns of brain activity in the right versus left hemispheres as previous research suggests that individuals who have left-hemisphere dominance have higher approach motivation and positive affect compared to those with right-hemisphere dominance who are often characterized as withdrawn [87]. Prior to intervention, the two groups of teens with ASD evidenced less left-dominant asymmetry compared to the TD group. When comparing post-intervention brain activity of teens who received PEERS to brain activity in the TD group, differences in left-hemispheric dominance were no longer observed. Differences between the WLC group and the TD group were still observed after the 14-week intervention period (during which the WLC group did not receive intervention). These findings suggest that participation in PEERS (compared to a waitlist condition) significantly changed patterns of brain activity to be increasingly left-hemisphere dominant, which is more similar to patterns observed in neurotypical teens. Venkataraman and colleagues [88] used a Bayesian probabilistic model to characterize fMRI activity in 19 children with ASD (mean age 5.87 years) before and after 16 weeks of PRT. Similar to Ventola et al. [82], the fMRI paradigm involved watching two types of light displays: biological motion and scrambled motion. The probabilistic model allowed the authors to measure PRT-induced changes in neural connectivity between neural regions of interest. Results indicated both reduced connectivity between the posterior cingulate cortex (PCC) and orbitofrontal cortex and increased connectivity between the PCC and areas of the ventral occipital temporal extrastriate cortex. These results are interesting given that the PCC is known for its role in social cognitive processes, the orbitofrontal cortex has been implicated in reward processes, and the ventral occipital and temporal cortices play a role in processing socially meaningful stimuli (including biological motion). These findings suggest that PRT causes a shift, in which connectivity between the PCC and orbitofrontal cortex is decreased while PCC and the ventral occipital-temporal cortex connection is strengthened.
Yang and colleagues [89] utilized fMRI as a pre-intervention predictor of intervention response in 20 children with ASD (mean age 5.90 years). The authors measured brain activity during a biological motion paradigm prior to 16 weeks of PRT. Results demonstrated that pre-intervention brain activity in areas implicated in: (1) social perception (fusiform gyrus, inferior temporal gyrus, middle temporal gyrus), (2) social attention (inferior parietal gyrus, superior parietal lobule), (3) emotion regulation and reward (orbitofrontal cortex, ventrolateral prefrontal cortex, anterior insula), and (4) social reward (putamen, pallidum, amygdala, hippocampus, ventral striatum) while viewing biological motion predicted behavioral improvements on the SRS from pre-to post-intervention. For all 4 regions, greater levels of pre-intervention brain activity were associated with increased behavioral improvements from pre-to post-intervention. Results from this study suggest that neuroscience methods may be able to predict which children are most likely to benefit from a specific intervention, as well as underscoring the importance of brain regions associated with social perception, emotion regulation, and reward for understanding how interventions may affect brain circuitry in ASD.
Baker and colleagues [90] used event-related potentials (ERP) to measure brain response to reward in ASD adolescents before and after participation in the PEERS program compared to a typically developing sample. Response to social and nonsocial rewards were measured using the reward-related positive component (RewP) in seven adolescents with ASD and seven TD adolescents, aged 10 to 17 years. Prior to intervention, patterns of reward-related brain activity (RewP mean amplitude) did not differ between groups. However, after intervention the ASD group demonstrated an enhanced sensitivity to rewards, regardless of social or nonsocial condition, compared to the TD group. Additionally, ASD participants with less robust responses to social rewards prior to the start of the PEERS intervention demonstrated the most gains in social behaviors (as measured via SRS-2 [91]; Social Skills Improvement System, [92]. Findings from this study suggest an enhancement of the neural response to rewards after teens with ASD receive training in social skills and additionally that teens with attenuated responses to social rewards may gain the most benefit from intervention. Taken together, these seven studies provide evidence for two critical concepts: (1) Behavioral Intervention can significantly change patterns and/or magnitude of brain activity in response to social stimuli for children with ASD, and (2) Neuroscience may be able to predict individual levels of behavioral intervention response and could eventually be used to advance "precision medicine" (e.g. to predict who is most likely to benefit from a specific intervention). The first concept is central to our theoretical framework, as it suggests that participation in empirically based interventions changes brain activity in children with ASD--particularly in response to social stimuli and in brain areas/patterns of activity related to social perception and reward. Neural changes in the reward system and other areas of the "social brain" after interventions provides an empirical basis for our hypothesis that the addition of pharmacological compounds that enhance brain activity/responses in those same regions might lead to increased benefits for children with ASD.

Current research combining oxytocin and behavioral interventions in ASD
Although there are no published studies (to our knowledge) that have conducted double-blind, placebo-controlled trials combining oxytocin administration with an empirically validated behavioral intervention in ASD, two trials are currently underway and recorded on ClinicalTrials.Gov (identifiers NCT02918864 and NCT03370510). We also note that a large-scale multi-site trial of the behavioral effects of oxytocin versus placebo is underway by the Study of Oxytocin in Autism to Improve Reciprocal Social Behaviors (SOARS-B; clinical trial identifier NCT01944046). One paper describing the rationale and methods for the trial has been published [93].

Does timing matter?
As noted above, we hypothesize that empirically based behavioral interventions designed to increase social-communication skills such as social initiation/motivation are more likely to be enhanced by oxytocin administration than interventions focused on other aspects of ASD (e.g. disruptive behaviors, anxiety). Such interventions are typically aimed towards young children (i.e. ESDM was developed for children ages 1-4) due to the targeted skills. We hypothesize that age is likely to play an important role in the efficacy of combining oxytocin with behavioral interventions such that younger children are more likely to benefit due to: (a) Early interventions in ASD often focus on improving social motivation and initiation, whereas interventions for older children are less likely to have such an emphasis. As oxytocin has been shown to affect the reward system, behavioral interventions designed to increase the reward value of social stimuli are likely to be the best candidates for this framework. (b) Increased neural plasticity in younger ages and the negative neuro-developmental sequelae described in the social motivation hypothesis. That is, young children have been hypothesized to benefit more from early intervention for ASD compared to older children due to neural plasticity, and the social motivation hypothesis posits that decreased reward value of social stimuli early in life leads to a negative cascade of developmental consequences [5,94]. It therefore seems likely that younger children are best positioned to benefit from combining behavioral intervention with oxytocin because early interventions may disrupt this negative developmental cascade and help children move back towards a typical trajectory. (c) Accumulating evidence that interventions for ASD are likely to be more effective if started earlier in life [94][95][96]. It is unclear, however, exactly how age will affect our theoretical model. Future research should pay close attention to how and if age is a predictor when measuring outcomes of combining oxytocin and behavioral intervention in ASD.

Conclusions
Taken together, considering extant research findings from behavioral and neural effects of oxytocin administration along with those on brain activity in response to behavioral interventions in children with ASD suggests that both oxytocin and behavioral interventions lead to measurable changes in regions of the "social brain" and reward network. These findings, combined with those suggesting that oxytocin administration may improve social-communication in children with ASD, provide the empirical basis of our hypothesis [57] that combining oxytocin administration with behavioral interventions may improve outcomes related to social-communication. We hypothesize that the administration of oxytocin prior to each session of an intervention may "prime" the neural reward system to be maximally responsive to the behavioral skills taught during the intervention session. Due to the central role of the reward system in our theoretical model we hypothesize that this combined approach may be most effective for social-communicative skills requiring social motivation (e.g. social initiation/approach) and be maximally beneficial for young children. See Figure 1 for hypothesized neural mechanisms underlying the efficacy of combining interventions with oxytocin to improve outcomes.
It is important to note, however, that this approach is unlikely to be equally effective in all children with ASD. As with all interventions, there does not appear to be any "one size fits all" approach to help children with ASD. It will be critical to measure and characterize individual differences that may explain variability in treatment efficacy from biological (e.g. levels of oxytocin concentration in plasma/ saliva before and after oxytocin administration, genetic expression of candidate genes relevant for oxytocin; OXTR, VAP1, brain activity in response to social stimuli prior to oxytocin administration, age), behavioral (e.g. social communicative symptom profile prior to oxytocin administration), and psychological (e.g. co-occurring diagnoses such as ADHD, anxiety, etc.) standpoints. Such individual differences will likely be central to our understanding of how, why, and for whom this combined approach will be maximally effective.

1) Neuroscience Outcomes after Behavioral Intervention for Children with ASD. This figure depicts
results from studies which measured brain activity before and after behavioral intervention. The brain areas highlighted in red indicate increased brain activity after behavioral intervention, whereas regions shown in blue indicate decreased brain activity after behavioral intervention. Results related to neural connectivity are shown with a dotted line. A. Dawson et al. [79], proposed neural generators of the Nc component (Reynolds & Richards, 2005)  Brain regions that have been implicated in both behavioral intervention and oxytocin administration are labeled. We propose that oxytocin administration will "prime" neural structures related to reward and/or social information processing, which will make behavioral intervention sessions more effective. [