Introduction
Difficulties in reciprocal social interactions and communication are among the core features of autism spectrum disorders (ASD), along with a restricted repertoire of activities and interests (American Psychiatric Association
2013). These social deficits have been documented in numerous studies showing that individuals with ASD have impairments in the ability to represent other people’s mental states (i.e., mentalizing; Baron-Cohen et al.
1985; Kaland et al.
2008) and in processing emotions of others (Adolphs et al.
2001; Hobson
1986; Uljarevic and Hamilton
2013). Neuroimaging studies have also revealed differences between individuals with ASD compared to typically developing (TD) individuals in brain areas relevant for social-affective functioning (Di Martino et al.
2009; Fishman et al.
2014; Frith
2001; Pelphrey et al.
2011; Philip et al.
2012; White et al.
2014). These studies suggest that social deficits in ASD are associated with atypical activation in brain areas involved in mentalizing, such as hypoactivation in the medial prefrontal cortex (mPFC) and temporoparietal junction (TPJ) (e.g., Castelli et al.
2002; Wang et al.
2007; Watanabe et al.
2012), as well as in brain areas relevant for processing and resonating with others’ emotions such as hypoactivation in the inferior frontal gyrus and both under- and overactivation in the amygdala (e.g., Greimel et al.
2010; Klapwijk et al.
2016a; Monk et al.
2010; Pelphrey et al.
2007; Swartz et al.
2013).
In most of the neuroimaging studies on social processing in ASD, participants are merely required to observe others or to think about their mental states (e.g., Kana et al.
2015; Schulte-Ruther et al.
2011; Vander Wyk et al.
2014). Although these studies have greatly advanced the understanding of the neurocognitive mechanisms associated with social deficits in ASD, most do not take more interactive elements of social exchange into account. Studying such elements, however, is essential, as responding towards others involves different cognitive processes than merely observing others’ behavior (Schilbach et al.
2013). This is especially important because a discrepancy has been reported between potentially normative performance on explicit social tasks in ASD versus difficulties in applying social abilities during social interactions (Klin et al.
2003). For example, although adults with ASD do not spontaneously attribute mental states to others, they are able to understand mental states of others when they are explicitly encouraged to mentalize (Moran et al.
2011; Senju et al.
2009).
Paradigms inspired by behavioral economics are increasingly used to investigate social cognitive processes underlying social interactions in psychiatric populations (Hasler
2012; Sharp et al.
2012) including ASD (Chiu et al.
2008; Sally and Hill
2006; Yoshida et al.
2010). These paradigms not only offer simplicity and experimental control, but also have the advantage that they model interactive elements of social exchanges (King-Casas and Chiu
2012; Rilling and Sanfey
2011). Previous experiments using economic games suggest that people with ASD are indeed impaired in executing mentalizing abilities during interactive games. For example, adolescents with ASD show a different response in the middle cingulate cortex compared to controls when deciding to reciprocate investments in the trust game, suggesting problems with mentalizing during online social interaction (Chiu et al.
2008; Frith and Frith
2008). In a different strategic game, the stag hunt game, players can cooperate to hunt highly valued stags or act alone and hunt rabbits of lower value. Yoshida et al. (
2010) used this game to estimate participants’ representations of the other player’s intentions for cooperation. They found that adults with ASD made less use of these representations than control participants when playing the game (Yoshida et al.
2010). Further evidence comes from a study in which children with ASD had to judge others’ morality and subsequently played a cooperative game both with the child they judged to be morally ‘nice’ and ‘bad’. This study showed that children with ASD (in contrast to TD children) did not distinguish between morally good and bad partners in the cooperative game but did correctly judge others’ morality in basic moral judgment stories (Li et al.
2014). These studies using economic games thus also suggest that individuals with ASD are able to make explicit inferences about others’ intentions but are less effective in using this information when making interactive decisions.
Although it has been suggested that individuals with ASD are impaired in processing emotions of others (Adolphs et al.
2001; Baron-Cohen et al.
1997; Harms et al.
2010), studies using economic games among individuals with ASD did not focus on the role of emotions in social interactions. However, many studies in healthy populations have shown that emotions expressed by others during interactions can influence subsequent behavior of the observer (van Kleef et al.
2010). For example, disappointed reactions of others might lead to fairer subsequent responses in observers than angry reactions of others (Lelieveld et al.
2012,
2013b), whereas during negotiations displays of happiness might signal satisfaction leading to lower offers (van Kleef et al.
2004). Currently, evidence suggests that individuals with ASD are less likely to integrate emotional contextual cues into their decision-making (De Martino et al.
2008). Yet little is known about how they make social decisions in response to emotions during social interaction. Therefore, in the current study we examined if emotions expressed by others influence fairness decisions and associated brain responses in boys with ASD compared with TD controls. While being scanned, participants were presented with written expressions of anger, disappointment and happiness by peers in response to an earlier decision about dividing tokens, after which they were given the opportunity to divide tokens again. A previous study using this paradigm found that TD adolescents reacted with more fair allocations after they read disappointed reactions compared with angry and happy reactions from their peers (Klapwijk et al.
2013). Neuroimaging studies that used this paradigm found that when TD participants received happy reactions they showed increased responses in the TPJ, a brain area that is important for mentalizing and attention (Klapwijk et al.
2016b; Lelieveld et al.
2013a).
Based on previous work showing that individuals with ASD made less use of social information when making decisions (Izuma et al.
2011; Li et al.
2014; Yoshida et al.
2010), we expected that they would be less likely to integrate emotional contextual information into their decision-making processes. This would be reflected in less differences in fairness decisions between the three emotions in the ASD versus TD group. Predictions for neuroimaging results were based on previous studies in ASD that revealed altered activation compared to controls in brain regions involved in social cognition. Whereas most previous studies used facial emotions, the current study used written emotions, and we therefore expected to find differences in frontotemporal brain regions involved both in social cognition and language processing. For example, reduced activation in the inferior frontal gyrus has been reported in ASD when presenting emotional faces (e.g., Baron-Cohen et al.
1999; Greimel et al.
2010; Holt et al.
2014) and altered activation in ASD in this region during mentalizing and social cognition has been identified in two meta-analyses (Di Martino et al.
2009; Philip et al.
2012). Furthermore, prior studies that used the same paradigm as in the current study showed that the TPJ is sensitive to happy reactions in TD controls (Klapwijk et al.
2016b; Lelieveld et al.
2013a). Given reports of reduced TPJ activation in social tasks in ASD (Castelli et al.
2002; Lombardo et al.
2011), we also expected group differences here.
Discussion
This is the first study focusing on the effects of emotions on fairness decisions and brain responses in ASD. Behavioral analyses showed that ASD participants were more unfair when dealing with angry compared to disappointed and happy peers, whereas TD participants more often were unfair when dealing with angry but also with happy peers compared to those that communicated disappointment. These group differences were mainly driven by differences in reactions to happy peers, as the TD group chose significantly more unfair offers after happy reactions than the ASD group. The imaging results showed reduced brain responses in the precental gyrus and middle frontal gyrus in the ASD versus TD group when receiving happy versus angry reactions. Additionally, more autistic traits in the ASD group were associated with more activity in the postcentral gyrus in the happiness versus anger and disappointment contrasts.
Although we hypothesized that the ASD group would be less likely to differentiate between the three emotions when making fairness decisions, this hypothesis was not supported as the behavioral results suggest that individuals with ASD did adjust their allocation behavior in response to the emotions of others. However, participants with ASD reacted less unfair than TD controls in response to happiness (and more unfair in response to anger compared to TD controls, although this difference failed to reach significance). The increase in unfairness in response to happiness of the TD participants is in line with findings from previous studies (Klapwijk et al.
2016b,
2013; van Kleef et al.
2004). When receiving a happy reaction after a previous unfair offer, one could infer that the other was already satisfied and would therefore be content with another unfair offer (Cacioppo and Gardner
1999; van Kleef et al.
2010). Possibly, our participants with ASD used different heuristics that require less such inferences about mental states since they did not choose to be more unfair in response to happiness compared to the TD participants. However, this interpretation could not be supported by altered activation in brain regions usually associated with mentalizing in the ASD group in the current study.
We did not find group differences in the specifically hypothesized brain regions that have been previously linked to atypical social-affective functioning in ASD such as the IFG and TPJ (Greimel et al.
2010; Lombardo et al.
2011). The absence of group differences in these areas might result from the specific task used in the current study, in which written emotions were presented and participants made fairness decisions subsequently. However, previous studies did report differences between ASD and TD controls in these regions in tasks using written stimuli (Lombardo et al.
2011) and the TPJ specifically has been implicated in previous studies using the same paradigm as in the current study (Klapwijk et al.
2016b; Lelieveld et al.
2013a). It might also be that individuals with ASD do not recruit these hypothesized social-affective brain regions differently from controls when making social decisions. The only other study that used fMRI to study social decisions in an economic game in ASD found group differences between individuals with ASD and controls in the middle cingulate gyrus (Chiu et al.
2008), and not in either IFG, mPFC, TPJ or amygdala. Given the sparse number of neuroimaging studies that employed economic games in ASD and the posited potential for understanding mental disorders using neuroeconomics (Hasler
2012; King-Casas and Chiu
2012; Kishida et al.
2010; Sharp et al.
2012), future studies are warranted to further test which brain regions are differentially recruited when making social decisions in ASD.
Interestingly, however, the reduced responses observed in the current study in the precentral gyrus and middle frontal gyrus, and also the postcentral gyrus activation related to autistic traits, align with results from recent meta-analyses of fMRI studies in ASD (Di Martino et al.
2009; Dickstein et al.
2013; Patriquin et al.
2016). Hypoactivation during social tasks in ASD versus controls was found in both the left and right precentral gyrus in the meta-analysis by Di Martino et al. (
2009) and in the left precentral gyrus in the Patriquin et al. (
2016) meta-analysis. Reduced responses in this area in ASD versus controls have been reported during imitation of emotional expressions and finger movements (Dapretto et al.
2006; Williams et al.
2006) and when observing fearful expressions (Deeley et al.
2007). Although the precentral gyrus is considered to be part of motor-related cortex, activity in this area has previously been associated with social-emotional functioning. Precentral gyrus activity has been found to increase when receiving empathic responses from others (Seehausen et al.
2014,
2016) and activity in this area is also related to self-reported affective empathy in social versus nonsocial emotional scenes (Hooker et al.
2010). Furthermore, atypical functional connectivity within the precentral gyrus has been associated not only with impaired motor skills but also with social deficits in ASD (Nebel et al.
2014). In the current study, reduced activation in the precental gyrus was found in the ASD versus TD group specifically when contrasting happy versus angry reactions. This might suggest that the ASD participants process the happy emotional information differently than the TD controls in this area and therefore also responded less unfair in response to happiness than the TD group. However, future studies are needed to further clarify the role of the precentral gyrus in social-emotional functioning. For example, the current paradigm does not allow inferring whether the different response to happiness in the ASD group is the result of less responsiveness to happy emotions in general or to a different cognitive appraisal of happiness that leads to increased fairness and decreased precentral gyrus activation. Experiments in which the emotional intensity of happiness is varied could resolve whether responsiveness to happiness is related to precentral gyrus activation or not. The current findings as well as the precentral gyrus hypoactivation in ASD during social tasks in two meta-analyses (Di Martino et al.
2009; Patriquin et al.
2016) might point to a relation between precentral gyrus dysfunctions and social deficits in ASD.
The current results additionally showed a positive association between autistic traits and activity in the postcentral gyrus in the ASD group in the happiness versus anger and disappointment contrasts. The postcentral gyrus is a somatosensory region that is also not usually discussed in the context of ASD social deficits, although it has consistently been revealed as a hyperactivated region in ASD meta-analyses of social tasks (Di Martino et al.
2009; Dickstein et al.
2013; Patriquin et al.
2016) and it has also been reported as a region being structurally altered in ASD (Hyde et al.
2010). Previous studies in healthy populations have reported the involvement of primary somatosensory cortex in affective touch (Gazzola et al.
2012), in processing facial and vocal emotions (Adolphs et al.
2000; Heberlein and Atkinson
2009) and in affective language use (Saxbe et al.
2013). The relation between autistic traits and postcentral gyrus activation in response to happy versus angry and disappointed emotions in the current paradigm might suggest a specific relation between somatosensory processing of positive emotions and ASD symptoms.
Several limitations to this study should be noted. First, although our sample size (N = 19 per group) is comparable with other task-related fMRI studies in ASD, this sample size is relatively small and may have limited the power to detect group differences in brain regions usually linked to social cognition and emotion processing. Second, since our sample contained adolescent boys only, we do not know whether our results are generalizable to girls and to children and adults with ASD. Third, the task design employed in the current study contained written preset emotions only. Future studies could further increase the amount of interaction by studying face-to-face interactions, for example by using virtual reality. Finally, it remains unclear why differences in ASD versus controls were found in the precentral gyrus, whilst a correlation with autistic traits was found in the postcentral gyrus but not in the precentral gyrus. It can be speculated that the relatively small sample size has limited the power to find a correlation between precentral gyrus activation and autistic traits. It is also possible that a correlation between autistic traits and brain activity within the ASD group does not necessarily imply group differences in the same region between the ASD and TD groups.
In conclusion, the current study provides an initial step in examining how explicit emotional feedback influences interactive decisions and associated brain responses in ASD. The results suggest that individuals with ASD do employ explicitly expressed emotional information when making social decisions, although responses towards happiness seemed atypical and were fairer than controls. The neuroimaging results might point to a possible role of precentral and postcentral gyrus in social-affective difficulties in ASD, although more research is needed to specify the neurocognitive mechanisms that are associated with these brain regions during social cognition. Future research in which the role of others’ expressed emotions is further investigated could help to refine models for social interactions in ASD.
Acknowledgments
Open access funding provided by Leiden University Medical Center (LUMC). The authors are grateful to all participants and their parents, to the participating centers (Centrum Autisme Rivierduinen, Curium-LUMC), and to Romy Emmerig and Simone van Montfort for their help with data collection. This study was supported by the Netherlands Organization for Scientific Research (NWO) Grant No. 056-23-011.