Introduction
Autism spectrum disorders (ASD) are characterized by impaired social communication as well as stereotyped and repetitive behaviors and interests (American Psychiatric Association,
2013). A recent review on cognitive impairments associated with ASD found that
mentalizing abilities were consistently attenuated in adults with ASD (Velikonja et al.,
2019).
Mentalizing summarizes a set of skills including abilities such as intention- or mental-state attribution, which are crucial for successful social interaction. Intention-attribution (IA), which falls under
mentalizing abilities, for instance, helps us understand and interpret other people’s behavior, or predict their next action. Impairments in
mentalizing abilities—particularly in IA—have been hypothesized to underlie deficits in social interactions and communication in indivduals with ASD (e.g. Lombardo et al.,
2011; Mason et al.,
2008). Studies examining IA in participants with ASD have often reported impairments in this ability (e.g. Schneider et al.,
2013; Schuwerk et al.,
2016; Vivanti et al.,
2011; Williams & Happé,
2010), but findings are inconsistent (e.g. Ciaramidaro et al.,
2015). In addition, the neural processes underlying IA remain understudied, particularly in ASD, although there is evidence that activation in brain areas associated with mentalizing abilities differs between participants with and without ASD (e.g. Ciaramidaro et al.,
2015; Kana et al.,
2014; Murdaugh et al.,
2014). Examining the underlying neural mechanisms could, therefore, provide important insights into how individuals with ASD process IA and whether this could impact their mentalizing ability.
A well-established method for investigating IA in the laboratory setting is the sequential comic strip paradigm (e.g. Vistoli et al.,
2011,
2015). Participants view comics that represent the different stages of an action or event and are asked to identify the congruous ending. The control condition, physical causality (PC) is very similar to IA, as it involves a sequence of events that logically lead to a certain outcome, but at the same time do not involve a person’s intentional actions. Predictions in the IA condition are based on the intention that is attributed to the character in the comic; predictions in the PC condition are based on various physical properties of objects and laws of physics, such as gravity. Therefore, both conditions draw on the same cognitive processes related to the processing of complex sequences of events and anticipation of consequences, but at the same time allow for the isolation of the processes that uniquely underlie IA.
Studies examining this paradigm using positron emission tomography (PET; Brunet et al.,
2000) and functional magnetic resonance imaging (fMRI; Ciaramidaro et al.,
2007; Walter et al.,
2004) in typically developing (TD) young adults have found differential activation patterns for IA compared to PC conditions with significantly increased activation for IA in frontal and temporal areas relevant for IA, such as the medial prefrontal cortex (mPFC) and temporo-parietal junction (TPJ).
In adults with ASD, fMRI studies comparing IA and PC with the sequential comic strip paradigm have found reduced activation compared to TD participants during IA in the TPJ, rIFG and left premotor cortex (PMC; Ciaramidaro et al.,
2015; Kana et al.,
2014; Murdaugh et al.,
2014), accompanied (with the exception of Ciaramidaro et al.,
2015) by an increased error rate for IA, but not PC trials. These findings indicate that attenuated IA performance in participants with ASD is associated with decreased activation in key brain regions underlying IA processing. However, due to the poor temporal resolution of fMRI, it remains unknown at which steps of IA processing these differences occur and which exact neural processes underlie IA impairments in individuals with ASD.
Neurophysiological methods, such as electroencephalography (EEG) or magnetoencephalography (MEG) can be used to determine a more precise time course of the neural signal underlying IA. Studies using MEG (Vistoli et al.,
2011) and EEG (Vistoli et al.,
2015) have implemented the comic strip paradigm in TD adults. In these studies, the comics were presented sequentially in order to compare neural activation between images, i.e. different processing stages related to IA. These comics were designed so that during the presentation of the 3rd image, the outcome of the story depicted in the comic could be predicted, either based on IA (i.e. the goal of the actor became clear) or PC (i.e. it became evident what would happen to the depicted object). Looking specifically at the 3rd image, they found that IA was associated with a significantly stronger positive activation approximately 200–600 ms after picture onset in the right posterior superior temporal sulcus (pSTS) and right TPJ, as well as the right inferior parietal lobule (IPL) compared to the PC condition. Additionally, the right intraparietal sulcus (IPS) also showed increased activation 240–540 ms after stimulus onset for IA (Vistoli et al.,
2011). Building on these findings, an EEG study found a significant difference between IA and PC in a positive component between 250–600 ms (with a peak at 300 ms) after onset of the 3rd image (i.e. enhanced amplitude for IA compared to PC) in bilateral posterior electrodes associated with IPL, TPJ and pSTS in TD adults (Vistoli et al.,
2015). This effect was only found for the 3rd, but not the first or second image, suggesting that this component may reflect increased difficulty associated with contextual integration (i.e. integrating the information from previous images) required for processing IA compared to PC (Vistoli et al.,
2015). Interestingly, the authors noted the similarity of this component with the P3-like component associated with context- and working-memory-updating processes (Friedman et al.,
2001; Ibanez et al.,
2012), and propose that IA is based on contextual integration (Brunet-Gouet et al.,
2011). Implementing this paradigm in ASD, therefore, will provide a more detailed understanding of which processes are affected in ASD, and whether impaired IA is rooted in aberrant contextual integration processes. Aberrant contextual integration has been hypothesized to underlie other ASD symptoms, such as repetitive behaviors and sensory abnormalities, within the framework of predictive coding theory (PCT; e.g. Pellicano & Burr,
2012).
Furthermore, the comic strip paradigm can also be used to examine neural correlates of the processing of information that is either congruous or incongruous to one’s own expectations, which is not only another important aspect of PCT, but also known to be affected in ASD (van de Cruys et al.,
2014). While previous ERP studies using the comic strip paradigm have not attempted to analyze neural processes related to the ending of the story, ERP studies applying other paradigms have demonstrated that several components related to incongruous information can be analyzed, as incongruous endings are expectation violations. Specifically, one ERP study examined ERPs for congruous vs. incongruous endings of stories and observed two distinct components (van der Cruyssen et al.,
2009): a positive frontal component around 200 ms after stimulus onset at electrodes associated with the mPFC, that was larger for incongruous endings, likely reflecting a redirection of attention toward the unexpected information, and a positive component in parietal electrodes associated with the TPJ around 300 ms after stimulus onset that was stronger for incongruous endings. This latter component is also consistent with ERP research assessing congruous/incongruous endings based on trait inferences (e.g. Bartholomow et al.,
2001; van Duynslaeger et al.,
2007,
2008). It is supposed to reflect a deliberate integration of the unexpected information into the preceding contextual information. In addition, continuous differential neural activation between congruous and incongruous endings was observed in medial or lateral frontal electrodes between 600 and up to 1200 ms after stimulus onset. Given that individuals with ASD have difficulties processing expectation violations, such as incongruous endings (for a review, see van de Cruys et al.,
2014), we aim to also examine activation during the 4th picture, in which the (congruous or incongruous) ending is shown.
In summary, the aim of the present study is to examine the P3-like ERP signature associated with IA during the presentation of the 3rd and 4th image of the sequential comic strip paradigm, and the later ERP associated with the processing of expectation violations during the presentation of the 4th image in participants with ASD. By comparing amplitude and latency of the ERPs we add time-related information on the neural processes underlying IA. In addition, we examine whether aberrant neural processing in ASD is specific to IA or may reflect a more general neural information processing deficit.
It should be noted that the abovementioned ERP-studies on IA and expectation violations have only examined adults. Maturation in areas relevant for social cognition, such as the mPFC, pSTS and TPJ, still takes place during adolescence (Blakemore,
2008,
2012; Choudhury et al.,
2008; Dumontheil,
2015; Mills et al.,
2014). This corresponds to reports of continuous improvement of social cognitive skills, for instance perspective-taking, face-processing ability and social attribution from childhood to early adulthood (Choudhury et al.,
2006,
2008; Hu et al.,
2010). In addition, an fMRI study contrasting activation between IA and PC in adolescents between 12 and 18 years and adults found a stronger activation in the mPFC in adolescence, while the difference in activation between IA and PC in adults was stronger in the right STS (Blakemore et al.,
2007), indicating a flexible involvement of social brain areas during development. To test whether the effects found by Vistoli et al., (
2011; 2014) could be replicated in children and adolescents, a pilot study with N = 6 children was conducted (see ‘
Methods’).
Based on the findings described above, we expected IA to be impaired in individuals with ASD. Therefore, the following hypotheses were tested:
1.
The ASD group will make more errors than the TD group when judging the ending in the IA, but not the PC condition.
2.
In accordance with previous studies, the P3-like component in posterior electrodes should be stronger for IA than PC in the TD group. If IA is based in contextual integration processes, and if these are impaired in individuals with ASD, this effect should not be found in the ASD group.
3.
Incongruous endings should cause a P3-like component in IA and PC conditions in TD, signifying deliberate contextual integration. If the ability to predict the ending in the IA condition is limited in individuals with ASD, this effect should not be observed in the IA condition in the ASD group.
4.
Finally, incongruous endings should cause a stronger late (approximately 700–1200 ms) ongoing positive component in both conditions in TD. If impaired IA in individuals with ASD is rooted in inflexible integration of expectation violations, the amplitude of this component should be stronger in the ASD group.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.