Neural Correlates of Reflection on Present and Past Selves in Autism Spectrum Disorder

A critical aspect of self-related cognition is the ability to remember events that occurred in one’s past (McAdams 2001; Conway 2005). It has been suggested that difficulties in accessing specific autobiographical memories in ASD may be due to problems in using the self as an effective memory organizational system (Crane et al. 2009). In a study examining narratives of self-defining and everyday autobiographical memories in adults with ASD, it was shown that individuals with ASD extracted less meaning from their memories than adult controls, which may be interpreted as a failure in using past experiences to update the self (Crane et al. 2010). However, no previous studies have examined the temporally-distant selves in ASD using a task other than memory tasks.

In the present fMRI study, we investigated the neural correlates of adjectives (referring to personal features) assignment to the present self, the past self, and a personally-relevant person in high functioning individuals with ASD. During fMRI scanning subjects were required to evaluate whether adjectives were suitable to describe their person at present, their person in the past (at the time of middle high school), and a close-other (friend or family member). The set of adjectives for each target of reflection included an equalized number of positive and negative features.

On the behavioral level, no significant between group effects were found while analyzing all experimental conditions together (present-self, past-self, close-other). However, analyses done for each condition separately revealed significant inter-group differences in the number of positive attributions made while evaluating personal features of the past self. In this case, individuals with ASD presented significantly weaker positivity bias (Mezulis et al. 2004) than the TD group. On the other hand, both groups presented similar positivity bias in the case of present self and the close-other conditions.

On the neural level, fMRI results for both groups of subjects collapsed together showed significant activations in MCG, ACG, Insula, and MFG, typically reported for self-related information processing and mentalizing (e.g. Lombardo et al. 2011; Murray et al. 2012; Kestemont et al. 2015) as well as for reflection on present and past selves (D’Argembeau et al. 2008). In addition, we found activations in the right IFG and precuneus that were previously reported specifically in the case of self-reflection across time (D’Argembeau et al. 2008). However, we did not found significant differences in brain activation between self-related and other person-related conditions. It is worth noting that in the current study, we compared brain activity during evaluation of the self with the close-other—not like in many of the previous studies with a famous (e.g. Kelley et al. 2002) or familiar but not personally known person (Pfeifer et al. 2007, 2013a). As it has been already shown, this distinction is important for the extent of differences between brain responses for the self and other. The areas of activity are highly common if the other is personally known, very significant person (Kennedy and Courchesne 2008; Tacikowski et al. 2013; Laurita et al. 2017). A lack of differentiation between the self and the other in cortical midline structures (CMS) has been observed in prior studies, most frequently when targets of reflection were either very close or similar to the self (Ochsner et al. 2005; Krienen et al. 2010). Interestingly, results of one of the studies on self-processing, where close friend was included as a condition in an evaluative perspective task, revealed the differences in brain activity between self and other person conditions only in the group of adolescents and not in the group of adults (Jankowski et al. 2014). It is also worth noting that in our study the contrasts for conditions were made for both groups ASD and TD collapsed together. Therefore, the effects specific for ASD group may bias our comparisons resulting in less typical activation patterns that had not been reported in previous research on general population. Specifically, it was previously shown that in contrast to TD groups, CMS activations in individuals with ASD were similar for the self and other (e.g. Kennedy and Courchesne 2008) or even stronger for other than the self condition (Lombardo et al. 2010).

Moreover, analysis of fMRI data revealed differences between the ASD and TD groups that were common for all experimental conditions (present-self, past-self, close-other) collapsed together. During the attribution process ASD subjects exhibited elevated activity in one cortical structure—the right posterior STG/TPJ. Interestingly, further exploration of this result suggested that this stronger activity was driven mainly by the inter-group difference in two of the analyzed conditions—past-self and close-other. Finally, analyses done for each condition separately showed significantly stronger activation in individuals with ASD when compared to TD group in the ‘past-self’ condition in a number of brain regions: the right insula, PCG, right and left posterior STG/TPJ, and left MTG. Importantly, percent signal change measures suggest that the inter-group differences in all of those regions are driven by the enhanced activity among ASD subjects and diminished activity in TD group.

The effect observed in the ASD group during retrieval of personal attributes about the past self could not be directly related to previous findings, as none of the earlier fMRI studies on ASD investigated the neural mechanisms underlying reflection on personal features of the past self. Nevertheless, there are some important links between our results and the published literature. For example, in a very recent fMRI study that included subclinical and clinical ASD groups, subjects were engaged in a causal attribution task that included self, another person, and situation conditions. Similarly to our findings, hyperactivity in the neural network including the posterior STS and TPJ was reported in individuals with ASD (Kestemont et al. 2016). Our results are also in line with several previous studies that revealed stronger activity or diminished task selectivity of relevant brain structures during performance of mentalizing about self and others (e.g. Mason et al. 2008; Lombardo et al. 2011; Schulte-Rüther et al. 2011).

Importantly, we observed atypical processing of information related to the past self on both the behavioral and neural levels. Behaviorally individuals with ASD were less inclined than their TD counterparts to assign positive attributes to their own person in the past. It was previously shown that TD youths make more positive self-appraisals than youths with ASD during self-evaluation tasks (Pfeifer et al. 2013b). ASD children were also characterized by diminished preference for memorizing positive over negative adjectives when those attributes had to be referred to themselves (Burrows et al. 2017). Moreover, in our study the ASD group was characterized by broad activity mainly in the posterior brain regions while reflecting on the past self. Importantly this extended network of activity includes areas that were disengaged in the TD group in the past-self condition. These results reflect highly atypical pattern of brain regions involved in retrieval of autobiographical memory and self-characteristics from the past in individuals with ASD. We hypothesize that stronger activations in the past-self condition observed in individuals with ASD may reflect the higher—in comparison to control subjects—level of difficulty of reflecting on one’s own person in the past. This notion is based on numerous studies showing that the more difficult the task the stronger and/or larger the activations (e.g. Gould et al. 2003; Erickson et al. 2007; Nowicka et al. 2011).

In line with the latter, enhanced engagement of right and left posterior temporal regions and parts of the TPJ may suggest the need for involvement of some additional brain resources by subjects with ASD to process more ambiguous social information. It was previously proposed that hyperactivity of the STS/TPJ may reflect the engagement of those areas that are involved in a more general cognitive processing which compensate for impairments in attribution judgements (Kestemont et al. 2016). In our study, apart from the right TPJ—the structure consequently linked to mentalizing abilities—we found activity in neighboring areas of the STS/STG and MTG. In previous research, those structures were linked to recognition of other people’s intentions mainly based on their body and gaze movement, and had direct connections with the mentalizing networks (Pelphrey et al. 2011). Importantly, activations in TPJ and adjacent posterior STG, found in the ASD group, may be also linked to reflected self-appraisals—a task involving mentalizing processes (Pfeifer et al. 2009; Pfeifer and Peake 2012). Although our participants were not instructed to think about other people’s perspectives on the self, they seemed to activate components of the social perception network including TPJ and posterior STG (Pfeifer et al. 2009, 2017). It was previously shown that neuro-typical adolescents when compared to adults during both direct and reflected self-appraisals more extensively engage brain regions involved in perspective-taking, like TPJ (Pfeifer et al. 2009). This may suggest a delayed development of cognitive networks involved in the past self processing in examined adult ASD subjects. It may be also hypothesized that in individuals with ASD their characteristic of the past self is based to a higher extent on information received from social environment (e.g. family members, therapists) than on the internalized knowledge about themselves. The later notion, however, is highly speculative in nature.

Additionally, we found a cluster of increased activity in our ASD group that included the very posterior part of the cingulate gyrus and cuneus. Activity in this area may suggest the involvement of imagery and memory of concrete episodic events using visual representations during retrieval of past self attributes (e.g. Addis et al. 2004). This enhanced activation of posterior brain regions found in the ASD group may also indicate that individuals with ASD seem to be ‘visual thinkers’ (Grandin 2009), i.e. individuals with ASD may tend to visualize one selves in the past in order to be able to describe/characterize themselves in the past.

All in all, findings of the current study may suggest that in our study individuals with ASD recruited additional areas to engage imagination and more embodied representations of their own behaviors and intentions from the past, which were crucial for recollection of personal traits describing the past self. They may also refer to the information about their past characteristics received from their social environment to a higher extent than the TD adults. Thus, recruitment of additional areas may serve as a kind of a compensatory mechanism for abnormalities in autobiographic memory reported among subjects with ASD (Brezis 2015).

While significant inter-group differences were found in the past-self condition, they were absent in the present-self conditions. This may also be viewed in the light of research on autobiographical memory in ASD. Studies on episodic autobiographical memory for different periods in subjects’ lifetimes found that in ASD memories for recent events were more specific and detailed than for remote events (Bruck et al. 2007; Crane and Goddard 2008; Tanweer et al. 2010; Goddard et al. 2014). Thus, one may assume that assignment of adjectives to one’s own person at present was much easier than to the self in the past; the latter forced individuals with ASD to ‘mentally move’ to their far past in order to judge whether some traits were suitable to characterize their person while attending middle high school. There is rather no doubt that this task can be characterized as episodic autobiographical memory. However, one may speculate that depending on the trait presented to participants, their personal experience and their self-awareness, they may sometimes refer also to their semantic knowledge. The reason is that some of the personal characteristics are rather stable over time (e.g. intelligent, blond) and other not necessarily (e.g. responsible, tall). Therefore, we assume that semantic and episodic memory may be interactively engaged, if a subject could not retrieve his specific characteristic in the past and had to ‘mentally move’ to the past situations using episodic memory.

Our supposition is directly supported by fMRI findings of the present study, i.e. stronger activation of posterior brain regions, including the PCG and bilateral temporal areas in the ASD group. Importantly, numerous previous studies have linked those structures with both episodic and semantic autobiographic memory retrieval (for meta-analysis see: Martinelli et al. 2013). Martinelli et al. proposed to include the conceptual self as one of the components of autobiographic memory that specifically relates to abstract self-representation and personal traits identified and internalized across the lifespan experience. However, the conceptual self was specifically related to activity in the frontal brain structures (Martinelli et al. 2013). Importantly, in the present study frontal cortical structures were not found among clusters of significantly stronger activity in individuals with ASD.

In conclusion, we propose that the specific difficulties observed in the ASD group in attributing personal traits to the past self resulted in extensive engagement of neural mechanisms related mostly to the semantic and/or episodic components of autobiographic memory (Crane and Goddard 2008; Brezis 2015), with additional components of imagery (e.g. Addis et al. 2004) and embodied representations of the past events and facts (Pelphrey et al. 2011).

Finally, we would like to comment on the limitations of the present study. First of all, the study is confined by a small sample size. Thus, the reported results should be treated with caution and they need to be further investigated in larger groups of individuals with ASD in order to enhance statistical power. Secondly the experimental group was limited to high functioning adult males. Therefore our behavioral and brain imaging results could not be generalized to all individuals with ASD, irrespective of sex. Moreover, as they were high functioning individuals, there is a possibility that the alternative strategies of activation of brain areas that they showed in this study would be different from those detectable in low functioning persons with autism. In addition, a higher number of experimental trials/events focused on mechanisms of past self-processing would also be beneficial; this would enable more detailed investigation of effects related to—for instance—positive and negative adjectives that could be detected in fMRI data. Moreover, it would be worth to include present and past aspects of the close-other into experimental design of future studies in this field. We propose that in such case the close-other should be pre-defined as a friend that had been the friend since the time of attending a high school. As a consequence, the definition of past self and past close-other would be the same and participants would be required to refer to their and the close-other’s personal characteristics at present and during the same time period in their past. This approach would enable comparisons that may reveal additional inter-group differences associated with processing of information about the distant past.