Sense of Self: Evidence from Neurotypical people and people with Autism Spectrum Conditions

The self has a significant influence on human cognition across different processing domains (memory, perception, attention, action). For example, research often finds self-biases in cognition - the phenomena whereby people show enhanced cognitive processing for information processed in relation to the self, than for information processed in relation to someone else, or in other ways. Given that the self influences cognition, it follows that any individual with an impoverished self-concept should have diminished self-bias capacity/profile. Potentially, this is highly relevant for our understanding of autism spectrum conditions (ASC), which is characterised by atypical development of both the understanding of self (meta-cognition) and of other people (mindreading). One possible reason for this might be that self-other distinction is impaired - e.g., people are processing self-referential information similarly to other-referential information. However, empirical evidence for impaired self-bias in ASC is inconsistent, and importantly, as of yet, no theory or explanation has been put forward that fully accounts for the discrepancy. Therefore, across five experimental chapters (a total of 11 experiments), this thesis investigates typical and atypical self-referential cognition by analysing evidence of self-biases in neurotypical adults and adults with ASC, in the domains of perception, action, and mental states. Chapter 2 (mental-state level) found that whereas neurotypical participants (N = 28) showed a moderate-to-large and statistically significant ownership effect in Experiment 1.2, autistic (N = 28) participants showed a small and statistically non-significant effect (i.e. comparable memory for items encoded in relation to the self and in relation to the experimenter). Importantly however, Chi-squared analysis revealed that there was no categorical difference between groups in the number of people who did/did not display an ownership effect, which suggests that although the size of the ownership effect was smaller in ASC (as shown by our continuous variable analysis), autistic participants were no less likely than neurotypical participants to display an ownership effect, thus suggesting that the ownership effect is not definitively impaired in autism. Furthermore, the ownership effect was unrelated to the number of autistic traits in either the case-control sample (Experiment 1.2: N = 56) or in a larger sample of neurotypical adults (Experiment 1.1: N = 100). Chapter 3 (mental-state level) investigated scene construction - the ability to mentally generate (i.e., imagine) and maintain a coherent spatial scene/event in mind (critical for episodic simulation), and found that scene construction ability was diminished (i.e., lower in quality) in autistic (N = 55) compared to neurotypical (N=62) adults, and was negatively associated with autistic traits. However, the ability to infer others' mental states (i.e. Theory of Mind; ToM) and to recognise emotions in oneself (i.e. alexithymia) were found to be better predictors of scene construction ability than an autism diagnosis. The frequency of descriptions of sensory experiences (which followed the same pattern in both groups: sight > sound > touch = smell > taste) and self-reference did not differ between groups. Chapter 4 (perceptual level) presented two experiments that aimed to test findings of undiminished perceptual self-bias in ASC when using unfamiliar stimuli Both in a case-control sample (pre-registered Experiment 3.2: N = 57), and in a sample of neurotypical adults (Experiment 3.1: N = 59), behavioural results showed that participants experienced both a self-bias and a familiarity effect. Importantly, the number of autistic traits was unrelated to the magnitude of self-bias and familiarity effects on RT or accuracy (in Experiments 3.1 and 3.2) and there was no between-group difference in the speed and accuracy of self-bias and familiarity effects (Experiment 3.2). Analysis of the ERP indices revealed clear evidence of a self-bias in both groups, with self trials eliciting larger N1 and P3 components than familiar ('friend') and unfamiliar ('stranger') other trials. Furthermore, the familiarity effect was largely absent on ERP measures, though the ASC group elicited some modulation of N2 reflecting familiarity conflict. Chapter 5 (mental-state level) investigated the malleability of self and other representations. In two neurotypical adult samples, we found evidence of both simulation induced malleability (Experiment 4.1; N = 163) and self-reflection induced malleability (Experiment 4.2; N = 162). Furthermore, the size of simulation induced malleability was positively associated with the number of autistic traits, but only when the simulated other was considered as the "average person" and not a "similar-other". Furthermore, in two pre-registered case-control experiments, there was greater simulation induced malleability in autistic than neurotypical adults (Experiment 4.3; N = 50) irrespective of similarity of the target, but no self-reflection induced malleability was observed in either diagnosis group (Experiment 4.4; N = 49). Chapter 6 (action level) measured participants' ability to distinguish their own motor repertoire from those displayed by other people of differing levels of similarity - specifically, the self vs. unfamiliar adult (similar-other) vs. unfamiliar child (dissimilar-other). We found a typical imitative compatibility effect in behavioural performance, which was not associated with the number of autistic traits (Experiments 5.1 and 5.2), and did not differ between neurotypical and autistic participants (Experiment 5.2). Furthermore, associated mu/alpha (8-13Hz) and beta (13-35Hz) desynchronisation - which are thought to reflect an index of mirror neuron activity - also did not reveal consistent differences between groups (Experiment 5.2). However, the key manipulation here, the referent effect, did not show any effect in behavioural performance (Experiment 5.1 and 5.2), and the imitative compatibility effect in mu/alpha and beta did not differ between referents (Experiment 5.2). The theoretical and practical implications of these findings for our understanding of self-awareness/self-referential cognition in ASC and in the neurotypical population, are discussed.