The challenges of neuroimaging the self,perspective-taking and psychosis risk

Park et al. (2020) report neuroimaging findings from a group at ultra high risk (UHR) for psychotic disorder. They were specifically interested in neural correlates of self-referential processing and perspective-taking, features that are often found to be disturbed in psychotic disorders. The 2 × 2 task design is useful, allowing inferences regarding self vs other-referential processing from both first-person and third-person perspectives. Findings indicated that UHR participants showed less neural activity in the left ventromedial prefrontal cortex/medial orbitofrontal cortex during self-referential processing compared to healthy controls and greater activity than healthy controls in the middle occipital gyrus during third-person perspective-taking. The former finding correlated with working memory deficits. There was no relationship between the imaging findings and disturbance of the basic sense of self.

Although the findings must necessarily be treated with caution given the small sample size (22 UHR participants), they are compelling because they indicate aberrant neural activity relating to self-referential processing and perspective-taking in patients deemed to be at risk of psychotic disorder (chiefly due to the presence of attenuated positive psychotic symptoms). This has not been sufficiently addressed in the psychosis risk studies to date and provides an important avenue for future research, particularly in studies that are sufficiently powered to examine neural characteristics and predictors of high-risk patients who progress to full threshold disorder. Another important consideration is to address the issue of specificity versus transdiagnostic correlates—that is, how specific are these findings to psychotic symptoms and to what extent are they present across symptom domains and diagnoses? It is difficult to adequately address this issue using a UHR sample as a study’s only clinical group given the clinical heterogeneity and high rates of comorbidity in this clinical population, including depression, anxiety and personality disorders (Beck et al., 2019), conditions in which the self-construct and perspective-taking are also highly relevant.

A more fundamental issue is the challenge of identifying neural correlates of the ‘basic self’. As the authors note, this refers to the implicit, pre-reflective aspect of consciousness, the fact that all phenomenal experience has a first-person perspective or ‘mineness’ built into it, like a silent coefficient or unnoticed bass line underpinning a piece of music (Parnas and Bovet, 2017). In the task used in this study, participants were asked to evaluate the extent to which certain adjectives described them well and the extent to which a family member would also describe them this way. While the object of inquiry here is clearly ‘the self’, the mental activity or process being elicited is one of evaluation, judgment, scrutiny. This would seem to be leaving the domain of the implicit and pre-reflective for the evaluative and reflective (i.e. from the ‘basic’ to the ‘narrative’ level of selfhood). In neural terms, this may be akin to the default mode network shifting to the task mode network (Carhart-Harris et al., 2013). As Legrand and Ruby (2009) have pointed out, many neuroimaging studies aimed at investigating the self show that self-relatedness evaluation involves a wide cerebral network the activation of which is not specific to self-processing and may be explained by the involvement of cognitive processes characterized by evaluation, specifically inferential processing and memory recall (they refer to this network as the Evaluation or E-network). In other words, the standard method of approaching neural substrates of ‘self’ through the lens of self-evaluation does not target the self in its specificity. Instead, drawing on phenomenological concepts, Legrand and Ruby propose that neural studies of self need to target the subjective perspective (i.e. how a representing subject relates to a represented object), rather than target self-evaluation. They propose that such self-specific subject–object relation is anchored to the sensorimotor integration of efference with reafference (i.e. the motor command of the subject’s action and its sensory consequence in the external world). Other promising (and overlapping) avenues can be found in research into predictive coding, the interoceptive–exteroceptive integrating function of the anterior insular cortex and the default mode network.

Finally, another pertinent issue in this fascinating field is that the sense of presence and synchronic and diachronic unity, all key features of basic selfhood, are unlikely to be as neat as an on/off switch and thereby easily parsed from task-oriented, evaluative or third-person modes. For instance, the first-person perspective is surely still active when asked to adopt a third-person perspective, because it remains my perspective of the other person’s perspective (i.e. the subject pole of the self-structure has not disappeared). Similarly, self-referential processing is a highly dynamic process, with first- and third-person perspectives braided around each other as we move frequently and seamlessly between first- and third-person modes (e.g. shifting in a millisecond from immersed activity to seeing oneself through another person’s eyes). Therefore, we are unlikely to identify neural correlates of basic selfhood that are simply activated or not depending on task. Rather, it may be useful for studies targeting the basic self to instead emphasize investigation of neural activity regardless of task (i.e. that which is present or common across mental orientations, even if attenuated in some tasks), highlighting features that are ‘hiding in plain sight’, and also to use symptom capture designs with participants who display profound disturbances of basic selfhood (e.g. subsets of patients with schizophrenia spectrum or dissociative disorders).