Testing a Bayesian model of auditory verbal hallucinations in the adolescent brain

点击翻译按钮获取中文摘要

Auditory verbal hallucinations (AVH) are a characteristic symptom of psychotic states such as schizophrenia. Adolescents are particular vulnerable to developing AVH and schizophrenia (Dalsgaard et al., 2020). It is imperative to characterise and model the developmental mechanisms of psychosis and to develop effective methods for early detection and diagnosis to optimize interventions. Little is certain about how the brain produces AVH. Some theories propose that AVH result from a failure to monitor one's own inner speech, verbal thoughts or memories so that they are perceived to be foreign or external (Frith, 2014). Others claim that AVH are intrusive memories (often from childhood traumatic events) that the brain fails to suppress (Waters et al., 2006). Because each of these models localises the pathology of AVH to one symptom dimension or cognitive mechanism, none can explain the full complexity of AVH alone. 'Bayesian inferencing' has become increasingly influential as it has the potential to unify multiple mechanisms of AVH under a generalised theoretical framework. The brain may use different mechanisms to compute the probability of a perception given prior experience and the current sensations. So whether speech is heard ultimately depends on the weight the brain attaches to previous experience (the priors) and to what is currently being heard. AVH may occur in the absence of a stimulus when the brain has a very strong and unchanging prior expectation (Corlett et al., 2019). The current project aims to test the Bayesian model of AVH to characterise and model the developmental mechanisms of AVH in adolescents. We will recruit people aged 15-25yo and: (1) prove the concept of a modelling approach by modelling experimentally manipulated prior beliefs (of hearing a speech) and sensory evidence (of whether the speech is present) in a series of signal detection tasks (Brookwell et al., 2013); (2) test if Bayesian modelling can separate healthy vs. at-risk and clinical groups more reliably than tasks measuring corollary discharge, source monitoring or memory inhibition; (3) localise the changes of prior beliefs to electrophysiological signatures in the brain using EEG (Yao et al., 2021), to identify potential biomarkers for non-invasive brain stimulation treatments. The PhD will benefit from interdisciplinary expertise in cognitive neuroscience and clinical psychology and receive comprehensive training in quantitative skills in Bayesian modelling, statistical analysis and EEG data processing, as well as in clinical assessment of psychotic symptoms and research governance