Overall Aim 1: Investigate the neuroanatomy and neurochemistry of self-blame-related overgeneralisation in major depressive disorder (MDD). Rationale for Hypothesis 1: Reductions in grey matter volume, abnormal metabolism and blood flow in the subgenual frontal area have been found in patients with unipolar and bipolar depression (Drevets et al., 1997). Direct brain stimulation of the subgenual cortex (BA25) reduces refractory symptoms of MDD in some cases (Mayberg et al., 2005). Individuals with depression show a propensity towards self-blaming emotions which are associated with feelings of worthlessness – a core symptom of depression (Harrison et al., 2022). Proneness to self-blaming emotions, including adaptive forms of guilt, has been found to be associated with increased fMRI blood oxygenation level dependent (BOLD) signal in the subgenual frontal area (Green et al., 2012; Lythe et al., 2022; Zahn et al., 2009) across remitted MDD and control groups. Overgeneralised forms of self-blame, such as self-hate, were associated with self-blame-selective decreases in functional connectivity between the subgenual cingulate cortex and the right superior anterior temporal Lobe (RSATL) (Green et al., 2012), a region shown to be important for differentiating social concepts (e.g. differentiating actions as ‘impolite’ or ‘absent-minded’ rather than overgeneralising them as ‘bad’) (Zahn et al., 2007, 2009). People with remitted MDD and higher self-hate showed reduced coupling for self- vs. other-blame between the RSATL seed with the anterior subgenual cingulate cortex, as well as with medial frontopolar, right hippocampal, and lateral hypothalamic areas. This was not due to medication effects or emotional intensity (Green et al., 2012). Furthermore, self-blame selective RSATL hyperconnectivity with a more posterior subgenual region, the subgenual cortex (BA25), has been found to predict MDD recurrence (Lythe et al., 2015). Recent work has led to a re-interpretation of the role of the subgenual area in self-blame (Lythe et al., 2022), suggesting that this area is not self-blame-selective per se, but is activated for blame irrespective of direction and its activation depends on individual variability in how frequently one blames oneself relative to others. As described in our review (Zahn et al., 2020), the posterior part of the subgenual area (BA25, subgenual cortex), is likely to serve different functions compared with the anterior part (subgenual cingulate cortex, i.e., subgenual parts of BA24/32). The RSATL is connected to the subgenual frontal area and to the limbic forebrain system (Nauta, 1960) via white matter fibres within the uncinate fasciculus (Vergani et al., 2016), which has been proposed to have a role in socio-emotional functions and mnemonic learning (Von Der Heide et al., 2013). Therefore, variations in the microstructure of this tract may contribute to a vulnerability to overgeneralisation and functional RSATL-subgenual hyperconnectivity found in MDD. Consistent with this suggestion, diffusion studies have reported lower fractional anisotropy and higher radial diffusivity in the uncinate fasciculus in MDD participants, suggesting that there may be reduced white matter microstructural complexity in MDD (de Kwaasteniet et al., 2013; Zhang et al., 2012). Consistent with the close anatomical connections of the subgenual cortex (BA25) with the hypothalamus (Nauta, 1960), the subgenual frontal area is likely to play a key role in modulating hypothalamic–pituitary–adrenal (HPA) - axis activity. Furthermore, metabolism within this region has been found to predict plasma cortisol levels (Jahn et al., 2010) and to affect cortisol dynamics during stress (Alexander et al., 2020). Cortisol has also been found to affect subgenual activation, for instance, administration of hydrocortisone has been shown to attenuate sadness-induced subgenual activity (Sudheimer et al., 2013) and higher stress-induced HPA-activity is associated with greater functional connectivity between the subgenual frontal area and the “salience network” in adolescents (Thomason et al., 2011). Chronic exposure to cortisol may also have a deleterious effect on brain structure and function (Echouffo-Tcheugui et al., 2018; McEwen & Gianaros, 2010) as cortisol levels have been associated with reduced volume and cortical thickness in prefrontal, anterior cingulate and hippocampal regions (Lee et al., 2002; MacLullich et al., 2006; Stomby et al., 2016). MDD has been found to be associated with HPA-axis dysregulation and increased cortisol secretion (Pariante & Lightman, 2008). Patients with current and remitted MDD showed flatter diurnal cortisol curves, however the sensitivity to stress or natural challenge (i.e., awakening) is more complex with both increases and decreases reported (Jarcho et al., 2013; Nandam et al., 2020). Rumination and self-blame were associated with flatter diurnal cortisol curves and lower cortisol awakening responses (Hilt et al., 2017; Kuehner et al., 2007; Lucas-Thompson & Hostinar, 2013), and exaggerated self-blame for stressful life events has been found to be associated with lower latent trait cortisol in adolescents, an effect that was moderated by maternal warmth (Stroud et al., 2020). Neuroanatomical studies have suggested that neglect and abuse in childhood and adolescence (Choi et al., 2009; Huang et al., 2012), and higher cortisol reactivity in early life may be associated with reductions in white matter microstructural complexity (Sheikh et al., 2014), and that positive early caregiving may moderate some of these effects (Sheikh et al., 2014). It is unclear how brain structure relates to functional connectivity, but one may speculate that weakened anatomical connectivity may render functional connectivity more vulnerable to stress-related neurochemical dynamics, which affect the long-term potentiation/long-term depression ratios (Holderbach et al., 2007) that underpin plasticity and integration of information in neural networks. For example, it has been shown that an optimal signal-to-noise ratio in frontal networks is modulated by neurotransmitters, such as dopamine, and appears to follow an inverted U-shape rather than linear function (Cools & D’Esposito, 2011). Therefore, it may be speculated that strong anatomical connectivity may render neural networks more resilient to stress-related changes in neurotransmitters, enabling networks to function effectively even with suboptimal levels of critical neurotransmitters. Overall Aim 2: Investigate the neuroanatomy of overgeneralised social knowledge in major depressive disorder (MDD) Background: We previously hypothesised that self-blame-selective overgeneralisation of social knowledge contributes to self-blaming biases and vulnerability to MDD. Social concepts when represented in a coarse, less differentiated manner e.g., interpreting one’s behaviour as “bad”, rather than “clumsy” were shown to have a more pronounced effect on negative feelings about oneself (Green et al., 2013). This may also contribute to less differentiated descriptions of negative emotions in depression (Demiralp et al., 2012), although self-reference or –agency is usually not controlled for in this literature, these effects may well have arisen from implicit self-related overgeneralisations. It has also been found that people with MDD and in remission from MDD showed overgeneral autobiographical memory, tending to retrieve fewer specific autobiographical memories, instead recalling non-specific events, such as repeated events, categories of events, extended events or personal semantic facts (Hallford et al., 2021; Weiss-Cowie et al., 2023). It has been suggested that this overgeneralisation may relate to functional avoidance of negative/traumatic memories or that rumination may interfere with memory retrieval (Sumner, 2012). Having more-differentiated social concepts and memories protects against overgenerlised self-blame-related emotions and rumination. Consistent with this view, higher emotional differentiation has been found to be associated with lower negative emotion intensity, lower depressive symptoms, lower neuroticism and higher self-esteem (Erbas et al., 2014; Willroth et al., 2019), and MDD patients have been found to feel more negatively when overgeneralising social concepts in the context of self-agency (Green et al., 2013). The descriptiveness (i.e. ‘richness’ of the description) of social concepts has previously been shown to engage the superior ATL on fMRI (Zahn et al., 2007) with a right hemispheric dominance shown using repetitive Transcranial Magnetic Stimulation (Pobric et al., 2016) and finding that right anterior temporal neurodegeneration led to selective impairments on social versus non-social concepts (Zahn et al., 2009, 2017), showing that this region is critical for representing social concepts, I.e. the context-independent, more abstract forms of social semantic knowledge. Green et al., (2013) found that social conceptual knowledge overgeneralisation was associated with overgeneralised self-blame and that the interdependence of conceptual overgeneralisation and high negative emotional intensity was associated with functional disconnection between the RSATL and right dorsolateral prefrontal cortex. Unpublished data in a larger sample of people (described in Lythe et al., 2022) with medication-free remitted MDD compared with a control group with no personal or family history of mood disorders, replicated these findings, but also showed a negative self-agency-selective overgeneralisation of conceptual social knowledge.