top of page

Self-blame in Depression: Can we use a brain training approach to tackle it?

Freud, the founder of psychoanalysis, is widely credited with highlighting the importance of self-blame-related feelings in depression. Although, critical of many aspects of Freudian theories of depression, cognitive theory proposed by American psychiatrist Aaron Beck, is very much focussed on self-critical thinking as well and this theory underpins cognitive behavioural therapy (CBT), which is the most widely available type of therapy for depression in the UK. Beck had already proposed in his early work that cognitions (thinking) and affective disturbance (depressive feelings) are separable and criticised that Freudian theories of depression focussed too much on feelings of guilt rather than self-critical cognitions, which he hypothesised to be the primary vulnerability factors for subsequent depressive feelings. Teasdale, a British psychologist, proposed a modification to Beck’s cognitive model which assumes that depressive cognitions and emotions are intertwined and can activate each other in both directions. I am a Reader in Mood disorders at King’s College London and an Honorary Consultant Psychiatrist at the South London and Maudsley NHS Trust. I must admit that I have struggled to be convinced by the dualism between cognition and emotion, in that any emotionally relevant cognition (e.g. “I am always failing at things”) will always be inextricably linked with emotion, and any complex emotion (e.g. guilt) will always entail cognition (e.g. the outcome of my action). I was fortunate to work with two scientists during my time as a postdoc at the US National Institutes of Health in Bethesda who used brain imaging to understand this fundamental question. Brain imaging is the use of quantitative techniques to study the structure and function of the brain, in a non-invasive manner. My supervisor, Jordan Grafman, had formulated a model of the frontal cortex (located near the front of our heads) as a long-term memory for sequences of events and actions. Based on his model, we predicted that representations of preceding and subsequent events of social actions in the frontal cortex will be crucial for understanding blame attributions. My postdoc colleague, Jorge Moll, had just carried out pioneering functional magnetic resonance imaging (fMRI) studies of moral feelings such as guilt. fMRI measures brain activity by detecting changes associated with blood flow using brain scanners, and helps detect abnormalities as well as the normal function of different parts of our brain. Our collaboration led to the observation that people who were prone to feelings of guilt had a higher activation in the subgenual a part of the frontal cortex, which sits in the midline in the depth of the front part of our brain (see image below) and had been previously associated with depression.

This is a three-dimensional reconstruction of a structural MRI of a brain. The blue line indicates a cut where the outside of the brain has been removed, so that one can see the depth of the frontal brain, more precisely the midline. The front of the brain points to the right in this picture.

In a separate set of studies, I was interested to see whether there is a part of our brain which specialises in representing the meaning of social behaviour, (e.g. what it means to act in a “stingy” or “generous” way).

Our work built on evidence from patients with a particular form of dementia in which the tip of the part of the brain beneath our temples, called anterior temporal lobes (as shown in the image below) were shrinking. This change resulted in a loss of understanding the meaning of things in these patients and was accompanied by inappropriate social behaviour.

A group in Cambridge (led by Professors Patterson and Hodges) and Manchester (led by Professor Lambon-Ralph, whom I later joined) had produced a body of evidence showing that this loss of understanding was not because of a language impairment, but because these patients lost a more abstract representation of meaning independently of whether it was language- or picture-based. We extended this work by showing that the upper right part of the anterior temporal lobe was more relevant for social than general understanding.

This is a picture of the right outer part of a brain with the skull around it. Depicted is activation for social concepts in the upper part of the right anterior temporal region. Described in this publication.

Based on the aforementioned work, we hypothesised that the integration of information between the right upper anterior temporal lobe and the subgenual frontal region is important for helping people interpret social behaviour in a nuanced way and thereby may protect them against gross overgeneralisations about themselves as Beck observed happening in people with depression (e.g., thinking of oneself as a total failure for a minor error and blaming oneself for things that are outside of one’s control).

Indeed, we found evidence for abnormal crosstalk between the anterior temporal and subgenual region, when imagining self-blame-evoking situations in people with major depressive disorder who, despite recovering from symptoms, subsequently developed another depressive episode over the next year. This was measured by investigating how the fMRI signal in one region correlated with the signal in the other region over time and is called functional connectivity.

This finding prompted us to wonder whether one could use brain training approaches to help people with depression change the way these two brain regions exchange information when feeling self-blame.

In our first study, undertaken by Jorge Moll’s team at the D’Or Institute in Rio de Janeiro, we were able to show that people who had recovered from depression were indeed able to change the fMRI correlation between these brain regions in a single training session. This was done by giving people cue words which they had previously defined as being associated with a self-blame-related memory and measuring the level of correlation between the brain regions to provide visual feedback on a screen in the fMRI scanner to people for brain training. We showed that the active brain training, also called neurofeedback, changed the brain region correlations and increased people’s self-esteem which is what we predicted given that nuanced interpretations of one’s actions protect one’s self-esteem from devaluating oneself as a whole.

Furthermore, in a second study which we carried out at King’s College London and published recently, Tanja Jaeckle recruited people who had not benefitted from a standard depression treatment and randomised them to two intervention groups:

  1. A solely psychological intervention, in which they were given strategies to choose from to tackle self-blame-related memories in three sessions,

  2. A functional MRI neurofeedback intervention, where these strategies were supported by giving people feedback about their subgenual-anterior temporal brain correlations in the scanner.

Surprisingly, both groups had reduced depression levels by 46% after 5–6 weeks, and there was no added benefit of using our neurofeedback approach. However, on further analysis, we found an important clue why this may be the case.

People with a depressive episode with anxious distress, a novel subtype of depression defined in the latest classification of the American Psychiatric Association (DSM-5), responded significantly better to the solely psychological intervention than people without anxious distress, who benefitted much more from neurofeedback. So, we concluded that the brain training signature we had been using may only be relevant for non-anxious depression and may be irrelevant for anxious depression, which could explain why people with anxious depression appeared to be distracted rather than aided by the brain training.

If this is true, then future studies could be successful at tackling depression using brain training by providing different brain training signatures to different patients depending on their symptoms and fMRI patterns. For example, other promising neurofeedback approaches to depression developed by Drs Young, Bodurka and Prof. Linden’s groups are based on increasing brain responses to positive memories and images, respectively, and may be of particular relevance to anxious depression. We hope to be contributing to this exciting yet highly complex line of research over the next decade.


Header image by Atormentado on Shutterstock


bottom of page