Understanding the neural and cognitive mechanisms of attributional styles and credit assignment in depression

Depression is the most common psychiatric illness. Yet, its causes remain poorly understood, despite considerable amount of research. I propose a novel approach to understanding depression, namely to investigate the role of credit assignment - or causal attribution - in depression. Credit assignment is about learning the causes of events when there are different potential causes. My experiments will involve measuring credit assignment behaviour in human participants, as well as assessing the underlying neural processes using modern brain imaging techniques.

In a nutshell, I will examine whether in depression, patients have problems with correctly attributing positive or negative outcomes to different sources - do they have a tendency to falsely belief that they are the source of negative outcomes? But that on the other hand, positive outcomes are due to external causes, such as the environment or other people? Do they only have problems identifying the source for outcomes external to them or does this extend to their own emotions, such as low mood? Any of these aberrant credit assignment processes could explain how depression maintains itself by continuously reinforcing negative views and attitudes about the self, thus leading to low mood.

To investigate what types of credit assignment are changed in depression, I propose a set of paradigms each testing experimentally a different aspect of credit assignment in depression. For this, I will use a computational psychiatry approach. This means, I will develop models describing the complex process of credit assignment in my paradigms in terms of mathematically defined aspects. For example, firstly making a decision in the expectation that it will lead to a certain outcome. Then assessing how the actual outcome compares to your expectation. And finally assigning credit to the internal or external causes that seem most likely. I can then assess which of the different aspects are more specifically affected in depression.

To test the neural effects, I will combine these measurements of behaviour with measurements of brain activity using functional magnetic resonance imaging (fMRI). I will use fMRI to identify the brain areas that are particularly engaged by specific aspects of credit assignment. Subsequently, I can test how deficits in credit assignment in depression can be linked to specific brain areas. In the past, cognitive neuroscience has been particularly fruitful in dissociating cognitive sub-processes as they often map onto very distinct brain regions.

To summarise, I am proposing a series of experiments looking at how patients with depression assign credit to different outcomes when there are multiple causes. I will investigate this question neurally and using quantitative computational modelling. Thus my proposal is an attempt to bridge the gap between psychiatric concepts and mechanistic models from cognitive neuroscience and psychology. In particular, my work will contribute to understanding some of the neural and behavioural mechanisms of the changes occurring in depression and attempts to treat it, using a multi-disciplinary approach.

Depression is the most common mental illness. In the past, depression has been linked to negative biases in perception, attention or memory. In particular, cognitive theories of depression have suggested negative attributional styles as the root cause of depression. However, the cognitive and neural mechanisms underlying this are not yet known. Recent advances in computational cognitive neuroscience have put forward quantitative descriptions of credit assignment, i.e. learning about the causes of events. I propose to investigate the cognitive and neural mechanisms of different aspects of credit assignment and test whether it can be used to provide an account of neural activity and behaviour associated with attributional styles in depression.

I will compare the behaviour of depressed patients with non-depressed participants and measure the neural mechanisms associated with the behaviour. The experiments I propose, focus on 1) how positive or negative events are attributed, to either to the self, environmental features or remain unassigned, 2) how attributions are made to the self or other people and how this information influences social decisions, 3) how emotional responses (as opposed to simpler, more concrete outcomes) are attributed, and how they might, if they unattributed or incorrectly attributed, interfere with other behaviours.

To address these questions, I will combine approaches from cognitive neuroscience and psychology. To characterize the behavioural effects, approaches will include computational modelling using approaches from machine learning, Bayesian optimal learning theory and hierarchical modelling. The neural mechanisms and changes in depression will be probed using computational analyses of functional magnetic resonance imaging (fMRI) data involving modelling and pattern classification techniques. The causal involvement of the areas identified using fMRI will be further probed using transcranial direct current stimulation (tDCS).

10-20% of the population will suffer from depression throughout their lives. As depression can be a very debilitating disorder, it is both a quality of life as well as economic issue. The aim of the research I propose is to understand attributional styles in depression better at the cognitive and neural level. As part of this I will design computerised tasks that can measure different aspects of attributional styles in depression. I can envision several potential practical applications of this work:

Firstly, a commercial application, in the mid-term future (within about 5-10 years), could be to use these tasks that I will develop as a test-battery to screen or refine new potential treatments for depression by e.g. pharmaceutical companies or researchers at universities. As a specific example, based on our neural results, we could suggest, again at some point in the future after appropriate piloting, brain areas that might be a target for non-invasive interventions such as transcranial magnetic stimulation (TMS) or transcranial direct current stimulation (tDCS).

Secondly, the computerized tasks could also be modified in the future to work as 'computer brain-training' interventions; if successful in clinical studies, these tasks could be widely distributed to patients with depression (or other patient groups with problems in attributional styles, such as anxiety) over the internet as an additional treatment. The advantage of this approach is that it is very up-scalable. Again, initial small-scale studies of this approach could be carried out within about 5-10 years. In the longer term then, the results of this work may benefit patients with depression by developing new treatments or complementing and enhancing current treatments.

The skills that I will develop during this fellowship will also have a wider impact on further mental health research that I am planning to carry out. The computational modeling and neuroimaging techniques will be applicable to other problems as they are a general approach for putting questions about symptoms of mental illnesses into experiments. This approach allows understanding symptoms that patients report in mechanistic terms and find out more about their basis in the brain. Thus the skills I will learn during this fellowship will be very valuable for my future career as a researcher.