Understanding Delusions: The Belief Learning and Memory Lab

Phil Corlett

I’m interested in beliefs. Specifically, how the brain is involved in normal and abnormal belief formation. For example, I study delusions, the often bizarre and fixed false beliefs that characterize serious mental illnesses like schizophrenia. I’m a cognitive neuroscientist, which means I use data from brains to make inferences about minds. 

 

I take what many consider to be a radically reductionist approach to beliefs. I think they might be related to simple behaviors like Pavlovian and instrumental learning. These processes can be observed in very simple organisms and I try to apply what we know about them to study beliefs.

Central to our understanding of learning, and I argue belief formation, is the concept of salience or importance. We learn and remember information about important events so we can respond appropriately if the same circumstances recur in the future. I think beliefs are one way that such learning and memory is manifest. If beliefs are so central to our comportment in the world, it is clear that there may be disastrous consequences when they are mis-specified. The salience hypothesis of psychosis rests on the simple but profound observation that subtle alterations in the way that we perceive and experience stimuli have important consequences for how important these stimuli become for us, how much they draw our attention, how they embed themselves in our memory and, ultimately, how they shape our beliefs. The original hypothesis went on to explore the putative neurobiology of incentive salience attribution and how excessive dopamine function in the striatum could lead to inappropriate salience attribution and ultimately to positive psychotic symptoms.

Whilst the theory is appealing in its simplicity and elegance, empirical work related to the salience hypothesis paints a more complex picture. Specifically, dopamine is, of course not the only neurochemical mediator of salience attribution, the basal ganglia are not the only site of salience processing and salience itself can be parsed into numerous sub-processes. The cognitive neuropsychiatry of delusions is much indebted to the aberrant salience hypothesis but the devil is in the details. I have an alternate view of the neurochemistry, neurobiology and phenomenology of psychosis related to, but different from, salience attribution; this view focuses on prediction error.

Prediction error represents the mismatch between what we expect in a given context and what we actually experience. Normally this mismatch acts as a teaching signal for belief updating. However, if prediction error is registered inappropriately, attention is driven toward irrelevant stimuli, thoughts, and percepts; and delusions are constructed to explain away such unpredictable experiences. My lab has generated empirical data with functional neuroimaging, psychopharmacological and behavioral studies that implicate prediction error in delusion formation. Furthermore, the prediction error model entails novel predictions about key aspects of delusions hitherto unaddressed by cognitive neuroscience; the fixity and elasticity of delusions in the face of contradictory evidence.

Our most recent work tests these predictions and finds that delusional beliefs do seem to be related in important ways to memory formation, memory maintenance and memory reconsolidation. For the first time, these puzzling symptoms are beginning to yield to insights from cognitive neuroscience; giving hope for better understanding and ultimately better treatments.