Brain imaging

Professor Paul Fletcher believes that exploring how the brain makes predictions about the world will help us to understand mental illness.

It is tempting, and perhaps comforting, to believe that our senses provide us with an accurate picture of the world. But they don’t.

It is tempting, and perhaps comforting, to believe that our senses provide us with an accurate picture of the world. But they don’t. Rather, what we perceive, and how we perceive it, is often determined by what we anticipate and what fits most comfortably with our prior expectations and biases.

The everyday experience of taking in data from the world, weighing it up and drawing conclusions implies that information flows exclusively in one direction: from perception to belief. Actually, it is a two-way street. It’s just that our beliefs about what is normal, predictable or logical may prevent us from experiencing the perceptions that violate our assumptions. This simple fact – that what we expect determines what we experience – has long been recognised by psychologists.

Professor Paul Fletcher, the Bernard Wolfe Professor of Health Neuroscience in the Department of Psychiatry, is investigating the ways in which we form beliefs about our environment, and how we use these to make predictions and decide what actions to take. Knowing how the brain normally does this may help us to understand why, under conditions of mental illness, people entertain perceptions and beliefs that may seem unusual and illogical. It may also tell us why we so frequently engage in behaviours that are detrimental to our overall health.

Deceiving ourselves

Why would our brain be so ready to refine, distort, add or remove sensory information when constructing our picture of the world? Perhaps it is because survival and success are not necessarily related to how accurately we represent the world but rather to how efficiently we can predict it.

If our brain tried to represent everything as accurately as possible, we would be paralysed by information. Since our aim must be to interact with our environment decisively, to gain reward and avoid pain, it seems better to sacrifice a detailed portrait of our surroundings in favour of quick snapshots of the important things, gleaned from a mixture of current information and prior experience. Allowing our knowledge to inform our perceptions helps us to achieve this while reducing the brain’s workload.

The implications of this view of perception and belief go well beyond the psychological laboratory. Our perceptions do not just represent our world, they create it: a self-conscious teenager hears ridicule in the innocent laughter of a passing group; an anxious man sees the outline of a bush as a waiting assailant; and wine from an expensively labelled bottle tastes better. In some mental illnesses, these perceptions can create a very frightening and bizarre world, one in which voices make critical and threatening comments and unseen persecutors control thoughts and actions. What happens in the brain to cause these altered perceptions?

Abnormal prediction

Advances in cognitive neuroscience have helped to understand how the brain learns about, and predicts and responds to, the world; this has begun to offer clues as to the processes that may be disrupted in mental illness. One process that Professor Fletcher believes may be particularly affected is the brain’s response to a mismatch between a predicted and an actual outcome: so-called ‘prediction error’.

Because prediction error effectively signals that we must learn something new about the world, it’s very useful in preventing us from becoming stereotyped, inflexible and unable to adapt to a changing environment. Persistent and inappropriate prediction error signal, though, would be detrimental. Imagine a world in which everything seemed to violate your expectation, everything vied for your attention because of its novelty and strangeness, even your own actions and thoughts. Such a world could rapidly become perplexing and threatening. Perhaps the only way of explaining such a change to yourself would be to conclude that you were the victim of some powerful persecutor – a common belief in some mental illnesses.

Professor Fletcher’s group, in collaboration with Professor Tony Dickinson, a learning expert based in the Department of Experimental Psychology, and Dr Philip Corlett at Yale University, has been studying the brain’s response to prediction error and examining the possibility that mental illnesses can be understood in terms of abnormal prediction error signals in the brain. Funded by the Wellcome Trust, they have been using functional brain imaging in patients and drug-induced, transient, psychosis-like experiences in healthy volunteers.

The results show that brain responses to violated expectations are indeed abnormal in psychosis and that the degree of abnormality seems to correlate well with the unusual beliefs and experiences that characterise the illness. It is even possible to predict in healthy volunteers undergoing brain imaging what sorts of symptoms they will experience when they subsequently receive a drug that disrupts prediction error signal. The work therefore provides evidence that a disrupted prediction error signal changes the relationship between prior beliefs and perceptions. This change could underlie some symptoms of mental illness.

Quest for reward

The symptoms of mental illness are one setting in which we might fruitfully examine changes in the relationships between perception and belief. But this framework also maps directly onto another domain that is core to human experience: the quest for reward.

Humans have evolved to be remarkably sensitive to the potential of any given environment to yield food. They respond to, and are motivated by, any stimulus that is predictive of such a reward. Indeed these stimuli, though not in themselves rewarding, may come to be valued and to enhance pleasure almost as much as the actual foods that they portend. The sights and smells of food and food-related stimuli drive behaviour very powerfully. In responding to foods, just as in trying to comprehend our environment, we remain reliant upon the complex interaction between our perceptions and our prior experience. Recognising this link could help us to understand the emergence of health-harming behaviours relating to food.

Professor Fletcher is therefore extending his work in collaboration with the Institute of Metabolic Science (IMS), and is working with Dr Sadaf Farooqi and Professor Steve O’Rahilly, leaders in the study of the genetics of obesity. This work, which is supported by the Bernard Wolfe Health Neuroscience Fund, sets out to examine how the brain’s ability to predict reward, and how the behaviours that emerge as a result of such predictions, may be useful to understanding overeating and its attendant health problems.

Once again, the use of functional brain imaging, informed by cognitive neuroscientific advances in our understanding of reward learning, is providing new information with potentially therapeutic value. Having explored the brain changes elicited by a widely prescribed anti-obesity drug, the researchers have taken the observations further to examine anti-obesity drugs of the future.

The potential of this approach has been recognised by the development of a unique ‘Academic Incubator’ involving GlaxoSmithKline, Professors Ed Bullmore and Fletcher in the Department of Psychiatry, Professors Barry Everitt and Trevor Robbins in the Department of Experimental Psychology, and Professor O’Rahilly in the IMS. This collaboration is investigating novel therapeutic drugs for obesity and addictive disorders. It represents a new way of capitalising on academic expertise in testing potential drug treatments and will allow a more direct route from basic cognitive neuroscience research through to practical benefits to patients.

For more information, please contact the author Professor Paul Fletcher ( at the Department of Psychiatry. Research described here takes place within the Behavioural and Clinical Neuroscience Institute, which is jointly funded by the Medical Research Council and the Wellcome Trust.


This work is licensed under a Creative Commons Licence. If you use this content on your site please link back to this page.