Our research

Decisions and decision making

Decisions are ubiquitous feature of our cognitive lives. We make decisions when selecting consumer products in the supermarket, or when choosing for whom to vote in elections. Psychologists and neuroscientists often think of decisions as occurring when animals make a commitment to a course of action, such a choosing to turn left rather than right at a fork in the road. Typically, animals make decisions that will maximise their outcomes, and/or allow them to achieve a goal. Less intuitively, we can also think of decisions are reflecting a commitment to a categorical proposition ('that is a chair'). Decisions of this sort, about perceptual stimuli, help us organise the sensory world into discrete categories. In the lab, we study information processing steps that occur as these commitments are made, charting the neural mechanisms that occur during human decision-making.

Computational approaches

One way of thinking about the brain is as a machine that processes information. We can try to understand the mechanisms that underlie decision-making by using computer simulations. We build mathematical models that emulate the information processing steps that we think characterise human decisions. We then present these models with inputs that resemble those experienced by human participants in our experiments, and compare their behaviours. Recently, we have begun to use biologically plausible neural network simulations to understand how humans learn to make decisions in novel settings.

Neural approaches

Behaviour is a consequence of the activity that occurs in neural circuits. In humans, this activity can be noninvasively measured with whole-brain neuroimaging techniques, such as functional magnetic resonance imaging (fMRI) and electroencephalograhy (EEG). Neuroimaging can help identify candidate brain regions involved in decision-making, but it can also help us arbitrate among different computational accounts of the decision process. To achieve this, we can use different computational models to make predictions about how brain activity should vary during decision-making. Using statistical methods, we can then compare simulated brain activity with observed human brain activity, and use this information to decide which model most closely resembles our human participants.

Do humans make good decisions?

Humans seem to be remarkably good at making decisions in some settings. Within minimal training, humans can make expert judgments about complex sensory stimuli, such as when recognising the face of a new acquaintance. However, in other settings, decisions seem to be biased and irrational. For example, humans will respond in different ways to the same question, depending on how it is framed. A key question for us is what cognitive mechanisms allow humans to make good decisions in some contexts and bad decisions in others. Understanding the quality of human decisions may help answer related questions, such as why some psychiatric disorders seem to be associated with poor-quality decision-making. Ultimately, we would like to work on interventions that will help everyone make better decisions.

Knowing when to decide

Decisions are based on evidence. For example, a doctor making a diagnosis might commission a series of tests, and base her judgment on all the evidence in favour of one condition over another. Evidence can take time to collect, and the time taken to accumulate evidence (or deliberate) can itself incur a cost. For example, if the doctor continues to commission more tests, the patient's condition might worsen. Knowing when to stop gathering evidence and commit to a choice is a hard problem to solve. A related problem is that not all evidence is equally reliable. For example, some test may be very diagnostic of a particular problem, and others less so. How does the doctor know which tests to believe? In the lab, we study how humans tackle these problems, which brain mechanisms allow them to do so, and they decide in a way that leads to the best outcomes.

Learning and decision-making

In order to make a good decisions, you need to have accurate knowledge about the world. The mechanisms by which we acquire information about the world (learning) are thus inextricably entangled with the mechanisms by which we make decisions. One challenging question is how humans make decisions in novel settings. Unlike current artificial systems, such as those based on advanced neural networks, humans seem to be remarkably good at negotiating novel environments, such as visiting a new city for the first time, where the language, currency and layout of the city may be unfamiliar. We think that this is because humans are able to form abstract concepts ('greeting', 'exchange', or 'taxi') that are transferable to novel settings where sensory information may be different. A key goal in the lab is to understand how these concepts are formed and represented in the human brain, and how they allow us to behave intelligently.