The role of action outcomes in cognitive control

The aim of our research is to understand how human thought and action are shaped by the goals and intentions we have, a function termed 'cognitive control' in psychology and neuroscience. The ability to act in a planned and purposeful way to achieve our goals is vital in everyday life, and is thought to rely particularly on the frontal lobes of the brain. This idea has significant implications: The frontal lobes are the part of the brain that is slowest to develop as we pass through childhood and adolescence, fastest to decline as we age, and is among the most affected regions by psychiatric conditions such as schizophrenia and depression. All of these groups struggle to some degree in situations requiring cognitive control: learning new skills, coping with many simultaneous demands on attention ('multi-tasking'), and ignoring distractions. Understanding the psychology of cognitive control and the neuroscience of the frontal lobes therefore promises to have important implications for education, treatment in psychiatry, and understanding the challenges of an aging population. Our research specifically investigates how people predict and learn from the consequences of their actions. To do this, we study how people perform carefully designed cognitive laboratory tasks, and in some experiments measure their brain activity as they do so, to answer the following questions: 1) Does focusing on the desired outcome of a task help us to perform the task better? Our hypothesis is that people should be better able to ignore distractions when the goal of their task is clear and understandable. Using neuroimaging, we expect to see brain activity relating to task goals even before those goals are reached, and from this activity we should be able to predict how well people will perform a task. 2) How do positive outcomes, such as financial rewards, help us to learn? We are testing the idea that, when rewards are earned, this reactivates brain regions that were involved in performing the task that led to the reward, strengthening their involvement and therefore improving task performance in future. By answering these questions, our research should help us to understand a central feature of human behaviour: our ability to direct our thoughts and activities toward achieving our goals. Our findings will not only increase scientific understanding of this basic human trait, but should also lead to new ideas about how to improve practice in the classroom, clinic, and beyond. For example, recent research has shown that focused training in cognitive control tasks can result in improvements in educational outcomes in schoolchildren, and can reduce some of the cognitive difficulties experienced by patients with schizophrenia. Better understanding of the brain systems responsible for cognitive control, and improved methods for measuring these abilities, should in the future lead to improvements in these applications of scientific research on cognitive control and the functions of the frontal lobes.

The proposed project investigates the role of action outcomes--rewards and sensory action effects--in the control of behaviour, exploring novel territory at the intersection of research on human cognitive control, instrumental action selection, and reinforcement learning. These literatures each focus on goal-driven behaviour, but differences in their theoretical and methodological approaches have limited cross-stimulation to date. As a consequence, little is currently known about (A) cognitive control of instrumental tasks in which actions are directed towards achieving particular outcomes and, conversely, about (B) the role of reinforcement in learning high-level task representations. The proposed research investigates these issues in two series of studies that extend behavioural, EEG and fMRI approaches used successfully in our prior work in these separate fields. Series A evaluates the hypothesis that anticipation of task outcomes--specifically, the sensory consequences of actions--contributes to effective online control of behaviour. Our approach contrasts with the traditional focus of cognitive control research on tasks in which the goal is defined in terms of the response required rather than the effect that the response yields in the environment. Series B evaluates the hypothesis that reinforcement drives learning by feeding back a teaching signal to reactivate areas that successfully fulfilled the task at hand. Extending previous research, we focus specifically on reactivation of high-level task representations. Both series use novel paradigms that incorporate meaningful outcomes into competing cognitive tasks. Using tasks with dissociable neural correlates, neuroimaging methods can be used to probe the modulation of task-selective cortical regions according to anticipated and experienced action outcomes. Together these studies will reveal the pivotal role of action outcomes in the online control of human action and in the acquisition of complex behaviours.

The primary goal of this project is to advance scientific theory and practice in the study of human voluntary action and cognitive control. The underlying neurocognitive mechanisms of goal-directed behaviour and learning are crucial to most higher-level mental faculties. Our findings therefore promise to have significant implications for measuring, understanding, and improving human learning and behaviour in a variety of basic science and applied contexts. Our ongoing collaborations and local collaborative opportunities provide specific pathways to impact for the proposed work. 1. Improving predictive cognitive markers throughout the lifespan Lab measures of cognitive control such as inhibition and working memory are predictive of educational success in childhood and of cognitive decline in ageing. Deepening our scientific understanding of underlying mechanisms will lead to more powerful tools for identifying key cognitive factors and targeting appropriate interventions. In particular, the central thesis of the present proposal is that existing approaches overlook a crucial factor in cognitive control: the nature of the goals to which our actions are directed. If confirmed, this idea would imply that current methods could be improved to more accurately probe cognitive control as it contributes to educational outcomes and quality of life in ageing. The Oxford Depts of Experimental Psychology and Education provide collaborative opportunities to develop this pathway to impact on evidence-based policy making through education and ageing research. 2. Probing decision making in gambling Studies of the mechanisms of reward and learning have contributed substantially to understanding pathological decision making in addiction and gambling. Our proposal provides valuable extensions of this research to investigate higher-level control of behaviour that begin to bridge between simplified lab paradigms and the complexity of real-world cognition and behaviour. Yeung's collaborative research investigating decision making and learning in gamblers provides a pathway for translating findings from the proposed research into more applied research with implications for improving wellbeing and quality of life. 3. Identifying cognitive markers for brain-computer interfaces (BCIs) BCI devices are finding increasing use in medical, military, and commercial settings. Neural markers of cognitive control and learning are particularly relevant to 'augmented cognition' devices that use real-time neuroimaging to optimise human performance. Yeung has an established industry collaboration on BCI development; for example, using neuroimaging markers of cognitive load in a BCI device that modulates the rate of information flow to intelligence analysts searching satellite imagery. This example illustrates the opportunities for commercialising our scientific findings to attract R&D investment from private sector business and industry for cutting edge technologies. 4. Increasing international collaboration The project would establish a new British-French collaboration, complementing and extending the PIs' existing collaborations with researchers in the UK, US, Canada, Germany, and the Netherlands towards developing a strong international research network in this field. 5. Increasing public understanding A primary aim of the project is to forge stronger links between behaviours studied in the psychology lab and behaviour in the real world, extending lab paradigms to incorporate the crucial feature of everyday actions that they are fundamentally directed towards achieving identifiable outcomes. As such, the project offers valuable opportunities to increase public understanding of science in this field via presentations of the work to non-academic audiences.