Cognitive Control of Choices and Actions

We review model-based neuroscience work on cognitive control of choices and actions. We consider both strategically deployed executive processes and more automatic influences, first in binary choice tasks and then in more complex tasks. These include “conflict” tasks, where automatic and executive control processes sometimes act in opposition; delay discounting tasks, which require self-control to obtain larger rewards; and tasks where routine actions are occasionally interrupted by cues requiring different action or the inhibition of action. For all of these tasks, dynamic cognitive models have been developed based on the idea of accumulating evidence. They have also been studied by traditional neuroscience methods, but direct links to the cognitive models have not always been made. We detail the way in which progress has been made with model-based neuroscience methods in some cases and in others highlight how this points the way towards opportunities for progress. We emphasise generative Bayesian estimation methods that are well suited to the complexities of model-based neuroscience and provide exercises with open-source code that allow readers to develop skills with models relevant to cognitive control.