Abstract

Decision-making frameworks highlight a number of important computations that may be necessary for guiding optimal behaviour. The brain needs to represent what alternatives are available and compute the variables that will enable efficient choice between alternatives, such as the costs and benefits associated with each alternative. Once the choice has been made the brain must compute the value of the obtained outcome and generate a prediction error if the actual outcome deviates from the predicted outcome. These signals could serve to modify and maintain the values associated with different alternatives, thereby ensuring that future choices are optimal and adaptive. Here I will discuss recent research from our lab suggesting that single neurons in different frontal cortex regions – in particular the anterior cingulate cortex (ACC), lateral prefrontal cortex (LPFC) and orbital frontal cortex (OFC) – encode different types of decision-making computations. Moreover, I will argue that the value representations encoded by neurons in these regions are necessary for reinforcement learning and optimal decision-making, as discrete lesions of ACC or OFC cause severe (but dissociable) learning and decision-making impairments. Critically however, while many single neurons encode decision value, they often do so with opponent encoding schemes such that value information averages away in population analyses. Such averaging effects may have implications for human neuroscience methods which average activity across neuronal populations.