Mechanisms of cognitive control: Testing a neurocomputational model using emotion and reward in a challenge paradigm
In this project, we use an existing neural network computational model to generate hypotheses and predictions about the neural substrates and mechanisms of working memory and cognitive control. A powerful tool for elucidating mechanisms is to use a task-irrelevant challenge to tax cognitive processing to the point at which some aspect of its underlying structure is revealed, e.g., through dissociations. In our previous research, we have used dual task methods, normal aging, psychopathology, and administration of pharmacological agents in this way. Here, we use normal affect as such a tool, one likely to be additionally informative given its strong psychological validity and functionally specific influences on self-regulation. In behavioral and brain imaging studies, we will test predictions of the model under two related but distinct types of affective challenge: induced emotional states, and the delivery of unexpected rewards. We anticipate that the affective challenge paradigm will yield data that will not only confirm our conceptual model but also necessitate an extension. Specifically, finding a divergent influence of emotion on spatial vs. verbal processing would require more than a single kind of internal representation. Thus, we will:
1. Test the role of prefrontal cortex (PFC) in verbal and spatial processing during emotional states
We hypothesize that activity in lateralized PFC subsystems specialized for active maintenance of verbal and spatial information will vary by positive (approach) and negative (withdrawal) emotional state. Study 1.1 will use behavioral measures to test two counterintuitive hypotheses supported by 3 preliminary studies: 1) that a negative emotional state will impair verbal active memory but, counterintuitively, will improve spatial active memory; and 2) that a positive emotional state will produce the opposite pattern of performance (improved verbal but impaired spatial active memory). Study 1.2 will use fMRI to test directly the hypothesis that the dissociation between verbal and spatial active memory revealed by emotional states is mediated by lateralized PFC activity. Study 1.3 will use computational analyses tosimulate these effects of emotion as a shift in the balance between two competing neuromodulatory sub-systems. The simulations will test whether this extension to the model can capture the empirical dissociations in both behavioral performance and brain activation.
2. Test the role of dopamine (DA) and PFC in cognitive control after an unexpected reward
We predict that unexpected rewards will also have specific influences on cognitive control, based on the dual role we postulate for the DA system inreward prediction and active memory regulation. Study 2.1 will use behavioral measures to test 3 hypotheses about the effects of task-irrelevant reward on performance: 1) delivery of reward during the initial presentation of an item to be actively maintained will facilitate active maintenance of the item; 2) delivery of reward during the delay period in which an item must be actively maintained will disrupt active maintenance; and 3) delivery of reward during presentation of interfering distractors will lead to a replacement of the item by the distractor. Study 2.2 will use state-of-the-art event-related fMRI to test whether reward delivery modulates active maintenance in PFC. We predict that: 1) rewards delivered during the initial presentation of the item will enhance the event-related PFC response; and 2) rewards delivered during the delay period will lead to decay in the event-related PFC response. Study 2.3 will be computational analyses of the relationship between unexpected rewards and DA activity dynamics, and the consequences of reward-mediated DA modulation for the active maintenance of information items in PFC. The simulations will test whether the model can capture detailed aspects of the behavioral and brain imaging data.
These studies have the potential to further scientific knowledge by both testing and extending an existing model of cognitive control, refining our understanding of how the DA and PFC systems mediate cognitive control. This work could enhance our understanding of human performance, and lead to improved techniques for optimizing performance using normal affect. In addition, these studies could provide another step toward a detailed, mechanistic understanding of the interactions between cognitive and affective systems.