Our research focuses on the neural mechanisms that shape social and economic decision making. To examine these issues, we use a combination of behavioral modeling, functional magnetic resonance imaging (fMRI), and transcranial current stimulation (TCS). Our current work delineates corticostriatal pathways involved in reward processing and uses TCS to manipulate responses to reward. We are also examining how healthy aging and psychopathology impact reward-dependent corticostriatal connectivity. Much of our past, present, and future work falls under three related themes.
1) Neural Mechanisms of Reward Processing and Decision Making. Much of our work has used neuroimaging to address key questions within neuroeconomics and decision neuroscience. For example, our work has examined how we compare disparate rewards. Although economists theorize that these decisions require each incentive to be transformed into a common currency, evidence for such signals in the brain have remained elusive. Our work provided the first evidence that a neural common currency signal was represented in a posterior region of ventromedial prefrontal cortex (Smith et al., 2010, Journal of Neuroscience). Our recent work extends these observations by quantifying how value signals within ventromedial prefrontal cortex rely on connectivity with other brain regions (Smith et al., 2014, Social Cognitive and Affective Neuroscience). We have also used brain connectivity to distinguish reward signals tied to affect and information. While both types of reward signals evoked activation within the striatum, we found that connectivity between the striatum and ventrolateral prefrontal cortex distinguished affective and informative reward signals (Smith et al., 2016, Scientific Reports). Finally, we have also used neuroimaging data to refine popular psychological theories of decision making. For instance, we tested the standard interpretation of the gain-loss framing effect (i.e., a competition between reason and emotion) against alternative explanations that would predict different activation patterns. Our study concluded that the framing effect results from an unexpected source: differential cognitive engagement across frames (Li*, Smith*, et al., 2017, Journal of Neuroscience).
2) Methodological Issues in Neuroimaging Analyses. Our work has aimed to improve and validate analytical approaches within the neuroimaging community. We have identified key differences between within- and cross-participant multivariate pattern analysis (MVPA) classification approaches (Clithero et al., 2011, NeuroImage). We also pioneered efforts to apply MVPA to lesion mapping, which helps clinicians to overcome several issues that plague univariate analyses (Smith et al., 2013, PNAS). Our other studies extend this theme of examining how multiple brain regions contribute to behavior by improving functional connectivity analyses. Specifically, we have used independent component analysis (ICA) combined with dual-regression analysis to estimate connectivity with large-scale neural networks. Our work demonstrated that ICA combined with dual regression predicts individual differences better than canonical approaches that only consider specific nodes of a neural network (Smith et al., 2014, NeuroImage). Finally, we have also used coordinate-based meta-analyses to validate the consistency and specificity of psychophysiological interaction (PPI) analysis, a popular brain connectivity analysis approach (Smith et al., 2016, Human Brain Mapping). Our lab is also committed to efforts that enhance rigor and reproducibility within the neuroimaging community, including sharing of code, raw data, and statistical maps.
3) Functional Significance of Large-Scale Neural Networks. We have also examined the functional significance of large-scale neural networks across distinct decision-making tasks and processing states. In one study, we demonstrated that a specific portion of the precuneus exhibits increased connectivity with the default-mode network (DMN) during resting states and increased connectivity with a fronto-parietal network during task states (Utevsky et al., 2014, Journal of Neuroscience). We have also shown that reduced synchrony between large-scale networks contributes to variation in autistic traits, extending theories that autism is associated with neural underconnectivity (Young*, Smith*, et al., Frontiers in Human Neuroscience). Connectivity with large-scale networks also plays an important role in decision making and behavioral change following feedback. While previous work indicates that the medial prefrontal cortex (MPFC) promotes behavioral change, it remains unclear whether these changes are due to the DMN or the executive control network because the MPFC is at the intersection of both networks. We demonstrated that behavioral changes were associated with distinct patterns of network connectivity (Smith*, Sip*, Delgado, 2015, Human Brain Mapping). Our recent work has integrated large-scale networks with PPI, which has allowed us to demonstrate that the receipt of reward enhances connectivity between the ventral striatum and DMN (Dobryakova & Smith, in preparation).