| The experiments comprising this thesis aimed to characterize the neural and behavioral correlates of learning about and responding to the predictive information conveyed by social cues. Facial expressions are cues that convey critical information about the environment, and that we can use to shape our behavior. While previous research has focused on valence (e.g., negativity) and emotional arousal to categorize these cues, we explored the hypothesis that threat related facial expressions differ in terms of their 'predictive certainty,' or the amount of information about the environment that they convey to viewers. We hypothesized that since fearful and angry facial expressions differ in their predictive certainty, they should have different effects on attention and memory. Experiments 1--3 showed that viewing angry faces (i.e., certain threat) elicited focused attention toward the face and augmented memory for the individuals producing the expression. In contrast, fearful faces, which do not provide information about the source of their eliciting event (i.e., uncertain threat) prompted more diffuse attention and enhanced memory for contextual information. In Experiments 4 and 5, we extended this notion to surprised faces, which are even more uncertain in that they could predict either a positive or negative event, and provide no information about the source of that event. This predictive uncertainty makes interpretations of surprised faces susceptible to contextual manipulations. We showed that temporal unpredictability biased participants to interpret surprised faces more negatively, and that amygdala responses to these faces were consistent with valence interpretations.;Throughout this thesis, we suggest that these effects are related to the fact that facial expressions are naturally conditioned stimuli, in that people represent and respond to these cues based on their prior experiences with them. Since we cannot control for an individual's previous experiences with facial expressions, in Experiment 6, we designed an experiment where we could control for the reinforcement history of social cues. We showed increased amygdala activity as participants learned about individuals predicting negative or positive social outcomes, and liken these responses to a rich literature in non-human animals delineating the amygdala's role in learning about cues that predict biologically relevant events. |
