Adult Anxiety
Computational Modeling of Decision Making During High Choice-Conflict Scenarios in an Approach Avoidance Task
Nicole Moughrabi, B.S.
Research Associate
The University of Texas at Austin
Austin, Texas
Jaryd Hiser, Ph.D.
Postdoctoral Fellow
University of Wisconsin-Madison
Middleton, Wisconsin
Kevin M. Crombie, Ph.D.
Postdoctoral Research Fellow
The University of Texas at Austin
Austin, Texas
Chloe Botsford, B.S.
Clinical Research Coordinator II
University of Wisconsin - Madison
Madison, Wisconsin
Tijana Sagorac Gruichich, B.S.
Clinical Research Coordinator
University of Wisconsin - Madison
Madison, Wisconsin
Ameera F. Azar, B.S.
Research Associate
The University of Texas at Austin
Austin, Texas
Nicole Bernal, B.A.
Research Coordinator
The University of Texas at Austin
Austin, Texas
Joshua Cisler, Ph.D.
associate professor
University of Texas at Austin
AUSTIN, Texas
Anxiety is universally experienced and is often characterized by heightened avoidance of threatening situations. Additionally, individuals with anxiety disorders may sacrifice rewarding activities for the sake of avoiding threat which further contributes to functional impairment. As such, an aspect of decision making that has received increased research focus is approach-avoidance conflict (AAC; Aupperle & Paulus, 2010), which is a framework for understanding this trade off of sacrificing rewarding situations to avoid associated perceived threats. It is still unclear, however, what computational mechanisms are involved in choices made during these conflicting situations. To address these limitations, we previously developed a novel task (Weaver et al., 2020) that separately manipulates both reward and threat contingencies and have now adapted this task for a neuroimaging environment, using computational models of learning in order to identify neurocircuitry encoding during AAC. We hypothesized that large-scale networks, specifically the striatum, salience, and frontoparietal networks, would be separately engaged in reward, threat, and approach-avoidance learning. In this task, three response options were associated with both a rewarding outcome (i.e. points) and threatening outcome (i.e., electric shock). Conflict between approach and avoidance goals were manipulated by separately varying the reward and threat contingencies, such that sometimes the high reward choice was least likely to lead to shock (low AAC) and sometimes the high reward choice was most likely to lead to shock (high AAC). Participants (n=25) were selected from a community sample with no history of mental health disorders. Adaptations of the Rescorla–Wagner (RW) model with Hierarchical Bayesian Inference analyses described learning and decision-making patterns among participants, with resulting estimates from the best fitting model used in imaging analyses. Independent component analyses were used to characterize network activation in response to the computational parameters for reward learning, threat learning, and AAC. The striatum was significantly engaged for anticipation of both reward and threat, with no differences between the two. By contrast, the salience network was significantly engaged for both as well, but significantly more so for anticipating threatening vs rewarding outcomes. The left frontoparietal network was only significantly engaged during anticipation of rewarding outcomes, and both right and left frontoparietal networks were significantly engaged during approach-avoidance conflict. The current results demonstrated significant activation in the striatum, salience, and frontoparietal networks with differences in engagement in the salience and frontoparietal networks based on reward vs threat. These results shed light on neural mechanisms contributing to learning in contexts with both reward and threat and during approach-avoidance conflict. Further elucidation of these processes will lead to better understanding of biased resolution of approach-avoidance conflict in individuals with PTSD and anxiety disorders.