Amitai Shenhav, Ph.D. - US grants

Summary This project seeks to elucidate the mechanistic underpinnings of aversive experiences that arise from choosing between conflicting alternatives. Choice conflict/uncertainty can induce stress as the value and number of one's options increase, and these stressful experiences can persist after a choice is made, as the individual weighs counterfactual choices. Indecisiveness and intolerance of uncertainty also constitute important transdiagnostic factors in Obsessive-Compulsive Disorder (OCD) and Generalized Anxiety Disorder (GAD), disorders that will affect more than 1 in 20 Americans over their lifetimes. However, the basis for stressful experiences of choice conflict, and what their potential may be for negatively impacting ongoing cognition, are still poorly understood. We aim to ground conflict-related neural signals and their subjective sequelae (increased anxiety and decreased choice confidence) in terms of the evidence accumulated for different choice options before and after a decision, measured using fMRI and EEG while human participants choose between salient rewards. In order to better understand cognitive impairments that result from heightened uncertainty and worry, we will further test whether choice conflict signals interfere with ongoing cognition (as evidenced by behavioral and neural measures during a concurrent working memory task); whether this interference is enhanced with increasing trait anxiety and diminished by targeted decision strategies; and what patterns of neural connectivity give rise to the interference. Providing a more complete account of choice conflict in terms of ongoing processes of evidence accumulation will lay the groundwork for understanding varieties of choice paralysis. Furthermore, these efforts will contribute directly to our understanding of neural circuits whose dysregulation exacerbates negative experiences of choice conflict, providing a transdiagnostic mechanism for GAD, OCD, and related disorders. By focusing on the common dysfunctions in neural circuit computations across these disorders, this project aligns well with the criteria of the RDoC initiative, and with its goal of offering potential mechanistic targets for diagnosis and risk assessment. Such contributions will in turn inform approaches to pharmacological and therapeutic intervention aimed at reducing maladaptive responses to everyday occurrences of decision conflict in individuals with these disorders.

PROJECT SUMMARY In the United States, psychotic disorders (PDs) are the leading medical cause of functional disability, resulting in extremely high rates of unemployment and high financial costs to the public healthcare system. Due to the role that negative symptoms play in functional disability in PDs, it is critical that progress be made in understanding mechanisms leading to negative symptoms. Current theoretical models of negative symptoms postulate that dysfunctional cortico-striatal interactions lead to impairments in several aspects of reward processing that prevent PD patients from utilizing decision-making processes needed to perform goal directed behavior. Paramount among these reward processing abnormalities is effort-cost computation. Impaired effort-cost computation has been repeatedly demonstrated in PDs and linked to greater negative symptom severity; however, the processes underlying this association and its ties to real-world functioning are unclear. In the current study, we aim to test the novel hypothesis that negative symptoms and poor functional outcome are associated with the failure of effort to become a secondary reinforcer. There is evidence from studies on animals and humans indicating that if high effort is consistently paired with high reward, this can form a conditioned association, whereby effort itself takes on the status of a secondary reinforcer (i.e., effort itself is learned to have value, even if it is not rewarded). Building upon translational neuroscience paradigms developed in rodents, we will administer a novel effort as a secondary reinforcer task to outpatients with PDs (n = 45) and healthy controls (n = 45). The task consists of four phases: 1) difficulty scaling: participants will perform 4 cognitive tasks (anagrams, Gabor patches, mental arithmetic, trail making) and a staircase procedure will be used to determine individualized difficulty levels set for low and high effort options utilized in subsequent phases ; 2) baseline: a two-forced choice effort decision-making task will be performed in relation to four cognitive tasks to determine baseline levels of effort avoidance; 3) effort training: in a two- forced choice decision making task, selection of a high effort option will be probabilistically reinforced over several training blocks performed for 3 tasks; 4) test phase: near and far transfer effects of training will be examined for the 3 tasks participants were reinforced to select the high effort option on (near transfer), as well as a novel task they were not trained on (far transfer). In addition to measuring decision-making behavior, pupil dilation will be recorded continuously throughout the four phases as an objective marker of cognitive effort. A novel computational model will be applied to evaluate mechanisms involved with the failure of effort to become a secondary reinforcer, which evaluates reward sensitivity, learning rate, and effort cost. We will explore the novel hypothesis that avolition is associated with the failure of effort to become a secondary reinforcer, and that modeling components associated with learning rate, effort cost, and reward sensitivity will predict this deficit. Completing this R21 will provide a new mechanistic understanding of avolition.

PROJECT SUMMARY/ABSTRACT Most daily tasks demand cognitive control, but people vary in their motivation to meet the control demands required of those tasks. Motivational impairments are a common and transdiagnostic feature of a wide range of psychiatric and neurological disorders?including major depression, schizophrenia, and Alzheimer?s?severely compromising the daily functioning and overall wellbeing of individuals with these disorders. Unfortunately, little is known about the neurocomputational mechanisms that drive these impairments. We recently developed a computational model of how people make decisions about control allocation based on an evaluation of the costs and benefits (the Expected Value of Control [EVC] model). Our model points to several potential sources of motivational impairments and their putative neural substrates. These include deficits in learning about incentives, signaling those incentives when expected, and/or properly utilizing those incentives when making decisions about control allocation. The model suggests that dorsal anterior cingulate (dACC) is responsible for integrating incentive information in order to motivate the level of cognitive control that is most worthwhile. Our model further points to two dissociable components of the incentives for control: (1) the expected efficacy of control (the extent to which control is necessary to reach a particular goal) and (2) the expected reward for reaching that goal. Previous research has primarily focused on the latter component. It is therefore largely unknown how efficacy is learned and anticipated; how it is integrated with reward to guide control allocation; and to what extent motivational impairments are caused by deficits in the processing of efficacy. We have developed and validated a set of tasks that tease apart the independent influences of reward and efficacy on effort allocation. We will have adult participants perform these tasks while undergoing EEG or fMRI, to characterize the neurocomputational mechanisms by which expected reward and efficacy are (1) signaled, (2) utilized to determine effort allocation, (3) updated based on feedback, and (4) generalized to novel stimuli. We predict that dACC will integrate reward and efficacy information from separate frontoparietal inputs, to determine the amount and type of control that is most worthwhile. This control allocation will be enacted through dACC?s interactions with goal-specific prefrontal and subcortical regions. We also predict that reward- and efficacy-selective regions of frontostriatal and frontoparietal circuits will interact to guide learning and generalization of task incentives. We will test these predictions with model-based analyses of behavior and neural activity, using our EVC model to generate participant-specific estimates of incentive processing and control allocation across trials. This research will offer critical new insight into the computations and circuits underlying the motivation of cognitive control. It therefore has the potential to inform our understanding of the mechanisms of evaluation and motivation more generally, and to provide a path towards improving diagnosis and treatment for impairments that are both prevalent and transdiagnostic.