20 February 2019: The depressed decision-maker

Joseph Kable
Department of Psychology
University of Pennsylvania

Depression is characterized by obvious changes in decision-making that cause distress and impairment. Here I will discuss some of our recent work aimed at understanding the specificity and dimensionality of decisional impairments in depression. We tested a large group of participants diagnosed with Major Depressive Disorder (MDD, n = 64) and healthy controls (n=64) on a comprehensive battery of nine value-based decision-making tasks which yield ten distinct measures. MDD participants were worse at learning from rewards and punishments, had more pessimistic predictions about the future, and were less persistent; but showed no differences in time or risk preferences or social bargaining behavior. Performance on learning, expectation, and persistence tasks each loaded on unique dimensions in a factor analysis and accounted for unique variance in predicting depressed status. Computational modeling of reward and punishment learning further revealed evidence for both hyposensitivity to outcomes and reduced learning rates in MDD relative to healthy controls. Our results highlight several learning and decision making processes that are the impacted in MDD whose more detailed study could shed light on distinct dimensions of this heterogeneous disorder.

6 February 2019: How does memory guide decisions: implications for psychiatry

Akram Bakkour
Department of Psychology
Zuckerman Mind Brain Behavior Institute
Columbia University

Memory is central to adaptive behavior, allowing past experience to guide decisions and actions. Indeed, decisions are often informed by memories. However, the neurobiological mechanisms by which episodic memory guides decisions and the consequences for behavior remain poorly understood. We use a computational framework developed for the study of perceptual decisions and adapt it to better understand the cognitive and neural mechanisms of value-based decision making. In this work, we use computational models to generate predictions that we test using fMRI and behavior in patient populations. This framework was adopted to better understand the basic mechanisms by which memory enters the decision process, whether value is constructed, and how preferences may be manipulated. Several psychiatric disorders are characterized by maladaptive decisions that lead to adverse outcomes. I demonstrate the utility of our framework for psychiatry by characterizing choices about food in patients with anorexia nervosa, an eating disorder typified by persistent and stereotypical choices of low-fat, low-calorie foods to the point of starvation. Using computational models to generate predictions that are testable using the tools of cognitive neuroscience offers promise in the search for novel interventions in the treatment of psychiatric disorders.

14 November 2018: The social brain: how we detect norms and influence others

Xiaosi Gu
Departments of Psychiatry & Neuroscience
Icahn School of Medicine at Mount Sinai

To maintain the normal functioning of a society, individuals must be able to learn to adapt to norms. More importantly, norms are not static construls but instead, can be changed and updated; in other words, individuals have the ability to influence social others. In this talk, I will present our recent work using computational models of decision-making, in conjunction with fMRI and lesion approaches, that investigates how humans learn to adapt to norms and how they influence others. Our findings suggest that different neural regions (e.g. vmPFC, insula) compute dissociable parameters that drive norm learning. Furthermore, humans are able to engage model-based planning to influence others, a process involving both mesolimbic and lateral frontal control regions. Taken together, these findings reveal the non-static nature of human interactions and the proactive feature of social exchange behaviors.

31 October 2018: Deficient belief updating as a convergent computational mechanism of psychosis

Guillermo Horga
Department of Psychiatry
Columbia University

Predictive coding and related models of perceptual inference provide a concrete framework to understand how expectations are integrated into subjective perceptual experiences and how sensory information affects the formation and updating of beliefs about hidden states. Recent work has begun to link specific computations underlying these processes to psychotic phenomena such as hallucinations and delusions. However, it remains unclear how the computations underlying perceptual inference relate to the well-established increase in nigro-striatal dopamine function in psychosis, what is the specific nature of abnormalities in these computations, and whether they can explain the concurrence of hallucinations and delusions that characterizes the psychotic syndrome of schizophrenia. During my presentation, I will discuss recent work from our group that capitalizes on computational models of inference and which suggests that deficient belief updating is a core mechanism underlying both hallucinations and delusions in schizophrenia, a mechanism that depends on nigro-striatal dopamine. Our findings further suggest the existence of symptom-specific pathways consistent with a hierarchical belief-updating model whereby lower-level deficits relate to hallucinations and higher-level deficits relate to delusions. I will also discuss the implications of this work for generating novel downstream targets that are more proximal to specific symptoms of psychosis relative to upstream abnormalities in dopaminergic nigro-striatal pathways.

17 October 2018: Connectome-based prediction of substance-use

Sarah W. Yip
Department of Psychiatry
Yale School of Medicine

Despite advances, the effectiveness of most substance-use interventions remains highly variable across individuals and multiple quit-attempts are standard. This talk will present recent findings demonstrating the ability of connectome-based predictive modeling (CPM)¬—a data-driven method of identifying neural networks subserving specific behaviors—to predict abstinence during substance-use treatment. Evidence for largely dissociable neural substrates of cocaine versus opioid use in poly-addicted individuals will also be presented and clinical implications will be discussed.

3 October 2018: A learning and memory model of major depression

Rivka Cohen
Department of Psychology
University of Pennsylvania

Memory processes prioritizing the activation of negative cognitions, including thoughts, images, and memories, have long been implicated in major depression. We introduce a model of major depression that characterizes the role of developmental history, emotional context, and retrieval processes in persistent emotional states. Building from the Context Maintenance and Retrieval (CMR) family of models, our theory characterizes how emotional attributes combine with other attributes within the cognitive system, how they are encoded and retrieved, and the influence of these processes on the persistence of emotional attributes within a person’s internal contextual state. The model presents a novel computational account of the development and maintenance of depression, as well as cognitive resilience factors and the time-course of recovery. Finally, the model accounts for the mechanisms underlying empirically validated psychotherapies and factors contributing to subsequent relapse.

19 September 2018: Decision-making impairment in long-term opioid users

Kathryn Biernacki
Centre for Molecular and Behavioral Neuroscience
Rutgers University Newark

Opioid users often demonstrate substantial decision-making impairments, which not only impair everyday functioning, but may also hinder treatment success and contribute to relapse in this population. However, it is still unclear what underlying factors may contribute to this deficit. In this talk, I will discuss the results of a meta-analysis of decision-making in opioid use disorder (OUD), as well as the results behavioral and psychophysiological experiments examining the potential underlying contributors to the decision-making impairment in long-term opioid users. Briefly, the results of the meta-analysis indicated that the decision-making impairment in opioid users is relatively severe. Further, the psychophysiological experiment revealed that the impairment is not due to an impairment in emotional signaling, while the behavioral experiment demonstrated that this impairment may be more restricted to specific decision-making scenarios (i.e., risky compared to ambiguous situations). I will also present some preliminary data regarding computational modelling of decision-making in OUD, as well as other substance use disorders.  By taking a computational psychiatry approach to OUD, this research program may help to identify specific components of decision-making that become impaired in this group and may inform future treatment practice to better support opioid users.

2 May 2018: A computational and neural model for mood dynamics

Robb Rutledge
Max Planck UCL Centre for Computational Psychiatry and Ageing Research
University College London

The subjective well-being or happiness of individuals is an important metric for societies, but we know little about how the cumulative influence of daily life events are aggregated into subjective feelings. Using computational modeling, I show that momentary happiness in a decision-making task is explained not by task earnings, but by the combined influence of past rewards and expectations. The robustness of this account was evident in a large-scale smartphone-based replication. I use a combination of neuroimaging and pharmacology to investigate the neural basis of mood dynamics, finding that it relates to dopamine. I then show that this computational approach can be used to investigate the link between mood and behavior and the dynamics of mood in psychiatric disorders including major depression and bipolar disorder.