13 October 2021: Neural computation underlying subjective value construction

Kyo Iigaya
Department of Psychiatry
Columbia University Irving Medical Center

It is an open question how humans construct the subjective value of complex stimuli, such as artistic paintings or photographs. While great progress has been made toward understanding how the brain updates the value of known stimuli e.g., through reinforcement learning, little is known about how the value arises in the brain in the first place. Here, we propose that the brain constructs the value of a novel stimulus by extracting and assembling common features shared across stimuli. Notably, because those features are shared across a broad range of stimuli, we show that simple linear regression in the feature space can work as a single mechanism to construct the value across stimulus domains. In large-scale behavioral experiments with human participants, we show that a model of feature abstraction and linear summation can predict the subjective value of paintings, photographs, as well as shopping items whose values change according to different goals. The model shows a remarkable generalization across stimulus types and participants, e.g., when trained on liking ratings for photographs, the model successfully predicts a completely different set of art painting ratings. Also, we show that these general features emerge in a deep convolutional neural network, without explicit training on the features, suggesting that features relevant for value computation could arise spontaneously. Furthermore, using fMRI, we found evidence that the brain performs value computation hierarchically by transforming low-level visual features into high-level abstract features which in turn are transformed into valuation. Our findings suggest the feature-based value computation can be a general neural principle enabling us to make flexible and reliable value computations for a wide range of complex stimuli.

View this recorded session here.

6 October 2021: Context-Dependence Induces False Memories of Economic Values: A Test Across Three Decision-Making Modalities and Four Preference Elicitation Methods

Stefano Palminteri
Department of Cognitive Studies
École normale supérieure

I will present results from seven experiments (N=100 each) demonstrating that, in the context of human learning decision-making, the way in which options are arranged (i.e., the choice architecture) significantly affects the resulting memory representations of economic values. More specifically, economic values stored in memory do not reflect objective values, but are generally consistent with a partial range adaptation process. The results are robust across preference elicitation methods (choices or ratings), decision-making modalities (experience-based or description-based), and across days.

View this recorded session here.

29 September 2021: Executive contributions to reinforcement learning computations in humans

Anne Collins
Department of Psychology
University of California, Berkeley

The study of the neural processes that support reinforcement learning has been greatly successful. It has characterized a simple brain network (including cortico-basal ganglia loops and dopaminergic signaling) that enables animals to learn to make valuable choices, using valenced outcomes. However, increasing evidence shows that the story is more complex in humans, where additional processes also contribute importantly to learning. In this talk, I will show three examples of how prefrontal-dependent executive processes are essential to reinforcement learning in humans, operating both in parallel to the brain’s reinforcement learning network, as well as feeding this network information.

View this recorded session here.

22 September 2021: Cognitive dysfunction in depression: implications for risk, maintenance, and treatment

Kean Hsu
Department of Psychiatry
Georgetown University

Impairments in basic cognitive processes like attention and executive functioning are common, significant, and an unmet treatment need that has broad downstream effects for depressed individuals. However, the nature of these impairments and how they might lead to negative affect or clinical disorder remains poorly understood. This talk presents empirical data addressing three lines of questioning: 1) Is cognitive impairment a scar left by depression or does it potentially precede depression (or both); 2) how are difficulties with basic cognitive processes associated with the phenomenology of depression; and 3) does an experimental manipulation of these processes impact depression maintenance? To address these questions, I have assessed cognition in a variety of depressed populations, including monozygotic twin pairs discordant for lifetime depression, currently depressed, formerly depressed, and never-depressed individuals drawn from the community, and a sample of depressed individuals specifically expressing a cognitive process of interest. Future directions for this program of research, including identification of which cognitive processes contribute to the risk for, maintenance of, and impairment from emotional disorders, as well as how we can translate these findings in applied settings, will be reviewed.

View this recorded session here. Password for recording is: mr+6E.!t

23 June 2021: Using quantitative models to improve treatments for destructive behavior in children with autism

Wayne Fisher
Rutgers University Center for Autism Research, Education, and Services (RU-CARES)
Robert Wood Johnson Medical School

The most important advancement in the treatment of destructive behavior has been the development of functional analysis (FA), which is used to prescribe effective treatments, such as functional communication training (FCT). With FCT, the consequence that historically reinforced destructive behavior is delivered contingent on an appropriate communication response and problem behavior is correlated with extinction. Although this approach can be highly effective, many pitfalls and practical challenges arise when this treatment is implemented by caregivers in natural community settings. In this presentation, I will present data and describe a line of research routed in behavioral momentum theory and the generalized matching law aimed at increasing the effectiveness, generality, and durability of FCT for individuals with ASD who display destructive behavior in typical community settings. Specifically, I focus on: (a) applications based on the matching law that can be used to prevent extinction bursts when treatment is initiated; (b) stimulus-control procedures that can be used to promote the rapid transfer of treatment effects to novel therapists, contexts, and caregivers without reemergence of destructive behavior; and (c) stimulus- and consequence-control procedures that can be used as “behavioral inoculation” to prevent resurgence of problem when caregivers do not implement treatment procedures with pristine procedural integrity.

View this recorded session here.

26 May 2021: Computational phenotyping in Borderline Personality Disorder using a Social Hierarchy Probe

Iris Vilares
Department of Psychology
University of Minnesota

Dysfunction in social interactions is a hallmark feature of several psychiatric disorders. However, due to their complexity, social interactions are also extremely hard to study. Experimental economic games have been successfully used to quantitatively analyze behaviors and motives relevant to social interactions, and the emerging field of computational psychiatry has been applying these techniques to understand apparent aberrant social decision-making in people with psychiatric disorders.

One important aspect of social interactions that has scantly been studied with experimental economic games is how much people value being in a more dominant social hierarchy position. A mal-adaptive reaction to social dominance may present a significant source of vulnerability for neuropsychiatric disorders, and can be particularly relevant for personality disorders that have trouble sustaining social relations, such as borderline personality disorder (BorPD).

Here, we were interested in knowing how people with BorPD value and behave in social interactions when there are differences in social dominance, and how (or if) these differ from controls. Moreover, we were interested in finding computational phenotypes of these behaviors. For this, we had participants (169 controls and 312 BorPD patients) play a multi-round Social Hierarchy game where money could be used to increase (or maintain) social status and applied computational models to the obtained behavior.

We found no difference between BorPD patients and Controls in the amount of money spent to become (or remain) in the dominant position, the challenge rate, or the number of rounds in the dominant position. However, we found that BorPDs in the dominant position transferred more money to the other player when first alpha. In addition, they finished the game with a more equitable distribution of points between them and the game partner. The computational model revealed promising computational phenotypes and suggests that BorPD patients may have a higher disutility from losing their status, and this was associated with higher self-reported feelings of shame and aggression tendencies. Overall, our results indicate that BorPDs and Controls value social dominance similarly but that they may be particularly sensitive to losing status once they have it. In addition, our results suggest BorPDs, when in control, may be especially prosocial, and offer specific computational parameters that can be used to quantitatively characterize and phenotype each individual.

View this recorded session here.

12 May 2021: Risky business: the effects of reward cues on decision making in models of addiction

Mariya V. Cherkasova
Department of Psychology
West Virginia University

Reward cues can potently influence behaviour. In people with addictive disorders cue-reactivity predicts addictive behaviour such as drug-seeking as well as relapse. In this talk I will present studies looking at the effects of reward cues on decision making as a candidate mechanism whereby exposure to cues may bias behavior. Mirroring earlier findings in rodents, my work in humans suggests that reward cues can promote riskier choice and that these effects may depend on dopamine signaling. Individual differences in the propensity to attribute motivational salience to reward cues may modulate these risk-promoting effects.

View this recorded session here.

14 April 2021: Why Modeling Time and Context is Critical for (Some) Mental Health Problems

Peter Hitchcock
Department of Cognitive, Linguistic, & Psychological Sciences
Brown University

Computational psychiatry has made important advances and proof-of-principle demonstrations, but it still seems far away from influencing routine clinical practice. Why? I will argue that the field has had difficulty recognizing the variability among mental health problems—and, consequently, the need to model context and temporal dynamics for many problems. Modeling context and temporal dynamics is challenging conceptually and logistically; it would be much easier not to do so. I will suggest three heuristics for deciding whether such modeling is necessary for a given mental health problem. As a case study, I will apply a critical lens to my own developing research program on rumination and worry and their relations to depression and anxiety disorders. I will argue that modeling time and context is indeed critical for these disorders. I will draw out the implications for my research, with an eye toward general principles for modeling problems of sufficient complexity that they are best understood as interacting elements unfolding in context over time.

View this recorded session here.