Author: Johanne

Nancy Lundin
Department of Psychiatry and Behavioral Health
The Ohio State University

Thought disorder is a cardinal feature of psychotic disorders, described as disorganized, tangential, illogical, and/or impoverished thought observed through speech. Higher levels of thought disorder relate to poorer functioning and long-term outcomes among individuals with psychotic disorders, yet the cognitive and neural underpinnings of this set of symptoms remain poorly understood. “Semantic foraging” is a new theoretical framework for examining potential processes underlying thought disorder with bases in behavioral ecology and cognitive science. Just as non-human animals search for food and mates by exploiting local patches of the environment and leaving to explore new patches when resources are depleted, humans may search for related concepts in memory using similar explore-exploit trade-offs. This has been studied using the verbal fluency test, in which individuals name items from a particular category or beginning with a particular letter in a limited time and often “cluster” (exploit) and “switch” (explore) between related groupings of items to optimize performance. In this seminar, I will discuss research examining search strategies during verbal fluency word retrieval tasks among individuals with psychotic disorders indicating preliminary evidence of alterations in semantic foraging, such as lower semantic similarity and longer interitem response times during search for related items than comparison groups. I will also discuss findings of neural signatures of semantic foraging processes from a study in which participants without a psychiatric diagnosis performed verbal fluency tests during functional magnetic resonance imaging (fMRI). We found that switching to new patches of related items was associated with activation in the hippocampus or posterior cerebellum depending on the way switching was operationalized (i.e., switches designated post hoc by participants versus computational metrics of similarity). Moreover, these regions exhibited activity which linearly increased during within-cluster search and re-set once a participant switched to a new patch, suggesting potential cognitive monitoring processes guiding the search. I will close with discussion of several future directions of this work. These include the need for replication of findings and consolidation of methodologies, with a particular focus on addressing the challenges to defining patch boundaries for when someone has stopped exploiting and started exploring, given the multifaceted ways in which concepts are related. Additional future directions include investigating the neural correlates of altered semantic foraging patterns among individuals with psychotic disorders and, most importantly, testing whether these explore-exploit patterns serve as a reliable marker of the production of cohesive speech to identify treatment targets for thought disorder.

View a recording of this session here.

Roy Salomon
Department of Cognitive Science
University of Haifa

We all rely on our senses to convey veridical information about the world around us, which we call “Reality”. However, distortions of reality in the form of hallucinations or illusions originating from neurological, psychiatric, pharmacological or psychological origins are prevalent. Thus, the sense of reality, the capacity to discriminate between true and false perceptions of the world, is a central criterion for neurological and psychiatric health. Despite the critical role of the sense of reality in our daily life, little is known about how this is formed in the mind and brain. The UnReal project explores the phenomenological, computational and neural processes underlying the Sense of Reality (SoR). We employ a specialized immersive virtual reality environment along with psychophysical paradigms to construct personalized “virtual hallucinations”. These “virtual hallucinations” are then used to modify the sense of reality while recording behavioral, ocular and neurophysiological responses to these perturbations of reality. Our results show differential behavioral and physiological responses to virtual hallucinations across different domains indicating divergent underlying models of reality. Developmental experiments along with computational modeling allowed us insight into the evolution of the sense of reality. This work may help to define a neurocognitive model of reality across healthy and clinical populations, and shed new light upon the fundamental philosophical question of “how do we know what is real?”.

View a recording of this session here.

Shirley Wang
Department of Psychology
Yale University

Suicide is a leading cause of death worldwide, claiming over 800,000 lives each year. Unfortunately, whereas scientific advances have led to declines in other leading causes of death over time, the current suicide rate is nearly identical to what it was 100 years ago. In order to make meaningful progress, new methods and approaches are needed that can capture and model the immense complexity of suicide and other forms of self-harm. In the first half of this talk, I will present a series of studies harnessing machine learning and real-time monitoring to predict who is at risk for suicide and when risk is highest. In the second half of this talk, I will take a complementary theory-driven computational approach to investigate why and how suicide and self-harm arise by building formal mathematical models of these phenomena as complex dynamical systems. Together, this work may help us advance the understanding, prediction, and prevention of suicide and self-harm.

View a recording of this session here.

Andrew Westbrook
Robert Wood Johnson Medical School
Rutgers University

Cognitive control is effortful, causing people to avoid demanding tasks, and undermining their goal-directed behavior. Yet the nature of effort costs, and the systems biasing decisions to exert or withhold effort are largely unknown. Striatal dopamine promotes physical effort for reward, by increasing sensitivity to reward benefits and decreasing sensitivity to effort costs. In the first part of my talk, I will discuss evidence that, as with physical action, striatal dopamine boosts motivation for cognitive effort as well. In a study combining [18F]-DOPA PET imaging of dopamine synthesis capacity with dopamine transport blocker methylphenidate, the D2 agent sulpiride, and placebo, we find that greater striatal dopamine signaling biases participants to choose harder working memory tasks to earn more money over easier tasks for less money. Moreover, gaze-informed drift diffusion modeling supports the inference that dopamine signaling increases willingness to exert effort by making people more sensitive to the benefits, and less sensitive to the costs of cognitive work.

In the second part of this talk, I consider a novel hypothesis about the nature of effort costs. Namely, that subjective cognitive effort is a phenomenological readout of divergence from criticality in the brain. Brains at rest exhibit emergent properties indicative of a dynamical system near a critical point – regulated by the balance of cortical excitation to inhibition. These properties are monotonically suppressed with increasing cognitive load, reflecting increasing divergence from criticality. Importantly, because criticality maximizes functional flexibility and information processing capacity, divergence implies computational costs. In my talk, I will discuss the rationale for the hypothesis linking subjective effort and criticality. I will also discuss a first study examining subjective cognitive effort and critical dynamics in EEG data while participants perform various levels of the N-back working memory task.

View a recording of this session here.

Emma Herms
Department of Psychological & Brain Sciences
Indiana University

Risk-taking is an integral part of decision-making. We take small risks throughout our daily life, from not feeding a parking meter to asking someone on a date. Taking risks in moderation can be advantageous – nothing ventured, nothing gained – and is an important part of learning. In contrast, taking too much or even too little risk may be disadvantageous. Thus, we are interested in the extent to which individuals with psychosis pursue risky rewards. A major limitation of task-based risky decision-making in psychosis-spectrum samples is that cognitive impairments are not typically controlled for, or explicitly examined (Purcell et al., 2021). In the current talk, I will present evidence from the Balloon Analogue Risk Task (BART), an uncertain-risk task where participants integrate information across trials to learn and weigh relative outcome probabilities. In this case, the likelihood that a balloon will explode. The BART is particularly interesting to examine because there is consistent evidence that individuals with psychosis pursue risky rewards less. However, given the complexities of the BART and confounds in psychosis samples, specific processes that contribute to this disadvantageous behavior remain poorly understood. Utilizing computational modeling, uncertain-risk decision-making behavior was parsed into subprocesses and examined for relationships with cognition, self-reported risk-specific processes, and non-risk specific personality traits to determine which of these external measures best explained group and individual differences in risk-taking.

View a recording of this session here.

Mor Nahum
School of Occupational Therapy, Faculty of Medicine
The Hebrew University

Cognitive control, the mental operations underlying our goal-directed behaviors, have been suggested as a potential key mechanism in mental health, contribution to mental resilience, emotion regulation and reduction of depressive symptoms. However, although cognitive control ability has been shown to fluctuate, it is often captured in a one-time lab-based measurement. Here, we use ecological momentary assessment (EMA) of inhibitory control, a component of cognitive control, and mood in real life to assess their contribution to mental health. In Study 1, conducted in a group of 156 young adults during their basic combat training in IDF, we show that for those with higher levels of resilience, inhibitory control is associated with their momentary mood, such that better inhibitory control predicts better mood. In Study 2, which included a group of 106 participant with pre-clinical depression, we show that a one-time measurement of inhibitory control does not predict depressive symptoms. Instead, variability of inhibitory control across the EMA sessions predicted depressive symptoms one week later. In addition, reduced levels of inhibition predicted worse mood for those with higher baseline depressive symptoms. Finally, in Study 3, which included a group of 132 young adults with and without ADHD, we show that emotion dysregulation at baseline is not associated with a one-time measurement of inhibitory control. Instead, those with higher levels of emotion dysregulation have reduced mean EMA of inhibitory control and increased inhibition variability. These results suggest that the stability of cognitive control over time may be more relevant for prediction of mental health related outcomes. We discuss these results in light of cognitive models related to mental health and their potential clinical implications.

View a recording of this session here.

Hyojeong Kim
Institute of Cognitive Science
University of Colorado Boulder

Failing to control the flow of our thoughts can impact our success in everyday life, and it is characteristic of multiple psychiatric disorders including depression and post-traumatic stress disorder. It has been demonstrated that thoughts can be removed from working memory (WM) in three differential methods (i.e., replacing a thought with something else, suppressing a specific thought, and clearing the mind of all thoughts), which recruit distinct neural regions involved in cognitive control (Banich et al., 2015). Our research has focused on how these different removal operations impact representational changes in WM and WM capacity as a result of the changes. In our removal task, participants were presented a target picture (a face, a fruit, or a scene) and then performed one of the removal operations based on a given instruction after the item offset. Employing multi-voxel pattern analysis (MVPA) on fMRI data, we were able to demonstrate that replacing, suppressing, and clearing are distinct neural processes that act upon WM information in different ways. By tracking the information contents of WM during the removal operations, we found that replacing an item was the quickest way to diminish information, followed by clearing and then suppressing. However, the representation of the removed item was mostly altered in WM when it was suppressed, followed by clearing and then replacing. Furthermore, these operations had different impacts on the encoding of new information. While maintaining an item interfered with the encoding of the next item as predicted, only suppressing an item, which displayed the most alteration of that item during removal, led to release from this proactive interference and even facilitated subsequent encoding (Kim et al., 2020). These results suggest that removal operations modulate WM representations in different ways, resulting in distinct consequences of representational alteration and WM capacity.

View a recording of this session here.

Matthew Rushworth
Department of Experimental Psychology and Wellcome Trust Centre for Integrative Neuroimaging (WIN)
University of Oxford, UK

Voluntary actions are ones that appear to be initiated by people and other animals in the absence of any external instruction or cue.  However, the likelihood that voluntary actions are initiated depends on identifiable features of both the current and the recent environment, recent behaviour, and the consequences that will ensure if the action is made (for example, will it lead to reward or not).  These factors mediate their influence on the initiation of voluntary action via a distributed neural circuit spanning cortical regions such as the anterior cingulate and insula cortex and subcortical nuclei including habenula, basal forebrain, ventral tegmental area, substantia nigra, and raphe nucleus.  I review a series of recent studies conducted in non-human and human primates using a combination of neuroimaging, temporary inactivation via ultrasound stimulation, and pharmacological manipulation that begin to dissect these different influences on voluntary behaviour and to trace the anatomical pathways through which each operates on voluntary behaviour.  The timing of action initiation depends both on reward expectations and on recent behaviour and these influences are governed by changes in activity in the basal forebrain and anterior cingulate cortex, altered by disruption of activity in either of these areas, and by cholinergic manipulations.  However, whether or not an action is initiated also depends on the richness/sparseness of opportunities in the environment in general and this is tracked by activity in the raphe nucleus and altered by serotonergic manipulation.

View a recording of this session here.

Ryan Smith
Laureate Institute for Brain Research
University of Tulsa

Computational modeling is a promising approach for parsing behavioral processes and dysfunctions in individuals with substance use disorders (SUDs), but it is unclear how much these processes change during the recovery period. In this talk, I will describe a study in which we assessed baseline and 1-year follow-up data on a sample of healthy controls (HCs; N = 48) and treatment-seeking individuals with one or more SUDs (alcohol, cannabis, sedatives, stimulants, hallucinogens, and/or opioids; N = 83) who completed a standard ‘three-armed bandit’ task designed to assess explore-exploit behavior. I will also describe a pre-registered replication study with a new sample of 168 individuals with SUDs and 99 HCs. Computational models – based on the notion of active learning – were fit to behavior on the task. Relative to HCs, participants with SUDs were found at baseline to show slower learning rates in response to negative outcomes and less precise action selection. We then repeated these analyses when the same individuals returned and re-performed the task 1 year later to assess the stability of these baseline differences. We also examined whether baseline modelling measures could predict symptoms at follow-up. Bayesian analyses indicated that: (a) group differences in learning rates were stable over time (posterior probability = 1); (b) relationships between model parameters at baseline and follow-up were all significant and ranged from small to moderate (.25 < ICCs < .54); and (c) learning rates and/or information-seeking values at baseline were associated with substance use severity at 1-year follow-up in stimulant and opioid users (.36 < rs < .43, .002 < ps < .02). Differences in learning rates for losses replicated in the second sample, and model parameters could jointly differentiate specific substance disorders when combining samples. These findings suggest that processing dysfunctions involving learning to arbitrate between exploration and exploitation may show some stability throughout the recovery period. At the same time, individual computational differences at baseline had some predictive value for changes in substance use severity. Taken together, these results suggest active learning models may allow measurement of trait dysfunctions that could have predictive utility for substance use severity.

View a recording of this session here.

Ignacio Saez
Departments of Neuroscience, Neurology and Neurosurgery
Icahn School of Medicine at Mount Sinai

My research focuses on the study of the neural basis of human cognition, particularly decision-making, through the examination of invasive brain activity in neurosurgical patients. I combine human electrophysiological and electrochemical recordings during decision-making tasks with computational models of reward and choice. This experimental combination provides access to fast, highly detailed, high signal-to-noise, multi-areal neural activity from the human brain, which is highly suited to studying the relationship of distributed neural activity to aspects of cognition and the use of neurostimulation techniques to modulate behavior and treat cognitive aspects of disease. In this talk, I will cover the different methodologies we employ in the lab, as well as current research efforts in the study of decision-making and the combination of computational psychiatry and intracranial approaches. 

View a recording of this session here.