Timothy Verstynen

Imagination is thought to boost exploration and problem solving by simulating states that have not been directly experienced as a means of reasoning through future plans. Here we tested this idea in the context of an impartial combinatorial game with a ground truth optional strategy. Adopting a stumbler-strategist reinforcement learning architecture, we showed how imagination can boost optimal strategy learning over memory replay. These model predictions were validated experimentally in humans (N=255) using an intervention design where participants were tasked with either imagining unplayed scenarios, reviewing prior games, or engaging in a simple visual attention task (control). These convergent results show how imagination uniquely boosts strategy learning in partially explored tasks.

Optimal patch foraging requires the ability to discern valuable resources and harvest them efficiently. This process requires integrating across multiple learning systems, including perceptual, structural, and value-based mechanisms. We explored how these varied learning systems contribute to foraging efficiency through a novel virtual foraging videogame. Human participants (N=60) navigated a virtual world with four neighborhoods (patches) populated with houses, searching for “bitter Oomplets,” characterized by the union of specific body colors and textures. Each neighborhood had a different overall probability of containing bitter Oomplets and, within each, specific house locations predicted the relative rate of bitter Oomplets. Participants proficiently learned to discriminate Oomplet types and strategically allocate their time to higher-probability neighborhoods. However, they did not learn the spatial predictors within neighborhoods. Importantly, we found no interaction between these two distinct learning abilities in determining foraging ability, suggesting that the impact of the perceptual and reward learning systems are independent of one another in human foraging efficiency.

Although the cortico-basal ganglia-thalamic (CBGT) network is identified as a central circuit for decision-making, the dynamic interplay of multiple control pathways within this network in shaping decision trajectories remains poorly understood. Here we develop and apply a novel computational framework—CLAW (Circuit Logic Assessed via Walks)—for tracing the instantaneous flow of neural activity as it progresses through CBGT networks engaged in a virtual decision-making task. Our CLAW analysis reveals that the complex dynamics of network activity is functionally dissectible into two critical phases: deliberation and commitment. These two phases are governed by distinct contributions of underlying CBGT pathways, with indirect and pallidostriatal pathways influencing deliberation, while the direct pathway drives action commitment. We translate CBGT dynamics into the evolution of decision-related policies, based on three previously identified control ensembles (responsiveness, pliancy, and choice) that encapsulate the relationship between CBGT activity and the evidence accumulation process. Our results demonstrate two contrasting strategies for decision-making. Fast decisions, with direct pathway dominance, feature an early response in both boundary height and drift rate, leading to a rapid collapse of decision boundaries and a clear directional bias. In contrast, slow decisions, driven by indirect and pallidostriatal pathway dominance, involve delayed changes in both decision policy parameters, allowing for an extended period of deliberation before commitment to an action. These analyses provide important insights into how the CBGT circuitry can be tuned to adopt various decision strategies and how the decision-making process unfolds within each regime.

Mammals flexibly adjust their decision strategies in dynamic environments based on prior experience. The cortico-basal ganglia-thalamic (CBGT) circuit is recognized as a critical driver of this adaptability, yet how plasticity-induced modifications to CBGT dynamics translate into modifications of decision policies remains poorly understood. Here we simulate learning in a biologically-grounded spiking CBGT model that learns to select a rewarded action via dopamine-dependent plasticity at corticostriatal synapses. Relying on three previously identified control ensembles (responsiveness, pliancy, and choice) within CBGT circuits, we disentangle the distinct roles these subnetworks have in reshaping decision trajectories across learning. Control ensemble dynamics were mapped onto the evolution of evidence accumulation, revealing within-trial parameter adjustments that shape decision dynamics and outcomes. Our results emphasize that learning optimizes not only what choice is favored, but also how the phases within the decision unfold. Early in the decision, learning accelerates evidence accumulation by driving the activity of corticothalamic and direct pathways. During later deliberation, this drive is temporarily restrained, and the indirect and pallidostriatal pathways become more critical for maintaining decision thresholds, preventing premature commitment despite the increased choice bias. As the system approaches the decision point, the direct pathway regains dominance, triggering boundary collapse to facilitate action selection. This mechanism effectively shifts decisions from deliberative to committed reward-directed choices, improving both speed and accuracy while preserving system stability and control throughout the process.

All mammals exhibit flexible decision policies that depend, at least in part, on the cortico-basal ganglia-thalamic (CBGT) pathways. Yet understanding how the complex connectivity, dynamics, and plasticity of CBGT circuits translate into experience-dependent shifts of decision policies represents a longstanding challenge in neuroscience. Here we present the results of a computational approach to address this problem. Specifically, we simulated decisions during the early learning process driven by CBGT circuits under baseline, unrewarded conditions using a spiking neural network, and fit an evidence accumulation model to the resulting behavior. Using canonical correlation analysis, we then replicated the identification of three control ensembles ( responsiveness , pliancy and choice ) within CBGT circuits, with each of these subnetworks mapping to a specific configuration of the evidence accumulation process. We subsequently simulated learning in a simple two-choice task with one optimal (i.e., rewarded) target and found that, during early stages of learning, feedback-driven dopaminergic plasticity on cortico-striatal synapses effectively increases reward rate over time. The learning-related changes in the decision policy can be decomposed in terms of the contributions of each control ensemble, whose influence is driven by sequential reward prediction errors on individual trials. Our results provide a clear and simple mechanism for how dopaminergic plasticity shifts subnetworks within CBGT circuits so as to increase reward rate by strategically modulating how evidence is used to drive decisions.

Background: Cardiovascular responses to psychological stressors have been separately associated with preclinical atherosclerosis and hemodynamic brain activity patterns across different studies and cohorts; however, what has not been established is whether cardiovascular stress responses reliably link indicators of stressor‐evoked brain activity and preclinical atherosclerosis that have been measured in the same individuals. Accordingly, the present study used cross‐validation and predictive modeling to test for the first time whether stressor‐evoked systolic blood pressure responses statistically mediated the association between concurrently measured brain activity and a vascular marker of preclinical atherosclerosis in the carotid arteries. Methods and Results: Six hundred twenty‐four midlife adults (aged 28–56 years, 54.97% women) from 2 different cohorts underwent 2 information‐conflict functional magnetic resonance imaging tasks, with concurrent systolic blood pressure measures collected. Carotid artery intima‐media thickness was measured by ultrasonography. A mediation framework that included harmonization, cross‐validation, and penalized principal component regression was then used. Brain areas where functional magnetic resonance imaging activity exhibited reliable direct and indirect effects were identified through bootstrapping. Sensitivity analysis further tested the robustness of findings after accounting for prevailing levels of cardiovascular disease risk and brain imaging data quality. Task‐averaged patterns of functional magnetic resonance imaging activity across distributed brain areas exhibited a generalizable association with carotid artery intima‐media thickness, which was reliably mediated by an area under the curve measure of aggregate systolic blood pressure reactivity. Importantly, this effect held in sensitivity analyses. Implicated brain areas in this mediation included the ventromedial prefrontal cortex, anterior cingulate cortex, insula, and amygdala. Conclusions: These novel findings support a link between stressor‐evoked brain activity and preclinical atherosclerosis, which is accounted for by individual differences in corresponding levels of stressor‐evoked cardiovascular reactivity.

The modern study of perceptual learning across humans, non-human animals, and artificial agents requires large-scale datasets with flexible, customizable, and controllable features for distinguishing between categories. To support this research, we developed the Oomplet Dataset Toolkit (ODT), an open-source, publicly available toolbox capable of generating 9.1 million unique visual stimuli across ten feature dimensions. Each stimulus is a cartoon-like humanoid character, termed an “Oomplet,” designed to be an instance within clearly defined visual categories that are engaging and suitable for use with diverse groups, including children. Experiments show that adults can use four to five of the ten dimensions as single classification criteria in simple perceptual discrimination tasks, underscoring the toolkit’s flexibility. With the ODT, researchers can dynamically generate large, novel stimulus sets to study perceptual learning across biological and artificial contexts.

Although the cortico-basal ganglia-thalamic (CBGT) network is identified as a central circuit for decision-making, the dynamic interplay of multiple control pathways within this network in shaping decision trajectories remains poorly understood. Here we develop and apply a novel computational framework — CLAW (Circuit Logic Assessed via Walks) — for tracing the instantaneous flow of neural activity as it progresses through CBGT networks engaged in a virtual decision-making task. Our CLAW analysis reveals that the complex dynamics of network activity is functionally dissectible into two critical phases: deliberation and commitment. These two phases are governed by distinct contributions of underlying CBGT pathways, with indirect and pallidostriatal pathways influencing deliberation, while the direct pathway drives action commitment. We translate CBGT dynamics into the evolution of decision-related policies, based on three previously identified control ensembles (responsiveness, pliancy, and choice) that encapsulate the relationship between CBGT activity and the evidence accumulation process. Our results demonstrate two contrasting strategies for decision-making. Fast decisions, with direct pathway dominance, feature an early response in both boundary height and drift rate, leading to a rapid collapse of decision boundaries and a clear directional bias. In contrast, slow decisions, driven by indirect and pallidostriatal pathway dominance, involve delayed changes in both decision policy parameters, allowing for an extended period of deliberation before commitment to an action. These analyses provide important insights into how the CBGT circuitry can be tuned to adopt various decision strategies and how the decision-making process unfolds within each regime. Author summary: We investigate how the cortico-basal ganglia-thalamic (CBGT) network coordinates decision-making through its interconnected pathways. Using a novel Circuit Logic Assessed via Walks (CLAW) framework, we trace instantaneous neural activity through virtual CBGT networks as they engage in forced choice decisions. This analysis uncovers two key phases of a decision: deliberation, shaped by the indirect and pallidostriatal pathways, and commitment, driven by the direct pathway. We also demonstrate that CBGT activity supports two distinct decision styles: fast decisions involve an early decision boundary collapse and strong directional preference, while slow decisions feature minimal changes during an extended deliberation phase. These findings reveal the dynamic mechanisms within the CBGT network that underlie the different decision processes and how these can be tuned to adapt decision-making across varying demands and contexts.

Objective: It is theorized that appraisals of perceived psychological stress are represented in the brain. However, a neural signature that reliably predicts perceived stress has yet to be fully characterized. Accordingly, the present preregistered study tested whether whole-brain resting-state functional connectivity patterns predict individual differences in perceived stress. Methods: Participants (N = 417; 53% female; 24.2% non-White; aged 30–54 years) completed the 10-item Perceived Stress Scale and underwent a 5-minute resting-state functional magnetic resonance imaging scan. Functional connectivity (FC) was computed between areas distributed across the brain. In total, 19,900 functional connections (edges) were retained for analyses. Cross-validated and multivariate machine learning methods were implemented. Using this approach, two penalized regression models with cross-validation—elastic net and ridge—were conducted to predict perceived stress from the edges. Results: Across the elastic net and ridge regression models, whole-brain resting-state FC patterns failed to predict individual differences in perceived stress. However, in exploratory analyses, they successfully generalized in cross-validation to predict age for both models (elastic net: r = 0.193, p < .0001, 95% CI = 0.099–0.284, RMSE = 6.661, MAE = 5.715, R ² = 0.037; ridge: r = 0.197, p < .0001, 95% CI = 0.103–0.287), RMSE = 6.613, MAE = 5.8140, R ² = 0.039). Conclusion: These results suggest that resting-state FC patterns may not reliably predict individual differences in self-reported perceived stress among midlife adults.

Social groups represent a collective identity defined by a distinct consensus of concepts (e.g., ideas, values, and goals) whose structural relationship varies between groups. Here we set out to measure how a set of inter-concept semantic associations, comprising what we refer to as a concept graph, covaries between established social groups, based on racial identity, and how this effect is mediated by information ecosystems, contextualized as news sources. Group differences among racial identity (278 Black and 294 white Americans) and informational ecosystems (Left- and Right- leaning news sources) are present in subjective judgments of how the meaning of concepts such as healthcare, police, and voting relate to each other. These racial group differences in concept graphs were partially mediated by the bias of news sources that individuals get their information from. This supports the idea of groups being defined by common conceptual semantic relationships that partially arise from shared information ecosystems.