Xiaosi Gu

This study shows human claustrum neurons encode uncertainty and prediction errors during aversive learning, with distinct temporal profiles compared to anterior cingulate cortex.

The paper investigates how amygdala beta bursts influence hippocampal activity during emotional processing in humans, finding they suppress hippocampal firing during negative emotional stimuli.

The paper records single neuron activity in the substantia nigra during social learning tasks and finds dopaminergic neurons encode norm prediction error valence more strongly during human vs computer interactions.

This study links reinforcement learning computations to memory by showing hippocampal-cingulate dynamics in humans during a gambling task.

This paper shows that dopamine and serotonin changes during deep brain stimulation predict symptom relief in treatment-resistant depression.

This paper develops a multiclass classification framework to identify shared and substance-specific neural circuits in cannabis and alcohol use disorder using functional connectivity features from cue-induced craving tasks.

This study shows human substantia nigra neurons encode reward expectations through firing rates modulated by trial outcomes and linked to reaction times.

The paper shows that nicotine-related beliefs in smokers induce dose-like thalamic responses and functional coupling with prefrontal cortex, similar to pharmacological effects.

This study identifies a computational mechanism linking momentary craving and decision-making in substance use disorders, showing differential effects between alcohol and cannabis users.

The paper investigates neural dynamics for belief updating with and without explicit outcomes, finding distinct networks for confidence and uncertainty encoding.

The paper shows humans use forward thinking to guide social choices, with neural activity reflecting control beliefs in specific brain regions.

This paper develops an interpretable decoding model using functional connectivity to classify chronic marijuana users and non-users with high accuracy.

This study identifies an insula-driven neural network that tracks decision uncertainty and predicts abstinence in chronic cocaine users.

This paper introduces a reinforcement learning model to show how environmental complexity and bounded rationality contribute to non-pharmacological addiction by causing cognitive overload and suboptimal decision-making.

This study demonstrates that the anterior insular cortex is critical for interoceptive attention by showing its activation and connectivity patterns in healthy individuals and impaired function in patients with AIC lesions.

This paper simulates endophenotypic variations in addiction using neural and reinforcement learning models, finding that imbalances between cortico-striatal systems affect addiction outcomes and treatment responses.

The paper uses fMRI to show that GPe activity interacts with reward and dopamine levels, supporting a computational model of value encoding in the basal ganglia.