Sam Cong

Key Responsibilities

• Model probabilistic selection task behavior with reinforcement learning drift diffusion models to test whether anhedonia reflects selective impairment in learning from positive feedback.
• Find reduced positive learning rates associated with higher anhedonic depression scores across community samples and individuals with major depressive disorder.
• Conduct fMRI analyses using whole-brain contrasts, RLDDM-derived parametric modulators, and reward-network ROI tests to map reinforcement-learning signals in MDD.

Key Responsibilities

• Led digital phenotyping analyses of greenspace exposure and stress dynamics, showing stress-buffering effects among healthy controls but limited benefits among clinical high-risk individuals for psychosis.
• Applied continuous-time structural equation modeling and network analysis to characterize affective-behavioral dynamics during hybrid cognitive behavioral therapy.
• Collaborated on Bayesian hidden Markov modeling of smartphone accelerometer and screen-state data to estimate sleep patterns in real-world psychiatric research.

Key Responsibilities

• Developed a hierarchical Bayesian modeling pipeline for stop-signal task data, spanning forward simulation, likelihood specification, inference, and model diagnostics.
• Implemented trial-specific likelihoods under the Independent Race Model with JAX acceleration and validated the pipeline against benchmark BEESTS estimates.
• Applied model-derived inhibitory-control parameters to a computational psychiatry study predicting 18-month suicidal ideation trajectories in teenagers.

Key Responsibilities

• Designed a master thesis study integrating visual creativity tasks with AI-based process measures to test how mood and cognitive flexibility shape originality.
• Built a JsPsych experiment combining mood induction, incomplete-shape drawing, and narrative reports from 90 participants across three mood conditions.
• Used stroke embeddings, semantic integration metrics, and automated drawing assessment to identify adaptive switching as a key predictor of creative originality.

Key Responsibilities

• Studied how feature-based representations support generalizable predictive knowledge by enabling inferences about distant future outcomes.
• Modeled a multi-phase reinforcement learning task in which participants learned robot transitions and rewards, then generalized mappings to novel robots with recombined features.
• Fit successor representation models to compare conjunction-based and feature-based learning, finding that feature-based learners generalized best across novel stimuli.

Key Responsibilities

• Investigated sex-specific effects of intranasal oxytocin on threat reversal learning to clarify its therapeutic potential for anxiety disorders.
• Modeled a double-blind, placebo-controlled threat reversal study with hierarchical Bayesian Pearce-Hall models using skin conductance responses from 180 healthy adults.
• Identified impaired threat reversal learning among females under placebo and female-specific enhancement of reversal learning following oxytocin administration.

Key Responsibilities

• Led a study on perceived academic stress and sleep quality among Chinese college students in the context of intensified academic competition.
• Validated a two-factor perceived academic stress scale distinguishing lasting competitive stress from episodic workload-related stress.
• Modeled serial mediation pathways showing that academic stress predicted poorer sleep through social comparison and bedtime procrastination, with distinct effects of emotion-regulation strategies.