Zhewei Zhang (张哲伟),
Ph.D.,
Cognitive/System/Computational Neuroscience
Open to tenure-track faculty positions！
zhzhewei36 at gmail dot com
zhewei.zhang at nih dot gov
My name is Zhewei Zhang, and I am currently a post-doctoral fellow at the National Institutes of Health (NIH) working with Dr. Geoffrey Schoenbaum. My research focuses on the neural circuitry involved in state representation and prediction error encoding. I completed my Ph.D. under the supervision of Dr. Tianming Yang at the Institute of Neuroscience (ION), Chinese Academy of Sciences.
Learning is essential for animals to optimize decisions in dynamic environments. The foundation of learning lies in prediction - the ability to anticipate future events based on past observations and actions. This predictive process relies on two critical components: 1) the efficient representation of past experiences, ignoring irrelevant details for rapid learning and generalization, preserving essential information, and organizing these states reasonably for successful learning. 2) and the mechanisms that transform these representations into predictions. My research program aims to uncover the neural principles of these processes. Through a unique combination of sophisticated behavioral paradigms, neural recordings, and computational modeling, I have made significant discoveries about how different brain regions coordinate to support learning and decision-making.
Education（教育经历）
Jun. 2021 - now Post-Doctoral Fellow, National Institutes of Health (美国国立卫生研究院)
• Advisor: Geoffrey Schoenbaum, MD, Ph.D.
Sep. 2014 - Jun. 2021 Ph.D., Institute of Neuroscience, Chinese Academy of Sciences (中国科学院神经科学研究所)
• Advisor: Tianming Yang (杨天明), Ph.D.
Sep. 2010 - Jun. 2014 B.S., School of Life Science, Sun Yat-Sen University (中山大学)
Publications（发表文章）
Reviews or Opinions
1. Zhang, Z., Costa, K. M., Langdon., J. A., and Schoenbaum, G. (2025) The devilish details affecting TDRL models in dopamine research. Trends in Cognitive Sciences. (Online)
Research Articles
1. Zhang, Z., Costa, K. M., Zhuo, Y., Li, G., Li, Y., & Schoenbaum, G. (2025). Accumbal acetylcholine signals associative salience. bioRxiv, 2025-01. (In Submission)
2. Takahashi, K. Y., Zhang, Z., Kahnt ,T., and Schoenbaum, G. (2024) Dopaminergic responses to identity prediction errors depend differently on the orbitofrontal cortex and hippocampus. bioRxiv, 2024-12. (In Submission)
3. Zong, W., Zhou, J., Gardner, M. P., Zhang, Z., Costa, K. M., & Schoenbaum, G. (2025) Schema cell formation in orbitofrontal cortex is suppressed by hippocampal output. Nature Neuroscience. (Accepted)
4. Costa, K. M.*, Zhang, Z*., Zhuo, Y., Li, G., Li, Y., & Schoenbaum, G. (2025) Dopamine and acetylcholine correlations in the nucleus accumbens depend on behavioral task states. Current Biology. (Online)
5. Zhang, Z.*, Takahashi, K. Y.*, Cartegena, M. M,. Kahnt ,T., Langdon., J. A., and Schoenbaum, G. (2024) Expectancy-related changes in firing of dopamine neurons depend on hippocampus. Nature Communications. 15(1), 8911. (Featured Paper of the Month in NIDA, https://irp.nida.nih.gov/featured-paper-january-2025/)
6. Zhang, Z., Yin, C. & Yang, T. (2022) Evidence accumulation occurs locally in the parietal cortex. Nature Communications. 13, 4426.
7. Liu, D., Deng, J., Zhang, Z., Zhang, Z., Sun, Y., Yang, T., and Yao, H. (2020) Orbitofrontal control of visual cortex gain promotes visual associative learning. Nature Communications. 11(1): 1-14.
8. Zhang, Z., Cheng, H., and Yang, T. (2020) A Recurrent Neural Network Model for Flexible and Adaptive Decision Making based on Sequence Learning. PLOS Computational Biology. 16(11): e1008342.
9. Zhang, Z*., Cheng, Z*., Lin, Z., Nie, C., and Yang, T. (2018) A neural network model for the orbitofrontal cortex and task space acquisition during reinforcement learning. PLOS Computational Biology. 14(1): e1005925.
Comments
1. Zhang, Z., Xie, Y. (2020) Understanding Commonalities and Discrepancies between Feature and Spatial Attention Effect in the Context of a Normalization Model. Journal of Neuroscience. 40 (5), 955-957
Research Experience（科研经历）
2021-present
Postdoctoral Fellow, National Institutes of Health, Baltimore, MD, USA
Advisor: Dr. Geoffrey Schoenbaum.
a) State learning and representation in rats.
Recorded midbrain dopaminergic neurons from rats with hippopcampus lesions.
Extended temporal difference reinforcement learning models in a partially observable semi-Markov decision framework.
Discovered HC shapes our task map and is necessary for estimating upper-level hidden states, compared to OFC, which provides information local to the trial.
Wrote an Opinion article highlighting the importance of state representation in interpreting dopaminergic neuron activity, an aspect ignored by many studies.
b) The role of acetylcholine during learning and its local interaction with dopamine.
Recorded accumbal dopamine and acetylcholine signals simultaneously in rats executing a task involving motivated approaching.
Discovered dopamine increases were not always coincidental with changes in acetylcholine and their relationship depends on the task phase.
Advanced hybrid attentional associative learning models which the salience of cues affected the learning speed. Discovered the salience predicted by a hybrid associative learning model captured acetylcholine dynamics during multiple tasks in stationary and volatile environments.
c) The offline replay in the orbitofrontal cortex and its interaction with hippocampus. (In progress)
2014-2021
Graduate Research Associate, University of Chinese Academy of Sciences
Advisor: Dr. Tianming Yang.
a) Neural basis of evidence accumulation in parietal cortex of the rhesus monkey.
Developed and implemented a complicated decision making task in which evidence is determined by two features of a stimulus, and recorded single neurons in partial cortex.
Discovered neurons in partial cortex represented accumulated evidence and transiently encoded variables about visual features that were critical for evaluating evidence.
Suggested neurons in LIP are responsible for both evidence accumulation and evidence evaluation in the decisionmaking process.
b) The computational principles underlying the neural basis of flexible and adaptive behaviors.
Developed a reservoir network with biologically plausible learning rules and demonstrated that the OFC can be viewed as a reservoir network to support model-based-like behaviors.
Proposed a recurrent neural network predicting future stimuli and reward outcomes sequence can facilitate multiple flexible and adaptive behaviors.
Discovered the units in neural networks trained solely on sequence prediction spontaneously reproduced diverse neuronal response patterns. Suggested that sequence prediction may capture key characteristics of how the brain works.