Research Areas
We combine behavioral tasks, large-scale electrophysiology, biosensor imaging, and computational modeling to understand the neural mechanisms of learning and decision-making.
Decision Making
Approaches
Conceptual model
From evidence to choice
Our lab studies how the brain evaluates options, weighs evidence, and commits to actions under uncertainty. We combine behavioral paradigms with large-scale neural recordings to understand how value representations emerge across cortical and subcortical circuits during decision-making. By tracking neural activity across multiple brain regions simultaneously, we aim to reveal how distributed circuits coordinate to produce adaptive choices.
In plain terms
We ask how the brain compares possibilities and chooses what to do next when the best answer is uncertain.
Key Questions
- How does the brain represent and compare values of different options?
- What neural mechanisms underlie the speed-accuracy tradeoff in decisions?
- How do neuromodulators shape the decision process in real time?
Working hypothesis
Adaptive choices emerge from interactions between distributed neural representations of value, uncertainty, and action.
Why it matters
Understanding those interactions can explain why the same experience produces flexible behavior in one setting and rigid behavior in another.
Neuromodulation
Approaches
Conceptual model
Signals are transformed locally
Neuromodulators including dopamine, norepinephrine, acetylcholine, and serotonin help regulate learning, motivation, attention, and cognitive flexibility. Our lab investigates how rapid neuromodulator dynamics reshape local circuit computations across behavioral contexts. We use fiber photometry, biosensor imaging, electrophysiology, and causal manipulations to connect these chemical signals to neural activity and behavior.
In plain terms
We track the brain’s chemical messages to learn how they change attention, motivation, and behavior from moment to moment.
Key Questions
- What information do rapid dopamine fluctuations encode beyond reward prediction errors?
- How do dopamine signals differ across brain regions and behavioral contexts?
- How do dopamine, norepinephrine, acetylcholine, and serotonin coordinate across cognitive demands?
Working hypothesis
Neuromodulators do not broadcast a single global message; local circuit mechanisms reshape their signals for distinct cognitive demands.
Why it matters
Separating broadcast and locally regulated signals may clarify how dopamine can support both learning and moment-to-moment motivation.
Reinforcement Learning
Approaches
Conceptual model
Learning across timescales
We develop and test computational models of reinforcement learning to understand how the brain updates predictions, learns action values, and adapts behavior based on reward history. By comparing model predictions with neural data recorded during learning, we reveal the algorithms implemented by biological circuits and identify where they diverge from classical theoretical frameworks.
In plain terms
We study how outcomes change the brain’s expectations so that future choices become better adapted to the environment.
Key Questions
- How does the brain compute and update reward prediction errors?
- What neural substrates implement different components of RL algorithms?
- How do model-based and model-free learning strategies interact in the brain?
Working hypothesis
The brain learns with a spectrum of update rules and time horizons rather than a single universal reward-prediction error.
Why it matters
Multi-timescale learning can help explain how animals adapt both to immediate outcomes and to slowly changing environments.
Interested in our research?
Explore our publications for detailed findings, or learn about opportunities to join the team.