Mitsuko Watabe‐Uchida

This study investigates how distributional reinforcement learning is implemented in the basal ganglia by analyzing dopamine neuron activity and striatal representations.

This study investigates how dopamine systems in the ventral and tail of the striatum contribute to reward seeking and threat avoidance in mice during a foraging task.

The paper shows that hunger increases exploration by modulating dopamine signaling in the tail of the striatum via AgRP neurons.

The paper demonstrates that TD learning computations are hardwired into a dopamine circuit by showing optogenetic stimulation of NAc D1 neurons drives dopamine neuron spiking according to TD error.

The paper introduces face-rhythm, an unsupervised pipeline for analyzing rodent facial movements, and shows it can decode task variables and cortical activity.

The paper shows that dopamine and direct/indirect pathway neurons in the tail of the striatum regulate threat avoidance at different stages of learning.

Opposing glutamatergic MPN Isolation and mostly GABAergic MPN Reunion neurons in the mouse preoptic hypothalamus encode social need and satiety, with touch as a likely cue.

The paper shows glutamate inputs to dopamine neurons signal reward prediction errors but not negative value of aversive stimuli, with opioid analgesics altering negative responses.

The paper develops improved GRABDA sensors to monitor dopaminergic activity in vivo, showing high sensitivity and spatially resolved DA release in mice during behaviors.

The paper uses machine learning to analyze novelty-induced behaviors and finds that dopamine in the striatum correlates with individual variability and is explained by a reinforcement learning model of threat prediction.

This paper shows that dopamine responses shift backward in time during associative learning, aligning with TD prediction error models.

The paper shows dopamine axon activity in three striatal regions is similar but differs in reward response bias, which relates to habit and skill.

This study demonstrates that ramping dopamine signals correspond to reward prediction errors rather than state values, unifying rapid phasic and slow ramping signals as derivative-like computations.