Nigrostriatal Dopamine Signals Sequence-Specific Action-Outcome Prediction Errors

Molecular Neurobiology Lab, Xin Jin
Salk Institute for Biological Studies
San Diego, CA

My current interest lies in the neurobiology underlying goal-directed behavior and learning, which motivated me to investigate nigrostriatal dopamine release in self-initiated actions for reinforcement in transgenic mice. To study this, I utilized mice expressing channelrhodopsin in their dopamine neurons to provoke reinforcement learning by optically stimulating the substantia nigra pars compacta when animals completed a specific heterogeneous lever task. With analysis, I found the duration between correct lever presses and latency to reinitiate the sequence decrease over training, indicating mice successfully learned the action sequence. With the robust behavior in place, I now chased after the dopamine dynamics. Due to the sub-second temporal resolution, I implanted chronic fast scan-cyclic voltammetry electrodes and was able to record dopamine release in the dorsal striatum during behavioral sessions. I now questioned two things; what does dopamine transmission look like when the animal receives reinforcement as a product of their own action? And do dopamine dynamics differ during the sequence? To answer my first question, I recorded voltammetry sessions with two phases: in the first phase, the animals perform the learned sequence to receive stimulation (self-stimulation) which was followed by the second phase, that has levers retract and stimulation is delivered non-contingently upon the animal’s action (passive playback). Upon analysis, data demonstrated that dopamine release is markedly lower during self-stimulation than during passive playback, indicating self-initiated action suppresses striatal dopamine release in relation to that action’s outcome. Collectively, this is consistent with the notion of feed forward inhibition triggered by the goal-directed action, suspected through an efference copy mechanism. For my second question, I examined dopamine transmission at the individual action level by probing the initiating and terminating lever presses. My hypothesis that both lever presses would exhibit action-induced suppression was incorrect; in fact, the data indicated dopamine responded differently at the elemental-level. These findings suggest dopamine is selective towards each action when conducting a sequence for reinforcement, and therefore may give insight to understanding maladaptive learning presented in addiction. With excitement, this work can be found via BioRxiv.

Nigrostriatal dopamine signals sequence-specific action-outcome prediction errors

Nick G. Hollon, Elora W. Williams, Christopher D. Howard, Hao Li, Tavish I. Traut, Xin Jin