To ask how dopamine might impart structure onto spontaneous behavior, we performed dLight photometry in the dorsolateral striatum in mice running around an open field, while we simultaneously identified ongoing behavioral syllables (i.e., brief 3D motifs of behavior like turning or rearing) and sequences using MoSeq.

A lot packed in the paper, but take home is that in the DLS DA fluctuates a ton during spontaneous behavior, and that these fluctuations causally structure ongoing action, even without task structure or rewards.

Importantly, DLS DA during free behavior doesn’t seem to casually determine movement kinematics or initiation, but instead appears to specify both digital aspects of behavior — which behavioral syllable to use and how much — and analog aspects of behavior — like the vigor of each syllable.

DA does this not by directly specifying which syllable to use at a given moment, but rather through reinforcement, suggesting that the same circuits/mechs that govern low-D choices in structured tasks are relevant during high-D spontaneous behavior without structure/rewards.

It is unclear what is driving the fluctuations in DLS DA during spontaneous behavior – DA could be implementing an action plan, or may be signaling some sort of action prediction error (as has been beautifully argued by Ashok Litwin-Kumar and Marcus Stephenson Jones recently).

Paper is an amazing collab with many labs (Bernardo Sabatini, Nao Uchida, Scott Linderman) and would have been impossible if not for the help of many others.