How striatal neurons influence decision-making depends on broader circuit context: evidence from theory-driven cell type identification

Neural cell types are often conceived as having fixed computational roles, yet their behavioral contributions may depend on the broader circuit state. In the striatum, for example, direct and indirect pathway spiny projection neurons (SPNs) are typically described as facilitating or suppressing actions, respectively. However, emerging observations of striatal activity, such as modulation related to action timing, action vigor, and reactive inhibition, call this dichotomy into question. Here we develop a state-space framework linking striatal population activity to decision dynamics by combining a spiking cortico-basal ganglia-thalamic (CBGT) model with large-scale in vivo recordings from mice performing a two-choice task. Modeled activity patterns guided the inference of putative striatal SPN subtypes in the empirical recordings, enabling reconstruction of decision trajectories through an underlying state space. Across simulated and experimental data, these trajectories revealed a shared set of decision dynamics and identified three state-dependent roles of indirect pathway activity: preserving uncertainty when active alone, signifying choice commitment when coactive with the direct pathway in the same action channel, and slowing decisions when recruited in the opposing channel. To interpret these roles mechanistically, we mapped state transitions onto previously described basal ganglia control ensembles (responsiveness, pliancy, and choice) that translate circuit activity into the evidence accumulation process, and we found that each role was associated with a distinct pattern of control ensemble engagement that tunes decision policy. These analyses support a context-dependent view of SPN function, in which their moment-to-moment impact emerges from the evolving profile of dynamics throughout the CBGT circuit.