Striatal dynamics facilitate sensorimotor transformations through associative learning
Date
2025
DOI
Version
OA Version
Citation
Abstract
Instrumental behaviors can be defined as actions that, often elicited by external stimuli, are taken in pursuit of particular goals or rewards. To engage in appropriate instrumental behaviors, it is necessary to distinguish and discriminate relevant stimuli and associate them with appropriate actions. This process, known as sensorimotor transformations, requires the brain to transform the perceived stimulus into a motor-related signal that can modify behavior. The basal ganglia, a set of subcortical nuclei, is particularly well positioned to perform these transformations. The striatum, its main input nucleus, receives and integrates sensory, motor and cognitive inputs from a multitude of cortical, subcortical and limbic regions. The output from the striatum can then directly modulate action through a series of pathways that interconnects it with other subcortical structures and eventually with cortex. The dorsomedial region of the striatum (DMS) is one of particular interest since it receives direct projections from somatosensory associative and motor cortices. Within the basal ganglia field, the involvement of the dorsomedial striatum in instrumental behaviors has been focused, primarily, in action-outcome associations. Studies looking into reward-based action selection have been extensive, from single action behavioral paradigms i.e. lever pressing, to more complex behaviors like dual port, magazine entries. However, there is an increasing need to incorporate more ethological behaviors that incorporate the critical interaction between stimuli, actions and outcomes. In an effort to expand on this literature, this study puts forward an investigation into the cell-type specific role of the dorsomedial striatum during sensory-guided instrumental behavior. To that end, my thesis puts forth four studies that link the activity of the striatum to the necessary sensorimotor transformations that produce instrumental behaviors. The first study utilized bilateral optogenetic inhibition of the DMS while expert mice performed a two-choice, visually guided task. Findings from this experiment established the necessity of the striatum in the performance of these behaviors. Additionally, we demonstrated that the computations occurring in this region during task performance were not only motoric but associative in nature. Here, we propose the DMS as a locus for visually guided action selection.
In the second study, we utilized two-photon imaging of DMS activity to assess the functional dynamics of the two main striatal pathways, the direct and indirect pathways. Our study demonstrates that learning of the stimulus-action associations governing the task reshapes the intrinsic cue-evoked activity of cells from the direct (dSPNs) and indirect (iSPNs) pathways. This learning effect creates cue-evoked movement information in SPNs, providing evidence of the sensorimotor transformations that produce appropriate instrumental actions. The third study investigates how the activity of individual cue-responsive cells changes with learning by comparing isolated cue responses across different stages of learning. This study poses the emergence of lateralized preferences in cells from both pathways as the source for these transformations. Altogether, these studies propose that learning reshapes the individual preference for lateralized movement of individual cells, leading to stimulus-evoked movement direction representation. We suggest these representations play an active role in instrumental action-selection.
Lastly, our final study aims to propose a mechanism for how the co-activation of the direct and indirect pathways selects and invigorates instrumental actions. We utilized the reliability of activation of individual cells across trials as an overall measure for levels of activation in both pathways. Comparison of dSPN and iSPN reliability show the emergence of a learning-related imbalance that favors the activation of the direct pathway. In accordance with classical model of basal ganglia function, we propose that after learning, higher levels of activation in the direct pathway ensure the selection of the appropriate lateralized response. Altogether, the body of work presented here proposes a model in which the striatum acquires the ability to select actions based on sensory stimuli through the imbalanced recruitment of cue-responsive SPNs that encode lateralized movement prior to the initiation of locomotion.
Description
2025