Motor control in mammals is traditionally viewed as a hierarchy of descending spinal-targeting pathways, with frontal cortex at the top 1–3. Many redundant muscle patterns can solve a given task, and this high dimensionality allows flexibility but poses a problem for efficient learning 4. Although a feasible solution invokes subcortical innate motor patterns, or primitives, to reduce the dimensionality of the control problem, how cortex learns to utilize such primitives remains an open question 5–7. To address this, we studied cortical and subcortical interactions as head-fixed mice learned contextual control of innate hindlimb extension behavior. Naïve mice performed reactive extensions to turn off a cold air stimulus within seconds and, using predictive cues, learned to avoid the stimulus altogether in tens of trials. Optogenetic inhibition of large areas of rostral cortex completely prevented avoidance behavior, but did not impair hindlimb extensions in reaction to the cold air stimulus. Remarkably, mice covertly learned to avoid the cold stimulus even without any prior experience of successful, cortically-mediated avoidance. These findings support a dynamic, heterarchical model in which the dominant locus of control can change, on the order of seconds, between cortical and subcortical brain areas. We propose that cortex can leverage periods when subcortex predominates as demonstrations, to learn parameterized control of innate behavioral primitives.

Summary The interplay between two major forebrain structures—cortex and subcortical striatum—is critical for flexible, goal-directed action. Traditionally, it has been proposed that striatum is critical for selecting what type of action is initiated, while the primary motor cortex is involved in specifying the continuous parameters of an upcoming/ongoing movement. Recent data indicate that striatum may also be involved in specification. These alternatives have been difficult to reconcile because comparing very distinct actions, as is often done, makes essentially indistinguishable predictions. Here, we develop quantitative models to reveal a somewhat paradoxical insight: only comparing neural activity across similar actions makes strongly distinguishing predictions. We thus developed a novel reach-to-pull task in which mice reliably selected between two similar but distinct reach targets and pull forces. Simultaneous cortical and subcortical recordings were uniquely consistent with a model in which cortex and striatum jointly specify continuous parameters governing movement execution.

Effective classification of neuronal cell types requires both molecular and morphological descriptors to be collected in situ at single cell resolution. However, current spatial transcriptomics techniques are not compatible with imaging workflows that successfully reconstruct the morphology of complete axonal projections. Here, we introduce a new methodology that combines tissue clearing, submicron whole-brain two photon imaging, and Expansion-Assisted Iterative Fluorescence In Situ Hybridization (EASI-FISH) to assign molecular identities to fully reconstructed neurons in the mouse brain, which we call morphoFISH. We used morphoFISH to molecularly identify a previously unknown population of cingulate neurons projecting ipsilaterally to the dorsal striatum and contralaterally to higher-order thalamus. By pairing whole-brain morphometry, improved techniques for nucleic acid preservation and spatial gene expression, morphoFISH offers a quantitative solution for discovery of multimodal cell types and complements existing techniques for characterization of increasingly fine-grained cellular heterogeneity in brain circuits.Competing Interest StatementThe authors have declared no competing interest.

In natural environments, animals must efficiently allocate their choices across multiple concurrently available resources when foraging, a complex decision-making process not fully captured by existing models. To understand how rodents learn to navigate this challenge we developed a novel paradigm in which untrained, water-restricted mice were free to sample from six options rewarded at a range of deterministic intervals and positioned around the walls of a large ( 2m) arena. Mice exhibited rapid learning, matching their choices to integrated reward ratios across six options within the first session. A reinforcement learning model with separate states for staying or leaving an option and a dynamic, global learning rate was able to accurately reproduce mouse learning and decision-making. Fiber photometry recordings revealed that dopamine in the nucleus accumbens core (NAcC), but not dorsomedial striatum (DMS), more closely reflected the global learning rate than local error-based updating. Altogether, our results provide insight into the neural substrate of a learning algorithm that allows mice to rapidly exploit multiple options when foraging in large spatial environments.

The nervous system evolved to enable navigation throughout the environment in the pursuit of resources. Evolutionarily newer structures allowed increasingly complex adaptations but necessarily added redundancy. A dominant view of movement neuroscientists is that there is a one-to-one mapping between brain region and function. However, recent experimental data is hard to reconcile with the most conservative interpretation of this framework, suggesting a degree of functional redundancy during the performance of well-learned, constrained behaviors. This apparent redundancy likely stems from the bidirectional interactions between the various cortical and subcortical structures involved in motor control. We posit that these bidirectional connections enable flexible interactions across structures that change depending upon behavioral demands, such as during acquisition, execution or adaptation of a skill. Observing the system across both multiple actions and behavioral timescales can help isolate the functional contributions of individual structures, leading to an integrated understanding of the neural control of movement.

The interplay between two major forebrain structures - cortex and subcortical striatum - is critical for flexible, goal-directed action. Traditionally, it has been proposed that striatum is critical for selecting what type of action is initiated while the primary motor cortex is involved in the online control of movement execution. Recent data indicates that striatum may also be critical for specifying movement execution. These alternatives have been difficult to reconcile because when comparing very distinct actions, as in the vast majority of work to date, they make essentially indistinguishable predictions. Here, we develop quantitative models to reveal a somewhat paradoxical insight: only comparing neural activity during similar actions makes strongly distinguishing predictions. We thus developed a novel reach-to-pull task in which mice reliably selected between two similar, but distinct reach targets and pull forces. Simultaneous cortical and subcortical recordings were uniquely consistent with a model in which cortex and striatum jointly specify flexible parameters of action during movement execution.

Recent success in training artificial agents and robots derives from a combination of direct learning of behavioural policies and indirect learning through value functions. Policy learning and value learning use distinct algorithms that optimize behavioural performance and reward prediction, respectively. In animals, behavioural learning and the role of mesolimbic dopamine signalling have been extensively evaluated with respect to reward prediction; however, so far there has been little consideration of how direct policy learning might inform our understanding. Here we used a comprehensive dataset of orofacial and body movements to understand how behavioural policies evolved as naive, head-restrained mice learned a trace conditioning paradigm. Individual differences in initial dopaminergic reward responses correlated with the emergence of learned behavioural policy, but not the emergence of putative value encoding for a predictive cue. Likewise, physiologically calibrated manipulations of mesolimbic dopamine produced several effects inconsistent with value learning but predicted by a neural-network-based model that used dopamine signals to set an adaptive rate, not an error signal, for behavioural policy learning. This work provides strong evidence that phasic dopamine activity can regulate direct learning of behavioural policies, expanding the explanatory power of reinforcement learning models for animal learning.

Animals learn trajectories to rewards in both spatial, navigational contexts and relational, non-navigational contexts. Synchronous reactivation of hippocampal activity is thought to be critical for recall and evaluation of trajectories for learning. Do hippocampal representations differentially contribute to experience-dependent learning of trajectories across spatial and relational contexts? In this study, we trained mice to navigate to a hidden target in a physical arena or manipulate a joystick to a virtual target to collect delayed rewards. In a navigational context, calcium imaging in freely moving mice revealed that synchronous CA1 reactivation was retrospective and important for evaluation of prior navigational trajectories. In a non-navigational context, reactivation was prospective and important for initiation of joystick trajectories, even in the same animals trained in both contexts. Adaptation of trajectories to a new target was well-explained by a common learning algorithm in which hippocampal activity makes dissociable contributions to reinforcement learning computations depending upon spatial context.

Recent success in training artificial agents and robots derives from a combination of direct learning of behavioral policies and indirect learning via value functions. Policy learning and value learning employ distinct algorithms that optimize behavioral performance and reward prediction, respectively. In animals, behavioral learning and the role of mesolimbic dopamine signaling have been extensively evaluated with respect to reward prediction; however, to date there has been little consideration of how direct policy learning might inform our understanding. Here we used a comprehensive dataset of orofacial and body movements to understand how behavioral policies evolve as naive, head-restrained mice learned a trace conditioning paradigm. Individual differences in initial dopaminergic reward responses correlated with the emergence of learned behavioral policy, but not the emergence of putative value encoding for a predictive cue. Likewise, physiologically-calibrated manipulations of mesolimbic dopamine produced multiple effects inconsistent with value learning but predicted by a neural network-based model that used dopamine signals to set an adaptive rate, not an error signal, for behavioral policy learning. This work provides strong evidence that phasic dopamine activity can regulate direct learning of behavioral policies, expanding the explanatory power of reinforcement learning models for animal learning.

The interaction of descending neocortical outputs and subcortical premotor circuits is critical for shaping skilled movements. Two broad classes of motor cortical output projection neurons provide input to many subcortical motor areas: pyramidal tract (PT) neurons, which project throughout the neuraxis, and intratelencephalic (IT) neurons, which project within the cortex and subcortical striatum. It is unclear whether these classes are functionally in series or whether each class carries distinct components of descending motor control signals. Here, we combine large-scale neural recordings across all layers of motor cortex with cell type-specific perturbations to study cortically dependent mouse motor behaviors: kinematically variable manipulation of a joystick and a kinematically precise reach-to-grasp. We find that striatum-projecting IT neuron activity preferentially represents amplitude, whereas pons-projecting PT neurons preferentially represent the variable direction of forelimb movements. Thus, separable components of descending motor cortical commands are distributed across motor cortical projection cell classes.