Unveiling the neural codes for posture and action in 3D

One of the great questions facing neuroscience in the 21st century is how the brain produces volitional, goal-directed behaviors. In working toward this goal, neuroscience has in recent years placed an overarching emphasis on the use of advanced recording and perturbation technologies to interrogate neural activity, often in genetically defined neural circuits, while animals perform tasks that are pre-configured and parameterized by human experimenters. This has led to the development of remarkable tools to study neural circuit function but has detracted from a purposeful focus on the richness and temporal structure of natural behavior that the brain evolved to produce. The lack of advanced methods for quantifying behavior in unrestrained subjects limits the field's ability to make meaningful links between neural activity and behavior and collars our understanding of behavioral phenotypes in genetic disease models. The current project aims to address these shortcomings by making transformative advances in behavioral tracking, and linking time-resolved behavioral data back to neuronal population coding in neocortex. To do this we will develop novel approaches that integrate facial and 3D tracking of freely moving animals, allowing us to measure how animals actively sense the environment and use that input to dynamically shape the posture and kinematics of the body. This will be coupled with recording 100's to 1000's of neurons across cortical regions simultaneously, allowing us to build a map of how dynamic behavior is encoded over long-range anatomical distances in the neocortex. These efforts stem from our prior work showing that higher motor centers in cortex, including posterior parietal and frontal motor regions, exhibit strong tuning to 3D posture and movement during spatial behavioral tasks (Mimica et al., Science 2018). In the current project, we find that neural encoding of 3D posture and movement is *not* unique to higher motor areas, but extends to primary sensory areas, including visual, auditory and somatosensory cortex, as well as primary motor cortex. Using time-frequency decomposition of the postural variables and low-dimensional embedding, we further find that postural features co-vary through time, and are combined into discrete behavioral elements, or 'syllables', corresponding to simple behaviors such as grooming, rearing, running or sampling the environment. Through studying how the brains of animals represent natural patterns of behavior in 3D, we hope to extract general neural computational principles of how different neocortical systems encode self-generated actions which, in the not too distant future, could inform and improve the development of brain-controlled prosthetic devices in patients suffering from paralysis. The results could also be applied in the optimization of generative movement algorithms in robotics-- a field which is intent on producing machines to assist humans in extreme or hazardous environments, or rescue situations.

One of the great scientific challenges facing this century is understanding how the brain generates cognition and behavior. Pursuit of this goal has motivated the development of extraordinary technologies for recording, manipulating and mapping cells throughout the brain. By comparison, far less focus has been placed on tools to quantify animal behavior objectively and precisely, which limits our ability to draw meaningful links between neural activity and behavior, and constrains our understanding of behavioral phenotypes in genetic disease models. Solving such challenges will require novel frameworks for relating neural activity to natural behavior, tracked precisely and at relevant spatiotemporal scales. The goal of this proposal is to therefore make transformative advances in measuring neural tuning to behavior in rodents by integrating eye- and whisker-tracking with a 3D tracking platform pioneered in my lab. The project will have as its starting point our recent discovery of postural coding in the posterior parietal cortex (PPC) and frontal motor cortex (M2), areas critical for coordinating movement through space. The first experiments will test if postural tuning is unique to these regions, or if it is a universal feature of cortical processing, a question that is prompted by our pilot observations of 3D postural tuning in visual cortex (V1). Higher-risk experiments will then incorporate eye tracking to measure how postural and visual signals are integrated in V1, and to characterize visual tuning properties in freely behaving animals. Together, these experiments will provide an unprecedented, comprehensive view on the neural coding of intact behavior using a platform that will be open-sourced and applicable to any disease model in mice or rats.