Task-driven neural network models predict neural dynamics of proprioception.

Proprioception tells the brain the state of the body based on distributed sensory neurons. Yet, the principles that govern proprioceptive processing are poorly understood. Here, we employ a task-driven modeling approach to investigate the neural code of proprioceptive neurons in cuneate nucleus (CN) and somatosensory cortex area 2 (S1). We simulated muscle spindle signals through musculoskeletal modeling and generated a large-scale movement repertoire to train neural networks based on 16 hypotheses, each representing different computational goals. We found that the emerging, task-optimized internal representations generalize from synthetic data to predict neural dynamics in CN and S1 of primates. Computational tasks that aim to predict the limb position and velocity were the best at predicting the neural activity in both areas. Since task optimization develops representations that better predict neural activity during active than passive movements, we postulate that neural activity in the CN and S1 is top-down modulated during goal-directed movements.

Modern Machine Learning as a Benchmark for Fitting Neural Responses.

Neuroscience has long focused on finding encoding models that effectively ask "what predicts neural spiking?" and generalized linear models (GLMs) are a typical approach. It is often unknown how much of explainable neural activity is captured, or missed, when fitting a model. Here we compared the predictive performance of simple models to three leading machine learning methods: feedforward neural networks, gradient boosted trees (using XGBoost), and stacked ensembles that combine the predictions of several methods. We predicted spike counts in macaque motor (M1) and somatosensory (S1) cortices from standard representations of reaching kinematics, and in rat hippocampal cells from open field location and orientation. Of these methods, XGBoost and the ensemble consistently produced more accurate spike rate predictions and were less sensitive to the preprocessing of features. These methods can thus be applied quickly to detect if feature sets relate to neural activity in a manner not captured by simpler methods. Encoding models built with a machine learning approach accurately predict spike rates and can offer meaningful benchmarks for simpler models.

Hip and ankle responses for reactive balance emerge from varying priorities to reduce effort and kinematic excursion: A simulation study.

Although standing balance is important in many daily activities, there has been little effort in developing detailed musculoskeletal models and simulations of balance control compared to other whole-body motor activities. Our objective was to develop a musculoskeletal model of human balance that can be used to predict movement patterns in reactive balance control. Similar to prior studies using torque-driven models, we investigated how movement patterns during a reactive balance response are affected by high-level task goals (e.g., reducing center-of-mass movement, maintaining vertical trunk orientation, and minimizing effort). We generated 23 forward dynamics simulations where optimal muscle excitations were found using cost functions with different weights on minimizing these high-level goals. Variations in hip and ankle angles observed experimentally (peak hip flexion=7.9-53.1°, peak dorsiflexion=0.5-4.7°) could be predicted by varying the priority of these high-level goals. More specifically, minimizing center-of-mass motion produced a hip strategy (peak hip flexion and ankle dorsiflexion angles of 45.5° and 2.3°, respectively) and the response shifted towards an ankle strategy as the priority to keep the trunk vertical was increased (peak hip and ankle angles of 13.7° and 8.5°, respectively). We also found that increasing the priority to minimize muscle stress always favors a hip strategy. These results are similar to those from sagittal-plane torque-driven models. Our muscle-actuated model facilitates the investigation of neuromechanical interactions governing reactive balance control to predict muscle activity and movement patterns based on interactions between neuromechanical elements such as spinal reflexes, muscle short-range stiffness, and task-level sensorimotor feedback.