International Business Department           Liu Bojia           April 1, 2023

  What is your first thought when you talk about the "sixth sense"? Is it magnetic senses, which have become more popular in recent years, or is it the gut-brain axis, as some studies have suggested? The scientific community has been trying to add a new sixth sense to the senses of sight, hearing, smell, taste, and touch, but there are still a lot of conflicting opinions and it's hard to come up with a conclusion.

  In addition to the magnetic and gut-brain senses, a new concept has also been mentioned, which is called proprioception, a sensory system that helps us collect information about the movement and position of our muscles and joints, and then passes it on to the central system for processing. The existence of proprioception allows us to know where we are in real time, coordinate our limbs, and innervate our muscles. Relevant scholars believe that this sense is unconscious but indispensable. In the event of an injury, the body's movements can become uncoordinated and even produce movement disorders.

  Currently, there is still not much research on proprioception in the scientific community, which makes it difficult to understand this potential "sixth sense" well. In a recent paper published in the journal Cell from the École Polytechnique Fédérale de Lausanne, a team of researchers explored proprioception by simulating musculoskeletal models and training task-driven neural networks, and found that the brain prioritises the integration of muscle spindle information from all parts of the body to aid in the understanding of movement and position.

  Muscle spindles are typically found in skeletal muscle and are used to sense stretching and contracting stimuli, as well as receptors for proprioception. In the study, the authors generated a large number of proprioceptive signals generated by natural movements, including limb length and movement speed, by simulating musculoskeletal models, based on which the authors constructed a collection of movement instructions. These instructions were also used to train task-driven neural networks on 16 computational tasks, where the tasks in question were built based on hypotheses of proprioceptive signalling from previous literature.

  Task-driven neural networks have been used in the study of a variety of sensory systems, such as predicting visually driven responses in the ventral visual pathway. Previously, it was not clear whether the model was applicable to proprioceptive studies. For this reason, the authors additionally collected data from non-human primates during active and passive limb movements and used the trained neural network to predict activity in specific brain regions, which they further analysed to see if these predictions were similar to the actual activity.

  According to the relevant data, the task-driven model neural network can accurately predict the neurodynamic processes in a single trial, and the task-driven model is superior to the classical coding model. They found that proprioceptive neurons present in the cuneate fascicular nucleus of the brain and in the somatomotor sensory cortex2 are specifically used to encode the motor state of the limb. In turn, signals from the muscle shuttles are transmitted through the dorsal root ganglia, brainstem, thalamus, and primary and secondary somatosensory cortex, building out the proprioceptive pathway.

  The new study confirms for the first time that neural networks trained by predicting limb position and velocity tasks are the most effective when studying proprioceptive systems. In addition, representations built on task-driven models better predict neural activity during active limb movements than passive limb movements, so the authors hypothesise that neural activity in the cuneate fasciculus nucleus as well as in the somatomotor sensory cortex is regulated top-down when goal-directed movements occur. These results have important implications for our understanding of proprioception and for unravelling the mystery of the "sixth sense".