Encoding of movement primitives and body posture through distributed proprioception in walking and climbing insects

Abstract

Targeted reaching movements and spatial coordination of footfall patterns are prime examples of spatial coordination of limbs in insects. To explain this, both physiological and computational studies have suggested the use of movement primitives or the existence of an internal body representation, much like they are assumed to occur in vertebrates. Since insects lack a dedicated posture-sensing organ or vestibular system, it is hypothesized that they derive high-level postural information from low-level proprioceptive cues, integrated across their limbs. The present study tests the extent to which a multi-layer spiking neural network can extract high-level information about limb movement and whole-body posture from information provided by distributed local proprioceptors. In a preceding part of the study, we introduced the phasic-tonic encoding of joint angles by strictly local proprioceptive hair field afferents, as well as high-accuracy encoding of joint angles and angular velocities in first-order interneurons. Here, we extend this model by second-order interneurons that use coincidence detection from two or three leg-local inputs to encode movement primitives of a single leg. Using experimental data on whole-body kinematics of unrestrained walking and climbing stick insects, we show that these movement primitives can be used to signal particular step cycle phases, but also step cycle transitions such as leg lift-off. Additionally, third-order interneurons are introduced to indicate climbing behaviour, for example by encoding the body pitch angle from 6 x 3 local leg joints. All encoding properties are validated against annotated experimental data, allowing for relevance rating of particular leg types and/or leg joint actions for all measures encoded. Our results demonstrate that simple combinations of two or three position/velocity inputs from disjunct hair field arrays are sufficient to encode high-order movement information about step cycle phases. The resulting movement primitive encoding may converge to represent particular locomotor states and whole-body posture. Author summaryInsect behaviours such as navigation or climbing involve complex movement sequences that have led scientists to postulate the existence of an internal body representation. As insects lack a dedicated organ for monitoring body posture, a major problem in computational neuroscience and biomimetic robotics is how high-level information about body posture and coordinated movement may be extracted from distributed, local, low-level sensory measures, such as joint angles or angular velocities. To solve this problem, we developed a spiking neural network model. The model was tuned and evaluated with experimental data on complex climbing sequences of stick insects, with detailed information about 6 x 3 joint angle time courses. In a preceding study, we focused on how joint angle sensors encode this information at various body parts and how it is processed to represent local joint position and movement. Here, we extend the model to include neurons that signal particular phases of a legs movement cycle. Other neurons encode whole-body movement, using the body pitch angle as an example parameter. We show that a straight-forward combination of movement signals from various body parts can indicate the timing of particular step cycle events, as well as provide an internal representation of the full bodys posture.