The organization of movement is a complex and difficult problem, in part because of a "degrees of freedom problem" in motor control. The richness of an animal's movement possibilities makes its choice of movement controls complex. However, unlike current robots, animals cope efficiently with their degrees of freedom. A newborn wildebeest calf walks with the herd within a few hours of birth. A frog or a turtle, using just its spinal cord, can control complex goal-directed trajectories. The spinal cord can also rapidly correct such movements if they are perturbed. It has been argued that these remarkable capacities are modular, constructed with small sets of primitives or motor building blocks. How such primitives arise and are used is the focus of this project.
The concepts of modularity and motor primitives have provided useful descriptions of the organization of spinal motor systems. Modular organization has been shown to support spinal behaviors, and may help to "bootstrap" motor learning. Nonetheless, modularity is controversial at many levels. Spinal primitives might need to be supplanted or augmented in order to perform complex, voluntary behaviors. This project attacks this problem in frog prey strike behaviors, a voluntary and adapted behavior in a system that is fundamentally important to the animal, and has also been well characterized in previous studies of modularity. The neuromechanics of prey strike is examined from a multi-disciplinary perspective. The importance of modular organization in neuroscience and behavior extends well beyond biological motor control, with ramifications in evolutionary and cognitive psychology. Biological strategies and solutions are also highly relevant to future technologies and robotics.
A computer model of prey strike will be developed using a novel approach based on Cosserat strand-elements. The model will be developed by a team of four investigators: Simon Giszter (neurophysiology) and Jonathan Nissanov (anatomy, imaging) at Drexel University, Dinesh Pai (computer science, biomechanical modeling) at the University of British Columbia, and Kiisa Nishikawa (neuromechanics) at Northern Arizona University. Cryoplane microscopy will be used to reconstruct bullfrog sensorimotor anatomy in detail. These structures will be modeled using a strand-based approach to incorporate this detail. Experimental and model analyses of prey strike using these data will inform one another to establish the benefits and limits of fixed or adaptive modular mechanisms, and the biological implementation used in frogs.
Intellectual merit: The purpose of this project was to further our understanding of how muscles, sense organs, and the brain interact to produce coordinated movements. A current hypothesis is that flexible and adaptable motor primitives can be used to describe the organization of motor systems, particularly the spinal control of reflex behavior. The goal of our multi-disciplinary collaborative studies was to ask whether the concept of motor primitives can usefully be applied to complex, voluntary behavior. The prey strike of frogs is an ideal system in which to pursue this goal. The work was performed by collaborators at Drexel University (Simon Giszer), Northern Arizona University (Kiisa Nishikawa), and the University of British Columbia (Dinesh Pai). The specific aims were: (1) to use high-speed imaging with synchronized electromyography for kinematic analysis of the frog’s prey strike; (2) cineradiography to image motion of the skeleton during the prey strike; (3) anatomical reconstruction using cryoplane microscopy and computed tomography (CT); (4) servo-motor force-lever studies to characterize viscoelastic properties of muscles during rapid unloading; (5) development of a muscle model that accounts for the viscoelastic properties of muscle during rapid unloading, as well as conventional contractile properties of muscle in standard physiological tests; and (6) simulation of the frog’s prey strike using a strand-based biomechanical model. Using conventional CT, cryoplane microscopy, and microCT, we developed a 3D mesh model of the frog. The registration procedure we used is promising for high-resolution modeling of organisms that are too large to be modeled using microCT alone. To model ballistic mouth opening, we developed a load-clamp test using a servomotor force lever that mimics in vivo storage and recovery of elastic potential energy by actively shortening muscle. These experiments inspired a new hypothesis for muscle contraction that involves mechanical engagement of the titin spring upon muscle activation, following by winding of titin on the thin filaments during isometric force development. Simulations of isovelocity experiments demonstrate that our model performs better than traditional muscle models at accounting for the non-linear, load-dependent force output of active muscles. The model, which reproduces the non-linear force output of human muscles, was used to design algorithms and actuators for orthotic and prosthetic applications. An international patent application is pending, and collaboration with two small businesses is underway to improve control of a foot-ankle prosthesis. Our results demonstrate that the brain of the frog uses the same spinal motor primitives to construct both reflex and voluntary movement. During voluntary movements, spinal motor primitives are combined to build more complex movements. Furthermore, each leg is controlled independently, even in jumps using both legs. Synchronized high-speed image analysis and electromyography demonstrate that the jaw muscles that power ballistic mouth opening are pre-loaded, storing elastic potential energy prior to movement. The stored energy is recovered during ballistic mouth opening, amplifying the power output of the mouth opening muscles more than 100-fold. Kinematic analyses also demonstrate that frogs modulate prey capture movements depending on prey size, shape, and azimuthal location. The results have important implications for understanding the prey strike and other voluntary movements. Broader Impacts: Two postdoctoral associates, two biology graduate students, and one graduate student in computer science participated in the proposed studies. Nine undergraduate students and two high school teachers also participated in the research. The results were disseminated broadly in a variety of venues, local, regional and national popular media. A related grant on "Emulating Biological Actuation" was obtained from NSF’s Industrial Engineering Program to use the muscle model to improve the control and maneuverability of a commercially available foot-ankle prosthesis.
