AGILE ANIMALS

Abstract

An octopus can be a canny customer - by shortening, bending, elongating and twisting its highly dexterous arms it can catch prey, build a shelter and even open jars. The dazzling array of delicate tasks that they can perform is well documented, but how exactly do they do it? Yoram Yekutieli and colleagues at the Hebrew University and Wiezmann Institute of Science in Israel built a series of octopus arm models to help answer how an octopus moves its arms and controls arm movement. The team wanted to find out if the control of the movements in an octopus' arm is as complicated as their dexterity suggests.Octopuses have a dynamic skeleton called a muscular hydrostat, where all the muscles in the limb generate forces in different directions and provide skeletal support. Because the volume of the limb is always maintained, force can be transferred from one set of muscles to another. The range of movements is much greater than a limb made of bone and muscle, which is physically restricted in the range of movements it can make.In the first of two papers, the team built a computer model to describe octopus arm reaching movements. The computer arm was divided up into 20 segments with a constant volume, connected to each other by virtual muscles. The model took into account the forces acting on the arm both internally and externally, including muscle force, gravity and drag. When a live octopus reaches for an object, a bend is created in its arm, usually near the base,which then travels up the arm towards the tip. This is likely to be caused by a wave of muscle activation that propels the bend to the end of the arm. Rather than contracting different sets of muscles out of phase with each other to produce a bend, the team's model suggests that all the muscles in an arm section are activated simultaneously to move the bend along the arm.In the second paper, the team refined the model to investigate the neural signal that causes arm movements and how these movements are coordinated. When the team tested the model with a simple neural signal that caused all of the muscles along the arm to contract, the model arm produced a movement very similar to that of a real octopus arm. They found that by controlling the signal amplitude (i.e. its height) and velocity along the arm, they could make their model re-enact the movements of a real arm. Above a minimum threshold level at which muscles would contract, the team discovered that further increases in the signal amplitude increased the force in the muscles but did not change the kinematics of the model arm's movement. These increased muscle forces mean that the arm is more stable in the face of external forces that could knock it off course.The model indicates that the control of octopus arm movement is very simple and robust and doesn't require complex coordination between different sets of muscles; a simple signal that causes all the muscles to contract can reproduce natural octopus arm movements. Understanding how an octopus produces different movements might aid the design of control systems for flexible robotic arms that need to perform a wide range of octopus-like movements.