119 — Embodied Organization of Octopus vulgaris Morphology, Vision, and Locomotion

Levy & Hochner (10.3389/fphys.2017.00164)

Read on 17 December 2017
#octopus  #locomotion  #motor-program  #body-model 

I was reading Other Minds: The Octopus, the Sea, and the Deep Origins of Consciousness by Peter Godfrey-Smith in which he mentions Binyamin Hochner, a researcher at Hebrew University.

And of course, Dr. Hochner has written many things! This paper was the first publication I saw that mentioned octopus vision, which I’ve read about before. (That paper was the sixteenth I read for my #365papers, which is just over 100 papers ago.)

Octopus locomotion is — not surprisingly — complicated for our human minds to understand. Levy and Hochner explain that while skeletal animals developed a mental model of their bodies in order to design top-down motion coordination, octopuses do not have a rigid joint system and so cannot design these motor programs in the same way. For example, a human squatting motion recruits motion around joints: If you know that you’re leaning too far forward, you can adjust your pelvis, your knees, or your ankles… but you don’t have to think about interpolating that adjustment and curving your femur. Octopuses do not have knees or ankles or hips, and they do not have the luxury of “ignoring” rigid areas of their body like we can.

Even other members of the invertebrates — insects — have rigid body structures that simplify the inverse kinematic formulas required to model a certain movement.

A question that arises: Do octopus nervous systems have a mental soft-body model that they can manipulate as we do our rigid-body models? Or do they have some other better-suited system? (Remember of course that the last human-octopus common ancestor was probably no more complicated than a simple worm, and so there is little reason to believe that our brains would have residual structures in common.)

To answer this, we can look to the higher motor centers of the octopus brain — the basal lobes: It appears that these regions do not maintain a central body model representation, and instead, motor control is distributed to the arms. The central motor system is in charge of directing the goal-oriented reaching and directing motions, but the octopus arm itself contains all of the machinery required to design and organize grasping and manipulating motor programs.

I wrote before about how insects’ legs contain the motor programs for walking and balancing movement. Unlike insects, however, the paper explains that octopus motion lacks central pattern generators (CPGs) and decisions for motion are made in a distributed, ad-hoc manner.

This was found by measuring octopus leg choice and motion during crawling locomotion: A FFT analysis shows no discernable “beat” or characteristic frequencies, required attributes of a CPG. (Insects, on the other hand, showed very robust characteristic frequencies of leg motion).

Insead, Levy and Hochner propose a probablistic control strategy for locomotion, where constant, distributed calculations allow the octopus to recruit the most useful arm for a given motion and goal at any given time.

The octopus is a very interesting model for neuro-inspired computation and for neuroscience: This system is an example of a highly distributed, highly parallel computation environment in which decisions are made on-the-fly and constant recalculations enable high-level, goal-seeking behavior. It is also a perfect example for studying the fundamentals of neuroscience. The only thing common between octopus and vertebrate nervous systems is the biochemistry of the neurons themselves (sorta). Anything else we have in common with the octopus is a function of convergent evolution, which gives us immense insight into the ways in which brains can work.